/usr/local/lib64/python3.6/site-packages/torch/lib
Edit: /usr/local/lib64/python3.6/site-packages/torch/lib/libnvrtc-builtins.so (4862944B)
ELF > @ J @ 8 @ J J Qtd Ptd J J J d d J j j ` p 0J j j 0J j j XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX b( b i ӯk ui j j j j j j j H7
H5j %j @ %j h HH5j HtHÐU=`j HATSubH=j t H=j Hj L%j L)HHXHj H9v D HHj AHj H9wj [A\f UH=j HtHj HtH=j IAÐHi H@ ÐH9@ H8 ÐHiA H ÐHB H ÐHC H ÐHD H ÐHE HD' ÐHF HD' ÐHIG Hd ÐH I Hd Ð tA#tF%tF2tF4t(< @ t==t>Ft>Ku1H j% D +6A[fqUHSH 4j HH)j Ht HHHuH[ÐHH /*
* Copyright 1993-2016 NVIDIA Corporation. All rights reserved.
*
* NOTICE TO LICENSEE:
*
* This source code and/or documentation ("Licensed Deliverables") are
* subject to NVIDIA intellectual property rights under U.S. and
* international Copyright laws.
*
* These Licensed Deliverables contained herein is PROPRIETARY and
* CONFIDENTIAL to NVIDIA and is being provided under the terms and
* conditions of a form of NVIDIA software license agreement by and
* between NVIDIA and Licensee ("License Agreement") or electronically
* accepted by Licensee. Notwithstanding any terms or conditions to
* the contrary in the License Agreement, reproduction or disclosure
* of the Licensed Deliverables to any third party without the express
* written consent of NVIDIA is prohibited.
*
* NOTWITHSTANDING ANY TERMS OR CONDITIONS TO THE CONTRARY IN THE
* LICENSE AGREEMENT, NVIDIA MAKES NO REPRESENTATION ABOUT THE
* SUITABILITY OF THESE LICENSED DELIVERABLES FOR ANY PURPOSE. IT IS
* PROVIDED "AS IS" WITHOUT EXPRESS OR IMPLIED WARRANTY OF ANY KIND.
* NVIDIA DISCLAIMS ALL WARRANTIES WITH REGARD TO THESE LICENSED
* DELIVERABLES, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY,
* NONINFRINGEMENT, AND FITNESS FOR A PARTICULAR PURPOSE.
* NOTWITHSTANDING ANY TERMS OR CONDITIONS TO THE CONTRARY IN THE
* LICENSE AGREEMENT, IN NO EVENT SHALL NVIDIA BE LIABLE FOR ANY
* SPECIAL, INDIRECT, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, OR ANY
* DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS,
* WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
* ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
* OF THESE LICENSED DELIVERABLES.
*
* U.S. Government End Users. These Licensed Deliverables are a
* "commercial item" as that term is defined at 48 C.F.R. 2.101 (OCT
* 1995), consisting of "commercial computer software" and "commercial
* computer software documentation" as such terms are used in 48
* C.F.R. 12.212 (SEPT 1995) and is provided to the U.S. Government
* only as a commercial end item. Consistent with 48 C.F.R.12.212 and
* 48 C.F.R. 227.7202-1 through 227.7202-4 (JUNE 1995), all
* U.S. Government End Users acquire the Licensed Deliverables with
* only those rights set forth herein.
*
* Any use of the Licensed Deliverables in individual and commercial
* software must include, in the user documentation and internal
* comments to the code, the above Disclaimer and U.S. Government End
* Users Notice.
*/
#define __CUDACC__ 1
#define __CUDANVVM__ 1
#if !defined(__cplusplus)
#error "unexpected!"
#endif
#if defined(__LP64__)
typedef unsigned long size_t;
typedef long int ptrdiff_t;
#else
typedef unsigned long long size_t;
typedef long long int ptrdiff_t;
#endif
typedef long clock_t;
#ifndef NULL
#define NULL 0
#endif
#if defined(__cplusplus)
extern "C" {
#endif
__attribute__((device)) int __nvvm_reflect(const char*);
#if defined(__cplusplus)
}
#endif
#if !defined(__CUDA_RUNTIME_H__)
#define __CUDA_RUNTIME_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_H__
#endif
#if !defined(__CUDACC_RTC__)
#if defined(__GNUC__)
#if defined(__clang__) || (!defined(__PGIC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 6)))
#pragma GCC diagnostic push
#endif
#if defined(__clang__) || (!defined(__PGIC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 2)))
#pragma GCC diagnostic ignored "-Wunused-function"
#endif
#elif defined(_MSC_VER)
#pragma warning(push)
#pragma warning(disable: 4820)
#endif
#endif
#ifdef __QNX__
#if (__GNUC__ == 4 && __GNUC_MINOR__ >= 7)
typedef unsigned size_t;
#endif
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_config.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_config.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_CONFIG_H__
#endif
#if !defined(__HOST_CONFIG_H__)
#define __HOST_CONFIG_H__
#if defined(__CUDACC__)
#if defined(__CUDACC_RTC__)
#define _CRTIMP
#define __THROW
#else
#if !defined(__GNUC__) && !defined(_WIN32)
#error --- !!! UNSUPPORTED COMPILER !!! ---
#endif
#if defined(__PGIC__)
#if !defined(__GNUC__) || !defined(__LP64__) || !defined(__linux__)
#error -- unsupported pgc++ configuration! pgc++ is supported only on Linux x86_64!
#endif
#endif
#if defined(__powerpc__)
#if !defined(__powerpc64__) || !defined(__LITTLE_ENDIAN__)
#error -- unsupported PPC platform! Only 64-bit little endian PPC is supported!
#endif
#endif
#if defined(__APPLE__) && defined(__MACH__) && !defined(__clang__)
#error -- clang and clang++ are the only supported host compilers on Mac OS X!
#endif
#if !__NV_NO_HOST_COMPILER_CHECK
#if defined(__ICC)
#if (__ICC != 1500 && __ICC != 1600 && __ICC != 1700 && __ICC != 1800 && __ICC != 1900) || !defined(__GNUC__) || !defined(__LP64__)
#error -- unsupported ICC configuration! Only ICC 15.0, ICC 16.0, ICC 17.0, ICC 18.0 and ICC 19.0 on Linux x86_64 are supported!
#endif
#endif
#if defined(__PGIC__)
#if ( (__PGIC__ != 18) && (__PGIC__ != 19) && !(__PGIC__ == 99 && __PGIC_MINOR__ == 99))
#error -- unsupported pgc++ configuration! Only pgc++ 18 and 19 are supported!
#endif
#endif
#if defined(__powerpc__)
#if defined(__ibmxl_vrm__) && !(__ibmxl_vrm__ >= 0x0d010000 && __ibmxl_vrm__ < 0x0d020000) && \
!(__ibmxl_vrm__ >= 0x10010000 && __ibmxl_vrm__ < 0x10020000)
#error -- unsupported xlC version! only xlC 13.1 and 16.1 are supported
#endif
#endif
#if defined(__GNUC__)
#if __GNUC__ > 8
#error -- unsupported GNU version! gcc versions later than 8 are not supported!
#endif
#if defined(__clang__) && !defined(__ibmxl_vrm__) && !defined(__ICC) && !defined(__HORIZON__) && !defined(__APPLE__)
#if (__clang_major__ >= 9) || (__clang_major__ < 3) || ((__clang_major__ == 3) && (__clang_minor__ < 3))
#error -- unsupported clang version! clang version must be less than 9 and greater than 3.2
#endif
#endif
#endif
#if defined(_WIN32)
#if _MSC_VER < 1700 || _MSC_VER >= 1930
#error -- unsupported Microsoft Visual Studio version! Only the versions between 2015 and 2019 (inclusive) are supported!
#elif _MSC_VER >= 1700 && _MSC_VER < 1900
#pragma message("support for this version of Microsoft Visual Studio has been deprecated! Only the versions between 2015 and 2019 (inclusive) are supported!")
#endif
#endif
#endif
#if defined(__APPLE__)
#define _CRTIMP
#define _ACRTIMP
#define __THROW
#if defined(__BLOCKS__)
#undef __BLOCKS__
#endif
#elif defined(__ANDROID__)
#define _CRTIMP
#define _ACRTIMP
#define __THROW
#elif defined(__QNX__)
#define _CRTIMP
#define _ACRTIMP
#define __THROW
#elif defined(__HORIZON__)
#define _CRTIMP
#define _ACRTIMP
#define __THROW
#elif defined(__GNUC__)
#define _CRTIMP
#define _ACRTIMP
#include
#elif defined(_WIN32)
#if _MSC_VER >= 1500
#undef _USE_DECLSPECS_FOR_SAL
#define _USE_DECLSPECS_FOR_SAL \
1
#endif
#if !defined(_CRT_NONSTDC_NO_WARNINGS)
#define _CRT_NONSTDC_NO_WARNINGS
#endif
#if !defined(_CRT_SECURE_NO_WARNINGS)
#define _CRT_SECURE_NO_WARNINGS
#endif
#if !defined(NOMINMAX)
#define NOMINMAX
#endif
#include
#if _MSC_VER >= 1900
#include
#endif
#define __THROW
#endif
#endif
#if defined(__cplusplus) && defined(__CUDA_ARCH__) && (defined(__PGIC__) || defined(__CUDACC_RTC__) || (defined(_WIN32) && defined(_MSC_VER)))
#if __CUDACC_RTC__
typedef char *va_list;
#else
#include
#endif
#undef va_start
#undef va_end
#undef va_arg
#ifdef __PGIC__
#undef __builtin_va_end
#define va_start(v,l) __builtin_alt_va_start(v,l)
#define va_end(v) __builtin_va_end(v)
#define va_arg(v,l) __builtin_alt_va_arg(v,l)
#if (__cplusplus >= 201103L)
#undef va_copy
#define va_copy(d,s) __builtin_va_copy(d,s)
#endif
#else
#define va_start(ap, x) (__cu_va_start(&ap, x))
#define va_end(ap) (__cu_va_end(&ap))
#define va_arg(ap, t) (*((t *)__cu_va_arg(&ap, (t *)0)))
#if (_MSC_VER >= 1800) || (defined(__CUDACC_RTC__) && (__cplusplus >= 201103L))
#undef va_copy
#define va_copy(apd, aps) (__cu_va_copy(&(apd), &(aps)))
#endif
#endif
#endif
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_CONFIG_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_CONFIG_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__LIBRARY_TYPES_H__)
#define __LIBRARY_TYPES_H__
typedef enum cudaDataType_t
{
CUDA_R_16F= 2,
CUDA_C_16F= 6,
CUDA_R_32F= 0,
CUDA_C_32F= 4,
CUDA_R_64F= 1,
CUDA_C_64F= 5,
CUDA_R_8I = 3,
CUDA_C_8I = 7,
CUDA_R_8U = 8,
CUDA_C_8U = 9,
CUDA_R_32I= 10,
CUDA_C_32I= 11,
CUDA_R_32U= 12,
CUDA_C_32U= 13
} cudaDataType;
typedef enum libraryPropertyType_t
{
MAJOR_VERSION,
MINOR_VERSION,
PATCH_LEVEL
} libraryPropertyType;
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_FUNCTIONS_H__)
#define __VECTOR_FUNCTIONS_H__
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#if defined(__CUDACC_RTC__)
#define __VECTOR_FUNCTIONS_DECL__ __host__ __device__
#else
#define __VECTOR_FUNCTIONS_DECL__ static __inline__ __host__ __device__
#endif
__VECTOR_FUNCTIONS_DECL__ char1 make_char1(signed char x);
__VECTOR_FUNCTIONS_DECL__ uchar1 make_uchar1(unsigned char x);
__VECTOR_FUNCTIONS_DECL__ char2 make_char2(signed char x, signed char y);
__VECTOR_FUNCTIONS_DECL__ uchar2 make_uchar2(unsigned char x, unsigned char y);
__VECTOR_FUNCTIONS_DECL__ char3 make_char3(signed char x, signed char y, signed char z);
__VECTOR_FUNCTIONS_DECL__ uchar3 make_uchar3(unsigned char x, unsigned char y, unsigned char z);
__VECTOR_FUNCTIONS_DECL__ char4 make_char4(signed char x, signed char y, signed char z, signed char w);
__VECTOR_FUNCTIONS_DECL__ uchar4 make_uchar4(unsigned char x, unsigned char y, unsigned char z, unsigned char w);
__VECTOR_FUNCTIONS_DECL__ short1 make_short1(short x);
__VECTOR_FUNCTIONS_DECL__ ushort1 make_ushort1(unsigned short x);
__VECTOR_FUNCTIONS_DECL__ short2 make_short2(short x, short y);
__VECTOR_FUNCTIONS_DECL__ ushort2 make_ushort2(unsigned short x, unsigned short y);
__VECTOR_FUNCTIONS_DECL__ short3 make_short3(short x,short y, short z);
__VECTOR_FUNCTIONS_DECL__ ushort3 make_ushort3(unsigned short x, unsigned short y, unsigned short z);
__VECTOR_FUNCTIONS_DECL__ short4 make_short4(short x, short y, short z, short w);
__VECTOR_FUNCTIONS_DECL__ ushort4 make_ushort4(unsigned short x, unsigned short y, unsigned short z, unsigned short w);
__VECTOR_FUNCTIONS_DECL__ int1 make_int1(int x);
__VECTOR_FUNCTIONS_DECL__ uint1 make_uint1(unsigned int x);
__VECTOR_FUNCTIONS_DECL__ int2 make_int2(int x, int y);
__VECTOR_FUNCTIONS_DECL__ uint2 make_uint2(unsigned int x, unsigned int y);
__VECTOR_FUNCTIONS_DECL__ int3 make_int3(int x, int y, int z);
__VECTOR_FUNCTIONS_DECL__ uint3 make_uint3(unsigned int x, unsigned int y, unsigned int z);
__VECTOR_FUNCTIONS_DECL__ int4 make_int4(int x, int y, int z, int w);
__VECTOR_FUNCTIONS_DECL__ uint4 make_uint4(unsigned int x, unsigned int y, unsigned int z, unsigned int w);
__VECTOR_FUNCTIONS_DECL__ long1 make_long1(long int x);
__VECTOR_FUNCTIONS_DECL__ ulong1 make_ulong1(unsigned long int x);
__VECTOR_FUNCTIONS_DECL__ long2 make_long2(long int x, long int y);
__VECTOR_FUNCTIONS_DECL__ ulong2 make_ulong2(unsigned long int x, unsigned long int y);
__VECTOR_FUNCTIONS_DECL__ long3 make_long3(long int x, long int y, long int z);
__VECTOR_FUNCTIONS_DECL__ ulong3 make_ulong3(unsigned long int x, unsigned long int y, unsigned long int z);
__VECTOR_FUNCTIONS_DECL__ long4 make_long4(long int x, long int y, long int z, long int w);
__VECTOR_FUNCTIONS_DECL__ ulong4 make_ulong4(unsigned long int x, unsigned long int y, unsigned long int z, unsigned long int w);
__VECTOR_FUNCTIONS_DECL__ float1 make_float1(float x);
__VECTOR_FUNCTIONS_DECL__ float2 make_float2(float x, float y);
__VECTOR_FUNCTIONS_DECL__ float3 make_float3(float x, float y, float z);
__VECTOR_FUNCTIONS_DECL__ float4 make_float4(float x, float y, float z, float w);
__VECTOR_FUNCTIONS_DECL__ longlong1 make_longlong1(long long int x);
__VECTOR_FUNCTIONS_DECL__ ulonglong1 make_ulonglong1(unsigned long long int x);
__VECTOR_FUNCTIONS_DECL__ longlong2 make_longlong2(long long int x, long long int y);
__VECTOR_FUNCTIONS_DECL__ ulonglong2 make_ulonglong2(unsigned long long int x, unsigned long long int y);
__VECTOR_FUNCTIONS_DECL__ longlong3 make_longlong3(long long int x, long long int y, long long int z);
__VECTOR_FUNCTIONS_DECL__ ulonglong3 make_ulonglong3(unsigned long long int x, unsigned long long int y, unsigned long long int z);
__VECTOR_FUNCTIONS_DECL__ longlong4 make_longlong4(long long int x, long long int y, long long int z, long long int w);
__VECTOR_FUNCTIONS_DECL__ ulonglong4 make_ulonglong4(unsigned long long int x, unsigned long long int y, unsigned long long int z, unsigned long long int w);
__VECTOR_FUNCTIONS_DECL__ double1 make_double1(double x);
__VECTOR_FUNCTIONS_DECL__ double2 make_double2(double x, double y);
__VECTOR_FUNCTIONS_DECL__ double3 make_double3(double x, double y, double z);
__VECTOR_FUNCTIONS_DECL__ double4 make_double4(double x, double y, double z, double w);
#undef __VECTOR_FUNCTIONS_DECL__
#if !defined(__CUDACC_RTC__)
#include "vector_functions.hpp"
#endif
#endif
#if defined(__CUDACC__)
#if defined(__CUDACC_RTC__)
#ifndef __NVRTC_DEVICE_RUNTIME_H__
#define __NVRTC_DEVICE_RUNTIME_H__
#if __cplusplus >= 201103L
namespace __nvrtc_util {
template
struct remove_ref { typedef T type; };
template
struct remove_ref { typedef T type; };
template
struct remove_ref { typedef T type; };
template
struct is_lvalue_ref { static const bool val = false; };
template
struct is_lvalue_ref { static const bool val = true; };
}
#if __NV_BUILTIN_MOVE_FORWARD
namespace std {
template
constexpr T&&
__device__ forward(typename __nvrtc_util::remove_ref::type &__in) noexcept
{ return static_cast(__in); }
template
constexpr T&&
__device__ forward(typename __nvrtc_util::remove_ref::type && __in) noexcept
{
static_assert(!__nvrtc_util::is_lvalue_ref::val,
"incorrect forward call!");
return static_cast(__in);
}
template
constexpr typename __nvrtc_util::remove_ref::type &&
__device__ move(T&& __in) noexcept
{
return static_cast::type &&>(__in);
}
}
#endif
#if __NV_BUILTIN_INITIALIZER_LIST
namespace std {
template
class initializer_list
{
public:
typedef T value_type;
typedef const T& reference;
typedef const T& const_reference;
typedef size_t size_type;
typedef const T* iterator;
typedef const T* const_iterator;
private:
const T* buf_ptr;
size_t length;
constexpr initializer_list(const_iterator in, size_type in_length)
: buf_ptr(in), length(in_length) { }
public:
constexpr initializer_list() noexcept
: buf_ptr(0), length(0) { }
constexpr size_type size() const noexcept { return length; }
constexpr const_iterator begin() const noexcept { return buf_ptr; }
constexpr const_iterator end() const noexcept { return buf_ptr + length; }
};
}
#endif
#endif
extern "C" inline __host__ __device__ void __cxa_vec_ctor(void *n,
size_t num,
size_t size,
void (*c) (void *),
void (*d) (void *))
{
if (!c)
return;
for (size_t i = 0; i < num; i++)
c((void*)((unsigned char *)n + i*size));
}
extern "C" inline __host__ __device__ void __cxa_vec_cctor(void *dest,
void *src,
size_t num,
size_t size,
void (*c) (void *, void *),
void (*d) (void *))
{
if (!c)
return;
for (size_t i = 0; i < num; i++) {
c((void*)((unsigned char *)dest + i*size),
(void*)((unsigned char *)src + i*size));
}
}
extern "C" inline __host__ __device__ void __cxa_vec_dtor(void *n,
size_t num,
size_t size,
void (*d) (void *))
{
if (!d)
return;
for (size_t i = num-1; i > 0; i--)
d((void*)((unsigned char *)n + i*size));
d((void*)((unsigned char *)n + 0*size));
}
extern "C" inline __host__ __device__ void * __cxa_vec_new2(size_t num,
size_t size,
size_t pad,
void (*c) (void *),
void (*d) (void *),
void *(*m) (size_t),
void (*f) (void *))
{
unsigned char *t = (unsigned char *)m(num*size + pad);
*(size_t*)t = num;
(void)__cxa_vec_ctor((void *)(t+pad), num, size, c, d);
return (void *)(t+pad);
}
#if defined(_MSC_VER)
extern "C" __host__ __device__ __cudart_builtin__ void* malloc(size_t);
extern "C" __host__ __device__ __cudart_builtin__ void free(void*);
#else
extern "C" __host__ __device__ __cudart_builtin__ __attribute__((used)) void* malloc(size_t);
extern "C" __host__ __device__ __cudart_builtin__ __attribute__((used)) void free(void*);
#endif
extern "C" inline __host__ __device__ void * __cxa_vec_new(size_t num,
size_t size,
size_t pad,
void (*c) (void *),
void (*d) (void *))
{
return __cxa_vec_new2(num, size, pad, c, d, malloc, free);
}
extern "C" inline __host__ __device__ void * __cxa_vec_new3(size_t num,
size_t size,
size_t pad,
void (*c) (void *),
void (*d) (void *),
void *(*m) (size_t),
void (*f)(void *, size_t))
{
return __cxa_vec_new2(num, size, pad, c, d, m, 0);
}
extern "C" inline __host__ __device__ void __cxa_vec_delete2(void *n,
size_t size,
size_t pad,
void (*d) (void *),
void (*f) (void *))
{
unsigned char *ptr = (unsigned char *)(n);
if (ptr) {
unsigned char *t = ptr - pad;
size_t num = *(size_t*)t;
__cxa_vec_dtor(ptr, num, size, d);
f((void *)t);
}
}
extern "C" inline __host__ __device__ void __cxa_vec_delete(void *n,
size_t size,
size_t pad,
void (*d) (void *))
{
__cxa_vec_delete2(n, size, pad, d, free);
}
extern "C" inline __host__ __device__ void __cxa_vec_delete3(void *n,
size_t size,
size_t pad,
void (*d)(void *),
void (*f)(void *,size_t))
{
unsigned char *ptr = (unsigned char *)(n);
if (ptr) {
unsigned char *t = ptr - pad; size_t num = *(size_t*)t;
size_t tsize = num*size+pad;
__cxa_vec_dtor(ptr, num, size, d);
f((void *)t, tsize);
}
}
extern "C" inline __host__ __device__ void *__gen_nvvm_memcpy_aligned1(void *dest,
void *src,
size_t size)
{
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 1);
return dest;
}
extern "C" inline __host__ __device__ void *__gen_nvvm_memcpy_aligned2(void *dest,
void *src,
size_t size)
{
if (((unsigned long long)(dest) % 2) == 0 &&
((unsigned long long)(src) % 2) == 0)
{
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 2);
} else {
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 1);
}
return dest;
}
extern "C" inline __host__ __device__ void *__gen_nvvm_memcpy_aligned4(void *dest,
void *src,
size_t size)
{
if (((unsigned long long)(dest) % 4) == 0 &&
((unsigned long long)(src) % 4) == 0)
{
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 4);
} else {
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 1);
}
return dest;
}
extern "C" inline __host__ __device__ void * __gen_nvvm_memcpy_aligned8(void *dest,
void *src,
size_t size)
{
if (((unsigned long long)(dest) % 8) == 0 &&
((unsigned long long)(src) % 8) == 0)
{
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 8);
} else {
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 1);
}
return dest;
}
extern "C" inline __host__ __device__ void * __gen_nvvm_memcpy_aligned16(void *dest,
void *src,
size_t size)
{
if (((unsigned long long)(dest) % 16) == 0 &&
((unsigned long long)(src) % 16) == 0)
{
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 16);
} else {
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, size, 1);
}
return dest;
}
extern "C" inline __host__ __device__ void __cxa_pure_virtual(void)
{
volatile int *ptr = 0;
*ptr = 1;
}
extern "C" inline __host__ __device__ void* __gen_nvvm_memset(void *dest, int c,
size_t n)
{
__nvvm_memset((unsigned char *)dest, (unsigned char)c, n,
1);
return dest;
}
extern "C" inline __host__ __device__ void* __gen_nvvm_memcpy(void *dest,
const void *src,
size_t n)
{
__nvvm_memcpy((unsigned char *)dest, (unsigned char *)src, n,
1);
return dest;
}
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/device_functions.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/device_functions.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_FUNCTIONS_H__
#endif
#if !defined(__DEVICE_FUNCTIONS_H__)
#define __DEVICE_FUNCTIONS_H__
#if defined(__cplusplus) && defined(__CUDACC__)
#if defined(__CUDACC_RTC__)
#define __DEVICE_FUNCTIONS_DECL__ __device__ __cudart_builtin__
