khronos-opencl-headers-1.2.orig/0000755000000000000500000000000012236503327015177 5ustar rootsrckhronos-opencl-headers-1.2.orig/opencl.h0000644000000000000500000000333212214174566016636 0ustar rootsrc/******************************************************************************* * Copyright (c) 2008-2012 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. ******************************************************************************/ /* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ #ifndef __OPENCL_H #define __OPENCL_H #ifdef __cplusplus extern "C" { #endif #ifdef __APPLE__ #include #include #include #include #else #include #include #include #include #endif #ifdef __cplusplus } #endif #endif /* __OPENCL_H */ khronos-opencl-headers-1.2.orig/cl_platform.h0000644000000000000500000012024712214174566017665 0ustar rootsrc/********************************************************************************** * Copyright (c) 2008-2012 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. **********************************************************************************/ /* $Revision: 11803 $ on $Date: 2010-06-25 10:02:12 -0700 (Fri, 25 Jun 2010) $ */ #ifndef __CL_PLATFORM_H #define __CL_PLATFORM_H #ifdef __APPLE__ /* Contains #defines for AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER below */ #include #endif #ifdef __cplusplus extern "C" { #endif #if defined(_WIN32) #define CL_API_ENTRY #define CL_API_CALL __stdcall #define CL_CALLBACK __stdcall #else #define CL_API_ENTRY #define CL_API_CALL #define CL_CALLBACK #endif #ifdef __APPLE__ #define CL_EXTENSION_WEAK_LINK __attribute__((weak_import)) #define CL_API_SUFFIX__VERSION_1_0 AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER #define CL_EXT_SUFFIX__VERSION_1_0 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER #define CL_API_SUFFIX__VERSION_1_1 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER #define GCL_API_SUFFIX__VERSION_1_1 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER #define CL_EXT_SUFFIX__VERSION_1_1 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER_BUT_DEPRECATED_IN_MAC_OS_X_VERSION_10_7 #ifdef AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER #define CL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER #define GCL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER #define CL_EXT_SUFFIX__VERSION_1_2 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_8_AND_LATER #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER_BUT_DEPRECATED_IN_MAC_OS_X_VERSION_10_8 #else #warning This path should never happen outside of internal operating system development. AvailabilityMacros do not function correctly here! #define CL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER #define GCL_API_SUFFIX__VERSION_1_2 AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER #define CL_EXT_SUFFIX__VERSION_1_2 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_7_AND_LATER #endif #else #define CL_EXTENSION_WEAK_LINK #define CL_API_SUFFIX__VERSION_1_0 #define CL_EXT_SUFFIX__VERSION_1_0 #define CL_API_SUFFIX__VERSION_1_1 #define CL_EXT_SUFFIX__VERSION_1_1 #define CL_API_SUFFIX__VERSION_1_2 #define CL_EXT_SUFFIX__VERSION_1_2 #ifdef __GNUC__ #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED #else #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED __attribute__((deprecated)) #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED #endif #ifdef CL_USE_DEPRECATED_OPENCL_1_1_APIS #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED #else #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED __attribute__((deprecated)) #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED #endif #elif _WIN32 #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED #else #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED __declspec(deprecated) #endif #ifdef CL_USE_DEPRECATED_OPENCL_1_1_APIS #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED #else #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED __declspec(deprecated) #endif #else #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED #define CL_EXT_PREFIX__VERSION_1_0_DEPRECATED #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED #endif #endif #if (defined (_WIN32) && defined(_MSC_VER)) /* scalar types */ typedef signed __int8 cl_char; typedef unsigned __int8 cl_uchar; typedef signed __int16 cl_short; typedef unsigned __int16 cl_ushort; typedef signed __int32 cl_int; typedef unsigned __int32 cl_uint; typedef signed __int64 cl_long; typedef unsigned __int64 cl_ulong; typedef unsigned __int16 cl_half; typedef float cl_float; typedef double cl_double; /* Macro names and corresponding values defined by OpenCL */ #define CL_CHAR_BIT 8 #define CL_SCHAR_MAX 127 #define CL_SCHAR_MIN (-127-1) #define CL_CHAR_MAX CL_SCHAR_MAX #define CL_CHAR_MIN CL_SCHAR_MIN #define CL_UCHAR_MAX 255 #define CL_SHRT_MAX 32767 #define CL_SHRT_MIN (-32767-1) #define CL_USHRT_MAX 65535 #define CL_INT_MAX 2147483647 #define CL_INT_MIN (-2147483647-1) #define CL_UINT_MAX 0xffffffffU #define CL_LONG_MAX ((cl_long) 0x7FFFFFFFFFFFFFFFLL) #define CL_LONG_MIN ((cl_long) -0x7FFFFFFFFFFFFFFFLL - 1LL) #define CL_ULONG_MAX ((cl_ulong) 0xFFFFFFFFFFFFFFFFULL) #define CL_FLT_DIG 6 #define CL_FLT_MANT_DIG 24 #define CL_FLT_MAX_10_EXP +38 #define CL_FLT_MAX_EXP +128 #define CL_FLT_MIN_10_EXP -37 #define CL_FLT_MIN_EXP -125 #define CL_FLT_RADIX 2 #define CL_FLT_MAX 340282346638528859811704183484516925440.0f #define CL_FLT_MIN 1.175494350822287507969e-38f #define CL_FLT_EPSILON 0x1.0p-23f #define CL_DBL_DIG 15 #define CL_DBL_MANT_DIG 53 #define CL_DBL_MAX_10_EXP +308 #define CL_DBL_MAX_EXP +1024 #define CL_DBL_MIN_10_EXP -307 #define CL_DBL_MIN_EXP -1021 #define CL_DBL_RADIX 2 #define CL_DBL_MAX 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.0 #define CL_DBL_MIN 2.225073858507201383090e-308 #define CL_DBL_EPSILON 2.220446049250313080847e-16 #define CL_M_E 2.718281828459045090796 #define CL_M_LOG2E 1.442695040888963387005 #define CL_M_LOG10E 0.434294481903251816668 #define CL_M_LN2 0.693147180559945286227 #define CL_M_LN10 2.302585092994045901094 #define CL_M_PI 3.141592653589793115998 #define CL_M_PI_2 1.570796326794896557999 #define CL_M_PI_4 0.785398163397448278999 #define CL_M_1_PI 0.318309886183790691216 #define CL_M_2_PI 0.636619772367581382433 #define CL_M_2_SQRTPI 1.128379167095512558561 #define CL_M_SQRT2 1.414213562373095145475 #define CL_M_SQRT1_2 0.707106781186547572737 #define CL_M_E_F 2.71828174591064f #define CL_M_LOG2E_F 1.44269502162933f #define CL_M_LOG10E_F 0.43429449200630f #define CL_M_LN2_F 0.69314718246460f #define CL_M_LN10_F 2.30258512496948f #define CL_M_PI_F 3.14159274101257f #define CL_M_PI_2_F 1.57079637050629f #define CL_M_PI_4_F 0.78539818525314f #define CL_M_1_PI_F 0.31830987334251f #define CL_M_2_PI_F 0.63661974668503f #define CL_M_2_SQRTPI_F 1.12837922573090f #define CL_M_SQRT2_F 1.41421353816986f #define CL_M_SQRT1_2_F 0.70710676908493f #define CL_NAN (CL_INFINITY - CL_INFINITY) #define CL_HUGE_VALF ((cl_float) 1e50) #define CL_HUGE_VAL ((cl_double) 1e500) #define CL_MAXFLOAT CL_FLT_MAX #define CL_INFINITY CL_HUGE_VALF #else #include /* scalar types */ typedef int8_t cl_char; typedef uint8_t cl_uchar; typedef int16_t cl_short __attribute__((aligned(2))); typedef uint16_t cl_ushort __attribute__((aligned(2))); typedef int32_t cl_int __attribute__((aligned(4))); typedef uint32_t cl_uint __attribute__((aligned(4))); typedef int64_t cl_long __attribute__((aligned(8))); typedef uint64_t cl_ulong __attribute__((aligned(8))); typedef uint16_t cl_half __attribute__((aligned(2))); typedef float cl_float __attribute__((aligned(4))); typedef double cl_double __attribute__((aligned(8))); /* Macro names and corresponding values defined by OpenCL */ #define CL_CHAR_BIT 8 #define CL_SCHAR_MAX 127 #define CL_SCHAR_MIN (-127-1) #define CL_CHAR_MAX CL_SCHAR_MAX #define CL_CHAR_MIN CL_SCHAR_MIN #define CL_UCHAR_MAX 255 #define CL_SHRT_MAX 32767 #define CL_SHRT_MIN (-32767-1) #define CL_USHRT_MAX 65535 #define CL_INT_MAX 2147483647 #define CL_INT_MIN (-2147483647-1) #define CL_UINT_MAX 0xffffffffU #define CL_LONG_MAX ((cl_long) 0x7FFFFFFFFFFFFFFFLL) #define CL_LONG_MIN ((cl_long) -0x7FFFFFFFFFFFFFFFLL - 1LL) #define CL_ULONG_MAX ((cl_ulong) 0xFFFFFFFFFFFFFFFFULL) #define CL_FLT_DIG 6 #define CL_FLT_MANT_DIG 24 #define CL_FLT_MAX_10_EXP +38 #define CL_FLT_MAX_EXP +128 #define CL_FLT_MIN_10_EXP -37 #define CL_FLT_MIN_EXP -125 #define CL_FLT_RADIX 2 #define CL_FLT_MAX 0x1.fffffep127f #define CL_FLT_MIN 0x1.0p-126f #define CL_FLT_EPSILON 0x1.0p-23f #define CL_DBL_DIG 15 #define CL_DBL_MANT_DIG 53 #define CL_DBL_MAX_10_EXP +308 #define CL_DBL_MAX_EXP +1024 #define CL_DBL_MIN_10_EXP -307 #define CL_DBL_MIN_EXP -1021 #define CL_DBL_RADIX 2 #define CL_DBL_MAX 0x1.fffffffffffffp1023 #define CL_DBL_MIN 0x1.0p-1022 #define CL_DBL_EPSILON 0x1.0p-52 #define CL_M_E 2.718281828459045090796 #define CL_M_LOG2E 1.442695040888963387005 #define CL_M_LOG10E 0.434294481903251816668 #define CL_M_LN2 0.693147180559945286227 #define CL_M_LN10 2.302585092994045901094 #define CL_M_PI 3.141592653589793115998 #define CL_M_PI_2 1.570796326794896557999 #define CL_M_PI_4 0.785398163397448278999 #define CL_M_1_PI 0.318309886183790691216 #define CL_M_2_PI 0.636619772367581382433 #define CL_M_2_SQRTPI 1.128379167095512558561 #define CL_M_SQRT2 1.414213562373095145475 #define CL_M_SQRT1_2 0.707106781186547572737 #define CL_M_E_F 2.71828174591064f #define CL_M_LOG2E_F 1.44269502162933f #define CL_M_LOG10E_F 0.43429449200630f #define CL_M_LN2_F 0.69314718246460f #define CL_M_LN10_F 2.30258512496948f #define CL_M_PI_F 3.14159274101257f #define CL_M_PI_2_F 1.57079637050629f #define CL_M_PI_4_F 0.78539818525314f #define CL_M_1_PI_F 0.31830987334251f #define CL_M_2_PI_F 0.63661974668503f #define CL_M_2_SQRTPI_F 1.12837922573090f #define CL_M_SQRT2_F 1.41421353816986f #define CL_M_SQRT1_2_F 0.70710676908493f #if defined( __GNUC__ ) #define CL_HUGE_VALF __builtin_huge_valf() #define CL_HUGE_VAL __builtin_huge_val() #define CL_NAN __builtin_nanf( "" ) #else #define CL_HUGE_VALF ((cl_float) 1e50) #define CL_HUGE_VAL ((cl_double) 1e500) float nanf( const char * ); #define CL_NAN nanf( "" ) #endif #define CL_MAXFLOAT CL_FLT_MAX #define CL_INFINITY CL_HUGE_VALF #endif #include /* Mirror types to GL types. Mirror types allow us to avoid deciding which 87s to load based on whether we are using GL or GLES here. */ typedef unsigned int cl_GLuint; typedef int cl_GLint; typedef unsigned int cl_GLenum; /* * Vector types * * Note: OpenCL requires that all types be naturally aligned. * This means that vector types must be naturally aligned. * For example, a vector of four floats must be aligned to * a 16 byte boundary (calculated as 4 * the natural 4-byte * alignment of the float). The alignment qualifiers here * will only function properly if your compiler supports them * and if you don't actively work to defeat them. For example, * in order for a cl_float4 to be 16 byte aligned in a struct, * the start of the struct must itself be 16-byte aligned. * * Maintaining proper alignment is the user's responsibility. */ /* Define basic vector types */ #if defined( __VEC__ ) #include /* may be omitted depending on compiler. AltiVec spec provides no way to detect whether the header is required. */ typedef vector unsigned char __cl_uchar16; typedef vector signed char __cl_char16; typedef vector unsigned short __cl_ushort8; typedef vector signed short __cl_short8; typedef vector unsigned int __cl_uint4; typedef vector signed int __cl_int4; typedef vector float __cl_float4; #define __CL_UCHAR16__ 1 #define __CL_CHAR16__ 1 #define __CL_USHORT8__ 1 #define __CL_SHORT8__ 1 #define __CL_UINT4__ 1 #define __CL_INT4__ 1 #define __CL_FLOAT4__ 1 #endif #if defined( __SSE__ ) #if defined( __MINGW64__ ) #include #else #include #endif #if defined( __GNUC__ ) typedef float __cl_float4 __attribute__((vector_size(16))); #else typedef __m128 __cl_float4; #endif #define __CL_FLOAT4__ 1 #endif #if defined( __SSE2__ ) #if defined( __MINGW64__ ) #include #else #include #endif #if defined( __GNUC__ ) typedef cl_uchar __cl_uchar16 __attribute__((vector_size(16))); typedef cl_char __cl_char16 __attribute__((vector_size(16))); typedef cl_ushort __cl_ushort8 __attribute__((vector_size(16))); typedef cl_short __cl_short8 __attribute__((vector_size(16))); typedef cl_uint __cl_uint4 __attribute__((vector_size(16))); typedef cl_int __cl_int4 __attribute__((vector_size(16))); typedef cl_ulong __cl_ulong2 __attribute__((vector_size(16))); typedef cl_long __cl_long2 __attribute__((vector_size(16))); typedef cl_double __cl_double2 __attribute__((vector_size(16))); #else typedef __m128i __cl_uchar16; typedef __m128i __cl_char16; typedef __m128i __cl_ushort8; typedef __m128i __cl_short8; typedef __m128i __cl_uint4; typedef __m128i __cl_int4; typedef __m128i __cl_ulong2; typedef __m128i __cl_long2; typedef __m128d __cl_double2; #endif #define __CL_UCHAR16__ 1 #define __CL_CHAR16__ 1 #define __CL_USHORT8__ 1 #define __CL_SHORT8__ 1 #define __CL_INT4__ 1 #define __CL_UINT4__ 1 #define __CL_ULONG2__ 1 #define __CL_LONG2__ 1 #define __CL_DOUBLE2__ 1 #endif #if defined( __MMX__ ) #include #if defined( __GNUC__ ) typedef cl_uchar __cl_uchar8 __attribute__((vector_size(8))); typedef cl_char __cl_char8 __attribute__((vector_size(8))); typedef cl_ushort __cl_ushort4 __attribute__((vector_size(8))); typedef cl_short __cl_short4 __attribute__((vector_size(8))); typedef cl_uint __cl_uint2 __attribute__((vector_size(8))); typedef cl_int __cl_int2 __attribute__((vector_size(8))); typedef cl_ulong __cl_ulong1 __attribute__((vector_size(8))); typedef cl_long __cl_long1 __attribute__((vector_size(8))); typedef cl_float __cl_float2 __attribute__((vector_size(8))); #else typedef __m64 __cl_uchar8; typedef __m64 __cl_char8; typedef __m64 __cl_ushort4; typedef __m64 __cl_short4; typedef __m64 __cl_uint2; typedef __m64 __cl_int2; typedef __m64 __cl_ulong1; typedef __m64 __cl_long1; typedef __m64 __cl_float2; #endif #define __CL_UCHAR8__ 1 #define __CL_CHAR8__ 1 #define __CL_USHORT4__ 1 #define __CL_SHORT4__ 1 #define __CL_INT2__ 1 #define __CL_UINT2__ 1 #define __CL_ULONG1__ 1 #define __CL_LONG1__ 1 #define __CL_FLOAT2__ 1 #endif #if defined( __AVX__ ) #if defined( __MINGW64__ ) #include #else #include #endif #if defined( __GNUC__ ) typedef cl_float __cl_float8 __attribute__((vector_size(32))); typedef cl_double __cl_double4 __attribute__((vector_size(32))); #else typedef __m256 __cl_float8; typedef __m256d __cl_double4; #endif #define __CL_FLOAT8__ 1 #define __CL_DOUBLE4__ 1 #endif /* Define alignment keys */ #if defined( __GNUC__ ) #define CL_ALIGNED(_x) __attribute__ ((aligned(_x))) #elif defined( _WIN32) && (_MSC_VER) /* Alignment keys neutered on windows because MSVC can't swallow function arguments with alignment requirements */ /* http://msdn.microsoft.com/en-us/library/373ak2y1%28VS.71%29.aspx */ /* #include */ /* #define CL_ALIGNED(_x) _CRT_ALIGN(_x) */ #define CL_ALIGNED(_x) #else #warning Need to implement some method to align data here #define CL_ALIGNED(_x) #endif /* Indicate whether .xyzw, .s0123 and .hi.lo are supported */ #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) /* .xyzw and .s0123...{f|F} are supported */ #define CL_HAS_NAMED_VECTOR_FIELDS 1 /* .hi and .lo are supported */ #define CL_HAS_HI_LO_VECTOR_FIELDS 1 #endif /* Define cl_vector types */ /* ---- cl_charn ---- */ typedef union { cl_char CL_ALIGNED(2) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_char x, y; }; __extension__ struct{ cl_char s0, s1; }; __extension__ struct{ cl_char lo, hi; }; #endif #if defined( __CL_CHAR2__) __cl_char2 v2; #endif }cl_char2; typedef union { cl_char CL_ALIGNED(4) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_char x, y, z, w; }; __extension__ struct{ cl_char s0, s1, s2, s3; }; __extension__ struct{ cl_char2 lo, hi; }; #endif #if defined( __CL_CHAR2__) __cl_char2 v2[2]; #endif #if defined( __CL_CHAR4__) __cl_char4 v4; #endif }cl_char4; /* cl_char3 is identical in size, alignment and behavior to cl_char4. See section 6.1.5. */ typedef cl_char4 cl_char3; typedef union { cl_char CL_ALIGNED(8) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_char x, y, z, w; }; __extension__ struct{ cl_char s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_char4 lo, hi; }; #endif #if defined( __CL_CHAR2__) __cl_char2 v2[4]; #endif #if defined( __CL_CHAR4__) __cl_char4 v4[2]; #endif #if defined( __CL_CHAR8__ ) __cl_char8 v8; #endif }cl_char8; typedef union { cl_char CL_ALIGNED(16) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_char x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_char s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_char8 lo, hi; }; #endif #if defined( __CL_CHAR2__) __cl_char2 v2[8]; #endif #if defined( __CL_CHAR4__) __cl_char4 v4[4]; #endif #if defined( __CL_CHAR8__ ) __cl_char8 v8[2]; #endif #if defined( __CL_CHAR16__ ) __cl_char16 v16; #endif }cl_char16; /* ---- cl_ucharn ---- */ typedef union { cl_uchar CL_ALIGNED(2) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_uchar x, y; }; __extension__ struct{ cl_uchar s0, s1; }; __extension__ struct{ cl_uchar lo, hi; }; #endif #if defined( __cl_uchar2__) __cl_uchar2 v2; #endif }cl_uchar2; typedef union { cl_uchar CL_ALIGNED(4) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_uchar x, y, z, w; }; __extension__ struct{ cl_uchar s0, s1, s2, s3; }; __extension__ struct{ cl_uchar2 lo, hi; }; #endif #if defined( __CL_UCHAR2__) __cl_uchar2 v2[2]; #endif #if defined( __CL_UCHAR4__) __cl_uchar4 v4; #endif }cl_uchar4; /* cl_uchar3 is identical in size, alignment and behavior to cl_uchar4. See section 6.1.5. */ typedef cl_uchar4 cl_uchar3; typedef union { cl_uchar CL_ALIGNED(8) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_uchar x, y, z, w; }; __extension__ struct{ cl_uchar s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_uchar4 lo, hi; }; #endif #if defined( __CL_UCHAR2__) __cl_uchar2 v2[4]; #endif #if defined( __CL_UCHAR4__) __cl_uchar4 v4[2]; #endif #if defined( __CL_UCHAR8__ ) __cl_uchar8 v8; #endif }cl_uchar8; typedef union { cl_uchar CL_ALIGNED(16) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_uchar x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_uchar s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_uchar8 lo, hi; }; #endif #if defined( __CL_UCHAR2__) __cl_uchar2 v2[8]; #endif #if defined( __CL_UCHAR4__) __cl_uchar4 v4[4]; #endif #if defined( __CL_UCHAR8__ ) __cl_uchar8 v8[2]; #endif #if defined( __CL_UCHAR16__ ) __cl_uchar16 v16; #endif }cl_uchar16; /* ---- cl_shortn ---- */ typedef union { cl_short CL_ALIGNED(4) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_short x, y; }; __extension__ struct{ cl_short s0, s1; }; __extension__ struct{ cl_short lo, hi; }; #endif #if defined( __CL_SHORT2__) __cl_short2 v2; #endif }cl_short2; typedef union { cl_short CL_ALIGNED(8) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_short x, y, z, w; }; __extension__ struct{ cl_short s0, s1, s2, s3; }; __extension__ struct{ cl_short2 lo, hi; }; #endif #if defined( __CL_SHORT2__) __cl_short2 v2[2]; #endif #if defined( __CL_SHORT4__) __cl_short4 v4; #endif }cl_short4; /* cl_short3 is identical in size, alignment and behavior to cl_short4. See section 6.1.5. */ typedef cl_short4 cl_short3; typedef union { cl_short CL_ALIGNED(16) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_short x, y, z, w; }; __extension__ struct{ cl_short s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_short4 lo, hi; }; #endif #if defined( __CL_SHORT2__) __cl_short2 v2[4]; #endif #if defined( __CL_SHORT4__) __cl_short4 v4[2]; #endif #if defined( __CL_SHORT8__ ) __cl_short8 v8; #endif }cl_short8; typedef union { cl_short CL_ALIGNED(32) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_short x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_short s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_short8 lo, hi; }; #endif #if defined( __CL_SHORT2__) __cl_short2 v2[8]; #endif #if defined( __CL_SHORT4__) __cl_short4 v4[4]; #endif #if defined( __CL_SHORT8__ ) __cl_short8 v8[2]; #endif #if defined( __CL_SHORT16__ ) __cl_short16 v16; #endif }cl_short16; /* ---- cl_ushortn ---- */ typedef union { cl_ushort CL_ALIGNED(4) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_ushort x, y; }; __extension__ struct{ cl_ushort s0, s1; }; __extension__ struct{ cl_ushort lo, hi; }; #endif #if defined( __CL_USHORT2__) __cl_ushort2 v2; #endif }cl_ushort2; typedef union { cl_ushort CL_ALIGNED(8) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_ushort x, y, z, w; }; __extension__ struct{ cl_ushort s0, s1, s2, s3; }; __extension__ struct{ cl_ushort2 lo, hi; }; #endif #if defined( __CL_USHORT2__) __cl_ushort2 v2[2]; #endif #if defined( __CL_USHORT4__) __cl_ushort4 v4; #endif }cl_ushort4; /* cl_ushort3 is identical in size, alignment and behavior to cl_ushort4. See section 6.1.5. */ typedef cl_ushort4 cl_ushort3; typedef union { cl_ushort CL_ALIGNED(16) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_ushort x, y, z, w; }; __extension__ struct{ cl_ushort s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_ushort4 lo, hi; }; #endif #if defined( __CL_USHORT2__) __cl_ushort2 v2[4]; #endif #if defined( __CL_USHORT4__) __cl_ushort4 v4[2]; #endif #if defined( __CL_USHORT8__ ) __cl_ushort8 v8; #endif }cl_ushort8; typedef union { cl_ushort CL_ALIGNED(32) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_ushort x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_ushort s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_ushort8 lo, hi; }; #endif #if defined( __CL_USHORT2__) __cl_ushort2 v2[8]; #endif #if defined( __CL_USHORT4__) __cl_ushort4 v4[4]; #endif #if defined( __CL_USHORT8__ ) __cl_ushort8 v8[2]; #endif #if defined( __CL_USHORT16__ ) __cl_ushort16 v16; #endif }cl_ushort16; /* ---- cl_intn ---- */ typedef union { cl_int CL_ALIGNED(8) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_int x, y; }; __extension__ struct{ cl_int s0, s1; }; __extension__ struct{ cl_int lo, hi; }; #endif #if defined( __CL_INT2__) __cl_int2 v2; #endif }cl_int2; typedef union { cl_int CL_ALIGNED(16) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_int x, y, z, w; }; __extension__ struct{ cl_int s0, s1, s2, s3; }; __extension__ struct{ cl_int2 lo, hi; }; #endif #if defined( __CL_INT2__) __cl_int2 v2[2]; #endif #if defined( __CL_INT4__) __cl_int4 v4; #endif }cl_int4; /* cl_int3 is identical in size, alignment and behavior to cl_int4. See section 6.1.5. */ typedef cl_int4 cl_int3; typedef union { cl_int CL_ALIGNED(32) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_int x, y, z, w; }; __extension__ struct{ cl_int s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_int4 lo, hi; }; #endif #if defined( __CL_INT2__) __cl_int2 v2[4]; #endif #if defined( __CL_INT4__) __cl_int4 v4[2]; #endif #if defined( __CL_INT8__ ) __cl_int8 v8; #endif }cl_int8; typedef union { cl_int CL_ALIGNED(64) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_int x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_int s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_int8 lo, hi; }; #endif #if defined( __CL_INT2__) __cl_int2 v2[8]; #endif #if defined( __CL_INT4__) __cl_int4 v4[4]; #endif #if defined( __CL_INT8__ ) __cl_int8 v8[2]; #endif #if defined( __CL_INT16__ ) __cl_int16 v16; #endif }cl_int16; /* ---- cl_uintn ---- */ typedef union { cl_uint CL_ALIGNED(8) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_uint x, y; }; __extension__ struct{ cl_uint s0, s1; }; __extension__ struct{ cl_uint lo, hi; }; #endif #if defined( __CL_UINT2__) __cl_uint2 v2; #endif }cl_uint2; typedef union { cl_uint CL_ALIGNED(16) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_uint x, y, z, w; }; __extension__ struct{ cl_uint s0, s1, s2, s3; }; __extension__ struct{ cl_uint2 lo, hi; }; #endif #if defined( __CL_UINT2__) __cl_uint2 v2[2]; #endif #if defined( __CL_UINT4__) __cl_uint4 v4; #endif }cl_uint4; /* cl_uint3 is identical in size, alignment and behavior to cl_uint4. See section 6.1.5. */ typedef cl_uint4 cl_uint3; typedef union { cl_uint CL_ALIGNED(32) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_uint x, y, z, w; }; __extension__ struct{ cl_uint s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_uint4 lo, hi; }; #endif #if defined( __CL_UINT2__) __cl_uint2 v2[4]; #endif #if defined( __CL_UINT4__) __cl_uint4 v4[2]; #endif #if defined( __CL_UINT8__ ) __cl_uint8 v8; #endif }cl_uint8; typedef union { cl_uint CL_ALIGNED(64) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_uint x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_uint s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_uint8 lo, hi; }; #endif #if defined( __CL_UINT2__) __cl_uint2 v2[8]; #endif #if defined( __CL_UINT4__) __cl_uint4 v4[4]; #endif #if defined( __CL_UINT8__ ) __cl_uint8 v8[2]; #endif #if defined( __CL_UINT16__ ) __cl_uint16 v16; #endif }cl_uint16; /* ---- cl_longn ---- */ typedef union { cl_long CL_ALIGNED(16) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_long x, y; }; __extension__ struct{ cl_long s0, s1; }; __extension__ struct{ cl_long lo, hi; }; #endif #if defined( __CL_LONG2__) __cl_long2 v2; #endif }cl_long2; typedef union { cl_long CL_ALIGNED(32) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_long x, y, z, w; }; __extension__ struct{ cl_long s0, s1, s2, s3; }; __extension__ struct{ cl_long2 lo, hi; }; #endif #if defined( __CL_LONG2__) __cl_long2 v2[2]; #endif #if defined( __CL_LONG4__) __cl_long4 v4; #endif }cl_long4; /* cl_long3 is identical in size, alignment and behavior to cl_long4. See section 6.1.5. */ typedef cl_long4 cl_long3; typedef union { cl_long CL_ALIGNED(64) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_long x, y, z, w; }; __extension__ struct{ cl_long s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_long4 lo, hi; }; #endif #if defined( __CL_LONG2__) __cl_long2 v2[4]; #endif #if defined( __CL_LONG4__) __cl_long4 v4[2]; #endif #if defined( __CL_LONG8__ ) __cl_long8 v8; #endif }cl_long8; typedef union { cl_long CL_ALIGNED(128) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_long x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_long s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_long8 lo, hi; }; #endif #if defined( __CL_LONG2__) __cl_long2 v2[8]; #endif #if defined( __CL_LONG4__) __cl_long4 v4[4]; #endif #if defined( __CL_LONG8__ ) __cl_long8 v8[2]; #endif #if defined( __CL_LONG16__ ) __cl_long16 v16; #endif }cl_long16; /* ---- cl_ulongn ---- */ typedef union { cl_ulong CL_ALIGNED(16) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_ulong x, y; }; __extension__ struct{ cl_ulong s0, s1; }; __extension__ struct{ cl_ulong lo, hi; }; #endif #if defined( __CL_ULONG2__) __cl_ulong2 v2; #endif }cl_ulong2; typedef union { cl_ulong CL_ALIGNED(32) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_ulong x, y, z, w; }; __extension__ struct{ cl_ulong s0, s1, s2, s3; }; __extension__ struct{ cl_ulong2 lo, hi; }; #endif #if defined( __CL_ULONG2__) __cl_ulong2 v2[2]; #endif #if defined( __CL_ULONG4__) __cl_ulong4 v4; #endif }cl_ulong4; /* cl_ulong3 is identical in size, alignment and behavior to cl_ulong4. See section 6.1.5. */ typedef cl_ulong4 cl_ulong3; typedef union { cl_ulong CL_ALIGNED(64) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_ulong x, y, z, w; }; __extension__ struct{ cl_ulong s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_ulong4 lo, hi; }; #endif #if defined( __CL_ULONG2__) __cl_ulong2 v2[4]; #endif #if defined( __CL_ULONG4__) __cl_ulong4 v4[2]; #endif #if defined( __CL_ULONG8__ ) __cl_ulong8 v8; #endif }cl_ulong8; typedef union { cl_ulong CL_ALIGNED(128) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_ulong x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_ulong s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_ulong8 lo, hi; }; #endif #if defined( __CL_ULONG2__) __cl_ulong2 v2[8]; #endif #if defined( __CL_ULONG4__) __cl_ulong4 v4[4]; #endif #if defined( __CL_ULONG8__ ) __cl_ulong8 v8[2]; #endif #if defined( __CL_ULONG16__ ) __cl_ulong16 v16; #endif }cl_ulong16; /* --- cl_floatn ---- */ typedef union { cl_float CL_ALIGNED(8) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_float x, y; }; __extension__ struct{ cl_float s0, s1; }; __extension__ struct{ cl_float lo, hi; }; #endif #if defined( __CL_FLOAT2__) __cl_float2 v2; #endif }cl_float2; typedef union { cl_float CL_ALIGNED(16) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_float x, y, z, w; }; __extension__ struct{ cl_float s0, s1, s2, s3; }; __extension__ struct{ cl_float2 lo, hi; }; #endif #if defined( __CL_FLOAT2__) __cl_float2 v2[2]; #endif #if defined( __CL_FLOAT4__) __cl_float4 v4; #endif }cl_float4; /* cl_float3 is identical in size, alignment and behavior to cl_float4. See section 6.1.5. */ typedef cl_float4 cl_float3; typedef union { cl_float CL_ALIGNED(32) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_float x, y, z, w; }; __extension__ struct{ cl_float s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_float4 lo, hi; }; #endif #if defined( __CL_FLOAT2__) __cl_float2 v2[4]; #endif #if defined( __CL_FLOAT4__) __cl_float4 v4[2]; #endif #if defined( __CL_FLOAT8__ ) __cl_float8 v8; #endif }cl_float8; typedef union { cl_float CL_ALIGNED(64) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_float x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_float s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_float8 lo, hi; }; #endif #if defined( __CL_FLOAT2__) __cl_float2 v2[8]; #endif #if defined( __CL_FLOAT4__) __cl_float4 v4[4]; #endif #if defined( __CL_FLOAT8__ ) __cl_float8 v8[2]; #endif #if defined( __CL_FLOAT16__ ) __cl_float16 v16; #endif }cl_float16; /* --- cl_doublen ---- */ typedef union { cl_double CL_ALIGNED(16) s[2]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_double x, y; }; __extension__ struct{ cl_double s0, s1; }; __extension__ struct{ cl_double lo, hi; }; #endif #if defined( __CL_DOUBLE2__) __cl_double2 v2; #endif }cl_double2; typedef union { cl_double CL_ALIGNED(32) s[4]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_double x, y, z, w; }; __extension__ struct{ cl_double s0, s1, s2, s3; }; __extension__ struct{ cl_double2 lo, hi; }; #endif #if defined( __CL_DOUBLE2__) __cl_double2 v2[2]; #endif #if defined( __CL_DOUBLE4__) __cl_double4 v4; #endif }cl_double4; /* cl_double3 is identical in size, alignment and behavior to cl_double4. See section 6.1.5. */ typedef cl_double4 cl_double3; typedef union { cl_double CL_ALIGNED(64) s[8]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_double x, y, z, w; }; __extension__ struct{ cl_double s0, s1, s2, s3, s4, s5, s6, s7; }; __extension__ struct{ cl_double4 lo, hi; }; #endif #if defined( __CL_DOUBLE2__) __cl_double2 v2[4]; #endif #if defined( __CL_DOUBLE4__) __cl_double4 v4[2]; #endif #if defined( __CL_DOUBLE8__ ) __cl_double8 v8; #endif }cl_double8; typedef union { cl_double CL_ALIGNED(128) s[16]; #if defined( __GNUC__) && ! defined( __STRICT_ANSI__ ) __extension__ struct{ cl_double