#define __DEVICE_FUNCTIONS_STATIC_DECL__ __device__ __cudart_builtin__
#else
#define __DEVICE_FUNCTIONS_DECL__ __device__ __cudart_builtin__
#define __DEVICE_FUNCTIONS_STATIC_DECL__ static __inline__ __device__ __cudart_builtin__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __mulhi(int x, int y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __umulhi(unsigned int x, unsigned int y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ long long int __mul64hi(long long int x, long long int y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned long long int __umul64hi(unsigned long long int x, unsigned long long int y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __int_as_float(int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __float_as_int(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __uint_as_float(unsigned int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __float_as_uint(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ void __syncthreads(void);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ void __prof_trigger(int);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ void __threadfence(void);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ void __threadfence_block(void);
__DEVICE_FUNCTIONS_DECL__
#if defined(__GNUC__) || defined(__CUDACC_RTC__)
__attribute__((__noreturn__))
#elif defined(_MSC_VER)
__declspec(noreturn)
#endif
__device_builtin__ void __trap(void);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ void __brkpt();
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __saturatef(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __sad(int x, int y, unsigned int z);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __usad(unsigned int x, unsigned int y, unsigned int z);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __mul24(int x, int y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __umul24(unsigned int x, unsigned int y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float fdividef(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fdividef(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ double fdivide(double x, double y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __sinf(float x) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __cosf(float x) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __tanf(float x) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ void __sincosf(float x, float *sptr, float *cptr) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __expf(float x) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __exp10f(float x) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __log2f(float x) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __log10f(float x) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __logf(float x) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ __cudart_builtin__ float __powf(float x, float y) __THROW;
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __float2int_rn(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __float2int_rz(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __float2int_ru(float);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __float2int_rd(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __float2uint_rn(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __float2uint_rz(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __float2uint_ru(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __float2uint_rd(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __int2float_rn(int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __int2float_rz(int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __int2float_ru(int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __int2float_rd(int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __uint2float_rn(unsigned int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __uint2float_rz(unsigned int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __uint2float_ru(unsigned int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __uint2float_rd(unsigned int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ long long int __float2ll_rn(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ long long int __float2ll_rz(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ long long int __float2ll_ru(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ long long int __float2ll_rd(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned long long int __float2ull_rn(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned long long int __float2ull_rz(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned long long int __float2ull_ru(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned long long int __float2ull_rd(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __ll2float_rn(long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __ll2float_rz(long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __ll2float_ru(long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __ll2float_rd(long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __ull2float_rn(unsigned long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __ull2float_rz(unsigned long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __ull2float_ru(unsigned long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __ull2float_rd(unsigned long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fadd_rn(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fadd_rz(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fadd_ru(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fadd_rd(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fsub_rn(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fsub_rz(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fsub_ru(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fsub_rd(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fmul_rn(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fmul_rz(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fmul_ru(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fmul_rd(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fmaf_rn(float x, float y, float z);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fmaf_rz(float x, float y, float z);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fmaf_ru(float x, float y, float z);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fmaf_rd(float x, float y, float z);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __frcp_rn(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __frcp_rz(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __frcp_ru(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __frcp_rd(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fsqrt_rn(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fsqrt_rz(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fsqrt_ru(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fsqrt_rd(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __frsqrt_rn(float x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fdiv_rn(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fdiv_rz(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fdiv_ru(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ float __fdiv_rd(float x, float y);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __clz(int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __ffs(int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __popc(unsigned int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __brev(unsigned int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __clzll(long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __ffsll(long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __popcll(unsigned long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned long long int __brevll(unsigned long long int x);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __byte_perm(unsigned int x, unsigned int y, unsigned int s);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __hadd(int, int);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __rhadd(int, int);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __uhadd(unsigned int, unsigned int);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __urhadd(unsigned int, unsigned int);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ int __double2int_rz(double);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __double2uint_rz(double);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ long long int __double2ll_rz(double);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned long long int __double2ull_rz(double);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __pm0(void);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __pm1(void);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __pm2(void);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __pm3(void);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vabs2(unsigned int a);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vabsss2(unsigned int a);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vadd2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vaddss2 (unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vaddus2 (unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vavgs2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vavgu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vhaddu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpeq2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpges2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpgeu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpgts2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpgtu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmples2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpleu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmplts2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpltu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpne2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vabsdiffu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vmaxs2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vmaxu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vmins2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vminu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vseteq2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetges2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetgeu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetgts2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetgtu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetles2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetleu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetlts2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetltu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetne2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsadu2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsub2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsubss2 (unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsubus2 (unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vneg2(unsigned int a);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vnegss2(unsigned int a);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vabsdiffs2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsads2(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vabs4(unsigned int a);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vabsss4(unsigned int a);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vadd4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vaddss4 (unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vaddus4 (unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vavgs4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vavgu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vhaddu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpeq4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpges4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpgeu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpgts4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpgtu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmples4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpleu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmplts4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpltu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vcmpne4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vabsdiffu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vmaxs4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vmaxu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vmins4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vminu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vseteq4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetles4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetleu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetlts4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetltu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetges4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetgeu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetgts4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetgtu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsetne4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsadu4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsub4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsubss4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsubus4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vneg4(unsigned int a);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vnegss4(unsigned int a);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vabsdiffs4(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_DECL__ __device_builtin__ unsigned int __vsads4(unsigned int a, unsigned int b);
}
__DEVICE_FUNCTIONS_STATIC_DECL__ int mulhi(int a, int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned int mulhi(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned int mulhi(int a, unsigned int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned int mulhi(unsigned int a, int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ long long int mul64hi(long long int a, long long int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned long long int mul64hi(unsigned long long int a, unsigned long long int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned long long int mul64hi(long long int a, unsigned long long int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned long long int mul64hi(unsigned long long int a, long long int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ int float_as_int(float a);
__DEVICE_FUNCTIONS_STATIC_DECL__ float int_as_float(int a);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned int float_as_uint(float a);
__DEVICE_FUNCTIONS_STATIC_DECL__ float uint_as_float(unsigned int a);
__DEVICE_FUNCTIONS_STATIC_DECL__ float saturate(float a);
__DEVICE_FUNCTIONS_STATIC_DECL__ int mul24(int a, int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned int umul24(unsigned int a, unsigned int b);
__DEVICE_FUNCTIONS_STATIC_DECL__ int float2int(float a, enum cudaRoundMode mode = cudaRoundZero);
__DEVICE_FUNCTIONS_STATIC_DECL__ unsigned int float2uint(float a, enum cudaRoundMode mode = cudaRoundZero);
__DEVICE_FUNCTIONS_STATIC_DECL__ float int2float(int a, enum cudaRoundMode mode = cudaRoundNearest);
__DEVICE_FUNCTIONS_STATIC_DECL__ float uint2float(unsigned int a, enum cudaRoundMode mode = cudaRoundNearest);
#undef __DEVICE_FUNCTIONS_DECL__
#undef __DEVICE_FUNCTIONS_STATIC_DECL__
#endif
#if !defined(__CUDACC_RTC__)
#include "device_functions.hpp"
#endif
#if !defined(__DEVICE_ATOMIC_FUNCTIONS_H__)
#define __DEVICE_ATOMIC_FUNCTIONS_H__
#if defined(__CUDACC_RTC__)
#define __DEVICE_ATOMIC_FUNCTIONS_DECL__ __device__
#else
#define __DEVICE_ATOMIC_FUNCTIONS_DECL__ static __inline__ __device__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
#ifdef __CUDA_ARCH__
extern "C"
{
extern __device__ __device_builtin__ int __iAtomicAdd(int *address, int val);
extern __device__ __device_builtin__ unsigned int __uAtomicAdd(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ int __iAtomicExch(int *address, int val);
extern __device__ __device_builtin__ unsigned int __uAtomicExch(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ float __fAtomicExch(float *address, float val);
extern __device__ __device_builtin__ int __iAtomicMin(int *address, int val);
extern __device__ __device_builtin__ unsigned int __uAtomicMin(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ int __iAtomicMax(int *address, int val);
extern __device__ __device_builtin__ unsigned int __uAtomicMax(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ unsigned int __uAtomicInc(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ unsigned int __uAtomicDec(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ int __iAtomicAnd(int *address, int val);
extern __device__ __device_builtin__ unsigned int __uAtomicAnd(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ int __iAtomicOr(int *address, int val);
extern __device__ __device_builtin__ unsigned int __uAtomicOr(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ int __iAtomicXor(int *address, int val);
extern __device__ __device_builtin__ unsigned int __uAtomicXor(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__ int __iAtomicCAS(int *address, int compare, int val);
extern __device__ __device_builtin__ unsigned int __uAtomicCAS(unsigned int *address, unsigned int compare, unsigned int val);
}
#endif
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicAdd(int *address, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicAdd(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicSub(int *address, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicSub(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicExch(int *address, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicExch(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ float atomicExch(float *address, float val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicMin(int *address, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicMin(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicMax(int *address, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicMax(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicInc(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicDec(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicAnd(int *address, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicAnd(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicOr(int *address, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicOr(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicXor(int *address, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicXor(unsigned int *address, unsigned int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ int atomicCAS(int *address, int compare, int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned int atomicCAS(unsigned int *address, unsigned int compare, unsigned int val) __DEF_IF_HOST
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#if defined(_WIN32)
# define __DEPRECATED__(msg) __declspec(deprecated(msg))
#elif (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 5 && !defined(__clang__))))
# define __DEPRECATED__(msg) __attribute__((deprecated))
#else
# define __DEPRECATED__(msg) __attribute__((deprecated(msg)))
#endif
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 700
#define __WSB_DEPRECATION_MESSAGE(x) #x"() is not valid on compute_70 and above, and should be replaced with "#x"_sync()."\
"To continue using "#x"(), specify virtual architecture compute_60 when targeting sm_70 and above, for example, using the pair of compiler options: -arch=compute_60 -code=sm_70."
#else
#define __WSB_DEPRECATION_MESSAGE(x) #x"() is deprecated in favor of "#x"_sync() and may be removed in a future release (Use -Wno-deprecated-declarations to suppress this warning)."