x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; }; __extension__ struct{ cl_double s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; }; __extension__ struct{ cl_double8 lo, hi; }; #endif #if defined( __CL_DOUBLE2__) __cl_double2 v2[8]; #endif #if defined( __CL_DOUBLE4__) __cl_double4 v4[4]; #endif #if defined( __CL_DOUBLE8__ ) __cl_double8 v8[2]; #endif #if defined( __CL_DOUBLE16__ ) __cl_double16 v16; #endif }cl_double16; /* Macro to facilitate debugging * Usage: * Place CL_PROGRAM_STRING_DEBUG_INFO on the line before the first line of your source. * The first line ends with: CL_PROGRAM_STRING_DEBUG_INFO \" * Each line thereafter of OpenCL C source must end with: \n\ * The last line ends in "; * * Example: * * const char *my_program = CL_PROGRAM_STRING_DEBUG_INFO "\ * kernel void foo( int a, float * b ) \n\ * { \n\ * // my comment \n\ * *b[ get_global_id(0)] = a; \n\ * } \n\ * "; * * This should correctly set up the line, (column) and file information for your source * string so you can do source level debugging. */ #define __CL_STRINGIFY( _x ) # _x #define _CL_STRINGIFY( _x ) __CL_STRINGIFY( _x ) #define CL_PROGRAM_STRING_DEBUG_INFO "#line " _CL_STRINGIFY(__LINE__) " \"" __FILE__ "\" \n\n" #ifdef __cplusplus } #endif #endif /* __CL_PLATFORM_H */ khronos-opencl-headers-1.2.orig/cl.h0000644000000000000500000017015112214174566015760 0ustar rootsrc/******************************************************************************* * Copyright (c) 2008 - 2012 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. ******************************************************************************/ #ifndef __OPENCL_CL_H #define __OPENCL_CL_H #ifdef __APPLE__ #include #else #include #endif #ifdef __cplusplus extern "C" { #endif /******************************************************************************/ typedef struct _cl_platform_id * cl_platform_id; typedef struct _cl_device_id * cl_device_id; typedef struct _cl_context * cl_context; typedef struct _cl_command_queue * cl_command_queue; typedef struct _cl_mem * cl_mem; typedef struct _cl_program * cl_program; typedef struct _cl_kernel * cl_kernel; typedef struct _cl_event * cl_event; typedef struct _cl_sampler * cl_sampler; typedef cl_uint cl_bool; /* WARNING! Unlike cl_ types in cl_platform.h, cl_bool is not guaranteed to be the same size as the bool in kernels. */ typedef cl_ulong cl_bitfield; typedef cl_bitfield cl_device_type; typedef cl_uint cl_platform_info; typedef cl_uint cl_device_info; typedef cl_bitfield cl_device_fp_config; typedef cl_uint cl_device_mem_cache_type; typedef cl_uint cl_device_local_mem_type; typedef cl_bitfield cl_device_exec_capabilities; typedef cl_bitfield cl_command_queue_properties; typedef intptr_t cl_device_partition_property; typedef cl_bitfield cl_device_affinity_domain; typedef intptr_t cl_context_properties; typedef cl_uint cl_context_info; typedef cl_uint cl_command_queue_info; typedef cl_uint cl_channel_order; typedef cl_uint cl_channel_type; typedef cl_bitfield cl_mem_flags; typedef cl_uint cl_mem_object_type; typedef cl_uint cl_mem_info; typedef cl_bitfield cl_mem_migration_flags; typedef cl_uint cl_image_info; typedef cl_uint cl_buffer_create_type; typedef cl_uint cl_addressing_mode; typedef cl_uint cl_filter_mode; typedef cl_uint cl_sampler_info; typedef cl_bitfield cl_map_flags; typedef cl_uint cl_program_info; typedef cl_uint cl_program_build_info; typedef cl_uint cl_program_binary_type; typedef cl_int cl_build_status; typedef cl_uint cl_kernel_info; typedef cl_uint cl_kernel_arg_info; typedef cl_uint cl_kernel_arg_address_qualifier; typedef cl_uint cl_kernel_arg_access_qualifier; typedef cl_bitfield cl_kernel_arg_type_qualifier; typedef cl_uint cl_kernel_work_group_info; typedef cl_uint cl_event_info; typedef cl_uint cl_command_type; typedef cl_uint cl_profiling_info; typedef struct _cl_image_format { cl_channel_order image_channel_order; cl_channel_type image_channel_data_type; } cl_image_format; typedef struct _cl_image_desc { cl_mem_object_type image_type; size_t image_width; size_t image_height; size_t image_depth; size_t image_array_size; size_t image_row_pitch; size_t image_slice_pitch; cl_uint num_mip_levels; cl_uint num_samples; cl_mem buffer; } cl_image_desc; typedef struct _cl_buffer_region { size_t origin; size_t size; } cl_buffer_region; /******************************************************************************/ /* Error Codes */ #define CL_SUCCESS 0 #define CL_DEVICE_NOT_FOUND -1 #define CL_DEVICE_NOT_AVAILABLE -2 #define CL_COMPILER_NOT_AVAILABLE -3 #define CL_MEM_OBJECT_ALLOCATION_FAILURE -4 #define CL_OUT_OF_RESOURCES -5 #define CL_OUT_OF_HOST_MEMORY -6 #define CL_PROFILING_INFO_NOT_AVAILABLE -7 #define CL_MEM_COPY_OVERLAP -8 #define CL_IMAGE_FORMAT_MISMATCH -9 #define CL_IMAGE_FORMAT_NOT_SUPPORTED -10 #define CL_BUILD_PROGRAM_FAILURE -11 #define CL_MAP_FAILURE -12 #define CL_MISALIGNED_SUB_BUFFER_OFFSET -13 #define CL_EXEC_STATUS_ERROR_FOR_EVENTS_IN_WAIT_LIST -14 #define CL_COMPILE_PROGRAM_FAILURE -15 #define CL_LINKER_NOT_AVAILABLE -16 #define CL_LINK_PROGRAM_FAILURE -17 #define CL_DEVICE_PARTITION_FAILED -18 #define CL_KERNEL_ARG_INFO_NOT_AVAILABLE -19 #define CL_INVALID_VALUE -30 #define CL_INVALID_DEVICE_TYPE -31 #define CL_INVALID_PLATFORM -32 #define CL_INVALID_DEVICE -33 #define CL_INVALID_CONTEXT -34 #define CL_INVALID_QUEUE_PROPERTIES -35 #define CL_INVALID_COMMAND_QUEUE -36 #define CL_INVALID_HOST_PTR -37 #define CL_INVALID_MEM_OBJECT -38 #define CL_INVALID_IMAGE_FORMAT_DESCRIPTOR -39 #define CL_INVALID_IMAGE_SIZE -40 #define CL_INVALID_SAMPLER -41 #define CL_INVALID_BINARY -42 #define CL_INVALID_BUILD_OPTIONS -43 #define CL_INVALID_PROGRAM -44 #define CL_INVALID_PROGRAM_EXECUTABLE -45 #define CL_INVALID_KERNEL_NAME -46 #define CL_INVALID_KERNEL_DEFINITION -47 #define CL_INVALID_KERNEL -48 #define CL_INVALID_ARG_INDEX -49 #define CL_INVALID_ARG_VALUE -50 #define CL_INVALID_ARG_SIZE -51 #define CL_INVALID_KERNEL_ARGS -52 #define CL_INVALID_WORK_DIMENSION -53 #define CL_INVALID_WORK_GROUP_SIZE -54 #define CL_INVALID_WORK_ITEM_SIZE -55 #define CL_INVALID_GLOBAL_OFFSET -56 #define CL_INVALID_EVENT_WAIT_LIST -57 #define CL_INVALID_EVENT -58 #define CL_INVALID_OPERATION -59 #define CL_INVALID_GL_OBJECT -60 #define CL_INVALID_BUFFER_SIZE -61 #define CL_INVALID_MIP_LEVEL -62 #define CL_INVALID_GLOBAL_WORK_SIZE -63 #define CL_INVALID_PROPERTY -64 #define CL_INVALID_IMAGE_DESCRIPTOR -65 #define CL_INVALID_COMPILER_OPTIONS -66 #define CL_INVALID_LINKER_OPTIONS -67 #define CL_INVALID_DEVICE_PARTITION_COUNT -68 /* OpenCL Version */ #define CL_VERSION_1_0 1 #define CL_VERSION_1_1 1 #define CL_VERSION_1_2 1 /* cl_bool */ #define CL_FALSE 0 #define CL_TRUE 1 #define CL_BLOCKING CL_TRUE #define CL_NON_BLOCKING CL_FALSE /* cl_platform_info */ #define CL_PLATFORM_PROFILE 0x0900 #define CL_PLATFORM_VERSION 0x0901 #define CL_PLATFORM_NAME 0x0902 #define CL_PLATFORM_VENDOR 0x0903 #define CL_PLATFORM_EXTENSIONS 0x0904 /* cl_device_type - bitfield */ #define CL_DEVICE_TYPE_DEFAULT (1 << 0) #define CL_DEVICE_TYPE_CPU (1 << 1) #define CL_DEVICE_TYPE_GPU (1 << 2) #define CL_DEVICE_TYPE_ACCELERATOR (1 << 3) #define CL_DEVICE_TYPE_CUSTOM (1 << 4) #define CL_DEVICE_TYPE_ALL 0xFFFFFFFF /* cl_device_info */ #define CL_DEVICE_TYPE 0x1000 #define CL_DEVICE_VENDOR_ID 0x1001 #define CL_DEVICE_MAX_COMPUTE_UNITS 0x1002 #define CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS 0x1003 #define CL_DEVICE_MAX_WORK_GROUP_SIZE 0x1004 #define CL_DEVICE_MAX_WORK_ITEM_SIZES 0x1005 #define CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR 0x1006 #define CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT 0x1007 #define CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT 0x1008 #define CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG 0x1009 #define CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT 0x100A #define CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE 0x100B #define CL_DEVICE_MAX_CLOCK_FREQUENCY 0x100C #define CL_DEVICE_ADDRESS_BITS 0x100D #define CL_DEVICE_MAX_READ_IMAGE_ARGS 0x100E #define CL_DEVICE_MAX_WRITE_IMAGE_ARGS 0x100F #define CL_DEVICE_MAX_MEM_ALLOC_SIZE 0x1010 #define CL_DEVICE_IMAGE2D_MAX_WIDTH 0x1011 #define CL_DEVICE_IMAGE2D_MAX_HEIGHT 0x1012 #define CL_DEVICE_IMAGE3D_MAX_WIDTH 0x1013 #define CL_DEVICE_IMAGE3D_MAX_HEIGHT 0x1014 #define CL_DEVICE_IMAGE3D_MAX_DEPTH 0x1015 #define CL_DEVICE_IMAGE_SUPPORT 0x1016 #define CL_DEVICE_MAX_PARAMETER_SIZE 0x1017 #define CL_DEVICE_MAX_SAMPLERS 0x1018 #define CL_DEVICE_MEM_BASE_ADDR_ALIGN 0x1019 #define CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE 0x101A #define CL_DEVICE_SINGLE_FP_CONFIG 0x101B #define CL_DEVICE_GLOBAL_MEM_CACHE_TYPE 0x101C #define CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE 0x101D #define CL_DEVICE_GLOBAL_MEM_CACHE_SIZE 0x101E #define CL_DEVICE_GLOBAL_MEM_SIZE 0x101F #define CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE 0x1020 #define CL_DEVICE_MAX_CONSTANT_ARGS 0x1021 #define CL_DEVICE_LOCAL_MEM_TYPE 0x1022 #define CL_DEVICE_LOCAL_MEM_SIZE 0x1023 #define CL_DEVICE_ERROR_CORRECTION_SUPPORT 0x1024 #define CL_DEVICE_PROFILING_TIMER_RESOLUTION 0x1025 #define CL_DEVICE_ENDIAN_LITTLE 0x1026 #define CL_DEVICE_AVAILABLE 0x1027 #define CL_DEVICE_COMPILER_AVAILABLE 0x1028 #define CL_DEVICE_EXECUTION_CAPABILITIES 0x1029 #define CL_DEVICE_QUEUE_PROPERTIES 0x102A #define CL_DEVICE_NAME 0x102B #define CL_DEVICE_VENDOR 0x102C #define CL_DRIVER_VERSION 0x102D #define CL_DEVICE_PROFILE 0x102E #define CL_DEVICE_VERSION 0x102F #define CL_DEVICE_EXTENSIONS 0x1030 #define CL_DEVICE_PLATFORM 0x1031 #define CL_DEVICE_DOUBLE_FP_CONFIG 0x1032 /* 0x1033 reserved for CL_DEVICE_HALF_FP_CONFIG */ #define CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF 0x1034 #define CL_DEVICE_HOST_UNIFIED_MEMORY 0x1035 #define CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR 0x1036 #define CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT 0x1037 #define CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 0x1038 #define CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 0x1039 #define CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT 0x103A #define CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE 0x103B #define CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF 0x103C #define CL_DEVICE_OPENCL_C_VERSION 0x103D #define CL_DEVICE_LINKER_AVAILABLE 0x103E #define CL_DEVICE_BUILT_IN_KERNELS 0x103F #define CL_DEVICE_IMAGE_MAX_BUFFER_SIZE 0x1040 #define CL_DEVICE_IMAGE_MAX_ARRAY_SIZE 0x1041 #define CL_DEVICE_PARENT_DEVICE 0x1042 #define CL_DEVICE_PARTITION_MAX_SUB_DEVICES 0x1043 #define CL_DEVICE_PARTITION_PROPERTIES 0x1044 #define CL_DEVICE_PARTITION_AFFINITY_DOMAIN 0x1045 #define CL_DEVICE_PARTITION_TYPE 0x1046 #define CL_DEVICE_REFERENCE_COUNT 0x1047 #define CL_DEVICE_PREFERRED_INTEROP_USER_SYNC 0x1048 #define CL_DEVICE_PRINTF_BUFFER_SIZE 0x1049 #define CL_DEVICE_IMAGE_PITCH_ALIGNMENT 0x104A #define CL_DEVICE_IMAGE_BASE_ADDRESS_ALIGNMENT 0x104B /* cl_device_fp_config - bitfield */ #define CL_FP_DENORM (1 << 0) #define CL_FP_INF_NAN (1 << 1) #define CL_FP_ROUND_TO_NEAREST (1 << 2) #define CL_FP_ROUND_TO_ZERO (1 << 3) #define CL_FP_ROUND_TO_INF (1 << 4) #define CL_FP_FMA (1 << 5) #define CL_FP_SOFT_FLOAT (1 << 6) #define CL_FP_CORRECTLY_ROUNDED_DIVIDE_SQRT (1 << 7) /* cl_device_mem_cache_type */ #define CL_NONE 0x0 #define CL_READ_ONLY_CACHE 0x1 #define CL_READ_WRITE_CACHE 0x2 /* cl_device_local_mem_type */ #define CL_LOCAL 0x1 #define CL_GLOBAL 0x2 /* cl_device_exec_capabilities - bitfield */ #define CL_EXEC_KERNEL (1 << 0) #define CL_EXEC_NATIVE_KERNEL (1 << 1) /* cl_command_queue_properties - bitfield */ #define CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE (1 << 0) #define CL_QUEUE_PROFILING_ENABLE (1 << 1) /* cl_context_info */ #define CL_CONTEXT_REFERENCE_COUNT 0x1080 #define CL_CONTEXT_DEVICES 0x1081 #define CL_CONTEXT_PROPERTIES 0x1082 #define CL_CONTEXT_NUM_DEVICES 0x1083 /* cl_context_properties */ #define CL_CONTEXT_PLATFORM 0x1084 #define CL_CONTEXT_INTEROP_USER_SYNC 0x1085 /* cl_device_partition_property */ #define CL_DEVICE_PARTITION_EQUALLY 0x1086 #define CL_DEVICE_PARTITION_BY_COUNTS 0x1087 #define CL_DEVICE_PARTITION_BY_COUNTS_LIST_END 0x0 #define CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN 0x1088 /* cl_device_affinity_domain */ #define CL_DEVICE_AFFINITY_DOMAIN_NUMA (1 << 0) #define CL_DEVICE_AFFINITY_DOMAIN_L4_CACHE (1 << 1) #define CL_DEVICE_AFFINITY_DOMAIN_L3_CACHE (1 << 2) #define CL_DEVICE_AFFINITY_DOMAIN_L2_CACHE (1 << 3) #define CL_DEVICE_AFFINITY_DOMAIN_L1_CACHE (1 << 4) #define CL_DEVICE_AFFINITY_DOMAIN_NEXT_PARTITIONABLE (1 << 5) /* cl_command_queue_info */ #define CL_QUEUE_CONTEXT 0x1090 #define CL_QUEUE_DEVICE 0x1091 #define CL_QUEUE_REFERENCE_COUNT 0x1092 #define CL_QUEUE_PROPERTIES 0x1093 /* cl_mem_flags - bitfield */ #define CL_MEM_READ_WRITE (1 << 0) #define CL_MEM_WRITE_ONLY (1 << 1) #define CL_MEM_READ_ONLY (1 << 2) #define CL_MEM_USE_HOST_PTR (1 << 3) #define CL_MEM_ALLOC_HOST_PTR (1 << 4) #define CL_MEM_COPY_HOST_PTR (1 << 5) // reserved (1 << 6) #define CL_MEM_HOST_WRITE_ONLY (1 << 7) #define CL_MEM_HOST_READ_ONLY (1 << 8) #define CL_MEM_HOST_NO_ACCESS (1 << 9) /* cl_mem_migration_flags - bitfield */ #define CL_MIGRATE_MEM_OBJECT_HOST (1 << 0) #define CL_MIGRATE_MEM_OBJECT_CONTENT_UNDEFINED (1 << 1) /* cl_channel_order */ #define CL_R 0x10B0 #define CL_A 0x10B1 #define CL_RG 0x10B2 #define CL_RA 0x10B3 #define CL_RGB 0x10B4 #define CL_RGBA 0x10B5 #define CL_BGRA 0x10B6 #define CL_ARGB 0x10B7 #define CL_INTENSITY 0x10B8 #define CL_LUMINANCE 0x10B9 #define CL_Rx 0x10BA #define CL_RGx 0x10BB #define CL_RGBx 0x10BC #define CL_DEPTH 0x10BD #define CL_DEPTH_STENCIL 0x10BE /* cl_channel_type */ #define CL_SNORM_INT8 0x10D0 #define CL_SNORM_INT16 0x10D1 #define CL_UNORM_INT8 0x10D2 #define CL_UNORM_INT16 0x10D3 #define CL_UNORM_SHORT_565 0x10D4 #define CL_UNORM_SHORT_555 0x10D5 #define CL_UNORM_INT_101010 0x10D6 #define CL_SIGNED_INT8 0x10D7 #define CL_SIGNED_INT16 0x10D8 #define CL_SIGNED_INT32 0x10D9 #define CL_UNSIGNED_INT8 0x10DA #define CL_UNSIGNED_INT16 0x10DB #define CL_UNSIGNED_INT32 0x10DC #define CL_HALF_FLOAT 0x10DD #define CL_FLOAT 0x10DE #define CL_UNORM_INT24 0x10DF /* cl_mem_object_type */ #define CL_MEM_OBJECT_BUFFER 0x10F0 #define CL_MEM_OBJECT_IMAGE2D 0x10F1 #define CL_MEM_OBJECT_IMAGE3D 0x10F2 #define CL_MEM_OBJECT_IMAGE2D_ARRAY 0x10F3 #define CL_MEM_OBJECT_IMAGE1D 0x10F4 #define CL_MEM_OBJECT_IMAGE1D_ARRAY 0x10F5 #define CL_MEM_OBJECT_IMAGE1D_BUFFER 0x10F6 /* cl_mem_info */ #define CL_MEM_TYPE 0x1100 #define CL_MEM_FLAGS 0x1101 #define CL_MEM_SIZE 0x1102 #define CL_MEM_HOST_PTR 0x1103 #define CL_MEM_MAP_COUNT 0x1104 #define CL_MEM_REFERENCE_COUNT 0x1105 #define CL_MEM_CONTEXT 0x1106 #define CL_MEM_ASSOCIATED_MEMOBJECT 0x1107 #define CL_MEM_OFFSET 0x1108 /* cl_image_info */ #define CL_IMAGE_FORMAT 0x1110 #define CL_IMAGE_ELEMENT_SIZE 0x1111 #define CL_IMAGE_ROW_PITCH 0x1112 #define CL_IMAGE_SLICE_PITCH 0x1113 #define CL_IMAGE_WIDTH 0x1114 #define CL_IMAGE_HEIGHT 0x1115 #define CL_IMAGE_DEPTH 0x1116 #define CL_IMAGE_ARRAY_SIZE 0x1117 #define CL_IMAGE_BUFFER 0x1118 #define CL_IMAGE_NUM_MIP_LEVELS 0x1119 #define CL_IMAGE_NUM_SAMPLES 0x111A /* cl_addressing_mode */ #define CL_ADDRESS_NONE 0x1130 #define CL_ADDRESS_CLAMP_TO_EDGE 0x1131 #define CL_ADDRESS_CLAMP 0x1132 #define CL_ADDRESS_REPEAT 0x1133 #define CL_ADDRESS_MIRRORED_REPEAT 0x1134 /* cl_filter_mode */ #define CL_FILTER_NEAREST 0x1140 #define CL_FILTER_LINEAR 0x1141 /* cl_sampler_info */ #define CL_SAMPLER_REFERENCE_COUNT 0x1150 #define CL_SAMPLER_CONTEXT 0x1151 #define CL_SAMPLER_NORMALIZED_COORDS 0x1152 #define CL_SAMPLER_ADDRESSING_MODE 0x1153 #define CL_SAMPLER_FILTER_MODE 0x1154 /* cl_map_flags - bitfield */ #define CL_MAP_READ (1 << 0) #define CL_MAP_WRITE (1 << 1) #define CL_MAP_WRITE_INVALIDATE_REGION (1 << 2) /* cl_program_info */ #define CL_PROGRAM_REFERENCE_COUNT 0x1160 #define CL_PROGRAM_CONTEXT 0x1161 #define CL_PROGRAM_NUM_DEVICES 0x1162 #define CL_PROGRAM_DEVICES 0x1163 #define CL_PROGRAM_SOURCE 0x1164 #define CL_PROGRAM_BINARY_SIZES 0x1165 #define CL_PROGRAM_BINARIES 0x1166 #define CL_PROGRAM_NUM_KERNELS 0x1167 #define CL_PROGRAM_KERNEL_NAMES 0x1168 /* cl_program_build_info */ #define CL_PROGRAM_BUILD_STATUS 0x1181 #define CL_PROGRAM_BUILD_OPTIONS 0x1182 #define CL_PROGRAM_BUILD_LOG 0x1183 #define CL_PROGRAM_BINARY_TYPE 0x1184 /* cl_program_binary_type */ #define CL_PROGRAM_BINARY_TYPE_NONE 0x0 #define CL_PROGRAM_BINARY_TYPE_COMPILED_OBJECT 0x1 #define CL_PROGRAM_BINARY_TYPE_LIBRARY 0x2 #define CL_PROGRAM_BINARY_TYPE_EXECUTABLE 0x4 /* cl_build_status */ #define CL_BUILD_SUCCESS 0 #define CL_BUILD_NONE -1 #define CL_BUILD_ERROR -2 #define CL_BUILD_IN_PROGRESS -3 /* cl_kernel_info */ #define CL_KERNEL_FUNCTION_NAME 0x1190 #define CL_KERNEL_NUM_ARGS 0x1191 #define CL_KERNEL_REFERENCE_COUNT 0x1192 #define CL_KERNEL_CONTEXT 0x1193 #define CL_KERNEL_PROGRAM 0x1194 #define CL_KERNEL_ATTRIBUTES 0x1195 /* cl_kernel_arg_info */ #define CL_KERNEL_ARG_ADDRESS_QUALIFIER 0x1196 #define CL_KERNEL_ARG_ACCESS_QUALIFIER 0x1197 #define CL_KERNEL_ARG_TYPE_NAME 0x1198 #define CL_KERNEL_ARG_TYPE_QUALIFIER 0x1199 #define CL_KERNEL_ARG_NAME 0x119A /* cl_kernel_arg_address_qualifier */ #define CL_KERNEL_ARG_ADDRESS_GLOBAL 0x119B #define CL_KERNEL_ARG_ADDRESS_LOCAL 0x119C #define CL_KERNEL_ARG_ADDRESS_CONSTANT 0x119D #define CL_KERNEL_ARG_ADDRESS_PRIVATE 0x119E /* cl_kernel_arg_access_qualifier */ #define CL_KERNEL_ARG_ACCESS_READ_ONLY 0x11A0 #define CL_KERNEL_ARG_ACCESS_WRITE_ONLY 0x11A1 #define CL_KERNEL_ARG_ACCESS_READ_WRITE 0x11A2 #define CL_KERNEL_ARG_ACCESS_NONE 0x11A3 /* cl_kernel_arg_type_qualifer */ #define CL_KERNEL_ARG_TYPE_NONE 0 #define CL_KERNEL_ARG_TYPE_CONST (1 << 0) #define CL_KERNEL_ARG_TYPE_RESTRICT (1 << 1) #define CL_KERNEL_ARG_TYPE_VOLATILE (1 << 2) /* cl_kernel_work_group_info */ #define CL_KERNEL_WORK_GROUP_SIZE 0x11B0 #define CL_KERNEL_COMPILE_WORK_GROUP_SIZE 0x11B1 #define CL_KERNEL_LOCAL_MEM_SIZE 0x11B2 #define CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE 0x11B3 #define CL_KERNEL_PRIVATE_MEM_SIZE 0x11B4 #define CL_KERNEL_GLOBAL_WORK_SIZE 0x11B5 /* cl_event_info */ #define CL_EVENT_COMMAND_QUEUE 0x11D0 #define CL_EVENT_COMMAND_TYPE 0x11D1 #define CL_EVENT_REFERENCE_COUNT 0x11D2 #define CL_EVENT_COMMAND_EXECUTION_STATUS 0x11D3 #define CL_EVENT_CONTEXT 0x11D4 /* cl_command_type */ #define CL_COMMAND_NDRANGE_KERNEL 0x11F0 #define CL_COMMAND_TASK 0x11F1 #define CL_COMMAND_NATIVE_KERNEL 0x11F2 #define CL_COMMAND_READ_BUFFER 0x11F3 #define CL_COMMAND_WRITE_BUFFER 0x11F4 #define CL_COMMAND_COPY_BUFFER 0x11F5 #define CL_COMMAND_READ_IMAGE 0x11F6 #define CL_COMMAND_WRITE_IMAGE 0x11F7 #define CL_COMMAND_COPY_IMAGE 0x11F8 #define CL_COMMAND_COPY_IMAGE_TO_BUFFER 0x11F9 #define CL_COMMAND_COPY_BUFFER_TO_IMAGE 0x11FA #define CL_COMMAND_MAP_BUFFER 0x11FB #define CL_COMMAND_MAP_IMAGE 0x11FC #define CL_COMMAND_UNMAP_MEM_OBJECT 0x11FD #define CL_COMMAND_MARKER 0x11FE #define CL_COMMAND_ACQUIRE_GL_OBJECTS 0x11FF #define CL_COMMAND_RELEASE_GL_OBJECTS 0x1200 #define CL_COMMAND_READ_BUFFER_RECT 0x1201 #define CL_COMMAND_WRITE_BUFFER_RECT 0x1202 #define CL_COMMAND_COPY_BUFFER_RECT 0x1203 #define CL_COMMAND_USER 0x1204 #define CL_COMMAND_BARRIER 0x1205 #define CL_COMMAND_MIGRATE_MEM_OBJECTS 0x1206 #define CL_COMMAND_FILL_BUFFER 0x1207 #define CL_COMMAND_FILL_IMAGE 0x1208 /* command execution status */ #define CL_COMPLETE 0x0 #define CL_RUNNING 0x1 #define CL_SUBMITTED 0x2 #define CL_QUEUED 0x3 /* cl_buffer_create_type */ #define CL_BUFFER_CREATE_TYPE_REGION 0x1220 /* cl_profiling_info */ #define CL_PROFILING_COMMAND_QUEUED 0x1280 #define CL_PROFILING_COMMAND_SUBMIT 0x1281 #define CL_PROFILING_COMMAND_START 0x1282 #define CL_PROFILING_COMMAND_END 0x1283 /********************************************************************************************************/ /* Platform API */ extern CL_API_ENTRY cl_int CL_API_CALL clGetPlatformIDs(cl_uint /* num_entries */, cl_platform_id * /* platforms */, cl_uint * /* num_platforms */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetPlatformInfo(cl_platform_id /* platform */, cl_platform_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; /* Device APIs */ extern CL_API_ENTRY cl_int CL_API_CALL clGetDeviceIDs(cl_platform_id /* platform */, cl_device_type /* device_type */, cl_uint /* num_entries */, cl_device_id * /* devices */, cl_uint * /* num_devices */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetDeviceInfo(cl_device_id /* device */, cl_device_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clCreateSubDevices(cl_device_id /* in_device */, const cl_device_partition_property * /* properties */, cl_uint /* num_devices */, cl_device_id * /* out_devices */, cl_uint * /* num_devices_ret */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clRetainDevice(cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clReleaseDevice(cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2; /* Context APIs */ extern CL_API_ENTRY cl_context CL_API_CALL clCreateContext(const cl_context_properties * /* properties */, cl_uint /* num_devices */, const cl_device_id * /* devices */, void (CL_CALLBACK * /* pfn_notify */)(const char *, const void *, size_t, void *), void * /* user_data */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_context CL_API_CALL clCreateContextFromType(const cl_context_properties * /* properties */, cl_device_type /* device_type */, void (CL_CALLBACK * /* pfn_notify*/ )(const char *, const void *, size_t, void *), void * /* user_data */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clRetainContext(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clReleaseContext(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetContextInfo(cl_context /* context */, cl_context_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; /* Command Queue APIs */ extern CL_API_ENTRY cl_command_queue CL_API_CALL clCreateCommandQueue(cl_context /* context */, cl_device_id /* device */, cl_command_queue_properties /* properties */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clRetainCommandQueue(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clReleaseCommandQueue(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetCommandQueueInfo(cl_command_queue /* command_queue */, cl_command_queue_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; /* Memory Object APIs */ extern CL_API_ENTRY cl_mem CL_API_CALL clCreateBuffer(cl_context /* context */, cl_mem_flags /* flags */, size_t /* size */, void * /* host_ptr */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_mem CL_API_CALL clCreateSubBuffer(cl_mem /* buffer */, cl_mem_flags /* flags */, cl_buffer_create_type /* buffer_create_type */, const void * /* buffer_create_info */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; extern CL_API_ENTRY cl_mem CL_API_CALL clCreateImage(cl_context /* context */, cl_mem_flags /* flags */, const cl_image_format * /* image_format */, const cl_image_desc * /* image_desc */, void * /* host_ptr */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clRetainMemObject(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clReleaseMemObject(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetSupportedImageFormats(cl_context /* context */, cl_mem_flags /* flags */, cl_mem_object_type /* image_type */, cl_uint /* num_entries */, cl_image_format * /* image_formats */, cl_uint * /* num_image_formats */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetMemObjectInfo(cl_mem /* memobj */, cl_mem_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetImageInfo(cl_mem /* image */, cl_image_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clSetMemObjectDestructorCallback( cl_mem /* memobj */, void (CL_CALLBACK * /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), void * /*user_data */ ) CL_API_SUFFIX__VERSION_1_1; /* Sampler APIs */ extern CL_API_ENTRY cl_sampler CL_API_CALL clCreateSampler(cl_context /* context */, cl_bool /* normalized_coords */, cl_addressing_mode /* addressing_mode */, cl_filter_mode /* filter_mode */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clRetainSampler(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clReleaseSampler(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetSamplerInfo(cl_sampler /* sampler */, cl_sampler_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; /* Program Object APIs */ extern CL_API_ENTRY cl_program CL_API_CALL clCreateProgramWithSource(cl_context /* context */, cl_uint /* count */, const char ** /* strings */, const size_t * /* lengths */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_program CL_API_CALL clCreateProgramWithBinary(cl_context /* context */, cl_uint /* num_devices */, const cl_device_id * /* device_list */, const size_t * /* lengths */, const unsigned char ** /* binaries */, cl_int * /* binary_status */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_program CL_API_CALL clCreateProgramWithBuiltInKernels(cl_context /* context */, cl_uint /* num_devices */, const cl_device_id * /* device_list */, const char * /* kernel_names */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clRetainProgram(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clReleaseProgram(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clBuildProgram(cl_program /* program */, cl_uint /* num_devices */, const cl_device_id * /* device_list */, const char * /* options */, void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), void * /* user_data */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clCompileProgram(cl_program /* program */, cl_uint /* num_devices */, const cl_device_id * /* device_list */, const char * /* options */, cl_uint /* num_input_headers */, const cl_program * /* input_headers */, const char ** /* header_include_names */, void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), void * /* user_data */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_program CL_API_CALL clLinkProgram(cl_context /* context */, cl_uint /* num_devices */, const cl_device_id * /* device_list */, const char * /* options */, cl_uint /* num_input_programs */, const cl_program * /* input_programs */, void (CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */), void * /* user_data */, cl_int * /* errcode_ret */ ) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clUnloadPlatformCompiler(cl_platform_id /* platform */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clGetProgramInfo(cl_program /* program */, cl_program_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetProgramBuildInfo(cl_program /* program */, cl_device_id /* device */, cl_program_build_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; /* Kernel Object APIs */ extern CL_API_ENTRY cl_kernel CL_API_CALL clCreateKernel(cl_program /* program */, const char * /* kernel_name */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clCreateKernelsInProgram(cl_program /* program */, cl_uint /* num_kernels */, cl_kernel * /* kernels */, cl_uint * /* num_kernels_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clRetainKernel(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clReleaseKernel(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clSetKernelArg(cl_kernel /* kernel */, cl_uint /* arg_index */, size_t /* arg_size */, const void * /* arg_value */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetKernelInfo(cl_kernel /* kernel */, cl_kernel_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetKernelArgInfo(cl_kernel /* kernel */, cl_uint /* arg_indx */, cl_kernel_arg_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clGetKernelWorkGroupInfo(cl_kernel /* kernel */, cl_device_id /* device */, cl_kernel_work_group_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; /* Event Object APIs */ extern CL_API_ENTRY cl_int CL_API_CALL clWaitForEvents(cl_uint /* num_events */, const cl_event * /* event_list */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetEventInfo(cl_event /* event */, cl_event_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_event CL_API_CALL clCreateUserEvent(cl_context /* context */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1; extern CL_API_ENTRY cl_int CL_API_CALL clRetainEvent(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clReleaseEvent(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clSetUserEventStatus(cl_event /* event */, cl_int /* execution_status */) CL_API_SUFFIX__VERSION_1_1; extern CL_API_ENTRY cl_int CL_API_CALL clSetEventCallback( cl_event /* event */, cl_int /* command_exec_callback_type */, void (CL_CALLBACK * /* pfn_notify */)(cl_event, cl_int, void *), void * /* user_data */) CL_API_SUFFIX__VERSION_1_1; /* Profiling APIs */ extern CL_API_ENTRY cl_int CL_API_CALL clGetEventProfilingInfo(cl_event /* event */, cl_profiling_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; /* Flush and Finish APIs */ extern CL_API_ENTRY cl_int CL_API_CALL clFlush(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clFinish(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0; /* Enqueued Commands APIs */ extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueReadBuffer(cl_command_queue /* command_queue */, cl_mem /* buffer */, cl_bool /* blocking_read */, size_t /* offset */, size_t /* size */, void * /* ptr */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueReadBufferRect(cl_command_queue /* command_queue */, cl_mem /* buffer */, cl_bool /* blocking_read */, const size_t * /* buffer_offset */, const size_t * /* host_offset */, const size_t * /* region */, size_t /* buffer_row_pitch */, size_t /* buffer_slice_pitch */, size_t /* host_row_pitch */, size_t /* host_slice_pitch */, void * /* ptr */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueWriteBuffer(cl_command_queue /* command_queue */, cl_mem /* buffer */, cl_bool /* blocking_write */, size_t /* offset */, size_t /* size */, const void * /* ptr */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueWriteBufferRect(cl_command_queue /* command_queue */, cl_mem /* buffer */, cl_bool /* blocking_write */, const size_t * /* buffer_offset */, const size_t * /* host_offset */, const size_t * /* region */, size_t /* buffer_row_pitch */, size_t /* buffer_slice_pitch */, size_t /* host_row_pitch */, size_t /* host_slice_pitch */, const void * /* ptr */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueFillBuffer(cl_command_queue /* command_queue */, cl_mem /* buffer */, const void * /* pattern */, size_t /* pattern_size */, size_t /* offset */, size_t /* size */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueCopyBuffer(cl_command_queue /* command_queue */, cl_mem /* src_buffer */, cl_mem /* dst_buffer */, size_t /* src_offset */, size_t /* dst_offset */, size_t /* size */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueCopyBufferRect(cl_command_queue /* command_queue */, cl_mem /* src_buffer */, cl_mem /* dst_buffer */, const size_t * /* src_origin */, const size_t * /* dst_origin */, const size_t * /* region */, size_t /* src_row_pitch */, size_t /* src_slice_pitch */, size_t /* dst_row_pitch */, size_t /* dst_slice_pitch */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueReadImage(cl_command_queue /* command_queue */, cl_mem /* image */, cl_bool /* blocking_read */, const size_t * /* origin[3] */, const size_t * /* region[3] */, size_t /* row_pitch */, size_t /* slice_pitch */, void * /* ptr */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueWriteImage(cl_command_queue /* command_queue */, cl_mem /* image */, cl_bool /* blocking_write */, const size_t * /* origin[3] */, const size_t * /* region[3] */, size_t /* input_row_pitch */, size_t /* input_slice_pitch */, const void * /* ptr */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueFillImage(cl_command_queue /* command_queue */, cl_mem /* image */, const void * /* fill_color */, const size_t * /* origin[3] */, const size_t * /* region[3] */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueCopyImage(cl_command_queue /* command_queue */, cl_mem /* src_image */, cl_mem /* dst_image */, const size_t * /* src_origin[3] */, const size_t * /* dst_origin[3] */, const size_t * /* region[3] */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueCopyImageToBuffer(cl_command_queue /* command_queue */, cl_mem /* src_image */, cl_mem /* dst_buffer */, const size_t * /* src_origin[3] */, const size_t * /* region[3] */, size_t /* dst_offset */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueCopyBufferToImage(cl_command_queue /* command_queue */, cl_mem /* src_buffer */, cl_mem /* dst_image */, size_t /* src_offset */, const size_t * /* dst_origin[3] */, const size_t * /* region[3] */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY void * CL_API_CALL clEnqueueMapBuffer(cl_command_queue /* command_queue */, cl_mem /* buffer */, cl_bool /* blocking_map */, cl_map_flags /* map_flags */, size_t /* offset */, size_t /* size */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY void * CL_API_CALL clEnqueueMapImage(cl_command_queue /* command_queue */, cl_mem /* image */, cl_bool /* blocking_map */, cl_map_flags /* map_flags */, const size_t * /* origin[3] */, const size_t * /* region[3] */, size_t * /* image_row_pitch */, size_t * /* image_slice_pitch */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueUnmapMemObject(cl_command_queue /* command_queue */, cl_mem /* memobj */, void * /* mapped_ptr */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueMigrateMemObjects(cl_command_queue /* command_queue */, cl_uint /* num_mem_objects */, const cl_mem * /* mem_objects */, cl_mem_migration_flags /* flags */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueNDRangeKernel(cl_command_queue /* command_queue */, cl_kernel /* kernel */, cl_uint /* work_dim */, const size_t * /* global_work_offset */, const size_t * /* global_work_size */, const size_t * /* local_work_size */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueTask(cl_command_queue /* command_queue */, cl_kernel /* kernel */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueNativeKernel(cl_command_queue /* command_queue */, void (CL_CALLBACK * /*user_func*/)(void *), void * /* args */, size_t /* cb_args */, cl_uint /* num_mem_objects */, const cl_mem * /* mem_list */, const void ** /* args_mem_loc */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueMarkerWithWaitList(cl_command_queue /* command_queue */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueBarrierWithWaitList(cl_command_queue /* command_queue */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_2; /* Extension function access * * Returns the extension function address for the given function name, * or NULL if a valid function can not be found. The client must * check to make sure the address is not NULL, before using or * calling the returned function address. */ extern CL_API_ENTRY void * CL_API_CALL clGetExtensionFunctionAddressForPlatform(cl_platform_id /* platform */, const char * /* func_name */) CL_API_SUFFIX__VERSION_1_2; // Deprecated OpenCL 1.1 APIs extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL clCreateImage2D(cl_context /* context */, cl_mem_flags /* flags */, const cl_image_format * /* image_format */, size_t /* image_width */, size_t /* image_height */, size_t /* image_row_pitch */, void * /* host_ptr */, cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL clCreateImage3D(cl_context /* context */, cl_mem_flags /* flags */, const cl_image_format * /* image_format */, size_t /* image_width */, size_t /* image_height */, size_t /* image_depth */, size_t /* image_row_pitch */, size_t /* image_slice_pitch */, void * /* host_ptr */, cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL clEnqueueMarker(cl_command_queue /* command_queue */, cl_event * /* event */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL clEnqueueWaitForEvents(cl_command_queue /* command_queue */, cl_uint /* num_events */, const cl_event * /* event_list */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL clEnqueueBarrier(cl_command_queue /* command_queue */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int CL_API_CALL clUnloadCompiler(void) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED void * CL_API_CALL clGetExtensionFunctionAddress(const char * /* func_name */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; #ifdef __cplusplus } #endif #endif /* __OPENCL_CL_H */ khronos-opencl-headers-1.2.orig/cl_ext.h0000644000000000000500000003251612214174566016642 0ustar rootsrc/******************************************************************************* * Copyright (c) 2008-2013 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. ******************************************************************************/ /* $Revision: 11928 $ on $Date: 2010-07-13 09:04:56 -0700 (Tue, 13 Jul 2010) $ */ /* cl_ext.h contains OpenCL extensions which don't have external */ /* (OpenGL, D3D) dependencies. */ #ifndef __CL_EXT_H #define __CL_EXT_H #ifdef __cplusplus extern "C" { #endif #ifdef __APPLE__ #include #include #else #include #endif /* cl_khr_fp16 extension - no extension #define since it has no functions */ #define CL_DEVICE_HALF_FP_CONFIG 0x1033 /* Memory object destruction * * Apple extension for use to manage externally allocated buffers used with cl_mem objects with CL_MEM_USE_HOST_PTR * * Registers a user callback function that will be called when the memory object is deleted and its resources * freed. Each call to clSetMemObjectCallbackFn registers the specified user callback function on a callback * stack associated with memobj. The registered user callback functions are called in the reverse order in * which they were registered. The user callback functions are called and then the memory object is deleted * and its resources freed. This provides a mechanism for the application (and libraries) using memobj to be * notified when the memory referenced by host_ptr, specified when the memory object is created and used as * the storage bits for the memory object, can be reused or freed. * * The application may not call CL api's with the cl_mem object passed to the pfn_notify. * * Please check for the "cl_APPLE_SetMemObjectDestructor" extension using clGetDeviceInfo(CL_DEVICE_EXTENSIONS) * before using. */ #define cl_APPLE_SetMemObjectDestructor 1 cl_int CL_API_ENTRY clSetMemObjectDestructorAPPLE( cl_mem /* memobj */, void (* /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), void * /*user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; /* Context Logging Functions * * The next three convenience functions are intended to be used as the pfn_notify parameter to clCreateContext(). * Please check for the "cl_APPLE_ContextLoggingFunctions" extension using clGetDeviceInfo(CL_DEVICE_EXTENSIONS) * before using. * * clLogMessagesToSystemLog fowards on all log messages to the Apple System Logger */ #define cl_APPLE_ContextLoggingFunctions 1 extern void CL_API_ENTRY clLogMessagesToSystemLogAPPLE( const char * /* errstr */, const void * /* private_info */, size_t /* cb */, void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; /* clLogMessagesToStdout sends all log messages to the file descriptor stdout */ extern void CL_API_ENTRY clLogMessagesToStdoutAPPLE( const char * /* errstr */, const void * /* private_info */, size_t /* cb */, void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; /* clLogMessagesToStderr sends all log messages to the file descriptor stderr */ extern void CL_API_ENTRY clLogMessagesToStderrAPPLE( const char * /* errstr */, const void * /* private_info */, size_t /* cb */, void * /* user_data */ ) CL_EXT_SUFFIX__VERSION_1_0; /************************ * cl_khr_icd extension * ************************/ #define cl_khr_icd 1 /* cl_platform_info */ #define CL_PLATFORM_ICD_SUFFIX_KHR 0x0920 /* Additional Error Codes */ #define CL_PLATFORM_NOT_FOUND_KHR -1001 extern CL_API_ENTRY cl_int CL_API_CALL clIcdGetPlatformIDsKHR(cl_uint /* num_entries */, cl_platform_id * /* platforms */, cl_uint * /* num_platforms */); typedef CL_API_ENTRY cl_int (CL_API_CALL *clIcdGetPlatformIDsKHR_fn)( cl_uint /* num_entries */, cl_platform_id * /* platforms */, cl_uint * /* num_platforms */); /* Extension: cl_khr_image2D_buffer * * This extension allows a 2D image to be created from a cl_mem buffer without a copy. * The type associated with a 2D image created from a buffer in an OpenCL program is image2d_t. * Both the sampler and sampler-less read_image built-in functions are supported for 2D images * and 2D images created from a buffer. Similarly, the write_image built-ins are also supported * for 2D images created from a buffer. * * When the 2D image from buffer is created, the client must specify the width, * height, image format (i.e. channel order and channel data type) and optionally the row pitch * * The pitch specified must be a multiple of CL_DEVICE_IMAGE_PITCH_ALIGNMENT pixels. * The base address of the buffer must be aligned to CL_DEVICE_IMAGE_BASE_ADDRESS_ALIGNMENT pixels. */ /************************************* * cl_khr_initalize_memory extension * *************************************/ #define CL_CONTEXT_MEMORY_INITIALIZE_KHR 0x200E /************************************** * cl_khr_terminate_context extension * **************************************/ #define CL_DEVICE_TERMINATE_CAPABILITY_KHR 0x200F #define CL_CONTEXT_TERMINATE_KHR 0x2010 #define cl_khr_terminate_context 1 extern CL_API_ENTRY cl_int CL_API_CALL clTerminateContextKHR(cl_context /* context */) CL_EXT_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_int (CL_API_CALL *clTerminateContextKHR_fn)(cl_context /* context */) CL_EXT_SUFFIX__VERSION_1_2; /* * Extension: cl_khr_spir * * This extension adds support to create an OpenCL program object from a * Standard Portable Intermediate Representation (SPIR) instance */ /****************************************** * cl_nv_device_attribute_query extension * ******************************************/ /* cl_nv_device_attribute_query extension - no extension #define since it has no functions */ #define CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV 0x4000 #define CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV 0x4001 #define CL_DEVICE_REGISTERS_PER_BLOCK_NV 0x4002 #define CL_DEVICE_WARP_SIZE_NV 0x4003 #define CL_DEVICE_GPU_OVERLAP_NV 0x4004 #define CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV 0x4005 #define CL_DEVICE_INTEGRATED_MEMORY_NV 0x4006 /********************************* * cl_amd_device_attribute_query * *********************************/ #define CL_DEVICE_PROFILING_TIMER_OFFSET_AMD 0x4036 #ifdef CL_VERSION_1_1 /*********************************** * cl_ext_device_fission extension * ***********************************/ #define cl_ext_device_fission 1 extern CL_API_ENTRY cl_int CL_API_CALL clReleaseDeviceEXT( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; typedef CL_API_ENTRY cl_int (CL_API_CALL *clReleaseDeviceEXT_fn)( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; extern CL_API_ENTRY cl_int CL_API_CALL clRetainDeviceEXT( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; typedef CL_API_ENTRY cl_int (CL_API_CALL *clRetainDeviceEXT_fn)( cl_device_id /*device*/ ) CL_EXT_SUFFIX__VERSION_1_1; typedef cl_ulong cl_device_partition_property_ext; extern CL_API_ENTRY cl_int CL_API_CALL clCreateSubDevicesEXT( cl_device_id /*in_device*/, const cl_device_partition_property_ext * /* properties */, cl_uint /*num_entries*/, cl_device_id * /*out_devices*/, cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; typedef CL_API_ENTRY cl_int ( CL_API_CALL * clCreateSubDevicesEXT_fn)( cl_device_id /*in_device*/, const cl_device_partition_property_ext * /* properties */, cl_uint /*num_entries*/, cl_device_id * /*out_devices*/, cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; /* cl_device_partition_property_ext */ #define CL_DEVICE_PARTITION_EQUALLY_EXT 0x4050 #define CL_DEVICE_PARTITION_BY_COUNTS_EXT 0x4051 #define CL_DEVICE_PARTITION_BY_NAMES_EXT 0x4052 #define CL_DEVICE_PARTITION_BY_AFFINITY_DOMAIN_EXT 0x4053 /* clDeviceGetInfo selectors */ #define CL_DEVICE_PARENT_DEVICE_EXT 0x4054 #define CL_DEVICE_PARTITION_TYPES_EXT 0x4055 #define CL_DEVICE_AFFINITY_DOMAINS_EXT 0x4056 #define CL_DEVICE_REFERENCE_COUNT_EXT 0x4057 #define CL_DEVICE_PARTITION_STYLE_EXT 0x4058 /* error codes */ #define CL_DEVICE_PARTITION_FAILED_EXT -1057 #define CL_INVALID_PARTITION_COUNT_EXT -1058 #define CL_INVALID_PARTITION_NAME_EXT -1059 /* CL_AFFINITY_DOMAINs */ #define CL_AFFINITY_DOMAIN_L1_CACHE_EXT 0x1 #define CL_AFFINITY_DOMAIN_L2_CACHE_EXT 0x2 #define CL_AFFINITY_DOMAIN_L3_CACHE_EXT 0x3 #define CL_AFFINITY_DOMAIN_L4_CACHE_EXT 0x4 #define CL_AFFINITY_DOMAIN_NUMA_EXT 0x10 #define CL_AFFINITY_DOMAIN_NEXT_FISSIONABLE_EXT 0x100 /* cl_device_partition_property_ext list terminators */ #define CL_PROPERTIES_LIST_END_EXT ((cl_device_partition_property_ext) 0) #define CL_PARTITION_BY_COUNTS_LIST_END_EXT ((cl_device_partition_property_ext) 0) #define CL_PARTITION_BY_NAMES_LIST_END_EXT ((cl_device_partition_property_ext) 0 - 1) /********************************* * cl_qcom_ext_host_ptr extension *********************************/ #define CL_MEM_EXT_HOST_PTR_QCOM (1 << 29) #define CL_DEVICE_EXT_MEM_PADDING_IN_BYTES_QCOM 0x40A0 #define CL_DEVICE_PAGE_SIZE_QCOM 0x40A1 #define CL_IMAGE_ROW_ALIGNMENT_QCOM 0x40A2 #define CL_IMAGE_SLICE_ALIGNMENT_QCOM 0x40A3 #define CL_MEM_HOST_UNCACHED_QCOM 0x40A4 #define CL_MEM_HOST_WRITEBACK_QCOM 0x40A5 #define CL_MEM_HOST_WRITETHROUGH_QCOM 0x40A6 #define CL_MEM_HOST_WRITE_COMBINING_QCOM 0x40A7 typedef cl_uint cl_image_pitch_info_qcom; extern CL_API_ENTRY cl_int CL_API_CALL clGetDeviceImageInfoQCOM(cl_device_id device, size_t image_width, size_t image_height, const cl_image_format *image_format, cl_image_pitch_info_qcom param_name, size_t param_value_size, void *param_value, size_t *param_value_size_ret); typedef struct _cl_mem_ext_host_ptr { // Type of external memory allocation. // Legal values will be defined in layered extensions. cl_uint allocation_type; // Host cache policy for this external memory allocation. cl_uint host_cache_policy; } cl_mem_ext_host_ptr; /********************************* * cl_qcom_ion_host_ptr extension *********************************/ #define CL_MEM_ION_HOST_PTR_QCOM 0x40A8 typedef struct _cl_mem_ion_host_ptr { // Type of external memory allocation. // Must be CL_MEM_ION_HOST_PTR_QCOM for ION allocations. cl_mem_ext_host_ptr ext_host_ptr; // ION file descriptor int ion_filedesc; // Host pointer to the ION allocated memory void* ion_hostptr; } cl_mem_ion_host_ptr; #endif /* CL_VERSION_1_1 */ #ifdef __cplusplus } #endif #endif /* __CL_EXT_H */ khronos-opencl-headers-1.2.orig/cl_d3d10.h0000644000000000000500000001137312214174566016653 0ustar rootsrc/********************************************************************************** * Copyright (c) 2008-2012 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. **********************************************************************************/ /* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ #ifndef __OPENCL_CL_D3D10_H #define __OPENCL_CL_D3D10_H #include #include #include #ifdef __cplusplus extern "C" { #endif /****************************************************************************** * cl_khr_d3d10_sharing */ #define cl_khr_d3d10_sharing 1 typedef cl_uint cl_d3d10_device_source_khr; typedef cl_uint cl_d3d10_device_set_khr; /******************************************************************************/ // Error Codes #define CL_INVALID_D3D10_DEVICE_KHR -1002 #define CL_INVALID_D3D10_RESOURCE_KHR -1003 #define CL_D3D10_RESOURCE_ALREADY_ACQUIRED_KHR -1004 #define CL_D3D10_RESOURCE_NOT_ACQUIRED_KHR -1005 // cl_d3d10_device_source_nv #define CL_D3D10_DEVICE_KHR 0x4010 #define CL_D3D10_DXGI_ADAPTER_KHR 0x4011 // cl_d3d10_device_set_nv #define CL_PREFERRED_DEVICES_FOR_D3D10_KHR 0x4012 #define CL_ALL_DEVICES_FOR_D3D10_KHR 0x4013 // cl_context_info #define CL_CONTEXT_D3D10_DEVICE_KHR 0x4014 #define CL_CONTEXT_D3D10_PREFER_SHARED_RESOURCES_KHR 0x402C // cl_mem_info #define CL_MEM_D3D10_RESOURCE_KHR 0x4015 // cl_image_info #define CL_IMAGE_D3D10_SUBRESOURCE_KHR 0x4016 // cl_command_type #define CL_COMMAND_ACQUIRE_D3D10_OBJECTS_KHR 0x4017 #define CL_COMMAND_RELEASE_D3D10_OBJECTS_KHR 0x4018 /******************************************************************************/ typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromD3D10KHR_fn)( cl_platform_id platform, cl_d3d10_device_source_khr d3d_device_source, void * d3d_object, cl_d3d10_device_set_khr d3d_device_set, cl_uint num_entries, cl_device_id * devices, cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_0; typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10BufferKHR_fn)( cl_context context, cl_mem_flags flags, ID3D10Buffer * resource, cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10Texture2DKHR_fn)( cl_context context, cl_mem_flags flags, ID3D10Texture2D * resource, UINT subresource, cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D10Texture3DKHR_fn)( cl_context context, cl_mem_flags flags, ID3D10Texture3D * resource, UINT subresource, cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0; typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireD3D10ObjectsKHR_fn)( cl_command_queue command_queue, cl_uint num_objects, const cl_mem * mem_objects, cl_uint num_events_in_wait_list, const cl_event * event_wait_list, cl_event * event) CL_API_SUFFIX__VERSION_1_0; typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseD3D10ObjectsKHR_fn)( cl_command_queue command_queue, cl_uint num_objects, const cl_mem * mem_objects, cl_uint num_events_in_wait_list, const cl_event * event_wait_list, cl_event * event) CL_API_SUFFIX__VERSION_1_0; #ifdef __cplusplus } #endif #endif // __OPENCL_CL_D3D10_H khronos-opencl-headers-1.2.orig/cl_gl.h0000644000000000000500000001625712214174566016450 0ustar rootsrc/********************************************************************************** * Copyright (c) 2008 - 2012 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. **********************************************************************************/ #ifndef __OPENCL_CL_GL_H #define __OPENCL_CL_GL_H #ifdef __APPLE__ #include #else #include #endif #ifdef __cplusplus extern "C" { #endif typedef cl_uint cl_gl_object_type; typedef cl_uint cl_gl_texture_info; typedef cl_uint cl_gl_platform_info; typedef struct __GLsync *cl_GLsync; /* cl_gl_object_type = 0x2000 - 0x200F enum values are currently taken */ #define CL_GL_OBJECT_BUFFER 0x2000 #define CL_GL_OBJECT_TEXTURE2D 0x2001 #define CL_GL_OBJECT_TEXTURE3D 0x2002 #define CL_GL_OBJECT_RENDERBUFFER 0x2003 #define CL_GL_OBJECT_TEXTURE2D_ARRAY 0x200E #define CL_GL_OBJECT_TEXTURE1D 0x200F #define CL_GL_OBJECT_TEXTURE1D_ARRAY 0x2010 #define CL_GL_OBJECT_TEXTURE_BUFFER 0x2011 /* cl_gl_texture_info */ #define CL_GL_TEXTURE_TARGET 0x2004 #define CL_GL_MIPMAP_LEVEL 0x2005 #define CL_GL_NUM_SAMPLES 0x2012 extern CL_API_ENTRY cl_mem CL_API_CALL clCreateFromGLBuffer(cl_context /* context */, cl_mem_flags /* flags */, cl_GLuint /* bufobj */, int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_mem CL_API_CALL clCreateFromGLTexture(cl_context /* context */, cl_mem_flags /* flags */, cl_GLenum /* target */, cl_GLint /* miplevel */, cl_GLuint /* texture */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_2; extern CL_API_ENTRY cl_mem CL_API_CALL clCreateFromGLRenderbuffer(cl_context /* context */, cl_mem_flags /* flags */, cl_GLuint /* renderbuffer */, cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetGLObjectInfo(cl_mem /* memobj */, cl_gl_object_type * /* gl_object_type */, cl_GLuint * /* gl_object_name */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clGetGLTextureInfo(cl_mem /* memobj */, cl_gl_texture_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueAcquireGLObjects(cl_command_queue /* command_queue */, cl_uint /* num_objects */, const cl_mem * /* mem_objects */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; extern CL_API_ENTRY cl_int CL_API_CALL clEnqueueReleaseGLObjects(cl_command_queue /* command_queue */, cl_uint /* num_objects */, const cl_mem * /* mem_objects */, cl_uint /* num_events_in_wait_list */, const cl_event * /* event_wait_list */, cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0; // Deprecated OpenCL 1.1 APIs extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL clCreateFromGLTexture2D(cl_context /* context */, cl_mem_flags /* flags */, cl_GLenum /* target */, cl_GLint /* miplevel */, cl_GLuint /* texture */, cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_mem CL_API_CALL clCreateFromGLTexture3D(cl_context /* context */, cl_mem_flags /* flags */, cl_GLenum /* target */, cl_GLint /* miplevel */, cl_GLuint /* texture */, cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; /* cl_khr_gl_sharing extension */ #define cl_khr_gl_sharing 1 typedef cl_uint cl_gl_context_info; /* Additional Error Codes */ #define CL_INVALID_GL_SHAREGROUP_REFERENCE_KHR -1000 /* cl_gl_context_info */ #define CL_CURRENT_DEVICE_FOR_GL_CONTEXT_KHR 0x2006 #define CL_DEVICES_FOR_GL_CONTEXT_KHR 0x2007 /* Additional cl_context_properties */ #define CL_GL_CONTEXT_KHR 0x2008 #define CL_EGL_DISPLAY_KHR 0x2009 #define CL_GLX_DISPLAY_KHR 0x200A #define CL_WGL_HDC_KHR 0x200B #define CL_CGL_SHAREGROUP_KHR 0x200C extern CL_API_ENTRY cl_int CL_API_CALL clGetGLContextInfoKHR(const cl_context_properties * /* properties */, cl_gl_context_info /* param_name */, size_t /* param_value_size */, void * /* param_value */, size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0; typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetGLContextInfoKHR_fn)( const cl_context_properties * properties, cl_gl_context_info param_name, size_t param_value_size, void * param_value, size_t * param_value_size_ret); #ifdef __cplusplus } #endif #endif /* __OPENCL_CL_GL_H */ khronos-opencl-headers-1.2.orig/cl_gl_ext.h0000644000000000000500000000510612214174566017317 0ustar rootsrc/********************************************************************************** * Copyright (c) 2008-2012 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. **********************************************************************************/ /* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ /* cl_gl_ext.h contains vendor (non-KHR) OpenCL extensions which have */ /* OpenGL dependencies. */ #ifndef __OPENCL_CL_GL_EXT_H #define __OPENCL_CL_GL_EXT_H #ifdef __cplusplus extern "C" { #endif #ifdef __APPLE__ #include #else #include #endif /* * For each extension, follow this template * cl_VEN_extname extension */ /* #define cl_VEN_extname 1 * ... define new types, if any * ... define new tokens, if any * ... define new APIs, if any * * If you need GLtypes here, mirror them with a cl_GLtype, rather than including a GL header * This allows us to avoid having to decide whether to include GL headers or GLES here. */ /* * cl_khr_gl_event extension * See section 9.9 in the OpenCL 1.1 spec for more information */ #define CL_COMMAND_GL_FENCE_SYNC_OBJECT_KHR 0x200D extern CL_API_ENTRY cl_event CL_API_CALL clCreateEventFromGLsyncKHR(cl_context /* context */, cl_GLsync /* cl_GLsync */, cl_int * /* errcode_ret */) CL_EXT_SUFFIX__VERSION_1_1; #ifdef __cplusplus } #endif #endif /* __OPENCL_CL_GL_EXT_H */ khronos-opencl-headers-1.2.orig/cl_dx9_media_sharing.h0000644000000000000500000001204512214174566021413 0ustar rootsrc/********************************************************************************** * Copyright (c) 2008-2012 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. **********************************************************************************/ /* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ #ifndef __OPENCL_CL_DX9_MEDIA_SHARING_H #define __OPENCL_CL_DX9_MEDIA_SHARING_H #include #include #ifdef __cplusplus extern "C" { #endif /****************************************************************************** /* cl_khr_dx9_media_sharing */ #define cl_khr_dx9_media_sharing 1 typedef cl_uint cl_dx9_media_adapter_type_khr; typedef cl_uint cl_dx9_media_adapter_set_khr; #if defined(_WIN32) #include typedef struct _cl_dx9_surface_info_khr { IDirect3DSurface9 *resource; HANDLE shared_handle; } cl_dx9_surface_info_khr; #endif /******************************************************************************/ // Error Codes #define CL_INVALID_DX9_MEDIA_ADAPTER_KHR -1010 #define CL_INVALID_DX9_MEDIA_SURFACE_KHR -1011 #define CL_DX9_MEDIA_SURFACE_ALREADY_ACQUIRED_KHR -1012 #define CL_DX9_MEDIA_SURFACE_NOT_ACQUIRED_KHR -1013 // cl_media_adapter_type_khr #define CL_ADAPTER_D3D9_KHR 0x2020 #define CL_ADAPTER_D3D9EX_KHR 0x2021 #define CL_ADAPTER_DXVA_KHR 0x2022 // cl_media_adapter_set_khr #define CL_PREFERRED_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR 0x2023 #define CL_ALL_DEVICES_FOR_DX9_MEDIA_ADAPTER_KHR 0x2024 // cl_context_info #define CL_CONTEXT_ADAPTER_D3D9_KHR 0x2025 #define CL_CONTEXT_ADAPTER_D3D9EX_KHR 0x2026 #define CL_CONTEXT_ADAPTER_DXVA_KHR 0x2027 // cl_mem_info #define CL_MEM_DX9_MEDIA_ADAPTER_TYPE_KHR 0x2028 #define CL_MEM_DX9_MEDIA_SURFACE_INFO_KHR 0x2029 // cl_image_info #define CL_IMAGE_DX9_MEDIA_PLANE_KHR 0x202A // cl_command_type #define CL_COMMAND_ACQUIRE_DX9_MEDIA_SURFACES_KHR 0x202B #define CL_COMMAND_RELEASE_DX9_MEDIA_SURFACES_KHR 0x202C /******************************************************************************/ typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromDX9MediaAdapterKHR_fn)( cl_platform_id platform, cl_uint num_media_adapters, cl_dx9_media_adapter_type_khr * media_adapter_type, void * media_adapters, cl_dx9_media_adapter_set_khr media_adapter_set, cl_uint num_entries, cl_device_id * devices, cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromDX9MediaSurfaceKHR_fn)( cl_context context, cl_mem_flags flags, cl_dx9_media_adapter_type_khr adapter_type, void * surface_info, cl_uint plane, cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireDX9MediaSurfacesKHR_fn)( cl_command_queue command_queue, cl_uint num_objects, const cl_mem * mem_objects, cl_uint num_events_in_wait_list, const cl_event * event_wait_list, cl_event * event) CL_API_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseDX9MediaSurfacesKHR_fn)( cl_command_queue command_queue, cl_uint num_objects, const cl_mem * mem_objects, cl_uint num_events_in_wait_list, const cl_event * event_wait_list, cl_event * event) CL_API_SUFFIX__VERSION_1_2; #ifdef __cplusplus } #endif #endif // __OPENCL_CL_DX9_MEDIA_SHARING_H khronos-opencl-headers-1.2.orig/cl_d3d11.h0000644000000000000500000001136512214174566016655 0ustar rootsrc/********************************************************************************** * Copyright (c) 2008-2012 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. **********************************************************************************/ /* $Revision: 11708 $ on $Date: 2010-06-13 23:36:24 -0700 (Sun, 13 Jun 2010) $ */ #ifndef __OPENCL_CL_D3D11_H #define __OPENCL_CL_D3D11_H #include #include #include #ifdef __cplusplus extern "C" { #endif /****************************************************************************** * cl_khr_d3d11_sharing */ #define cl_khr_d3d11_sharing 1 typedef cl_uint cl_d3d11_device_source_khr; typedef cl_uint cl_d3d11_device_set_khr; /******************************************************************************/ // Error Codes #define CL_INVALID_D3D11_DEVICE_KHR -1006 #define CL_INVALID_D3D11_RESOURCE_KHR -1007 #define CL_D3D11_RESOURCE_ALREADY_ACQUIRED_KHR -1008 #define CL_D3D11_RESOURCE_NOT_ACQUIRED_KHR -1009 // cl_d3d11_device_source #define CL_D3D11_DEVICE_KHR 0x4019 #define CL_D3D11_DXGI_ADAPTER_KHR 0x401A // cl_d3d11_device_set #define CL_PREFERRED_DEVICES_FOR_D3D11_KHR 0x401B #define CL_ALL_DEVICES_FOR_D3D11_KHR 0x401C // cl_context_info #define CL_CONTEXT_D3D11_DEVICE_KHR 0x401D #define CL_CONTEXT_D3D11_PREFER_SHARED_RESOURCES_KHR 0x402D // cl_mem_info #define CL_MEM_D3D11_RESOURCE_KHR 0x401E // cl_image_info #define CL_IMAGE_D3D11_SUBRESOURCE_KHR 0x401F // cl_command_type #define CL_COMMAND_ACQUIRE_D3D11_OBJECTS_KHR 0x4020 #define CL_COMMAND_RELEASE_D3D11_OBJECTS_KHR 0x4021 /******************************************************************************/ typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetDeviceIDsFromD3D11KHR_fn)( cl_platform_id platform, cl_d3d11_device_source_khr d3d_device_source, void * d3d_object, cl_d3d11_device_set_khr d3d_device_set, cl_uint num_entries, cl_device_id * devices, cl_uint * num_devices) CL_API_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11BufferKHR_fn)( cl_context context, cl_mem_flags flags, ID3D11Buffer * resource, cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11Texture2DKHR_fn)( cl_context context, cl_mem_flags flags, ID3D11Texture2D * resource, UINT subresource, cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromD3D11Texture3DKHR_fn)( cl_context context, cl_mem_flags flags, ID3D11Texture3D * resource, UINT subresource, cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireD3D11ObjectsKHR_fn)( cl_command_queue command_queue, cl_uint num_objects, const cl_mem * mem_objects, cl_uint num_events_in_wait_list, const cl_event * event_wait_list, cl_event * event) CL_API_SUFFIX__VERSION_1_2; typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseD3D11ObjectsKHR_fn)( cl_command_queue command_queue, cl_uint num_objects, const cl_mem * mem_objects, cl_uint num_events_in_wait_list, const cl_event * event_wait_list, cl_event * event) CL_API_SUFFIX__VERSION_1_2; #ifdef __cplusplus } #endif #endif // __OPENCL_CL_D3D11_H khronos-opencl-headers-1.2.orig/cl.hpp0000644000000000000500000104554612231573665016334 0ustar rootsrc/******************************************************************************* * Copyright (c) 2008-2013 The Khronos Group Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and/or associated documentation files (the * "Materials"), to deal in the Materials without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Materials, and to * permit persons to whom the Materials are furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be included * in all copies or substantial portions of the Materials. * * THE MATERIALS ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS. ******************************************************************************/ /*! \file * * \brief C++ bindings for OpenCL 1.0 (rev 48), OpenCL 1.1 (rev 33) and * OpenCL 1.2 (rev 15) * \author Benedict R. Gaster, Laurent Morichetti and Lee Howes * * Additions and fixes from: * Brian Cole, March 3rd 2010 and April 2012 * Matt Gruenke, April 2012. * Bruce Merry, February 2013. * Tom Deakin and Simon McIntosh-Smith, July 2013 * * \version 1.2.6 * \date August 2013 * * Optional extension support * * cl * cl_ext_device_fission * #define USE_CL_DEVICE_FISSION */ /*! \mainpage * \section intro Introduction * For many large applications C++ is the language of choice and so it seems * reasonable to define C++ bindings for OpenCL. * * * The interface is contained with a single C++ header file \em cl.hpp and all * definitions are contained within the namespace \em cl. There is no additional * requirement to include \em cl.h and to use either the C++ or original C * bindings it is enough to simply include \em cl.hpp. * * The bindings themselves are lightweight and correspond closely to the * underlying C API. Using the C++ bindings introduces no additional execution * overhead. * * For detail documentation on the bindings see: * * The OpenCL C++ Wrapper API 1.2 (revision 09) * http://www.khronos.org/registry/cl/specs/opencl-cplusplus-1.2.pdf * * \section example Example * * The following example shows a general use case for the C++ * bindings, including support for the optional exception feature and * also the supplied vector and string classes, see following sections for * decriptions of these features. * * \code * #define __CL_ENABLE_EXCEPTIONS * * #if defined(__APPLE__) || defined(__MACOSX) * #include * #else * #include * #endif * #include * #include * #include * * const char * helloStr = "__kernel void " * "hello(void) " * "{ " * " " * "} "; * * int * main(void) * { * cl_int err = CL_SUCCESS; * try { * * std::vector platforms; * cl::Platform::get(&platforms); * if (platforms.size() == 0) { * std::cout << "Platform size 0\n"; * return -1; * } * * cl_context_properties properties[] = * { CL_CONTEXT_PLATFORM, (cl_context_properties)(platforms[0])(), 0}; * cl::Context context(CL_DEVICE_TYPE_CPU, properties); * * std::vector devices = context.getInfo(); * * cl::Program::Sources source(1, * std::make_pair(helloStr,strlen(helloStr))); * cl::Program program_ = cl::Program(context, source); * program_.build(devices); * * cl::Kernel kernel(program_, "hello", &err); * * cl::Event event; * cl::CommandQueue queue(context, devices[0], 0, &err); * queue.enqueueNDRangeKernel( * kernel, * cl::NullRange, * cl::NDRange(4,4), * cl::NullRange, * NULL, * &event); * * event.wait(); * } * catch (cl::Error err) { * std::cerr * << "ERROR: " * << err.what() * << "(" * << err.err() * << ")" * << std::endl; * } * * return EXIT_SUCCESS; * } * * \endcode * */ #ifndef CL_HPP_ #define CL_HPP_ #ifdef _WIN32 #include #include #include #include #if defined(__CL_ENABLE_EXCEPTIONS) #include #endif // #if defined(__CL_ENABLE_EXCEPTIONS) #pragma push_macro("max") #undef max #if defined(USE_DX_INTEROP) #include #include #endif #endif // _WIN32 // #if defined(USE_CL_DEVICE_FISSION) #include #endif #if defined(__APPLE__) || defined(__MACOSX) #include #include #include #else #include #include #endif // !__APPLE__ // To avoid accidentally taking ownership of core OpenCL types // such as cl_kernel constructors are made explicit // under OpenCL 1.2 #if defined(CL_VERSION_1_2) && !defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) #define __CL_EXPLICIT_CONSTRUCTORS explicit #else // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) #define __CL_EXPLICIT_CONSTRUCTORS #endif // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) // Define deprecated prefixes and suffixes to ensure compilation // in case they are not pre-defined #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED #endif // #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) #if !defined(CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED) #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED #endif // #if !defined(CL_EXT_PREFIX__VERSION_1_1_DEPRECATED) #if !defined(CL_CALLBACK) #define CL_CALLBACK #endif //CL_CALLBACK #include #include #if !defined(__NO_STD_VECTOR) #include #endif #if !defined(__NO_STD_STRING) #include #endif #if defined(linux) || defined(__APPLE__) || defined(__MACOSX) #include #include #include #endif // linux #include /*! \namespace cl * * \brief The OpenCL C++ bindings are defined within this namespace. * */ namespace cl { class Memory; /** * Deprecated APIs for 1.2 */ #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) #define __INIT_CL_EXT_FCN_PTR(name) \ if(!pfn_##name) { \ pfn_##name = (PFN_##name) \ clGetExtensionFunctionAddress(#name); \ if(!pfn_##name) { \ } \ } #endif // #if defined(CL_VERSION_1_1) #if defined(CL_VERSION_1_2) #define __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, name) \ if(!pfn_##name) { \ pfn_##name = (PFN_##name) \ clGetExtensionFunctionAddressForPlatform(platform, #name); \ if(!pfn_##name) { \ } \ } #endif // #if defined(CL_VERSION_1_1) class Program; class Device; class Context; class CommandQueue; class Memory; class Buffer; #if defined(__CL_ENABLE_EXCEPTIONS) /*! \brief Exception class * * This may be thrown by API functions when __CL_ENABLE_EXCEPTIONS is defined. */ class Error : public std::exception { private: cl_int err_; const char * errStr_; public: /*! \brief Create a new CL error exception for a given error code * and corresponding message. * * \param err error code value. * * \param errStr a descriptive string that must remain in scope until * handling of the exception has concluded. If set, it * will be returned by what(). */ Error(cl_int err, const char * errStr = NULL) : err_(err), errStr_(errStr) {} ~Error() throw() {} /*! \brief Get error string associated with exception * * \return A memory pointer to the error message string. */ virtual const char * what() const throw () { if (errStr_ == NULL) { return "empty"; } else { return errStr_; } } /*! \brief Get error code associated with exception * * \return The error code. */ cl_int err(void) const { return err_; } }; #define __ERR_STR(x) #x #else #define __ERR_STR(x) NULL #endif // __CL_ENABLE_EXCEPTIONS namespace detail { #if defined(__CL_ENABLE_EXCEPTIONS) static inline cl_int errHandler ( cl_int err, const char * errStr = NULL) { if (err != CL_SUCCESS) { throw Error(err, errStr); } return err; } #else static inline cl_int errHandler (cl_int err, const char * errStr = NULL) { (void) errStr; // suppress unused variable warning return err; } #endif // __CL_ENABLE_EXCEPTIONS } //! \cond DOXYGEN_DETAIL #if !defined(__CL_USER_OVERRIDE_ERROR_STRINGS) #define __GET_DEVICE_INFO_ERR __ERR_STR(clGetDeviceInfo) #define __GET_PLATFORM_INFO_ERR __ERR_STR(clGetPlatformInfo) #define __GET_DEVICE_IDS_ERR __ERR_STR(clGetDeviceIDs) #define __GET_PLATFORM_IDS_ERR __ERR_STR(clGetPlatformIDs) #define __GET_CONTEXT_INFO_ERR __ERR_STR(clGetContextInfo) #define __GET_EVENT_INFO_ERR __ERR_STR(clGetEventInfo) #define __GET_EVENT_PROFILE_INFO_ERR __ERR_STR(clGetEventProfileInfo) #define __GET_MEM_OBJECT_INFO_ERR __ERR_STR(clGetMemObjectInfo) #define __GET_IMAGE_INFO_ERR __ERR_STR(clGetImageInfo) #define __GET_SAMPLER_INFO_ERR __ERR_STR(clGetSamplerInfo) #define __GET_KERNEL_INFO_ERR __ERR_STR(clGetKernelInfo) #if defined(CL_VERSION_1_2) #define __GET_KERNEL_ARG_INFO_ERR __ERR_STR(clGetKernelArgInfo) #endif // #if defined(CL_VERSION_1_2) #define __GET_KERNEL_WORK_GROUP_INFO_ERR __ERR_STR(clGetKernelWorkGroupInfo) #define __GET_PROGRAM_INFO_ERR __ERR_STR(clGetProgramInfo) #define __GET_PROGRAM_BUILD_INFO_ERR __ERR_STR(clGetProgramBuildInfo) #define __GET_COMMAND_QUEUE_INFO_ERR __ERR_STR(clGetCommandQueueInfo) #define __CREATE_CONTEXT_ERR __ERR_STR(clCreateContext) #define __CREATE_CONTEXT_FROM_TYPE_ERR __ERR_STR(clCreateContextFromType) #define __GET_SUPPORTED_IMAGE_FORMATS_ERR __ERR_STR(clGetSupportedImageFormats) #define __CREATE_BUFFER_ERR __ERR_STR(clCreateBuffer) #define __COPY_ERR __ERR_STR(cl::copy) #define __CREATE_SUBBUFFER_ERR __ERR_STR(clCreateSubBuffer) #define __CREATE_GL_BUFFER_ERR __ERR_STR(clCreateFromGLBuffer) #define __CREATE_GL_RENDER_BUFFER_ERR __ERR_STR(clCreateFromGLBuffer) #define __GET_GL_OBJECT_INFO_ERR __ERR_STR(clGetGLObjectInfo) #if defined(CL_VERSION_1_2) #define __CREATE_IMAGE_ERR __ERR_STR(clCreateImage) #define __CREATE_GL_TEXTURE_ERR __ERR_STR(clCreateFromGLTexture) #define __IMAGE_DIMENSION_ERR __ERR_STR(Incorrect image dimensions) #endif // #if defined(CL_VERSION_1_2) #define __CREATE_SAMPLER_ERR __ERR_STR(clCreateSampler) #define __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR __ERR_STR(clSetMemObjectDestructorCallback) #define __CREATE_USER_EVENT_ERR __ERR_STR(clCreateUserEvent) #define __SET_USER_EVENT_STATUS_ERR __ERR_STR(clSetUserEventStatus) #define __SET_EVENT_CALLBACK_ERR __ERR_STR(clSetEventCallback) #define __WAIT_FOR_EVENTS_ERR __ERR_STR(clWaitForEvents) #define __CREATE_KERNEL_ERR __ERR_STR(clCreateKernel) #define __SET_KERNEL_ARGS_ERR __ERR_STR(clSetKernelArg) #define __CREATE_PROGRAM_WITH_SOURCE_ERR __ERR_STR(clCreateProgramWithSource) #define __CREATE_PROGRAM_WITH_BINARY_ERR __ERR_STR(clCreateProgramWithBinary) #if defined(CL_VERSION_1_2) #define __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR __ERR_STR(clCreateProgramWithBuiltInKernels) #endif // #if defined(CL_VERSION_1_2) #define __BUILD_PROGRAM_ERR __ERR_STR(clBuildProgram) #if defined(CL_VERSION_1_2) #define __COMPILE_PROGRAM_ERR __ERR_STR(clCompileProgram) #endif // #if defined(CL_VERSION_1_2) #define __CREATE_KERNELS_IN_PROGRAM_ERR __ERR_STR(clCreateKernelsInProgram) #define __CREATE_COMMAND_QUEUE_ERR __ERR_STR(clCreateCommandQueue) #define __SET_COMMAND_QUEUE_PROPERTY_ERR __ERR_STR(clSetCommandQueueProperty) #define __ENQUEUE_READ_BUFFER_ERR __ERR_STR(clEnqueueReadBuffer) #define __ENQUEUE_READ_BUFFER_RECT_ERR __ERR_STR(clEnqueueReadBufferRect) #define __ENQUEUE_WRITE_BUFFER_ERR __ERR_STR(clEnqueueWriteBuffer) #define __ENQUEUE_WRITE_BUFFER_RECT_ERR __ERR_STR(clEnqueueWriteBufferRect) #define __ENQEUE_COPY_BUFFER_ERR __ERR_STR(clEnqueueCopyBuffer) #define __ENQEUE_COPY_BUFFER_RECT_ERR __ERR_STR(clEnqueueCopyBufferRect) #define __ENQUEUE_FILL_BUFFER_ERR __ERR_STR(clEnqueueFillBuffer) #define __ENQUEUE_READ_IMAGE_ERR __ERR_STR(clEnqueueReadImage) #define __ENQUEUE_WRITE_IMAGE_ERR __ERR_STR(clEnqueueWriteImage) #define __ENQUEUE_COPY_IMAGE_ERR __ERR_STR(clEnqueueCopyImage) #define __ENQUEUE_FILL_IMAGE_ERR __ERR_STR(clEnqueueFillImage) #define __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR __ERR_STR(clEnqueueCopyImageToBuffer) #define __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR __ERR_STR(clEnqueueCopyBufferToImage) #define __ENQUEUE_MAP_BUFFER_ERR __ERR_STR(clEnqueueMapBuffer) #define __ENQUEUE_MAP_IMAGE_ERR __ERR_STR(clEnqueueMapImage) #define __ENQUEUE_UNMAP_MEM_OBJECT_ERR __ERR_STR(clEnqueueUnMapMemObject) #define __ENQUEUE_NDRANGE_KERNEL_ERR __ERR_STR(clEnqueueNDRangeKernel) #define __ENQUEUE_TASK_ERR __ERR_STR(clEnqueueTask) #define __ENQUEUE_NATIVE_KERNEL __ERR_STR(clEnqueueNativeKernel) #if defined(CL_VERSION_1_2) #define __ENQUEUE_MIGRATE_MEM_OBJECTS_ERR __ERR_STR(clEnqueueMigrateMemObjects) #endif // #if defined(CL_VERSION_1_2) #define __ENQUEUE_ACQUIRE_GL_ERR __ERR_STR(clEnqueueAcquireGLObjects) #define __ENQUEUE_RELEASE_GL_ERR __ERR_STR(clEnqueueReleaseGLObjects) #define __RETAIN_ERR __ERR_STR(Retain Object) #define __RELEASE_ERR __ERR_STR(Release Object) #define __FLUSH_ERR __ERR_STR(clFlush) #define __FINISH_ERR __ERR_STR(clFinish) #define __VECTOR_CAPACITY_ERR __ERR_STR(Vector capacity error) /** * CL 1.2 version that uses device fission. */ #if defined(CL_VERSION_1_2) #define __CREATE_SUB_DEVICES __ERR_STR(clCreateSubDevices) #else #define __CREATE_SUB_DEVICES __ERR_STR(clCreateSubDevicesEXT) #endif // #if defined(CL_VERSION_1_2) /** * Deprecated APIs for 1.2 */ #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) #define __ENQUEUE_MARKER_ERR __ERR_STR(clEnqueueMarker) #define __ENQUEUE_WAIT_FOR_EVENTS_ERR __ERR_STR(clEnqueueWaitForEvents) #define __ENQUEUE_BARRIER_ERR __ERR_STR(clEnqueueBarrier) #define __UNLOAD_COMPILER_ERR __ERR_STR(clUnloadCompiler) #define __CREATE_GL_TEXTURE_2D_ERR __ERR_STR(clCreateFromGLTexture2D) #define __CREATE_GL_TEXTURE_3D_ERR __ERR_STR(clCreateFromGLTexture3D) #define __CREATE_IMAGE2D_ERR __ERR_STR(clCreateImage2D) #define __CREATE_IMAGE3D_ERR __ERR_STR(clCreateImage3D) #endif // #if defined(CL_VERSION_1_1) #endif // __CL_USER_OVERRIDE_ERROR_STRINGS //! \endcond /** * CL 1.2 marker and barrier commands */ #if defined(CL_VERSION_1_2) #define __ENQUEUE_MARKER_WAIT_LIST_ERR __ERR_STR(clEnqueueMarkerWithWaitList) #define __ENQUEUE_BARRIER_WAIT_LIST_ERR __ERR_STR(clEnqueueBarrierWithWaitList) #endif // #if defined(CL_VERSION_1_2) #if !defined(__USE_DEV_STRING) && !defined(__NO_STD_STRING) typedef std::string STRING_CLASS; #elif !defined(__USE_DEV_STRING) /*! \class string * \brief Simple string class, that provides a limited subset of std::string * functionality but avoids many of the issues that come with that class. * \note Deprecated. Please use std::string as default or * re-define the string class to match the std::string * interface by defining STRING_CLASS */ class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED string CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED { private: ::size_t size_; char * str_; public: //! \brief Constructs an empty string, allocating no memory. string(void) : size_(0), str_(NULL) { } /*! \brief Constructs a string populated from an arbitrary value of * specified size. * * An extra '\0' is added, in case none was contained in str. * * \param str the initial value of the string instance. Note that '\0' * characters receive no special treatment. If NULL, * the string is left empty, with a size of 0. * * \param size the number of characters to copy from str. */ string(const char * str, ::size_t size) : size_(size), str_(NULL) { if( size > 0 ) { str_ = new char[size_+1]; if (str_ != NULL) { memcpy(str_, str, size_ * sizeof(char)); str_[size_] = '\0'; } else { size_ = 0; } } } /*! \brief Constructs a string populated from a null-terminated value. * * \param str the null-terminated initial value of the string instance. * If NULL, the string is left empty, with a size of 0. */ string(const char * str) : size_(0), str_(NULL) { if( str ) { size_= ::strlen(str); } if( size_ > 0 ) { str_ = new char[size_ + 1]; if (str_ != NULL) { memcpy(str_, str, (size_ + 1) * sizeof(char)); } } } void resize( ::size_t n ) { if( size_ == n ) { return; } if (n == 0) { if( str_ ) { delete [] str_; } str_ = NULL; size_ = 0; } else { char *newString = new char[n + 1]; int copySize = n; if( size_ < n ) { copySize = size_; } size_ = n; if(str_) { memcpy(newString, str_, (copySize + 1) * sizeof(char)); } if( copySize < size_ ) { memset(newString + copySize, 0, size_ - copySize); } newString[size_] = '\0'; delete [] str_; str_ = newString; } } const char& operator[] ( ::size_t pos ) const { return str_[pos]; } char& operator[] ( ::size_t pos ) { return str_[pos]; } /*! \brief Copies the value of another string to this one. * * \param rhs the string to copy. * * \returns a reference to the