#endif
extern "C"
{
#ifdef __CUDA_ARCH__
extern __device__ __device_builtin__ unsigned long long int __ullAtomicAdd(unsigned long long int *address, unsigned long long int val);
extern __device__ __device_builtin__ unsigned long long int __ullAtomicExch(unsigned long long int *address, unsigned long long int val);
extern __device__ __device_builtin__ unsigned long long int __ullAtomicCAS(unsigned long long int *address, unsigned long long int compare, unsigned long long int val);
#endif
extern __device__ __device_builtin__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__any)) int __any(int cond);
extern __device__ __device_builtin__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__all)) int __all(int cond);
}
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned long long int atomicAdd(unsigned long long int *address, unsigned long long int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned long long int atomicExch(unsigned long long int *address, unsigned long long int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ unsigned long long int atomicCAS(unsigned long long int *address, unsigned long long int compare, unsigned long long int val) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__any)) bool any(bool cond) __DEF_IF_HOST
__DEVICE_ATOMIC_FUNCTIONS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__all)) bool all(bool cond) __DEF_IF_HOST
#undef __DEPRECATED__
#undef __WSB_DEPRECATION_MESSAGE
#endif
#undef __DEF_IF_HOST
#undef __DEVICE_ATOMIC_FUNCTIONS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "device_atomic_functions.hpp"
#endif
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/device_double_functions.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/device_double_functions.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_DOUBLE_FUNCTIONS_H__
#endif
#if !defined(__DEVICE_DOUBLE_FUNCTIONS_H__)
#define __DEVICE_DOUBLE_FUNCTIONS_H__
#if defined(__cplusplus) && defined(__CUDACC__)
#if defined(__CUDACC_RTC__)
#define __DEVICE_DOUBLE_FUNCTIONS_DECL__ __device__
#else
#define __DEVICE_DOUBLE_FUNCTIONS_DECL__ static __inline__ __device__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __device_builtin__ long long int __double_as_longlong(double x);
extern __device__ __device_builtin__ double __longlong_as_double(long long int x);
extern __device__ __device_builtin__ double __fma_rn(double x, double y, double z);
extern __device__ __device_builtin__ double __fma_rz(double x, double y, double z);
extern __device__ __device_builtin__ double __fma_ru(double x, double y, double z);
extern __device__ __device_builtin__ double __fma_rd(double x, double y, double z);
extern __device__ __device_builtin__ double __dadd_rn(double x, double y);
extern __device__ __device_builtin__ double __dadd_rz(double x, double y);
extern __device__ __device_builtin__ double __dadd_ru(double x, double y);
extern __device__ __device_builtin__ double __dadd_rd(double x, double y);
extern __device__ __device_builtin__ double __dsub_rn(double x, double y);
extern __device__ __device_builtin__ double __dsub_rz(double x, double y);
extern __device__ __device_builtin__ double __dsub_ru(double x, double y);
extern __device__ __device_builtin__ double __dsub_rd(double x, double y);
extern __device__ __device_builtin__ double __dmul_rn(double x, double y);
extern __device__ __device_builtin__ double __dmul_rz(double x, double y);
extern __device__ __device_builtin__ double __dmul_ru(double x, double y);
extern __device__ __device_builtin__ double __dmul_rd(double x, double y);
extern __device__ __device_builtin__ float __double2float_rn(double x);
extern __device__ __device_builtin__ float __double2float_rz(double x);
extern __device__ __device_builtin__ float __double2float_ru(double x);
extern __device__ __device_builtin__ float __double2float_rd(double x);
extern __device__ __device_builtin__ int __double2int_rn(double x);
extern __device__ __device_builtin__ int __double2int_ru(double x);
extern __device__ __device_builtin__ int __double2int_rd(double x);
extern __device__ __device_builtin__ unsigned int __double2uint_rn(double x);
extern __device__ __device_builtin__ unsigned int __double2uint_ru(double x);
extern __device__ __device_builtin__ unsigned int __double2uint_rd(double x);
extern __device__ __device_builtin__ long long int __double2ll_rn(double x);
extern __device__ __device_builtin__ long long int __double2ll_ru(double x);
extern __device__ __device_builtin__ long long int __double2ll_rd(double x);
extern __device__ __device_builtin__ unsigned long long int __double2ull_rn(double x);
extern __device__ __device_builtin__ unsigned long long int __double2ull_ru(double x);
extern __device__ __device_builtin__ unsigned long long int __double2ull_rd(double x);
extern __device__ __device_builtin__ double __int2double_rn(int x);
extern __device__ __device_builtin__ double __uint2double_rn(unsigned int x);
extern __device__ __device_builtin__ double __ll2double_rn(long long int x);
extern __device__ __device_builtin__ double __ll2double_rz(long long int x);
extern __device__ __device_builtin__ double __ll2double_ru(long long int x);
extern __device__ __device_builtin__ double __ll2double_rd(long long int x);
extern __device__ __device_builtin__ double __ull2double_rn(unsigned long long int x);
extern __device__ __device_builtin__ double __ull2double_rz(unsigned long long int x);
extern __device__ __device_builtin__ double __ull2double_ru(unsigned long long int x);
extern __device__ __device_builtin__ double __ull2double_rd(unsigned long long int x);
extern __device__ __device_builtin__ int __double2hiint(double x);
extern __device__ __device_builtin__ int __double2loint(double x);
extern __device__ __device_builtin__ double __hiloint2double(int hi, int lo);
}
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double fma(double a, double b, double c, enum cudaRoundMode mode);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double dmul(double a, double b, enum cudaRoundMode mode = cudaRoundNearest);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double dadd(double a, double b, enum cudaRoundMode mode = cudaRoundNearest);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double dsub(double a, double b, enum cudaRoundMode mode = cudaRoundNearest);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ int double2int(double a, enum cudaRoundMode mode = cudaRoundZero);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ unsigned int double2uint(double a, enum cudaRoundMode mode = cudaRoundZero);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ long long int double2ll(double a, enum cudaRoundMode mode = cudaRoundZero);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ unsigned long long int double2ull(double a, enum cudaRoundMode mode = cudaRoundZero);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double ll2double(long long int a, enum cudaRoundMode mode = cudaRoundNearest);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double ull2double(unsigned long long int a, enum cudaRoundMode mode = cudaRoundNearest);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double int2double(int a, enum cudaRoundMode mode = cudaRoundNearest);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double uint2double(unsigned int a, enum cudaRoundMode mode = cudaRoundNearest);
__DEVICE_DOUBLE_FUNCTIONS_DECL__ double float2double(float a, enum cudaRoundMode mode = cudaRoundNearest);
#undef __DEVICE_DOUBLE_FUNCTIONS_DECL__
#endif
#if !defined(__CUDACC_RTC__)
#include "device_double_functions.hpp"
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_DOUBLE_FUNCTIONS_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_DOUBLE_FUNCTIONS_H__
#endif
#if !defined(__SM_20_ATOMIC_FUNCTIONS_H__)
#define __SM_20_ATOMIC_FUNCTIONS_H__
#if defined(__CUDACC_RTC__)
#define __SM_20_ATOMIC_FUNCTIONS_DECL__ __device__
#else
#define __SM_20_ATOMIC_FUNCTIONS_DECL__ static __inline__ __device__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
#ifdef __CUDA_ARCH__
extern "C"
{
extern __device__ __device_builtin__ float __fAtomicAdd(float *address, float val);
}
#endif
__SM_20_ATOMIC_FUNCTIONS_DECL__ float atomicAdd(float *address, float val) __DEF_IF_HOST
#endif
#undef __DEF_IF_HOST
#undef __SM_20_ATOMIC_FUNCTIONS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_20_atomic_functions.hpp"
#endif
#endif
#if !defined(__SM_32_ATOMIC_FUNCTIONS_H__)
#define __SM_32_ATOMIC_FUNCTIONS_H__
#if defined(__CUDACC_RTC__)
#define __SM_32_ATOMIC_FUNCTIONS_DECL__ __device__
#else
#define __SM_32_ATOMIC_FUNCTIONS_DECL__ static __inline__ __device__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ >= 320
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
#ifdef __CUDA_ARCH__
extern "C"
{
extern __device__ __device_builtin__ long long __illAtomicMin(long long *address, long long val);
extern __device__ __device_builtin__ long long __illAtomicMax(long long *address, long long val);
extern __device__ __device_builtin__ long long __llAtomicAnd(long long *address, long long val);
extern __device__ __device_builtin__ long long __llAtomicOr(long long *address, long long val);
extern __device__ __device_builtin__ long long __llAtomicXor(long long *address, long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicMin(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicMax(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicAnd(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicOr (unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicXor(unsigned long long *address, unsigned long long val);
}
#endif
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicMin(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicMax(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicAnd(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicOr(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicXor(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicMin(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicMax(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicAnd(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicOr(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicXor(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
#endif
#endif
#undef __DEF_IF_HOST
#undef __SM_32_ATOMIC_FUNCTIONS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_32_atomic_functions.hpp"
#endif
#endif
#if !defined(__SM_35_ATOMIC_FUNCTIONS_H__)
#define __SM_35_ATOMIC_FUNCTIONS_H__
#if !defined(__SM_32_ATOMIC_FUNCTIONS_H__)
#define __SM_32_ATOMIC_FUNCTIONS_H__
#if defined(__CUDACC_RTC__)
#define __SM_32_ATOMIC_FUNCTIONS_DECL__ __device__
#else
#define __SM_32_ATOMIC_FUNCTIONS_DECL__ static __inline__ __device__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ >= 320
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
#ifdef __CUDA_ARCH__
extern "C"
{
extern __device__ __device_builtin__ long long __illAtomicMin(long long *address, long long val);
extern __device__ __device_builtin__ long long __illAtomicMax(long long *address, long long val);
extern __device__ __device_builtin__ long long __llAtomicAnd(long long *address, long long val);
extern __device__ __device_builtin__ long long __llAtomicOr(long long *address, long long val);
extern __device__ __device_builtin__ long long __llAtomicXor(long long *address, long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicMin(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicMax(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicAnd(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicOr (unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__ unsigned long long __ullAtomicXor(unsigned long long *address, unsigned long long val);
}
#endif
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicMin(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicMax(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicAnd(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicOr(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ long long atomicXor(long long *address, long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicMin(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicMax(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicAnd(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicOr(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_32_ATOMIC_FUNCTIONS_DECL__ unsigned long long atomicXor(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
#endif
#endif
#undef __DEF_IF_HOST
#undef __SM_32_ATOMIC_FUNCTIONS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_32_atomic_functions.hpp"
#endif
#endif
#endif
#if !defined(__SM_60_ATOMIC_FUNCTIONS_H__)
#define __SM_60_ATOMIC_FUNCTIONS_H__
#if defined(__CUDACC_RTC__)
#define __SM_60_ATOMIC_FUNCTIONS_DECL__ __device__
#else
#define __SM_60_ATOMIC_FUNCTIONS_DECL__ static __inline__ __device__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ >= 600
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
#ifdef __CUDA_ARCH__
extern "C"
{
extern __device__ __device_builtin__ double __dAtomicAdd(double *address, double val);
extern __device__ __device_builtin__
int __iAtomicAdd_block(int *address, int val);
extern __device__ __device_builtin__
int __iAtomicAdd_system(int *address, int val);
extern __device__ __device_builtin__
unsigned int __uAtomicAdd_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicAdd_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicAdd_block(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicAdd_system(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
float __fAtomicAdd_block(float *address, float val);
extern __device__ __device_builtin__
float __fAtomicAdd_system(float *address, float val);
extern __device__ __device_builtin__
double __dAtomicAdd_block(double *address, double val);
extern __device__ __device_builtin__
double __dAtomicAdd_system(double *address, double val);
extern __device__ __device_builtin__
int __iAtomicExch_block(int *address, int val);
extern __device__ __device_builtin__
int __iAtomicExch_system(int *address, int val);
extern __device__ __device_builtin__
unsigned int __uAtomicExch_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicExch_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicExch_block(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicExch_system(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
float __fAtomicExch_block(float *address, float val);
extern __device__ __device_builtin__
float __fAtomicExch_system(float *address, float val);
extern __device__ __device_builtin__
int __iAtomicMin_block(int *address, int val);
extern __device__ __device_builtin__
int __iAtomicMin_system(int *address, int val);
extern __device__ __device_builtin__
long long __illAtomicMin_block(long long *address, long long val);
extern __device__ __device_builtin__
long long __illAtomicMin_system(long long *address, long long val);
extern __device__ __device_builtin__
unsigned int __uAtomicMin_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicMin_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicMin_block(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicMin_system(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
int __iAtomicMax_block(int *address, int val);
extern __device__ __device_builtin__
int __iAtomicMax_system(int *address, int val);
extern __device__ __device_builtin__
long long __illAtomicMax_block(long long *address, long long val);
extern __device__ __device_builtin__
long long __illAtomicMax_system(long long *address, long long val);
extern __device__ __device_builtin__
unsigned int __uAtomicMax_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicMax_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicMax_block(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicMax_system(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
unsigned int __uAtomicInc_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicInc_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicDec_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicDec_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
int __iAtomicCAS_block(int *address, int compare, int val);
extern __device__ __device_builtin__
int __iAtomicCAS_system(int *address, int compare, int val);
extern __device__ __device_builtin__
unsigned int __uAtomicCAS_block(unsigned int *address, unsigned int compare,
unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicCAS_system(unsigned int *address, unsigned int compare,
unsigned int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicCAS_block(unsigned long long int *address,
unsigned long long int compare,
unsigned long long int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicCAS_system(unsigned long long int *address,
unsigned long long int compare,
unsigned long long int val);
extern __device__ __device_builtin__
int __iAtomicAnd_block(int *address, int val);
extern __device__ __device_builtin__
int __iAtomicAnd_system(int *address, int val);
extern __device__ __device_builtin__
long long __llAtomicAnd_block(long long *address, long long val);
extern __device__ __device_builtin__
long long __llAtomicAnd_system(long long *address, long long val);
extern __device__ __device_builtin__
unsigned int __uAtomicAnd_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicAnd_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicAnd_block(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicAnd_system(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
int __iAtomicOr_block(int *address, int val);
extern __device__ __device_builtin__
int __iAtomicOr_system(int *address, int val);
extern __device__ __device_builtin__
long long __llAtomicOr_block(long long *address, long long val);
extern __device__ __device_builtin__
long long __llAtomicOr_system(long long *address, long long val);
extern __device__ __device_builtin__
unsigned int __uAtomicOr_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicOr_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicOr_block(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicOr_system(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
int __iAtomicXor_block(int *address, int val);
extern __device__ __device_builtin__
int __iAtomicXor_system(int *address, int val);
extern __device__ __device_builtin__
long long __llAtomicXor_block(long long *address, long long val);
extern __device__ __device_builtin__
long long __llAtomicXor_system(long long *address, long long val);
extern __device__ __device_builtin__
unsigned int __uAtomicXor_block(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned int __uAtomicXor_system(unsigned int *address, unsigned int val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicXor_block(unsigned long long *address, unsigned long long val);
extern __device__ __device_builtin__
unsigned long long __ullAtomicXor_system(unsigned long long *address, unsigned long long val);
}
#endif
__SM_60_ATOMIC_FUNCTIONS_DECL__ double atomicAdd(double *address, double val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicAdd_block(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicAdd_system(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicAdd_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicAdd_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicAdd_block(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicAdd_system(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
float atomicAdd_block(float *address, float val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
float atomicAdd_system(float *address, float val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
double atomicAdd_block(double *address, double val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
double atomicAdd_system(double *address, double val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicSub_block(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicSub_system(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicSub_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicSub_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicExch_block(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicExch_system(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicExch_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicExch_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicExch_block(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicExch_system(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
float atomicExch_block(float *address, float val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
float atomicExch_system(float *address, float val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicMin_block(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicMin_system(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicMin_block(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicMin_system(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicMin_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicMin_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicMin_block(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicMin_system(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicMax_block(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicMax_system(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicMax_block(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicMax_system(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicMax_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicMax_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicMax_block(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicMax_system(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicInc_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicInc_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicDec_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicDec_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicCAS_block(int *address, int compare, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicCAS_system(int *address, int compare, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicCAS_block(unsigned int *address, unsigned int compare,
unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicCAS_system(unsigned int *address, unsigned int compare,
unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long int atomicCAS_block(unsigned long long int *address,
unsigned long long int compare,
unsigned long long int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long int atomicCAS_system(unsigned long long int *address,
unsigned long long int compare,
unsigned long long int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicAnd_block(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicAnd_system(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicAnd_block(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicAnd_system(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicAnd_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicAnd_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicAnd_block(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicAnd_system(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicOr_block(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicOr_system(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicOr_block(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicOr_system(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicOr_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicOr_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicOr_block(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicOr_system(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicXor_block(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
int atomicXor_system(int *address, int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicXor_block(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
long long atomicXor_system(long long *address, long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicXor_block(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned int atomicXor_system(unsigned int *address, unsigned int val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicXor_block(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
__SM_60_ATOMIC_FUNCTIONS_DECL__
unsigned long long atomicXor_system(unsigned long long *address, unsigned long long val) __DEF_IF_HOST
#endif
#endif
#undef __SM_60_ATOMIC_FUNCTIONS_DECL__
#undef __DEF_IF_HOST
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_60_atomic_functions.hpp"
#endif
#endif
#if !defined(__SM_20_INTRINSICS_H__)
#define __SM_20_INTRINSICS_H__
#if defined(__CUDACC_RTC__)
#define __SM_20_INTRINSICS_DECL__ __device__
#else
#define __SM_20_INTRINSICS_DECL__ static __inline__ __device__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
#if defined(_WIN32)
# define __DEPRECATED__(msg) __declspec(deprecated(msg))
#elif (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 5 && !defined(__clang__))))
# define __DEPRECATED__(msg) __attribute__((deprecated))
#else
# define __DEPRECATED__(msg) __attribute__((deprecated(msg)))
#endif
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 700
#define __WSB_DEPRECATION_MESSAGE(x) #x"() is not valid on compute_70 and above, and should be replaced with "#x"_sync()."\
"To continue using "#x"(), specify virtual architecture compute_60 when targeting sm_70 and above, for example, using the pair of compiler options: -arch=compute_60 -code=sm_70."
#else
#define __WSB_DEPRECATION_MESSAGE(x) #x"() is deprecated in favor of "#x"_sync() and may be removed in a future release (Use -Wno-deprecated-declarations to suppress this warning)."
#endif
extern "C"
{
extern __device__ __device_builtin__ void __threadfence_system(void);
extern __device__ __device_builtin__ double __ddiv_rn(double x, double y);
extern __device__ __device_builtin__ double __ddiv_rz(double x, double y);
extern __device__ __device_builtin__ double __ddiv_ru(double x, double y);
extern __device__ __device_builtin__ double __ddiv_rd(double x, double y);
extern __device__ __device_builtin__ double __drcp_rn(double x);
extern __device__ __device_builtin__ double __drcp_rz(double x);
extern __device__ __device_builtin__ double __drcp_ru(double x);
extern __device__ __device_builtin__ double __drcp_rd(double x);
extern __device__ __device_builtin__ double __dsqrt_rn(double x);
extern __device__ __device_builtin__ double __dsqrt_rz(double x);
extern __device__ __device_builtin__ double __dsqrt_ru(double x);
extern __device__ __device_builtin__ double __dsqrt_rd(double x);
extern __device__ __device_builtin__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__ballot)) unsigned int __ballot(int);
extern __device__ __device_builtin__ int __syncthreads_count(int);
extern __device__ __device_builtin__ int __syncthreads_and(int);
extern __device__ __device_builtin__ int __syncthreads_or(int);
extern __device__ __device_builtin__ long long int clock64(void);
extern __device__ __device_builtin__ float __fmaf_ieee_rn(float, float, float);
extern __device__ __device_builtin__ float __fmaf_ieee_rz(float, float, float);
extern __device__ __device_builtin__ float __fmaf_ieee_ru(float, float, float);
extern __device__ __device_builtin__ float __fmaf_ieee_rd(float, float, float);
extern __device__ __device_builtin__ long long int __double_as_longlong(double x);
extern __device__ __device_builtin__ double __longlong_as_double(long long int x);
extern __device__ __device_builtin__ double __fma_rn(double x, double y, double z);
extern __device__ __device_builtin__ double __fma_rz(double x, double y, double z);
extern __device__ __device_builtin__ double __fma_ru(double x, double y, double z);
extern __device__ __device_builtin__ double __fma_rd(double x, double y, double z);
extern __device__ __device_builtin__ double __dadd_rn(double x, double y);
extern __device__ __device_builtin__ double __dadd_rz(double x, double y);
extern __device__ __device_builtin__ double __dadd_ru(double x, double y);
extern __device__ __device_builtin__ double __dadd_rd(double x, double y);
extern __device__ __device_builtin__ double __dsub_rn(double x, double y);
extern __device__ __device_builtin__ double __dsub_rz(double x, double y);
extern __device__ __device_builtin__ double __dsub_ru(double x, double y);
extern __device__ __device_builtin__ double __dsub_rd(double x, double y);
extern __device__ __device_builtin__ double __dmul_rn(double x, double y);
extern __device__ __device_builtin__ double __dmul_rz(double x, double y);
extern __device__ __device_builtin__ double __dmul_ru(double x, double y);
extern __device__ __device_builtin__ double __dmul_rd(double x, double y);
extern __device__ __device_builtin__ float __double2float_rn(double x);
extern __device__ __device_builtin__ float __double2float_rz(double x);
extern __device__ __device_builtin__ float __double2float_ru(double x);
extern __device__ __device_builtin__ float __double2float_rd(double x);
extern __device__ __device_builtin__ int __double2int_rn(double x);
extern __device__ __device_builtin__ int __double2int_ru(double x);
extern __device__ __device_builtin__ int __double2int_rd(double x);
extern __device__ __device_builtin__ unsigned int __double2uint_rn(double x);
extern __device__ __device_builtin__ unsigned int __double2uint_ru(double x);
extern __device__ __device_builtin__ unsigned int __double2uint_rd(double x);
extern __device__ __device_builtin__ long long int __double2ll_rn(double x);
extern __device__ __device_builtin__ long long int __double2ll_ru(double x);
extern __device__ __device_builtin__ long long int __double2ll_rd(double x);
extern __device__ __device_builtin__ unsigned long long int __double2ull_rn(double x);
extern __device__ __device_builtin__ unsigned long long int __double2ull_ru(double x);
extern __device__ __device_builtin__ unsigned long long int __double2ull_rd(double x);
extern __device__ __device_builtin__ double __int2double_rn(int x);
extern __device__ __device_builtin__ double __uint2double_rn(unsigned int x);
extern __device__ __device_builtin__ double __ll2double_rn(long long int x);
extern __device__ __device_builtin__ double __ll2double_rz(long long int x);
extern __device__ __device_builtin__ double __ll2double_ru(long long int x);
extern __device__ __device_builtin__ double __ll2double_rd(long long int x);
extern __device__ __device_builtin__ double __ull2double_rn(unsigned long long int x);
extern __device__ __device_builtin__ double __ull2double_rz(unsigned long long int x);
extern __device__ __device_builtin__ double __ull2double_ru(unsigned long long int x);
extern __device__ __device_builtin__ double __ull2double_rd(unsigned long long int x);
extern __device__ __device_builtin__ int __double2hiint(double x);
extern __device__ __device_builtin__ int __double2loint(double x);
extern __device__ __device_builtin__ double __hiloint2double(int hi, int lo);
}
__SM_20_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__ballot)) unsigned int ballot(bool pred) __DEF_IF_HOST
__SM_20_INTRINSICS_DECL__ int syncthreads_count(bool pred) __DEF_IF_HOST
__SM_20_INTRINSICS_DECL__ bool syncthreads_and(bool pred) __DEF_IF_HOST
__SM_20_INTRINSICS_DECL__ bool syncthreads_or(bool pred) __DEF_IF_HOST
#undef __DEPRECATED__
#undef __WSB_DEPRECATION_MESSAGE
__SM_20_INTRINSICS_DECL__ unsigned int __isGlobal(const void *ptr) __DEF_IF_HOST
__SM_20_INTRINSICS_DECL__ unsigned int __isShared(const void *ptr) __DEF_IF_HOST
__SM_20_INTRINSICS_DECL__ unsigned int __isConstant(const void *ptr) __DEF_IF_HOST
__SM_20_INTRINSICS_DECL__ unsigned int __isLocal(const void *ptr) __DEF_IF_HOST
#endif
#undef __DEF_IF_HOST
#undef __SM_20_INTRINSICS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_20_intrinsics.hpp"
#endif
#endif
#if !defined(__SM_30_INTRINSICS_H__)
#define __SM_30_INTRINSICS_H__
#if defined(__CUDACC_RTC__)
#define __SM_30_INTRINSICS_DECL__ __device__
#else
#define __SM_30_INTRINSICS_DECL__ static __device__ __inline__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ >= 300
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
#if !defined warpSize && !defined __local_warpSize
#define warpSize 32
#define __local_warpSize
#endif
#if defined(_WIN32)
# define __DEPRECATED__(msg) __declspec(deprecated(msg))
#elif (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 5 && !defined(__clang__))))
# define __DEPRECATED__(msg) __attribute__((deprecated))
#else
# define __DEPRECATED__(msg) __attribute__((deprecated(msg)))
#endif
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ < 700
#define __WSB_DEPRECATION_MESSAGE(x) #x"() is deprecated in favor of "#x"_sync() and may be removed in a future release (Use -Wno-deprecated-declarations to suppress this warning)."
#endif
__SM_30_INTRINSICS_DECL__ unsigned __fns(unsigned mask, unsigned base, int offset) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ void __barrier_sync(unsigned id) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ void __barrier_sync_count(unsigned id, unsigned cnt) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ void __syncwarp(unsigned mask=0xFFFFFFFF) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ int __all_sync(unsigned mask, int pred) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ int __any_sync(unsigned mask, int pred) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ int __uni_sync(unsigned mask, int pred) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned __ballot_sync(unsigned mask, int pred) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned __activemask() __DEF_IF_HOST
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ < 700
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl)) int __shfl(int var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl)) unsigned int __shfl(unsigned int var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_up)) int __shfl_up(int var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_up)) unsigned int __shfl_up(unsigned int var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_down)) int __shfl_down(int var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_down)) unsigned int __shfl_down(unsigned int var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_xor)) int __shfl_xor(int var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_xor)) unsigned int __shfl_xor(unsigned int var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl)) float __shfl(float var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_up)) float __shfl_up(float var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_down)) float __shfl_down(float var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_xor)) float __shfl_xor(float var, int laneMask, int width=warpSize) __DEF_IF_HOST
#endif
__SM_30_INTRINSICS_DECL__ int __shfl_sync(unsigned mask, int var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned int __shfl_sync(unsigned mask, unsigned int var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ int __shfl_up_sync(unsigned mask, int var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned int __shfl_up_sync(unsigned mask, unsigned int var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ int __shfl_down_sync(unsigned mask, int var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned int __shfl_down_sync(unsigned mask, unsigned int var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ int __shfl_xor_sync(unsigned mask, int var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned int __shfl_xor_sync(unsigned mask, unsigned int var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ float __shfl_sync(unsigned mask, float var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ float __shfl_up_sync(unsigned mask, float var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ float __shfl_down_sync(unsigned mask, float var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ float __shfl_xor_sync(unsigned mask, float var, int laneMask, int width=warpSize) __DEF_IF_HOST
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ < 700
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl)) unsigned long long __shfl(unsigned long long var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl)) long long __shfl(long long var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_up)) long long __shfl_up(long long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_up)) unsigned long long __shfl_up(unsigned long long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_down)) long long __shfl_down(long long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_down)) unsigned long long __shfl_down(unsigned long long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_xor)) long long __shfl_xor(long long var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_xor)) unsigned long long __shfl_xor(unsigned long long var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl)) double __shfl(double var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_up)) double __shfl_up(double var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_down)) double __shfl_down(double var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_xor)) double __shfl_xor(double var, int laneMask, int width=warpSize) __DEF_IF_HOST
#endif
__SM_30_INTRINSICS_DECL__ long long __shfl_sync(unsigned mask, long long var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned long long __shfl_sync(unsigned mask, unsigned long long var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ long long __shfl_up_sync(unsigned mask, long long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned long long __shfl_up_sync(unsigned mask, unsigned long long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ long long __shfl_down_sync(unsigned mask, long long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned long long __shfl_down_sync(unsigned mask, unsigned long long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ long long __shfl_xor_sync(unsigned mask, long long var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned long long __shfl_xor_sync(unsigned mask, unsigned long long var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ double __shfl_sync(unsigned mask, double var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ double __shfl_up_sync(unsigned mask, double var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ double __shfl_down_sync(unsigned mask, double var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ double __shfl_xor_sync(unsigned mask, double var, int laneMask, int width=warpSize) __DEF_IF_HOST
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ < 700
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl)) long __shfl(long var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl)) unsigned long __shfl(unsigned long var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_up)) long __shfl_up(long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_up)) unsigned long __shfl_up(unsigned long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_down)) long __shfl_down(long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_down)) unsigned long __shfl_down(unsigned long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_xor)) long __shfl_xor(long var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ __DEPRECATED__(__WSB_DEPRECATION_MESSAGE(__shfl_xor)) unsigned long __shfl_xor(unsigned long var, int laneMask, int width=warpSize) __DEF_IF_HOST
#endif
__SM_30_INTRINSICS_DECL__ long __shfl_sync(unsigned mask, long var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned long __shfl_sync(unsigned mask, unsigned long var, int srcLane, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ long __shfl_up_sync(unsigned mask, long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned long __shfl_up_sync(unsigned mask, unsigned long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ long __shfl_down_sync(unsigned mask, long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned long __shfl_down_sync(unsigned mask, unsigned long var, unsigned int delta, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ long __shfl_xor_sync(unsigned mask, long var, int laneMask, int width=warpSize) __DEF_IF_HOST
__SM_30_INTRINSICS_DECL__ unsigned long __shfl_xor_sync(unsigned mask, unsigned long var, int laneMask, int width=warpSize) __DEF_IF_HOST
#undef __DEPRECATED__
#undef __WSB_DEPRECATION_MESSAGE
#if defined(__local_warpSize)
#undef warpSize
#undef __local_warpSize
#endif
#endif
#endif
#undef __DEF_IF_HOST
#undef __SM_30_INTRINSICS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_30_intrinsics.hpp"
#endif
#endif
#if !defined(__SM_32_INTRINSICS_H__)
#define __SM_32_INTRINSICS_H__
#if defined(__CUDACC_RTC__)
#define __SM_32_INTRINSICS_DECL__ __device__
#else
#define __SM_32_INTRINSICS_DECL__ static __device__ __inline__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ >= 320
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
__SM_32_INTRINSICS_DECL__ long __ldg(const long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long __ldg(const unsigned long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char __ldg(const char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ signed char __ldg(const signed char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short __ldg(const short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int __ldg(const int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long long __ldg(const long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char2 __ldg(const char2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char4 __ldg(const char4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short2 __ldg(const short2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short4 __ldg(const short4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int2 __ldg(const int2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int4 __ldg(const int4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ longlong2 __ldg(const longlong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned char __ldg(const unsigned char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned short __ldg(const unsigned short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __ldg(const unsigned int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long long __ldg(const unsigned long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar2 __ldg(const uchar2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar4 __ldg(const uchar4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort2 __ldg(const ushort2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort4 __ldg(const ushort4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint2 __ldg(const uint2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint4 __ldg(const uint4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ulonglong2 __ldg(const ulonglong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float __ldg(const float *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double __ldg(const double *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float2 __ldg(const float2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float4 __ldg(const float4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double2 __ldg(const double2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long __ldcg(const long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long __ldcg(const unsigned long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char __ldcg(const char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ signed char __ldcg(const signed char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short __ldcg(const short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int __ldcg(const int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long long __ldcg(const long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char2 __ldcg(const char2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char4 __ldcg(const char4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short2 __ldcg(const short2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short4 __ldcg(const short4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int2 __ldcg(const int2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int4 __ldcg(const int4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ longlong2 __ldcg(const longlong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned char __ldcg(const unsigned char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned short __ldcg(const unsigned short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __ldcg(const unsigned int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long long __ldcg(const unsigned long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar2 __ldcg(const uchar2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar4 __ldcg(const uchar4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort2 __ldcg(const ushort2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort4 __ldcg(const ushort4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint2 __ldcg(const uint2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint4 __ldcg(const uint4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ulonglong2 __ldcg(const ulonglong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float __ldcg(const float *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double __ldcg(const double *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float2 __ldcg(const float2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float4 __ldcg(const float4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double2 __ldcg(const double2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long __ldca(const long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long __ldca(const unsigned long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char __ldca(const char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ signed char __ldca(const signed char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short __ldca(const short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int __ldca(const int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long long __ldca(const long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char2 __ldca(const char2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char4 __ldca(const char4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short2 __ldca(const short2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short4 __ldca(const short4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int2 __ldca(const int2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int4 __ldca(const int4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ longlong2 __ldca(const longlong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned char __ldca(const unsigned char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned short __ldca(const unsigned short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __ldca(const unsigned int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long long __ldca(const unsigned long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar2 __ldca(const uchar2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar4 __ldca(const uchar4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort2 __ldca(const ushort2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort4 __ldca(const ushort4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint2 __ldca(const uint2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint4 __ldca(const uint4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ulonglong2 __ldca(const ulonglong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float __ldca(const float *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double __ldca(const double *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float2 __ldca(const float2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float4 __ldca(const float4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double2 __ldca(const double2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long __ldcs(const long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long __ldcs(const unsigned long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char __ldcs(const char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ signed char __ldcs(const signed char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short __ldcs(const short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int __ldcs(const int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long long __ldcs(const long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char2 __ldcs(const char2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char4 __ldcs(const char4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short2 __ldcs(const short2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short4 __ldcs(const short4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int2 __ldcs(const int2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int4 __ldcs(const int4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ longlong2 __ldcs(const longlong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned char __ldcs(const unsigned char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned short __ldcs(const unsigned short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __ldcs(const unsigned int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long long __ldcs(const unsigned long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar2 __ldcs(const uchar2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar4 __ldcs(const uchar4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort2 __ldcs(const ushort2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort4 __ldcs(const ushort4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint2 __ldcs(const uint2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint4 __ldcs(const uint4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ulonglong2 __ldcs(const ulonglong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float __ldcs(const float *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double __ldcs(const double *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float2 __ldcs(const float2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float4 __ldcs(const float4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double2 __ldcs(const double2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __funnelshift_l(unsigned int lo, unsigned int hi, unsigned int shift) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __funnelshift_lc(unsigned int lo, unsigned int hi, unsigned int shift) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __funnelshift_r(unsigned int lo, unsigned int hi, unsigned int shift) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __funnelshift_rc(unsigned int lo, unsigned int hi, unsigned int shift) __DEF_IF_HOST
#endif
#endif
#undef __SM_32_INTRINSICS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_32_intrinsics.hpp"
#endif
#endif
#if !defined(__SM_35_INTRINSICS_H__)
#define __SM_35_INTRINSICS_H__
#if !defined(__SM_32_INTRINSICS_H__)
#define __SM_32_INTRINSICS_H__
#if defined(__CUDACC_RTC__)
#define __SM_32_INTRINSICS_DECL__ __device__
#else
#define __SM_32_INTRINSICS_DECL__ static __device__ __inline__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ >= 320
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
__SM_32_INTRINSICS_DECL__ long __ldg(const long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long __ldg(const unsigned long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char __ldg(const char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ signed char __ldg(const signed char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short __ldg(const short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int __ldg(const int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long long __ldg(const long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char2 __ldg(const char2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char4 __ldg(const char4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short2 __ldg(const short2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short4 __ldg(const short4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int2 __ldg(const int2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int4 __ldg(const int4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ longlong2 __ldg(const longlong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned char __ldg(const unsigned char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned short __ldg(const unsigned short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __ldg(const unsigned int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long long __ldg(const unsigned long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar2 __ldg(const uchar2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar4 __ldg(const uchar4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort2 __ldg(const ushort2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort4 __ldg(const ushort4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint2 __ldg(const uint2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint4 __ldg(const uint4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ulonglong2 __ldg(const ulonglong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float __ldg(const float *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double __ldg(const double *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float2 __ldg(const float2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float4 __ldg(const float4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double2 __ldg(const double2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long __ldcg(const long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long __ldcg(const unsigned long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char __ldcg(const char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ signed char __ldcg(const signed char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short __ldcg(const short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int __ldcg(const int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long long __ldcg(const long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char2 __ldcg(const char2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char4 __ldcg(const char4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short2 __ldcg(const short2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short4 __ldcg(const short4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int2 __ldcg(const int2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int4 __ldcg(const int4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ longlong2 __ldcg(const longlong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned char __ldcg(const unsigned char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned short __ldcg(const unsigned short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __ldcg(const unsigned int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long long __ldcg(const unsigned long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar2 __ldcg(const uchar2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar4 __ldcg(const uchar4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort2 __ldcg(const ushort2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort4 __ldcg(const ushort4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint2 __ldcg(const uint2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint4 __ldcg(const uint4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ulonglong2 __ldcg(const ulonglong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float __ldcg(const float *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double __ldcg(const double *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float2 __ldcg(const float2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float4 __ldcg(const float4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double2 __ldcg(const double2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long __ldca(const long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long __ldca(const unsigned long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char __ldca(const char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ signed char __ldca(const signed char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short __ldca(const short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int __ldca(const int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long long __ldca(const long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char2 __ldca(const char2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char4 __ldca(const char4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short2 __ldca(const short2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short4 __ldca(const short4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int2 __ldca(const int2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int4 __ldca(const int4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ longlong2 __ldca(const longlong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned char __ldca(const unsigned char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned short __ldca(const unsigned short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __ldca(const unsigned int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long long __ldca(const unsigned long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar2 __ldca(const uchar2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar4 __ldca(const uchar4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort2 __ldca(const ushort2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort4 __ldca(const ushort4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint2 __ldca(const uint2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint4 __ldca(const uint4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ulonglong2 __ldca(const ulonglong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float __ldca(const float *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double __ldca(const double *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float2 __ldca(const float2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float4 __ldca(const float4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double2 __ldca(const double2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long __ldcs(const long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long __ldcs(const unsigned long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char __ldcs(const char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ signed char __ldcs(const signed char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short __ldcs(const short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int __ldcs(const int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ long long __ldcs(const long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char2 __ldcs(const char2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ char4 __ldcs(const char4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short2 __ldcs(const short2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ short4 __ldcs(const short4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int2 __ldcs(const int2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ int4 __ldcs(const int4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ longlong2 __ldcs(const longlong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned char __ldcs(const unsigned char *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned short __ldcs(const unsigned short *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __ldcs(const unsigned int *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned long long __ldcs(const unsigned long long *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar2 __ldcs(const uchar2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uchar4 __ldcs(const uchar4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort2 __ldcs(const ushort2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ushort4 __ldcs(const ushort4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint2 __ldcs(const uint2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ uint4 __ldcs(const uint4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ ulonglong2 __ldcs(const ulonglong2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float __ldcs(const float *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double __ldcs(const double *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float2 __ldcs(const float2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ float4 __ldcs(const float4 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ double2 __ldcs(const double2 *ptr) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __funnelshift_l(unsigned int lo, unsigned int hi, unsigned int shift) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __funnelshift_lc(unsigned int lo, unsigned int hi, unsigned int shift) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __funnelshift_r(unsigned int lo, unsigned int hi, unsigned int shift) __DEF_IF_HOST
__SM_32_INTRINSICS_DECL__ unsigned int __funnelshift_rc(unsigned int lo, unsigned int hi, unsigned int shift) __DEF_IF_HOST
#endif
#endif
#undef __SM_32_INTRINSICS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_32_intrinsics.hpp"
#endif
#endif
#endif
#if !defined(__SM_61_INTRINSICS_H__)
#define __SM_61_INTRINSICS_H__
#if defined(__CUDACC_RTC__)
#define __SM_61_INTRINSICS_DECL__ __device__
#else
#define __SM_61_INTRINSICS_DECL__ static __device__ __inline__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ >= 610
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
__SM_61_INTRINSICS_DECL__ int __dp2a_lo(int srcA, int srcB, int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ unsigned int __dp2a_lo(unsigned int srcA, unsigned int srcB, unsigned int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ int __dp2a_lo(short2 srcA, char4 srcB, int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ unsigned int __dp2a_lo(ushort2 srcA, uchar4 srcB, unsigned int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ int __dp2a_hi(int srcA, int srcB, int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ unsigned int __dp2a_hi(unsigned int srcA, unsigned int srcB, unsigned int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ int __dp2a_hi(short2 srcA, char4 srcB, int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ unsigned int __dp2a_hi(ushort2 srcA, uchar4 srcB, unsigned int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ int __dp4a(int srcA, int srcB, int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ unsigned int __dp4a(unsigned int srcA, unsigned int srcB, unsigned int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ int __dp4a(char4 srcA, char4 srcB, int c) __DEF_IF_HOST
__SM_61_INTRINSICS_DECL__ unsigned int __dp4a(uchar4 srcA, uchar4 srcB, unsigned int c) __DEF_IF_HOST
#endif
#endif
#undef __DEF_IF_HOST
#undef __SM_61_INTRINSICS_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_61_intrinsics.hpp"
#endif
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/sm_70_rt.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/sm_70_rt.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_SM_70_RT_H__
#endif
#if !defined(__SM_70_RT_H__)
#define __SM_70_RT_H__
#if defined(__CUDACC_RTC__)
#define __SM_70_RT_DECL__ __host__ __device__
#else
#define __SM_70_RT_DECL__ static __device__ __inline__
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || __CUDA_ARCH__ >= 700
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#ifndef __CUDA_ARCH__
#define __DEF_IF_HOST { }
#else
#define __DEF_IF_HOST ;
#endif
__SM_70_RT_DECL__ unsigned int __match_any_sync(unsigned mask, unsigned value) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_any_sync(unsigned mask, int value) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_any_sync(unsigned mask, unsigned long value) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_any_sync(unsigned mask, long value) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_any_sync(unsigned mask, unsigned long long value) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_any_sync(unsigned mask, long long value) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_any_sync(unsigned mask, float value) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_any_sync(unsigned mask, double value) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_all_sync(unsigned mask, unsigned value, int *pred) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_all_sync(unsigned mask, int value, int *pred) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_all_sync(unsigned mask, unsigned long value, int *pred) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_all_sync(unsigned mask, long value, int *pred) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_all_sync(unsigned mask, unsigned long long value, int *pred) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_all_sync(unsigned mask, long long value, int *pred) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_all_sync(unsigned mask, float value, int *pred) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned int __match_all_sync(unsigned mask, double value, int *pred) __DEF_IF_HOST
__SM_70_RT_DECL__ void __nanosleep(unsigned int ns) __DEF_IF_HOST
__SM_70_RT_DECL__ unsigned short int atomicCAS(unsigned short int *address, unsigned short int compare, unsigned short int val) __DEF_IF_HOST
#endif
#endif
#undef __DEF_IF_HOST
#undef __SM_70_RT_DECL__
#if !defined(__CUDACC_RTC__) && defined(__CUDA_ARCH__)
#include "sm_70_rt.hpp"
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_SM_70_RT_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_SM_70_RT_H__
#endif
#if !defined(__SURFACE_FUNCTIONS_H__)
#define __SURFACE_FUNCTIONS_H__
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
#if !defined(__CUDA_SURFACE_TYPES_H__)
#define __CUDA_SURFACE_TYPES_H__
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__CUDA_RUNTIME_API_H__)
#define __CUDA_RUNTIME_API_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#define CUDART_VERSION 10020
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__DEVICE_TYPES_H__)
#define __DEVICE_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
enum __device_builtin__ cudaRoundMode
{
cudaRoundNearest,
cudaRoundZero,
cudaRoundPosInf,
cudaRoundMinInf
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DEVICE_TYPES_H__
#endif
#endif
#if !defined(__CUDACC_RTC__)
#endif
#if !defined(__SURFACE_TYPES_H__)
#define __SURFACE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaSurfaceType1D 0x01
#define cudaSurfaceType2D 0x02
#define cudaSurfaceType3D 0x03
#define cudaSurfaceTypeCubemap 0x0C
#define cudaSurfaceType1DLayered 0xF1
#define cudaSurfaceType2DLayered 0xF2
#define cudaSurfaceTypeCubemapLayered 0xFC
enum __device_builtin__ cudaSurfaceBoundaryMode
{
cudaBoundaryModeZero = 0,
cudaBoundaryModeClamp = 1,
cudaBoundaryModeTrap = 2
};
enum __device_builtin__ cudaSurfaceFormatMode
{
cudaFormatModeForced = 0,
cudaFormatModeAuto = 1
};
struct __device_builtin__ surfaceReference
{
struct cudaChannelFormatDesc channelDesc;
};
typedef __device_builtin__ unsigned long long cudaSurfaceObject_t;
#endif
#if !defined(__TEXTURE_TYPES_H__)
#define __TEXTURE_TYPES_H__
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#define cudaTextureType1D 0x01
#define cudaTextureType2D 0x02
#define cudaTextureType3D 0x03
#define cudaTextureTypeCubemap 0x0C
#define cudaTextureType1DLayered 0xF1
#define cudaTextureType2DLayered 0xF2
#define cudaTextureTypeCubemapLayered 0xFC
enum __device_builtin__ cudaTextureAddressMode
{
cudaAddressModeWrap = 0,
cudaAddressModeClamp = 1,
cudaAddressModeMirror = 2,
cudaAddressModeBorder = 3
};
enum __device_builtin__ cudaTextureFilterMode
{
cudaFilterModePoint = 0,
cudaFilterModeLinear = 1
};
enum __device_builtin__ cudaTextureReadMode
{
cudaReadModeElementType = 0,
cudaReadModeNormalizedFloat = 1
};
struct __device_builtin__ textureReference
{
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
int sRGB;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
int __cudaReserved[15];
};
struct __device_builtin__ cudaTextureDesc
{
enum cudaTextureAddressMode addressMode[3];
enum cudaTextureFilterMode filterMode;
enum cudaTextureReadMode readMode;
int sRGB;
float borderColor[4];
int normalizedCoords;
unsigned int maxAnisotropy;
enum cudaTextureFilterMode mipmapFilterMode;
float mipmapLevelBias;
float minMipmapLevelClamp;
float maxMipmapLevelClamp;
};
typedef __device_builtin__ unsigned long long cudaTextureObject_t;
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_DEVICE_RUNTIME_API_H__)
#define __CUDA_DEVICE_RUNTIME_API_H__
#if !defined(__CUDACC_RTC__)
#if !defined(__CUDACC_INTERNAL_NO_STUBS__) && !defined(__CUDACC_RDC__) && !defined(__CUDACC_EWP__) && defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 350) && !defined(__CUDADEVRT_INTERNAL__)
#if defined(__cplusplus)
extern "C" {
#endif
struct cudaFuncAttributes;
#if defined(_WIN32)
#define __NV_WEAK__ __declspec(nv_weak)
#else
#define __NV_WEAK__ __attribute__((nv_weak))
#endif
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaMalloc(void **p, size_t s)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *p, const void *c)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaGetDevice(int *device)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize)
{
return cudaErrorUnknown;
}
__device__ __NV_WEAK__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags)
{
return cudaErrorUnknown;
}
#undef __NV_WEAK__
#if defined(__cplusplus)
}
#endif
#endif
#endif
#if defined(__cplusplus) && defined(__CUDACC__)
#if !defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350)
#if !defined(__DRIVER_TYPES_H__)
#define __DRIVER_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__VECTOR_TYPES_H__)
#define __VECTOR_TYPES_H__
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__CUDACC__) && !defined(__CUDACC_RTC__) && \
defined(_WIN32) && !defined(_WIN64)
#pragma warning(push)
#pragma warning(disable: 4201 4408)
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ tag \
{ \
union \
{ \
struct { members }; \
struct { long long int :1,:0; }; \
}; \
}
#else
#define __cuda_builtin_vector_align8(tag, members) \
struct __device_builtin__ __align__(8) tag \
{ \
members \
}
#endif
struct __device_builtin__ char1
{
signed char x;
};
struct __device_builtin__ uchar1
{
unsigned char x;
};
struct __device_builtin__ __align__(2) char2
{
signed char x, y;
};
struct __device_builtin__ __align__(2) uchar2
{
unsigned char x, y;
};
struct __device_builtin__ char3
{
signed char x, y, z;
};
struct __device_builtin__ uchar3
{
unsigned char x, y, z;
};
struct __device_builtin__ __align__(4) char4
{
signed char x, y, z, w;
};
struct __device_builtin__ __align__(4) uchar4
{
unsigned char x, y, z, w;
};
struct __device_builtin__ short1
{
short x;
};
struct __device_builtin__ ushort1
{
unsigned short x;
};
struct __device_builtin__ __align__(4) short2
{
short x, y;
};
struct __device_builtin__ __align__(4) ushort2
{
unsigned short x, y;
};
struct __device_builtin__ short3
{
short x, y, z;
};
struct __device_builtin__ ushort3
{
unsigned short x, y, z;
};
__cuda_builtin_vector_align8(short4, short x; short y; short z; short w;);
__cuda_builtin_vector_align8(ushort4, unsigned short x; unsigned short y; unsigned short z; unsigned short w;);
struct __device_builtin__ int1
{
int x;
};
struct __device_builtin__ uint1
{
unsigned int x;
};
__cuda_builtin_vector_align8(int2, int x; int y;);
__cuda_builtin_vector_align8(uint2, unsigned int x; unsigned int y;);
struct __device_builtin__ int3
{
int x, y, z;
};
struct __device_builtin__ uint3
{
unsigned int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) int4
{
int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) uint4
{
unsigned int x, y, z, w;
};
struct __device_builtin__ long1
{
long int x;
};
struct __device_builtin__ ulong1
{
unsigned long x;
};
#if defined(_WIN32)
__cuda_builtin_vector_align8(long2, long int x; long int y;);
__cuda_builtin_vector_align8(ulong2, unsigned long int x; unsigned long int y;);
#else
struct __device_builtin__ __align__(2*sizeof(long int)) long2
{
long int x, y;
};
struct __device_builtin__ __align__(2*sizeof(unsigned long int)) ulong2
{
unsigned long int x, y;
};
#endif
struct __device_builtin__ long3
{
long int x, y, z;
};
struct __device_builtin__ ulong3
{
unsigned long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) long4
{
long int x, y, z, w;
};
struct __device_builtin__ __builtin_align__(16) ulong4
{
unsigned long int x, y, z, w;
};
struct __device_builtin__ float1
{
float x;
};
#if !defined(__CUDACC__) && defined(__arm__) && \
defined(__ARM_PCS_VFP) && __GNUC__ == 4 && __GNUC_MINOR__ == 6
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-pedantic"
struct __device_builtin__ __attribute__((aligned(8))) float2
{
float x; float y; float __cuda_gnu_arm_ice_workaround[0];
};
#pragma GCC poison __cuda_gnu_arm_ice_workaround
#pragma GCC diagnostic pop
#else
__cuda_builtin_vector_align8(float2, float x; float y;);
#endif
struct __device_builtin__ float3
{
float x, y, z;
};
struct __device_builtin__ __builtin_align__(16) float4
{
float x, y, z, w;
};
struct __device_builtin__ longlong1
{
long long int x;
};
struct __device_builtin__ ulonglong1
{
unsigned long long int x;
};
struct __device_builtin__ __builtin_align__(16) longlong2
{
long long int x, y;
};
struct __device_builtin__ __builtin_align__(16) ulonglong2
{
unsigned long long int x, y;
};
struct __device_builtin__ longlong3
{
long long int x, y, z;
};
struct __device_builtin__ ulonglong3
{
unsigned long long int x, y, z;
};
struct __device_builtin__ __builtin_align__(16) longlong4
{
long long int x, y, z ,w;
};
struct __device_builtin__ __builtin_align__(16) ulonglong4
{
unsigned long long int x, y, z, w;
};
struct __device_builtin__ double1
{
double x;
};
struct __device_builtin__ __builtin_align__(16) double2
{
double x, y;
};
struct __device_builtin__ double3
{
double x, y, z;
};
struct __device_builtin__ __builtin_align__(16) double4
{
double x, y, z, w;
};
#if !defined(__CUDACC__) && defined(_WIN32) && !defined(_WIN64)
#pragma warning(pop)
#endif
typedef __device_builtin__ struct char1 char1;
typedef __device_builtin__ struct uchar1 uchar1;
typedef __device_builtin__ struct char2 char2;
typedef __device_builtin__ struct uchar2 uchar2;
typedef __device_builtin__ struct char3 char3;
typedef __device_builtin__ struct uchar3 uchar3;
typedef __device_builtin__ struct char4 char4;
typedef __device_builtin__ struct uchar4 uchar4;
typedef __device_builtin__ struct short1 short1;
typedef __device_builtin__ struct ushort1 ushort1;
typedef __device_builtin__ struct short2 short2;
typedef __device_builtin__ struct ushort2 ushort2;
typedef __device_builtin__ struct short3 short3;
typedef __device_builtin__ struct ushort3 ushort3;
typedef __device_builtin__ struct short4 short4;
typedef __device_builtin__ struct ushort4 ushort4;
typedef __device_builtin__ struct int1 int1;
typedef __device_builtin__ struct uint1 uint1;
typedef __device_builtin__ struct int2 int2;
typedef __device_builtin__ struct uint2 uint2;
typedef __device_builtin__ struct int3 int3;
typedef __device_builtin__ struct uint3 uint3;
typedef __device_builtin__ struct int4 int4;
typedef __device_builtin__ struct uint4 uint4;
typedef __device_builtin__ struct long1 long1;
typedef __device_builtin__ struct ulong1 ulong1;
typedef __device_builtin__ struct long2 long2;
typedef __device_builtin__ struct ulong2 ulong2;
typedef __device_builtin__ struct long3 long3;
typedef __device_builtin__ struct ulong3 ulong3;
typedef __device_builtin__ struct long4 long4;
typedef __device_builtin__ struct ulong4 ulong4;
typedef __device_builtin__ struct float1 float1;
typedef __device_builtin__ struct float2 float2;
typedef __device_builtin__ struct float3 float3;
typedef __device_builtin__ struct float4 float4;
typedef __device_builtin__ struct longlong1 longlong1;
typedef __device_builtin__ struct ulonglong1 ulonglong1;
typedef __device_builtin__ struct longlong2 longlong2;
typedef __device_builtin__ struct ulonglong2 ulonglong2;
typedef __device_builtin__ struct longlong3 longlong3;
typedef __device_builtin__ struct ulonglong3 ulonglong3;
typedef __device_builtin__ struct longlong4 longlong4;
typedef __device_builtin__ struct ulonglong4 ulonglong4;
typedef __device_builtin__ struct double1 double1;
typedef __device_builtin__ struct double2 double2;
typedef __device_builtin__ struct double3 double3;
typedef __device_builtin__ struct double4 double4;
struct __device_builtin__ dim3
{
unsigned int x, y, z;
#if defined(__cplusplus)
#if __cplusplus >= 201103L
__host__ __device__ constexpr dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ constexpr dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ constexpr operator uint3(void) const { return uint3{x, y, z}; }
#else
__host__ __device__ dim3(unsigned int vx = 1, unsigned int vy = 1, unsigned int vz = 1) : x(vx), y(vy), z(vz) {}
__host__ __device__ dim3(uint3 v) : x(v.x), y(v.y), z(v.z) {}
__host__ __device__ operator uint3(void) const { uint3 t; t.x = x; t.y = y; t.z = z; return t; }
#endif
#endif
};
typedef __device_builtin__ struct dim3 dim3;
#undef __cuda_builtin_vector_align8
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_VECTOR_TYPES_H__
#endif
#endif
#if !defined(__CUDA_INTERNAL_COMPILATION__)
#if !defined(__CUDACC_RTC__)
#include
#include
#endif
#define cudaHostAllocDefault 0x00
#define cudaHostAllocPortable 0x01
#define cudaHostAllocMapped 0x02
#define cudaHostAllocWriteCombined 0x04
#define cudaHostRegisterDefault 0x00
#define cudaHostRegisterPortable 0x01
#define cudaHostRegisterMapped 0x02
#define cudaHostRegisterIoMemory 0x04
#define cudaPeerAccessDefault 0x00
#define cudaStreamDefault 0x00
#define cudaStreamNonBlocking 0x01
#define cudaStreamLegacy ((cudaStream_t)0x1)
#define cudaStreamPerThread ((cudaStream_t)0x2)
#define cudaEventDefault 0x00
#define cudaEventBlockingSync 0x01
#define cudaEventDisableTiming 0x02
#define cudaEventInterprocess 0x04
#define cudaDeviceScheduleAuto 0x00
#define cudaDeviceScheduleSpin 0x01
#define cudaDeviceScheduleYield 0x02
#define cudaDeviceScheduleBlockingSync 0x04
#define cudaDeviceBlockingSync 0x04
#define cudaDeviceScheduleMask 0x07
#define cudaDeviceMapHost 0x08
#define cudaDeviceLmemResizeToMax 0x10
#define cudaDeviceMask 0x1f
#define cudaArrayDefault 0x00
#define cudaArrayLayered 0x01
#define cudaArraySurfaceLoadStore 0x02
#define cudaArrayCubemap 0x04
#define cudaArrayTextureGather 0x08
#define cudaArrayColorAttachment 0x20
#define cudaIpcMemLazyEnablePeerAccess 0x01
#define cudaMemAttachGlobal 0x01
#define cudaMemAttachHost 0x02
#define cudaMemAttachSingle 0x04
#define cudaOccupancyDefault 0x00
#define cudaOccupancyDisableCachingOverride 0x01
#define cudaCpuDeviceId ((int)-1)
#define cudaInvalidDeviceId ((int)-2)
#define cudaCooperativeLaunchMultiDeviceNoPreSync 0x01
#define cudaCooperativeLaunchMultiDeviceNoPostSync 0x02
#endif
#if defined(__DOXYGEN_ONLY__) || defined(CUDA_ENABLE_DEPRECATED)
#define __CUDA_DEPRECATED
#elif defined(_MSC_VER)
#define __CUDA_DEPRECATED __declspec(deprecated)
#elif defined(__GNUC__)
#define __CUDA_DEPRECATED __attribute__((deprecated))
#else
#define __CUDA_DEPRECATED
#endif
enum __device_builtin__ cudaError
{
cudaSuccess = 0,
cudaErrorInvalidValue = 1,
cudaErrorMemoryAllocation = 2,
cudaErrorInitializationError = 3,
cudaErrorCudartUnloading = 4,
cudaErrorProfilerDisabled = 5,
cudaErrorProfilerNotInitialized = 6,
cudaErrorProfilerAlreadyStarted = 7,
cudaErrorProfilerAlreadyStopped = 8,
cudaErrorInvalidConfiguration = 9,
cudaErrorInvalidPitchValue = 12,
cudaErrorInvalidSymbol = 13,
cudaErrorInvalidHostPointer = 16,
cudaErrorInvalidDevicePointer = 17,
cudaErrorInvalidTexture = 18,
cudaErrorInvalidTextureBinding = 19,
cudaErrorInvalidChannelDescriptor = 20,
cudaErrorInvalidMemcpyDirection = 21,
cudaErrorAddressOfConstant = 22,
cudaErrorTextureFetchFailed = 23,
cudaErrorTextureNotBound = 24,
cudaErrorSynchronizationError = 25,
cudaErrorInvalidFilterSetting = 26,
cudaErrorInvalidNormSetting = 27,
cudaErrorMixedDeviceExecution = 28,
cudaErrorNotYetImplemented = 31,
cudaErrorMemoryValueTooLarge = 32,
cudaErrorInsufficientDriver = 35,
cudaErrorInvalidSurface = 37,
cudaErrorDuplicateVariableName = 43,
cudaErrorDuplicateTextureName = 44,
cudaErrorDuplicateSurfaceName = 45,
cudaErrorDevicesUnavailable = 46,
cudaErrorIncompatibleDriverContext = 49,
cudaErrorMissingConfiguration = 52,
cudaErrorPriorLaunchFailure = 53,
cudaErrorLaunchMaxDepthExceeded = 65,
cudaErrorLaunchFileScopedTex = 66,
cudaErrorLaunchFileScopedSurf = 67,
cudaErrorSyncDepthExceeded = 68,
cudaErrorLaunchPendingCountExceeded = 69,
cudaErrorInvalidDeviceFunction = 98,
cudaErrorNoDevice = 100,
cudaErrorInvalidDevice = 101,
cudaErrorStartupFailure = 127,
cudaErrorInvalidKernelImage = 200,
cudaErrorDeviceUninitialized = 201,
cudaErrorMapBufferObjectFailed = 205,
cudaErrorUnmapBufferObjectFailed = 206,
cudaErrorArrayIsMapped = 207,
cudaErrorAlreadyMapped = 208,
cudaErrorNoKernelImageForDevice = 209,
cudaErrorAlreadyAcquired = 210,
cudaErrorNotMapped = 211,
cudaErrorNotMappedAsArray = 212,
cudaErrorNotMappedAsPointer = 213,
cudaErrorECCUncorrectable = 214,
cudaErrorUnsupportedLimit = 215,
cudaErrorDeviceAlreadyInUse = 216,
cudaErrorPeerAccessUnsupported = 217,
cudaErrorInvalidPtx = 218,
cudaErrorInvalidGraphicsContext = 219,
cudaErrorNvlinkUncorrectable = 220,
cudaErrorJitCompilerNotFound = 221,
cudaErrorInvalidSource = 300,
cudaErrorFileNotFound = 301,
cudaErrorSharedObjectSymbolNotFound = 302,
cudaErrorSharedObjectInitFailed = 303,
cudaErrorOperatingSystem = 304,
cudaErrorInvalidResourceHandle = 400,
cudaErrorIllegalState = 401,
cudaErrorSymbolNotFound = 500,
cudaErrorNotReady = 600,
cudaErrorIllegalAddress = 700,
cudaErrorLaunchOutOfResources = 701,
cudaErrorLaunchTimeout = 702,
cudaErrorLaunchIncompatibleTexturing = 703,
cudaErrorPeerAccessAlreadyEnabled = 704,
cudaErrorPeerAccessNotEnabled = 705,
cudaErrorSetOnActiveProcess = 708,
cudaErrorContextIsDestroyed = 709,
cudaErrorAssert = 710,
cudaErrorTooManyPeers = 711,
cudaErrorHostMemoryAlreadyRegistered = 712,
cudaErrorHostMemoryNotRegistered = 713,
cudaErrorHardwareStackError = 714,
cudaErrorIllegalInstruction = 715,
cudaErrorMisalignedAddress = 716,
cudaErrorInvalidAddressSpace = 717,
cudaErrorInvalidPc = 718,
cudaErrorLaunchFailure = 719,
cudaErrorCooperativeLaunchTooLarge = 720,
cudaErrorNotPermitted = 800,
cudaErrorNotSupported = 801,
cudaErrorSystemNotReady = 802,
cudaErrorSystemDriverMismatch = 803,
cudaErrorCompatNotSupportedOnDevice = 804,
cudaErrorStreamCaptureUnsupported = 900,
cudaErrorStreamCaptureInvalidated = 901,
cudaErrorStreamCaptureMerge = 902,
cudaErrorStreamCaptureUnmatched = 903,
cudaErrorStreamCaptureUnjoined = 904,
cudaErrorStreamCaptureIsolation = 905,
cudaErrorStreamCaptureImplicit = 906,
cudaErrorCapturedEvent = 907,
cudaErrorStreamCaptureWrongThread = 908,
cudaErrorTimeout = 909,
cudaErrorGraphExecUpdateFailure = 910,
cudaErrorUnknown = 999,
cudaErrorApiFailureBase = 10000
};
enum __device_builtin__ cudaChannelFormatKind
{
cudaChannelFormatKindSigned = 0,
cudaChannelFormatKindUnsigned = 1,
cudaChannelFormatKindFloat = 2,
cudaChannelFormatKindNone = 3
};
struct __device_builtin__ cudaChannelFormatDesc
{
int x;
int y;
int z;
int w;
enum cudaChannelFormatKind f;
};
typedef struct cudaArray *cudaArray_t;
typedef const struct cudaArray *cudaArray_const_t;
struct cudaArray;
typedef struct cudaMipmappedArray *cudaMipmappedArray_t;
typedef const struct cudaMipmappedArray *cudaMipmappedArray_const_t;
struct cudaMipmappedArray;
enum __device_builtin__ cudaMemoryType
{
cudaMemoryTypeUnregistered = 0,
cudaMemoryTypeHost = 1,
cudaMemoryTypeDevice = 2,
cudaMemoryTypeManaged = 3
};
enum __device_builtin__ cudaMemcpyKind
{
cudaMemcpyHostToHost = 0,
cudaMemcpyHostToDevice = 1,
cudaMemcpyDeviceToHost = 2,
cudaMemcpyDeviceToDevice = 3,
cudaMemcpyDefault = 4
};
struct __device_builtin__ cudaPitchedPtr
{
void *ptr;
size_t pitch;
size_t xsize;
size_t ysize;
};
struct __device_builtin__ cudaExtent
{
size_t width;
size_t height;
size_t depth;
};
struct __device_builtin__ cudaPos
{
size_t x;
size_t y;
size_t z;
};
struct __device_builtin__ cudaMemcpy3DParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
struct cudaExtent extent;
enum cudaMemcpyKind kind;
};
struct __device_builtin__ cudaMemcpy3DPeerParms
{
cudaArray_t srcArray;
struct cudaPos srcPos;
struct cudaPitchedPtr srcPtr;
int srcDevice;
cudaArray_t dstArray;
struct cudaPos dstPos;
struct cudaPitchedPtr dstPtr;
int dstDevice;
struct cudaExtent extent;
};
struct __device_builtin__ cudaMemsetParams {
void *dst;
size_t pitch;
unsigned int value;
unsigned int elementSize;
size_t width;
size_t height;
};
#ifdef _WIN32
#define CUDART_CB __stdcall
#else
#define CUDART_CB
#endif
typedef void (CUDART_CB *cudaHostFn_t)(void *userData);
struct __device_builtin__ cudaHostNodeParams {
cudaHostFn_t fn;
void* userData;
};
enum __device_builtin__ cudaStreamCaptureStatus {
cudaStreamCaptureStatusNone = 0,
cudaStreamCaptureStatusActive = 1,
cudaStreamCaptureStatusInvalidated = 2
};
enum __device_builtin__ cudaStreamCaptureMode {
cudaStreamCaptureModeGlobal = 0,
cudaStreamCaptureModeThreadLocal = 1,
cudaStreamCaptureModeRelaxed = 2
};
struct cudaGraphicsResource;
enum __device_builtin__ cudaGraphicsRegisterFlags
{
cudaGraphicsRegisterFlagsNone = 0,
cudaGraphicsRegisterFlagsReadOnly = 1,
cudaGraphicsRegisterFlagsWriteDiscard = 2,
cudaGraphicsRegisterFlagsSurfaceLoadStore = 4,
cudaGraphicsRegisterFlagsTextureGather = 8
};
enum __device_builtin__ cudaGraphicsMapFlags
{
cudaGraphicsMapFlagsNone = 0,
cudaGraphicsMapFlagsReadOnly = 1,
cudaGraphicsMapFlagsWriteDiscard = 2
};
enum __device_builtin__ cudaGraphicsCubeFace
{
cudaGraphicsCubeFacePositiveX = 0x00,
cudaGraphicsCubeFaceNegativeX = 0x01,
cudaGraphicsCubeFacePositiveY = 0x02,
cudaGraphicsCubeFaceNegativeY = 0x03,
cudaGraphicsCubeFacePositiveZ = 0x04,
cudaGraphicsCubeFaceNegativeZ = 0x05
};
enum __device_builtin__ cudaResourceType
{
cudaResourceTypeArray = 0x00,
cudaResourceTypeMipmappedArray = 0x01,
cudaResourceTypeLinear = 0x02,
cudaResourceTypePitch2D = 0x03
};
enum __device_builtin__ cudaResourceViewFormat
{
cudaResViewFormatNone = 0x00,
cudaResViewFormatUnsignedChar1 = 0x01,
cudaResViewFormatUnsignedChar2 = 0x02,
cudaResViewFormatUnsignedChar4 = 0x03,
cudaResViewFormatSignedChar1 = 0x04,
cudaResViewFormatSignedChar2 = 0x05,
cudaResViewFormatSignedChar4 = 0x06,
cudaResViewFormatUnsignedShort1 = 0x07,
cudaResViewFormatUnsignedShort2 = 0x08,
cudaResViewFormatUnsignedShort4 = 0x09,
cudaResViewFormatSignedShort1 = 0x0a,
cudaResViewFormatSignedShort2 = 0x0b,
cudaResViewFormatSignedShort4 = 0x0c,
cudaResViewFormatUnsignedInt1 = 0x0d,
cudaResViewFormatUnsignedInt2 = 0x0e,
cudaResViewFormatUnsignedInt4 = 0x0f,
cudaResViewFormatSignedInt1 = 0x10,
cudaResViewFormatSignedInt2 = 0x11,
cudaResViewFormatSignedInt4 = 0x12,
cudaResViewFormatHalf1 = 0x13,
cudaResViewFormatHalf2 = 0x14,
cudaResViewFormatHalf4 = 0x15,
cudaResViewFormatFloat1 = 0x16,
cudaResViewFormatFloat2 = 0x17,
cudaResViewFormatFloat4 = 0x18,
cudaResViewFormatUnsignedBlockCompressed1 = 0x19,
cudaResViewFormatUnsignedBlockCompressed2 = 0x1a,
cudaResViewFormatUnsignedBlockCompressed3 = 0x1b,
cudaResViewFormatUnsignedBlockCompressed4 = 0x1c,
cudaResViewFormatSignedBlockCompressed4 = 0x1d,
cudaResViewFormatUnsignedBlockCompressed5 = 0x1e,
cudaResViewFormatSignedBlockCompressed5 = 0x1f,
cudaResViewFormatUnsignedBlockCompressed6H = 0x20,
cudaResViewFormatSignedBlockCompressed6H = 0x21,
cudaResViewFormatUnsignedBlockCompressed7 = 0x22
};
struct __device_builtin__ cudaResourceDesc {
enum cudaResourceType resType;
union {
struct {
cudaArray_t array;
} array;
struct {
cudaMipmappedArray_t mipmap;
} mipmap;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t sizeInBytes;
} linear;
struct {
void *devPtr;
struct cudaChannelFormatDesc desc;
size_t width;
size_t height;
size_t pitchInBytes;
} pitch2D;
} res;
};
struct __device_builtin__ cudaResourceViewDesc
{
enum cudaResourceViewFormat format;
size_t width;
size_t height;
size_t depth;
unsigned int firstMipmapLevel;
unsigned int lastMipmapLevel;
unsigned int firstLayer;
unsigned int lastLayer;
};
struct __device_builtin__ cudaPointerAttributes
{
__CUDA_DEPRECATED enum cudaMemoryType memoryType;
enum cudaMemoryType type;
int device;
void *devicePointer;
void *hostPointer;
__CUDA_DEPRECATED int isManaged;
};
struct __device_builtin__ cudaFuncAttributes
{
size_t sharedSizeBytes;
size_t constSizeBytes;
size_t localSizeBytes;
int maxThreadsPerBlock;
int numRegs;
int ptxVersion;
int binaryVersion;
int cacheModeCA;
int maxDynamicSharedSizeBytes;
int preferredShmemCarveout;
};
enum __device_builtin__ cudaFuncAttribute
{
cudaFuncAttributeMaxDynamicSharedMemorySize = 8,
cudaFuncAttributePreferredSharedMemoryCarveout = 9,
cudaFuncAttributeMax
};
enum __device_builtin__ cudaFuncCache
{
cudaFuncCachePreferNone = 0,
cudaFuncCachePreferShared = 1,
cudaFuncCachePreferL1 = 2,
cudaFuncCachePreferEqual = 3
};
enum __device_builtin__ cudaSharedMemConfig
{
cudaSharedMemBankSizeDefault = 0,
cudaSharedMemBankSizeFourByte = 1,
cudaSharedMemBankSizeEightByte = 2
};
enum __device_builtin__ cudaSharedCarveout {
cudaSharedmemCarveoutDefault = -1,
cudaSharedmemCarveoutMaxShared = 100,
cudaSharedmemCarveoutMaxL1 = 0
};
enum __device_builtin__ cudaComputeMode
{
cudaComputeModeDefault = 0,
cudaComputeModeExclusive = 1,
cudaComputeModeProhibited = 2,
cudaComputeModeExclusiveProcess = 3
};
enum __device_builtin__ cudaLimit
{
cudaLimitStackSize = 0x00,
cudaLimitPrintfFifoSize = 0x01,
cudaLimitMallocHeapSize = 0x02,
cudaLimitDevRuntimeSyncDepth = 0x03,
cudaLimitDevRuntimePendingLaunchCount = 0x04,
cudaLimitMaxL2FetchGranularity = 0x05
};
enum __device_builtin__ cudaMemoryAdvise
{
cudaMemAdviseSetReadMostly = 1,
cudaMemAdviseUnsetReadMostly = 2,
cudaMemAdviseSetPreferredLocation = 3,
cudaMemAdviseUnsetPreferredLocation = 4,
cudaMemAdviseSetAccessedBy = 5,
cudaMemAdviseUnsetAccessedBy = 6
};
enum __device_builtin__ cudaMemRangeAttribute
{
cudaMemRangeAttributeReadMostly = 1,
cudaMemRangeAttributePreferredLocation = 2,
cudaMemRangeAttributeAccessedBy = 3,
cudaMemRangeAttributeLastPrefetchLocation = 4
};
enum __device_builtin__ cudaOutputMode
{
cudaKeyValuePair = 0x00,
cudaCSV = 0x01
};
enum __device_builtin__ cudaDeviceAttr
{
cudaDevAttrMaxThreadsPerBlock = 1,
cudaDevAttrMaxBlockDimX = 2,
cudaDevAttrMaxBlockDimY = 3,
cudaDevAttrMaxBlockDimZ = 4,
cudaDevAttrMaxGridDimX = 5,
cudaDevAttrMaxGridDimY = 6,
cudaDevAttrMaxGridDimZ = 7,
cudaDevAttrMaxSharedMemoryPerBlock = 8,
cudaDevAttrTotalConstantMemory = 9,
cudaDevAttrWarpSize = 10,
cudaDevAttrMaxPitch = 11,
cudaDevAttrMaxRegistersPerBlock = 12,
cudaDevAttrClockRate = 13,
cudaDevAttrTextureAlignment = 14,
cudaDevAttrGpuOverlap = 15,
cudaDevAttrMultiProcessorCount = 16,
cudaDevAttrKernelExecTimeout = 17,
cudaDevAttrIntegrated = 18,
cudaDevAttrCanMapHostMemory = 19,
cudaDevAttrComputeMode = 20,
cudaDevAttrMaxTexture1DWidth = 21,
cudaDevAttrMaxTexture2DWidth = 22,
cudaDevAttrMaxTexture2DHeight = 23,
cudaDevAttrMaxTexture3DWidth = 24,
cudaDevAttrMaxTexture3DHeight = 25,
cudaDevAttrMaxTexture3DDepth = 26,
cudaDevAttrMaxTexture2DLayeredWidth = 27,
cudaDevAttrMaxTexture2DLayeredHeight = 28,
cudaDevAttrMaxTexture2DLayeredLayers = 29,
cudaDevAttrSurfaceAlignment = 30,
cudaDevAttrConcurrentKernels = 31,
cudaDevAttrEccEnabled = 32,
cudaDevAttrPciBusId = 33,
cudaDevAttrPciDeviceId = 34,
cudaDevAttrTccDriver = 35,
cudaDevAttrMemoryClockRate = 36,
cudaDevAttrGlobalMemoryBusWidth = 37,
cudaDevAttrL2CacheSize = 38,
cudaDevAttrMaxThreadsPerMultiProcessor = 39,
cudaDevAttrAsyncEngineCount = 40,
cudaDevAttrUnifiedAddressing = 41,
cudaDevAttrMaxTexture1DLayeredWidth = 42,
cudaDevAttrMaxTexture1DLayeredLayers = 43,
cudaDevAttrMaxTexture2DGatherWidth = 45,
cudaDevAttrMaxTexture2DGatherHeight = 46,
cudaDevAttrMaxTexture3DWidthAlt = 47,
cudaDevAttrMaxTexture3DHeightAlt = 48,
cudaDevAttrMaxTexture3DDepthAlt = 49,
cudaDevAttrPciDomainId = 50,
cudaDevAttrTexturePitchAlignment = 51,
cudaDevAttrMaxTextureCubemapWidth = 52,
cudaDevAttrMaxTextureCubemapLayeredWidth = 53,
cudaDevAttrMaxTextureCubemapLayeredLayers = 54,
cudaDevAttrMaxSurface1DWidth = 55,
cudaDevAttrMaxSurface2DWidth = 56,
cudaDevAttrMaxSurface2DHeight = 57,
cudaDevAttrMaxSurface3DWidth = 58,
cudaDevAttrMaxSurface3DHeight = 59,
cudaDevAttrMaxSurface3DDepth = 60,
cudaDevAttrMaxSurface1DLayeredWidth = 61,
cudaDevAttrMaxSurface1DLayeredLayers = 62,
cudaDevAttrMaxSurface2DLayeredWidth = 63,
cudaDevAttrMaxSurface2DLayeredHeight = 64,
cudaDevAttrMaxSurface2DLayeredLayers = 65,
cudaDevAttrMaxSurfaceCubemapWidth = 66,
cudaDevAttrMaxSurfaceCubemapLayeredWidth = 67,
cudaDevAttrMaxSurfaceCubemapLayeredLayers = 68,
cudaDevAttrMaxTexture1DLinearWidth = 69,
cudaDevAttrMaxTexture2DLinearWidth = 70,
cudaDevAttrMaxTexture2DLinearHeight = 71,
cudaDevAttrMaxTexture2DLinearPitch = 72,
cudaDevAttrMaxTexture2DMipmappedWidth = 73,
cudaDevAttrMaxTexture2DMipmappedHeight = 74,
cudaDevAttrComputeCapabilityMajor = 75,
cudaDevAttrComputeCapabilityMinor = 76,
cudaDevAttrMaxTexture1DMipmappedWidth = 77,
cudaDevAttrStreamPrioritiesSupported = 78,
cudaDevAttrGlobalL1CacheSupported = 79,
cudaDevAttrLocalL1CacheSupported = 80,
cudaDevAttrMaxSharedMemoryPerMultiprocessor = 81,
cudaDevAttrMaxRegistersPerMultiprocessor = 82,
cudaDevAttrManagedMemory = 83,
cudaDevAttrIsMultiGpuBoard = 84,
cudaDevAttrMultiGpuBoardGroupID = 85,
cudaDevAttrHostNativeAtomicSupported = 86,
cudaDevAttrSingleToDoublePrecisionPerfRatio = 87,
cudaDevAttrPageableMemoryAccess = 88,
cudaDevAttrConcurrentManagedAccess = 89,
cudaDevAttrComputePreemptionSupported = 90,
cudaDevAttrCanUseHostPointerForRegisteredMem = 91,
cudaDevAttrReserved92 = 92,
cudaDevAttrReserved93 = 93,
cudaDevAttrReserved94 = 94,
cudaDevAttrCooperativeLaunch = 95,
cudaDevAttrCooperativeMultiDeviceLaunch = 96,
cudaDevAttrMaxSharedMemoryPerBlockOptin = 97,
cudaDevAttrCanFlushRemoteWrites = 98,
cudaDevAttrHostRegisterSupported = 99,
cudaDevAttrPageableMemoryAccessUsesHostPageTables = 100,
cudaDevAttrDirectManagedMemAccessFromHost = 101
};
enum __device_builtin__ cudaDeviceP2PAttr {
cudaDevP2PAttrPerformanceRank = 1,
cudaDevP2PAttrAccessSupported = 2,
cudaDevP2PAttrNativeAtomicSupported = 3,
cudaDevP2PAttrCudaArrayAccessSupported = 4
};
#ifndef CU_UUID_HAS_BEEN_DEFINED
#define CU_UUID_HAS_BEEN_DEFINED
struct __device_builtin__ CUuuid_st {
char bytes[16];
};
typedef __device_builtin__ struct CUuuid_st CUuuid;
#endif
typedef __device_builtin__ struct CUuuid_st cudaUUID_t;
struct __device_builtin__ cudaDeviceProp
{
char name[256];
cudaUUID_t uuid;
char luid[8];
unsigned int luidDeviceNodeMask;
size_t totalGlobalMem;
size_t sharedMemPerBlock;
int regsPerBlock;
int warpSize;
size_t memPitch;
int maxThreadsPerBlock;
int maxThreadsDim[3];
int maxGridSize[3];
int clockRate;
size_t totalConstMem;
int major;
int minor;
size_t textureAlignment;
size_t texturePitchAlignment;
int deviceOverlap;
int multiProcessorCount;
int kernelExecTimeoutEnabled;
int integrated;
int canMapHostMemory;
int computeMode;
int maxTexture1D;
int maxTexture1DMipmap;
int maxTexture1DLinear;
int maxTexture2D[2];
int maxTexture2DMipmap[2];
int maxTexture2DLinear[3];
int maxTexture2DGather[2];
int maxTexture3D[3];
int maxTexture3DAlt[3];
int maxTextureCubemap;
int maxTexture1DLayered[2];
int maxTexture2DLayered[3];
int maxTextureCubemapLayered[2];
int maxSurface1D;
int maxSurface2D[2];
int maxSurface3D[3];
int maxSurface1DLayered[2];
int maxSurface2DLayered[3];
int maxSurfaceCubemap;
int maxSurfaceCubemapLayered[2];
size_t surfaceAlignment;
int concurrentKernels;
int ECCEnabled;
int pciBusID;
int pciDeviceID;
int pciDomainID;
int tccDriver;
int asyncEngineCount;
int unifiedAddressing;
int memoryClockRate;
int memoryBusWidth;
int l2CacheSize;
int maxThreadsPerMultiProcessor;
int streamPrioritiesSupported;
int globalL1CacheSupported;
int localL1CacheSupported;
size_t sharedMemPerMultiprocessor;
int regsPerMultiprocessor;
int managedMemory;
int isMultiGpuBoard;
int multiGpuBoardGroupID;
int hostNativeAtomicSupported;
int singleToDoublePrecisionPerfRatio;
int pageableMemoryAccess;
int concurrentManagedAccess;
int computePreemptionSupported;
int canUseHostPointerForRegisteredMem;
int cooperativeLaunch;
int cooperativeMultiDeviceLaunch;
size_t sharedMemPerBlockOptin;
int pageableMemoryAccessUsesHostPageTables;
int directManagedMemAccessFromHost;
};
#define cudaDevicePropDontCare \
{ \
{'\0'}, \
{{0}}, \
{'\0'}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
0, \
-1, \
-1, \
0, \
0, \
-1, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
{0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
0, \
{0, 0}, \
{0, 0, 0}, \
{0, 0}, \
{0, 0, 0}, \
0, \
{0, 0}, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
0, \
}
#define CUDA_IPC_HANDLE_SIZE 64
typedef __device_builtin__ struct __device_builtin__ cudaIpcEventHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcEventHandle_t;
typedef __device_builtin__ struct __device_builtin__ cudaIpcMemHandle_st
{
char reserved[CUDA_IPC_HANDLE_SIZE];
}cudaIpcMemHandle_t;
enum __device_builtin__ cudaExternalMemoryHandleType {
cudaExternalMemoryHandleTypeOpaqueFd = 1,
cudaExternalMemoryHandleTypeOpaqueWin32 = 2,
cudaExternalMemoryHandleTypeOpaqueWin32Kmt = 3,
cudaExternalMemoryHandleTypeD3D12Heap = 4,
cudaExternalMemoryHandleTypeD3D12Resource = 5,
cudaExternalMemoryHandleTypeD3D11Resource = 6,
cudaExternalMemoryHandleTypeD3D11ResourceKmt = 7,
cudaExternalMemoryHandleTypeNvSciBuf = 8
};
#define cudaExternalMemoryDedicated 0x1
#define cudaExternalSemaphoreSignalSkipNvSciBufMemSync 0x01
#define cudaExternalSemaphoreWaitSkipNvSciBufMemSync 0x02
#define cudaNvSciSyncAttrSignal 0x1
#define cudaNvSciSyncAttrWait 0x2
struct __device_builtin__ cudaExternalMemoryHandleDesc {
enum cudaExternalMemoryHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void *nvSciBufObject;
} handle;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryBufferDesc {
unsigned long long offset;
unsigned long long size;
unsigned int flags;
};
struct __device_builtin__ cudaExternalMemoryMipmappedArrayDesc {
unsigned long long offset;
struct cudaChannelFormatDesc formatDesc;
struct cudaExtent extent;
unsigned int flags;
unsigned int numLevels;
};
enum __device_builtin__ cudaExternalSemaphoreHandleType {
cudaExternalSemaphoreHandleTypeOpaqueFd = 1,
cudaExternalSemaphoreHandleTypeOpaqueWin32 = 2,
cudaExternalSemaphoreHandleTypeOpaqueWin32Kmt = 3,
cudaExternalSemaphoreHandleTypeD3D12Fence = 4,
cudaExternalSemaphoreHandleTypeD3D11Fence = 5,
cudaExternalSemaphoreHandleTypeNvSciSync = 6,
cudaExternalSemaphoreHandleTypeKeyedMutex = 7,
cudaExternalSemaphoreHandleTypeKeyedMutexKmt = 8
};
struct __device_builtin__ cudaExternalSemaphoreHandleDesc {
enum cudaExternalSemaphoreHandleType type;
union {
int fd;
struct {
void *handle;
const void *name;
} win32;
const void* nvSciSyncObj;
} handle;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreSignalParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
} keyedMutex;
} params;
unsigned int flags;
};
struct __device_builtin__ cudaExternalSemaphoreWaitParams {
struct {
struct {
unsigned long long value;
} fence;
union {
void *fence;
unsigned long long reserved;
} nvSciSync;
struct {
unsigned long long key;
unsigned int timeoutMs;
} keyedMutex;
} params;
unsigned int flags;
};
typedef __device_builtin__ enum cudaError cudaError_t;
typedef __device_builtin__ struct CUstream_st *cudaStream_t;
typedef __device_builtin__ struct CUevent_st *cudaEvent_t;
typedef __device_builtin__ struct cudaGraphicsResource *cudaGraphicsResource_t;
typedef __device_builtin__ enum cudaOutputMode cudaOutputMode_t;
typedef __device_builtin__ struct CUexternalMemory_st *cudaExternalMemory_t;
typedef __device_builtin__ struct CUexternalSemaphore_st *cudaExternalSemaphore_t;
typedef __device_builtin__ struct CUgraph_st *cudaGraph_t;
typedef __device_builtin__ struct CUgraphNode_st *cudaGraphNode_t;
enum __device_builtin__ cudaCGScope {
cudaCGScopeInvalid = 0,
cudaCGScopeGrid = 1,
cudaCGScopeMultiGrid = 2
};
struct __device_builtin__ cudaLaunchParams
{
void *func;
dim3 gridDim;
dim3 blockDim;
void **args;
size_t sharedMem;
cudaStream_t stream;
};
struct __device_builtin__ cudaKernelNodeParams {
void* func;
dim3 gridDim;
dim3 blockDim;
unsigned int sharedMemBytes;
void **kernelParams;
void **extra;
};
enum __device_builtin__ cudaGraphNodeType {
cudaGraphNodeTypeKernel = 0x00,
cudaGraphNodeTypeMemcpy = 0x01,
cudaGraphNodeTypeMemset = 0x02,
cudaGraphNodeTypeHost = 0x03,
cudaGraphNodeTypeGraph = 0x04,
cudaGraphNodeTypeEmpty = 0x05,
cudaGraphNodeTypeCount
};
typedef struct CUgraphExec_st* cudaGraphExec_t;
enum __device_builtin__ cudaGraphExecUpdateResult {
cudaGraphExecUpdateSuccess = 0x0,
cudaGraphExecUpdateError = 0x1,
cudaGraphExecUpdateErrorTopologyChanged = 0x2,
cudaGraphExecUpdateErrorNodeTypeChanged = 0x3,
cudaGraphExecUpdateErrorFunctionChanged = 0x4,
cudaGraphExecUpdateErrorParametersChanged = 0x5,
cudaGraphExecUpdateErrorNotSupported = 0x6
};
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_DRIVER_TYPES_H__
#endif
#undef __CUDA_DEPRECATED
#endif
#if !defined(__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__)
#if defined(_MSC_VER)
#pragma message("crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead.")
#else
#warning "crt/host_defines.h is an internal header file and must not be used directly. Please use cuda_runtime_api.h or cuda_runtime.h instead."
#endif
#define __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#define __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
#if !defined(__HOST_DEFINES_H__)
#define __HOST_DEFINES_H__
#if defined(__GNUC__) || (defined(__PGIC__) && defined(__linux__)) || defined(__CUDA_LIBDEVICE__) || defined(__CUDACC_RTC__)
#if defined(__CUDACC_RTC__)
#define __volatile__ volatile
#endif
#define __no_return__ \
__attribute__((noreturn))
#if defined(__CUDACC__) || defined(__CUDA_ARCH__) || defined(__CUDA_LIBDEVICE__)
#define __noinline__ \
__attribute__((noinline))
#endif
#define __forceinline__ \
__inline__ __attribute__((always_inline))
#define __align__(n) \
__attribute__((aligned(n)))
#define __thread__ \
__thread
#define __import__
#define __export__
#define __cdecl
#define __annotate__(a) \
__attribute__((a))
#define __location__(a) \
__annotate__(a)
#define CUDARTAPI
#elif defined(_MSC_VER)
#if _MSC_VER >= 1400
#define __restrict__ \
__restrict
#else
#define __restrict__
#endif
#define __inline__ \
__inline
#define __no_return__ \
__declspec(noreturn)
#define __noinline__ \
__declspec(noinline)
#define __forceinline__ \
__forceinline
#define __align__(n) \
__declspec(align(n))
#define __thread__ \
__declspec(thread)
#define __import__ \
__declspec(dllimport)
#define __export__ \
__declspec(dllexport)
#define __annotate__(a) \
__declspec(a)
#define __location__(a) \
__annotate__(__##a##__)
#define CUDARTAPI \
__stdcall
#else
#define __inline__
#if !defined(__align__)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for '__align__' !!! ---
#endif
#if !defined(CUDARTAPI)
#error --- !!! UNKNOWN COMPILER: please provide a CUDA compatible definition for 'CUDARTAPI' !!! ---
#endif
#endif
#if (defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 3 && !defined(__clang__)))) || \
(defined(_MSC_VER) && _MSC_VER < 1900) || \
(!defined(__GNUC__) && !defined(_MSC_VER))
#define __specialization_static \
static
#else
#define __specialization_static
#endif
#if !defined(__CUDACC__) && !defined(__CUDA_LIBDEVICE__)
#undef __annotate__
#define __annotate__(a)
#else
#define __launch_bounds__(...) \
__annotate__(launch_bounds(__VA_ARGS__))
#endif
#if defined(__CUDACC__) || defined(__CUDA_LIBDEVICE__) || \
defined(__GNUC__) || defined(_WIN64)
#define __builtin_align__(a) \
__align__(a)
#else
#define __builtin_align__(a)
#endif
#if defined(__CUDACC__) || !defined(__host__)
#define __host__ \
__location__(host)
#endif
#if defined(__CUDACC__) || !defined(__device__)
#define __device__ \
__location__(device)
#endif
#if defined(__CUDACC__) || !defined(__global__)
#define __global__ \
__location__(global)
#endif
#if defined(__CUDACC__) || !defined(__shared__)
#define __shared__ \
__location__(shared)
#endif
#if defined(__CUDACC__) || !defined(__constant__)
#define __constant__ \
__location__(constant)
#endif
#if defined(__CUDACC__) || !defined(__managed__)
#define __managed__ \
__location__(managed)
#endif
#if !defined(__CUDACC__)
#define __device_builtin__
#define __device_builtin_texture_type__
#define __device_builtin_surface_type__
#define __cudart_builtin__
#else
#define __device_builtin__ \
__location__(device_builtin)
#define __device_builtin_texture_type__ \
__location__(device_builtin_texture_type)
#define __device_builtin_surface_type__ \
__location__(device_builtin_surface_type)
#define __cudart_builtin__ \
__location__(cudart_builtin)
#endif
#endif
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_HOST_DEFINES_H__
#endif
extern "C"
{
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetAttribute(int *value, enum cudaDeviceAttr attr, int device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetLimit(size_t *pValue, enum cudaLimit limit);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetCacheConfig(enum cudaFuncCache *pCacheConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceGetSharedMemConfig(enum cudaSharedMemConfig *pConfig);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaDeviceSynchronize(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetLastError(void);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaPeekAtLastError(void);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorString(cudaError_t error);
extern __device__ __cudart_builtin__ const char* CUDARTAPI cudaGetErrorName(cudaError_t error);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDeviceCount(int *count);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaGetDevice(int *device);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamCreateWithFlags(cudaStream_t *pStream, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamDestroy(cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaStreamWaitEvent_ptsz(cudaStream_t stream, cudaEvent_t event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventCreateWithFlags(cudaEvent_t *event, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventRecord_ptsz(cudaEvent_t event, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaEventDestroy(cudaEvent_t event);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFuncGetAttributes(struct cudaFuncAttributes *attr, const void *func);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaFree(void *devPtr);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMalloc(void **devPtr, size_t size);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpyAsync_ptsz(void *dst, const void *src, size_t count, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy2DAsync_ptsz(void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t height, enum cudaMemcpyKind kind, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemcpy3DAsync_ptsz(const struct cudaMemcpy3DParms *p, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemsetAsync_ptsz(void *devPtr, int value, size_t count, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset2DAsync_ptsz(void *devPtr, size_t pitch, int value, size_t width, size_t height, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaMemset3DAsync_ptsz(struct cudaPitchedPtr pitchedDevPtr, int value, struct cudaExtent extent, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaRuntimeGetVersion(int *runtimeVersion);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBuffer(size_t alignment, size_t size);
extern __device__ __cudart_builtin__ void * CUDARTAPI cudaGetParameterBufferV2(void *func, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice_ptsz(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2_ptsz(void *parameterBuffer, cudaStream_t stream);
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) && defined(__CUDA_ARCH__)
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream)
{
return cudaLaunchDevice_ptsz(func, parameterBuffer, gridDimension, blockDimension, sharedMemSize, stream);
}
static __inline__ __device__ __cudart_builtin__ cudaError_t CUDARTAPI
cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream)
{
return cudaLaunchDeviceV2_ptsz(parameterBuffer, stream);
}
#else
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDevice(void *func, void *parameterBuffer, dim3 gridDimension, dim3 blockDimension, unsigned int sharedMemSize, cudaStream_t stream);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaLaunchDeviceV2(void *parameterBuffer, cudaStream_t stream);
#endif
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, const void *func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
extern __device__ __cudart_builtin__ unsigned long long CUDARTAPI cudaCGGetIntrinsicHandle(enum cudaCGScope scope);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronize(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGSynchronizeGrid(unsigned long long handle, unsigned int flags);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetSize(unsigned int *numThreads, unsigned int *numGrids, unsigned long long handle);
extern __device__ __cudart_builtin__ cudaError_t CUDARTAPI cudaCGGetRank(unsigned int *threadRank, unsigned int *gridRank, unsigned long long handle);
}
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaMalloc(T **devPtr, size_t size);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaFuncGetAttributes(struct cudaFuncAttributes *attr, T *entry);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessor(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize);
template static __inline__ __device__ __cudart_builtin__ cudaError_t cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(int *numBlocks, T func, int blockSize, size_t dynamicSmemSize, unsigned int flags);
#endif
#endif
#endif
#if defined(CUDA_API_PER_THREAD_DEFAULT_STREAM) || defined(__CUDA_API_VERSION_INTERNAL)
#define __CUDART_API_PER_THREAD_DEFAULT_STREAM
#define __CUDART_API_PTDS(api) api ## _ptds
#define __CUDART_API_PTSZ(api) api ## _ptsz
#else
#define __CUDART_API_PTDS(api) api
#define __CUDART_API_PTSZ(api) api
#endif
#if defined(__CUDART_API_PER_THREAD_DEFAULT_STREAM)
#define cudaMemcpy __CUDART_API_PTDS(cudaMemcpy)
#define cudaMemcpyToSymbol __CUDART_API_PTDS(cudaMemcpyToSymbol)
#define cudaMemcpyFromSymbol __CUDART_API_PTDS(cudaMemcpyFromSymbol)
#define cudaMemcpy2D __CUDART_API_PTDS(cudaMemcpy2D)
#define cudaMemcpyToArray __CUDART_API_PTDS(cudaMemcpyToArray)
#define cudaMemcpy2DToArray __CUDART_API_PTDS(cudaMemcpy2DToArray)
#define cudaMemcpyFromArray __CUDART_API_PTDS(cudaMemcpyFromArray)
#define cudaMemcpy2DFromArray __CUDART_API_PTDS(cudaMemcpy2DFromArray)
#define cudaMemcpyArrayToArray __CUDART_API_PTDS(cudaMemcpyArrayToArray)
#define cudaMemcpy2DArrayToArray __CUDART_API_PTDS(cudaMemcpy2DArrayToArray)
#define cudaMemcpy3D __CUDART_API_PTDS(cudaMemcpy3D)
#define cudaMemcpy3DPeer __CUDART_API_PTDS(cudaMemcpy3DPeer)
#define cudaMemset __CUDART_API_PTDS(cudaMemset)
#define cudaMemset2D __CUDART_API_PTDS(cudaMemset2D)
#define cudaMemset3D __CUDART_API_PTDS(cudaMemset3D)
#define cudaGraphLaunch __CUDART_API_PTSZ(cudaGraphLaunch)
#define cudaStreamBeginCapture __CUDART_API_PTSZ(cudaStreamBeginCapture)
#define cudaStreamEndCapture __CUDART_API_PTSZ(cudaStreamEndCapture)
#define cudaStreamIsCapturing __CUDART_API_PTSZ(cudaStreamIsCapturing)
#define cudaStreamGetCaptureInfo __CUDART_API_PTSZ(cudaStreamGetCaptureInfo)
#define cudaMemcpyAsync __CUDART_API_PTSZ(cudaMemcpyAsync)
#define cudaMemcpyToSymbolAsync __CUDART_API_PTSZ(cudaMemcpyToSymbolAsync)
#define cudaMemcpyFromSymbolAsync __CUDART_API_PTSZ(cudaMemcpyFromSymbolAsync)
#define cudaMemcpy2DAsync __CUDART_API_PTSZ(cudaMemcpy2DAsync)
#define cudaMemcpyToArrayAsync __CUDART_API_PTSZ(cudaMemcpyToArrayAsync)
#define cudaMemcpy2DToArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DToArrayAsync)
#define cudaMemcpyFromArrayAsync __CUDART_API_PTSZ(cudaMemcpyFromArrayAsync)
#define cudaMemcpy2DFromArrayAsync __CUDART_API_PTSZ(cudaMemcpy2DFromArrayAsync)
#define cudaMemcpy3DAsync __CUDART_API_PTSZ(cudaMemcpy3DAsync)
#define cudaMemcpy3DPeerAsync __CUDART_API_PTSZ(cudaMemcpy3DPeerAsync)
#define cudaMemsetAsync __CUDART_API_PTSZ(cudaMemsetAsync)
#define cudaMemset2DAsync __CUDART_API_PTSZ(cudaMemset2DAsync)
#define cudaMemset3DAsync __CUDART_API_PTSZ(cudaMemset3DAsync)
#define cudaStreamQuery __CUDART_API_PTSZ(cudaStreamQuery)
#define cudaStreamGetFlags __CUDART_API_PTSZ(cudaStreamGetFlags)
#define cudaStreamGetPriority __CUDART_API_PTSZ(cudaStreamGetPriority)
#define cudaEventRecord __CUDART_API_PTSZ(cudaEventRecord)
#define cudaStreamWaitEvent __CUDART_API_PTSZ(cudaStreamWaitEvent)
#define cudaStreamAddCallback __CUDART_API_PTSZ(cudaStreamAddCallback)
#define cudaStreamAttachMemAsync __CUDART_API_PTSZ(cudaStreamAttachMemAsync)
#define cudaStreamSynchronize __CUDART_API_PTSZ(cudaStreamSynchronize)
#define cudaLaunchKernel __CUDART_API_PTSZ(cudaLaunchKernel)
#define cudaLaunchHostFunc __CUDART_API_PTSZ(cudaLaunchHostFunc)
#define cudaMemPrefetchAsync __CUDART_API_PTSZ(cudaMemPrefetchAsync)
#define cudaLaunchCooperativeKernel __CUDART_API_PTSZ(cudaLaunchCooperativeKernel)
#define cudaSignalExternalSemaphoresAsync __CUDART_API_PTSZ(cudaSignalExternalSemaphoresAsync)
#define cudaWaitExternalSemaphoresAsync __CUDART_API_PTSZ(cudaWaitExternalSemaphoresAsync)
#endif
#if !defined(__dv)
#if defined(__cplusplus)
#define __dv(v) \
= v
#else
#define __dv(v)
#endif
#endif
#if (!defined(__CUDA_ARCH__) || (__CUDA_ARCH__ >= 350))
#define CUDART_DEVICE __device__
#else
#define CUDART_DEVICE
#endif
#if !defined(__CUDACC_RTC__)
#endif
#undef __dv
#undef __CUDA_DEPRECATED
#if defined(__UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__)
#undef __CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__
#undef __UNDEF_CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS_CUDA_RUNTIME_API_H__
#endif
#endif
template
struct __device_builtin_surface_type__ surface : public surfaceReference
{
#if !defined(__CUDACC_RTC__)
__host__ surface(void)
{
channelDesc = cudaCreateChannelDesc();
}
__host__ surface(struct cudaChannelFormatDesc desc)
{
channelDesc = desc;
}
#endif
};
template
struct __device_builtin_surface_type__ surface : public surfaceReference
{
#if !defined(__CUDACC_RTC__)
__host__ surface(void)
{
channelDesc = cudaCreateChannelDesc();
}
#endif
};
#endif
#endif
#ifdef __CUDA_ARCH__
template struct __nv_surf_trait { typedef void * cast_type; };
template<> struct __nv_surf_trait { typedef char * cast_type; };
template<> struct __nv_surf_trait { typedef signed char * cast_type; };
template<> struct __nv_surf_trait { typedef unsigned char * cast_type; };
template<> struct __nv_surf_trait { typedef char1 * cast_type; };
template<> struct __nv_surf_trait { typedef uchar1 * cast_type; };
template<> struct __nv_surf_trait { typedef char2 * cast_type; };
template<> struct __nv_surf_trait { typedef uchar2 * cast_type; };
template<> struct __nv_surf_trait { typedef char4 * cast_type; };
template<> struct __nv_surf_trait { typedef uchar4 * cast_type; };
template<> struct __nv_surf_trait { typedef short * cast_type; };
template<> struct __nv_surf_trait { typedef unsigned short * cast_type; };
template<> struct __nv_surf_trait { typedef short1 * cast_type; };
template<> struct __nv_surf_trait { typedef ushort1 * cast_type; };
template<> struct __nv_surf_trait { typedef short2 * cast_type; };
template<> struct __nv_surf_trait { typedef ushort2 * cast_type; };
template<> struct __nv_surf_trait { typedef short4 * cast_type; };
template<> struct __nv_surf_trait { typedef ushort4 * cast_type; };
template<> struct __nv_surf_trait { typedef int * cast_type; };
template<> struct __nv_surf_trait { typedef unsigned int * cast_type; };
template<> struct __nv_surf_trait { typedef int1 * cast_type; };
template<> struct __nv_surf_trait { typedef uint1 * cast_type; };
template<> struct __nv_surf_trait { typedef int2 * cast_type; };
template<> struct __nv_surf_trait { typedef uint2 * cast_type; };
template<> struct __nv_surf_trait { typedef int4 * cast_type; };
template<> struct __nv_surf_trait { typedef uint4 * cast_type; };
template<> struct __nv_surf_trait { typedef long long * cast_type; };
template<> struct __nv_surf_trait { typedef unsigned long long * cast_type; };
template<> struct __nv_surf_trait { typedef longlong1 * cast_type; };
template<> struct __nv_surf_trait { typedef ulonglong1 * cast_type; };
template<> struct __nv_surf_trait { typedef longlong2 * cast_type; };
template<> struct __nv_surf_trait { typedef ulonglong2 * cast_type; };
#if !defined(__LP64__)
template<> struct __nv_surf_trait { typedef int * cast_type; };
template<> struct __nv_surf_trait { typedef unsigned int * cast_type; };
template<> struct __nv_surf_trait { typedef int1 * cast_type; };
template<> struct __nv_surf_trait { typedef uint1 * cast_type; };
template<> struct __nv_surf_trait { typedef int2 * cast_type; };
template<> struct __nv_surf_trait { typedef uint2 * cast_type; };
template<> struct __nv_surf_trait { typedef uint4 * cast_type; };
template<> struct __nv_surf_trait { typedef int4 * cast_type; };
#endif
template<> struct __nv_surf_trait { typedef float * cast_type; };
template<> struct __nv_surf_trait { typedef float1 * cast_type; };
template<> struct __nv_surf_trait { typedef float2 * cast_type; };
template<> struct __nv_surf_trait { typedef float4 * cast_type; };
#endif
template
static __device__ __forceinline__ void surf1Dread(T *res, surface surf, int x, int s, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
__nv_tex_surf_handler("__surf1Dread_v2", (void *)res, s, surf, x, mode);
#endif
}
template
static __device__ __forceinline__ T surf1Dread(surface surf, int x, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
T temp;
__nv_tex_surf_handler("__surf1Dread_v2", (typename __nv_surf_trait::cast_type)&temp, (int)sizeof(T), surf, x, mode);
return temp;
#endif
}
template
static __device__ __forceinline__ void surf1Dread(T *res, surface surf, int x, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
*res = surf1Dread(surf, x, mode);
#endif
}
template
static __device__ __forceinline__ void surf2Dread(T *res, surface surf, int x, int y, int s, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
__nv_tex_surf_handler("__surf2Dread_v2", (void *)res, s, surf, x, y, mode);
#endif
}
template
static __device__ __forceinline__ T surf2Dread(surface surf, int x, int y, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
T temp;
__nv_tex_surf_handler("__surf2Dread_v2", (typename __nv_surf_trait::cast_type)&temp, (int)sizeof(T), surf, x, y, mode);
return temp;
#endif
}
template
static __device__ __forceinline__ void surf2Dread(T *res, surface surf, int x, int y, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
*res = surf2Dread(surf, x, y, mode);
#endif
}
template
static __device__ __forceinline__ void surf3Dread(T *res, surface surf, int x, int y, int z, int s, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
__nv_tex_surf_handler("__surf3Dread_v2", (void *)res, s, surf, x, y, z, mode);
#endif
}
template
static __device__ __forceinline__ T surf3Dread(surface surf, int x, int y, int z, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
T temp;
__nv_tex_surf_handler("__surf3Dread_v2", (typename __nv_surf_trait::cast_type)&temp, (int)sizeof(T), surf, x, y, z, mode);
return temp;
#endif
}
template
static __device__ __forceinline__ void surf3Dread(T *res, surface surf, int x, int y, int z, enum cudaSurfaceBoundaryMode mode = cudaBoundaryModeTrap)
{
#ifdef __CUDA_ARCH__
*res = surf3Dread(surf, x, y, z, mode);
#endif
}
template