modified instance. */ string& operator=(const string& rhs) { if (this == &rhs) { return *this; } if( str_ != NULL ) { delete [] str_; str_ = NULL; size_ = 0; } if (rhs.size_ == 0 || rhs.str_ == NULL) { str_ = NULL; size_ = 0; } else { str_ = new char[rhs.size_ + 1]; size_ = rhs.size_; if (str_ != NULL) { memcpy(str_, rhs.str_, (size_ + 1) * sizeof(char)); } else { size_ = 0; } } return *this; } /*! \brief Constructs a string by copying the value of another instance. * * \param rhs the string to copy. */ string(const string& rhs) : size_(0), str_(NULL) { *this = rhs; } //! \brief Destructor - frees memory used to hold the current value. ~string() { delete[] str_; str_ = NULL; } //! \brief Queries the length of the string, excluding any added '\0's. ::size_t size(void) const { return size_; } //! \brief Queries the length of the string, excluding any added '\0's. ::size_t length(void) const { return size(); } /*! \brief Returns a pointer to the private copy held by this instance, * or "" if empty/unset. */ const char * c_str(void) const { return (str_) ? str_ : "";} }; typedef cl::string STRING_CLASS; #endif // #elif !defined(__USE_DEV_STRING) #if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR) #define VECTOR_CLASS std::vector #elif !defined(__USE_DEV_VECTOR) #define VECTOR_CLASS cl::vector #if !defined(__MAX_DEFAULT_VECTOR_SIZE) #define __MAX_DEFAULT_VECTOR_SIZE 10 #endif /*! \class vector * \brief Fixed sized vector implementation that mirroring * * \note Deprecated. Please use std::vector as default or * re-define the vector class to match the std::vector * interface by defining VECTOR_CLASS * \note Not recommended for use with custom objects as * current implementation will construct N elements * * std::vector functionality. * \brief Fixed sized vector compatible with std::vector. * * \note * This differs from std::vector<> not just in memory allocation, * but also in terms of when members are constructed, destroyed, * and assigned instead of being copy constructed. * * \param T type of element contained in the vector. * * \param N maximum size of the vector. */ template class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED vector CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED { private: T data_[N]; unsigned int size_; public: //! \brief Constructs an empty vector with no memory allocated. vector() : size_(static_cast(0)) {} //! \brief Deallocates the vector's memory and destroys all of its elements. ~vector() { clear(); } //! \brief Returns the number of elements currently contained. unsigned int size(void) const { return size_; } /*! \brief Empties the vector of all elements. * \note * This does not deallocate memory but will invoke destructors * on contained elements. */ void clear() { while(!empty()) { pop_back(); } } /*! \brief Appends an element after the last valid element. * Calling this on a vector that has reached capacity will throw an * exception if exceptions are enabled. */ void push_back (const T& x) { if (size() < N) { new (&data_[size_]) T(x); size_++; } else { detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR); } } /*! \brief Removes the last valid element from the vector. * Calling this on an empty vector will throw an exception * if exceptions are enabled. */ void pop_back(void) { if (size_ != 0) { --size_; data_[size_].~T(); } else { detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR); } } /*! \brief Constructs with a value copied from another. * * \param vec the vector to copy. */ vector(const vector& vec) : size_(vec.size_) { if (size_ != 0) { assign(vec.begin(), vec.end()); } } /*! \brief Constructs with a specified number of initial elements. * * \param size number of initial elements. * * \param val value of initial elements. */ vector(unsigned int size, const T& val = T()) : size_(0) { for (unsigned int i = 0; i < size; i++) { push_back(val); } } /*! \brief Overwrites the current content with that copied from another * instance. * * \param rhs vector to copy. * * \returns a reference to this. */ vector& operator=(const vector& rhs) { if (this == &rhs) { return *this; } if (rhs.size_ != 0) { assign(rhs.begin(), rhs.end()); } else { clear(); } return *this; } /*! \brief Tests equality against another instance. * * \param vec the vector against which to compare. */ bool operator==(vector &vec) { if (size() != vec.size()) { return false; } for( unsigned int i = 0; i < size(); ++i ) { if( operator[](i) != vec[i] ) { return false; } } return true; } //! \brief Conversion operator to T*. operator T* () { return data_; } //! \brief Conversion operator to const T*. operator const T* () const { return data_; } //! \brief Tests whether this instance has any elements. bool empty (void) const { return size_==0; } //! \brief Returns the maximum number of elements this instance can hold. unsigned int max_size (void) const { return N; } //! \brief Returns the maximum number of elements this instance can hold. unsigned int capacity () const { return N; } /*! \brief Returns a reference to a given element. * * \param index which element to access. * * \note * The caller is responsible for ensuring index is >= 0 and < size(). */ T& operator[](int index) { return data_[index]; } /*! \brief Returns a const reference to a given element. * * \param index which element to access. * * \note * The caller is responsible for ensuring index is >= 0 and < size(). */ const T& operator[](int index) const { return data_[index]; } /*! \brief Assigns elements of the vector based on a source iterator range. * * \param start Beginning iterator of source range * \param end Enditerator of source range * * \note * Will throw an exception if exceptions are enabled and size exceeded. */ template void assign(I start, I end) { clear(); while(start != end) { push_back(*start); start++; } } /*! \class iterator * \brief Const iterator class for vectors */ class iterator { private: const vector *vec_; int index_; /** * Internal iterator constructor to capture reference * to the vector it iterates over rather than taking * the vector by copy. */ iterator (const vector &vec, int index) : vec_(&vec) { if( !vec.empty() ) { index_ = index; } else { index_ = -1; } } public: iterator(void) : index_(-1), vec_(NULL) { } iterator(const iterator& rhs) : vec_(rhs.vec_), index_(rhs.index_) { } ~iterator(void) {} static iterator begin(const cl::vector &vec) { iterator i(vec, 0); return i; } static iterator end(const cl::vector &vec) { iterator i(vec, vec.size()); return i; } bool operator==(iterator i) { return ((vec_ == i.vec_) && (index_ == i.index_)); } bool operator!=(iterator i) { return (!(*this==i)); } iterator& operator++() { ++index_; return *this; } iterator operator++(int) { iterator retVal(*this); ++index_; return retVal; } iterator& operator--() { --index_; return *this; } iterator operator--(int) { iterator retVal(*this); --index_; return retVal; } const T& operator *() const { return (*vec_)[index_]; } }; iterator begin(void) { return iterator::begin(*this); } iterator begin(void) const { return iterator::begin(*this); } iterator end(void) { return iterator::end(*this); } iterator end(void) const { return iterator::end(*this); } T& front(void) { return data_[0]; } T& back(void) { return data_[size_]; } const T& front(void) const { return data_[0]; } const T& back(void) const { return data_[size_-1]; } }; #endif // #if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR) namespace detail { #define __DEFAULT_NOT_INITIALIZED 1 #define __DEFAULT_BEING_INITIALIZED 2 #define __DEFAULT_INITIALIZED 4 /* * Compare and exchange primitives are needed for handling of defaults */ inline int compare_exchange(volatile int * dest, int exchange, int comparand) { #ifdef _WIN32 return (int)(InterlockedCompareExchange( (volatile long*)dest, (long)exchange, (long)comparand)); #elif defined(__APPLE__) || defined(__MACOSX) return OSAtomicOr32Orig((uint32_t)exchange, (volatile uint32_t*)dest); #else // !_WIN32 || defined(__APPLE__) || defined(__MACOSX) return (__sync_val_compare_and_swap( dest, comparand, exchange)); #endif // !_WIN32 } inline void fence() { _mm_mfence(); } }; // namespace detail /*! \brief class used to interface between C++ and * OpenCL C calls that require arrays of size_t values, whose * size is known statically. */ template class size_t { private: ::size_t data_[N]; public: //! \brief Initialize size_t to all 0s size_t() { for( int i = 0; i < N; ++i ) { data_[i] = 0; } } ::size_t& operator[](int index) { return data_[index]; } const ::size_t& operator[](int index) const { return data_[index]; } //! \brief Conversion operator to T*. operator ::size_t* () { return data_; } //! \brief Conversion operator to const T*. operator const ::size_t* () const { return data_; } }; namespace detail { // Generic getInfoHelper. The final parameter is used to guide overload // resolution: the actual parameter passed is an int, which makes this // a worse conversion sequence than a specialization that declares the // parameter as an int. template inline cl_int getInfoHelper(Functor f, cl_uint name, T* param, long) { return f(name, sizeof(T), param, NULL); } // Specialized getInfoHelper for VECTOR_CLASS params template inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, long) { ::size_t required; cl_int err = f(name, 0, NULL, &required); if (err != CL_SUCCESS) { return err; } T* value = (T*) alloca(required); err = f(name, required, value, NULL); if (err != CL_SUCCESS) { return err; } param->assign(&value[0], &value[required/sizeof(T)]); return CL_SUCCESS; } /* Specialization for reference-counted types. This depends on the * existence of Wrapper::cl_type, and none of the other types having the * cl_type member. Note that simplify specifying the parameter as Wrapper * does not work, because when using a derived type (e.g. Context) the generic * template will provide a better match. */ template inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, int, typename T::cl_type = 0) { ::size_t required; cl_int err = f(name, 0, NULL, &required); if (err != CL_SUCCESS) { return err; } typename T::cl_type * value = (typename T::cl_type *) alloca(required); err = f(name, required, value, NULL); if (err != CL_SUCCESS) { return err; } ::size_t elements = required / sizeof(typename T::cl_type); param->assign(&value[0], &value[elements]); for (::size_t i = 0; i < elements; i++) { if (value[i] != NULL) { err = (*param)[i].retain(); if (err != CL_SUCCESS) { return err; } } } return CL_SUCCESS; } // Specialized for getInfo template inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS* param, int) { cl_int err = f(name, param->size() * sizeof(char *), &(*param)[0], NULL); if (err != CL_SUCCESS) { return err; } return CL_SUCCESS; } // Specialized GetInfoHelper for STRING_CLASS params template inline cl_int getInfoHelper(Func f, cl_uint name, STRING_CLASS* param, long) { ::size_t required; cl_int err = f(name, 0, NULL, &required); if (err != CL_SUCCESS) { return err; } char* value = (char*) alloca(required); err = f(name, required, value, NULL); if (err != CL_SUCCESS) { return err; } *param = value; return CL_SUCCESS; } // Specialized GetInfoHelper for cl::size_t params template inline cl_int getInfoHelper(Func f, cl_uint name, size_t* param, long) { ::size_t required; cl_int err = f(name, 0, NULL, &required); if (err != CL_SUCCESS) { return err; } ::size_t* value = (::size_t*) alloca(required); err = f(name, required, value, NULL); if (err != CL_SUCCESS) { return err; } for(int i = 0; i < N; ++i) { (*param)[i] = value[i]; } return CL_SUCCESS; } template struct ReferenceHandler; /* Specialization for reference-counted types. This depends on the * existence of Wrapper::cl_type, and none of the other types having the * cl_type member. Note that simplify specifying the parameter as Wrapper * does not work, because when using a derived type (e.g. Context) the generic * template will provide a better match. */ template inline cl_int getInfoHelper(Func f, cl_uint name, T* param, int, typename T::cl_type = 0) { typename T::cl_type value; cl_int err = f(name, sizeof(value), &value, NULL); if (err != CL_SUCCESS) { return err; } *param = value; if (value != NULL) { err = param->retain(); if (err != CL_SUCCESS) { return err; } } return CL_SUCCESS; } #define __PARAM_NAME_INFO_1_0(F) \ F(cl_platform_info, CL_PLATFORM_PROFILE, STRING_CLASS) \ F(cl_platform_info, CL_PLATFORM_VERSION, STRING_CLASS) \ F(cl_platform_info, CL_PLATFORM_NAME, STRING_CLASS) \ F(cl_platform_info, CL_PLATFORM_VENDOR, STRING_CLASS) \ F(cl_platform_info, CL_PLATFORM_EXTENSIONS, STRING_CLASS) \ \ F(cl_device_info, CL_DEVICE_TYPE, cl_device_type) \ F(cl_device_info, CL_DEVICE_VENDOR_ID, cl_uint) \ F(cl_device_info, CL_DEVICE_MAX_COMPUTE_UNITS, cl_uint) \ F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS, cl_uint) \ F(cl_device_info, CL_DEVICE_MAX_WORK_GROUP_SIZE, ::size_t) \ F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_SIZES, VECTOR_CLASS< ::size_t>) \ F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR, cl_uint) \ F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT, cl_uint) \ F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT, cl_uint) \ F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG, cl_uint) \ F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT, cl_uint) \ F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE, cl_uint) \ F(cl_device_info, CL_DEVICE_MAX_CLOCK_FREQUENCY, cl_uint) \ F(cl_device_info, CL_DEVICE_ADDRESS_BITS, cl_uint) \ F(cl_device_info, CL_DEVICE_MAX_READ_IMAGE_ARGS, cl_uint) \ F(cl_device_info, CL_DEVICE_MAX_WRITE_IMAGE_ARGS, cl_uint) \ F(cl_device_info, CL_DEVICE_MAX_MEM_ALLOC_SIZE, cl_ulong) \ F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_WIDTH, ::size_t) \ F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_HEIGHT, ::size_t) \ F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_WIDTH, ::size_t) \ F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_HEIGHT, ::size_t) \ F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_DEPTH, ::size_t) \ F(cl_device_info, CL_DEVICE_IMAGE_SUPPORT, cl_bool) \ F(cl_device_info, CL_DEVICE_MAX_PARAMETER_SIZE, ::size_t) \ F(cl_device_info, CL_DEVICE_MAX_SAMPLERS, cl_uint) \ F(cl_device_info, CL_DEVICE_MEM_BASE_ADDR_ALIGN, cl_uint) \ F(cl_device_info, CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE, cl_uint) \ F(cl_device_info, CL_DEVICE_SINGLE_FP_CONFIG, cl_device_fp_config) \ F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, cl_device_mem_cache_type) \ F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE, cl_uint)\ F(cl_device_info, CL_DEVICE_GLOBAL_MEM_CACHE_SIZE, cl_ulong) \ F(cl_device_info, CL_DEVICE_GLOBAL_MEM_SIZE, cl_ulong) \ F(cl_device_info, CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE, cl_ulong) \ F(cl_device_info, CL_DEVICE_MAX_CONSTANT_ARGS, cl_uint) \ F(cl_device_info, CL_DEVICE_LOCAL_MEM_TYPE, cl_device_local_mem_type) \ F(cl_device_info, CL_DEVICE_LOCAL_MEM_SIZE, cl_ulong) \ F(cl_device_info, CL_DEVICE_ERROR_CORRECTION_SUPPORT, cl_bool) \ F(cl_device_info, CL_DEVICE_PROFILING_TIMER_RESOLUTION, ::size_t) \ F(cl_device_info, CL_DEVICE_ENDIAN_LITTLE, cl_bool) \ F(cl_device_info, CL_DEVICE_AVAILABLE, cl_bool) \ F(cl_device_info, CL_DEVICE_COMPILER_AVAILABLE, cl_bool) \ F(cl_device_info, CL_DEVICE_EXECUTION_CAPABILITIES, cl_device_exec_capabilities) \ F(cl_device_info, CL_DEVICE_QUEUE_PROPERTIES, cl_command_queue_properties) \ F(cl_device_info, CL_DEVICE_PLATFORM, cl_platform_id) \ F(cl_device_info, CL_DEVICE_NAME, STRING_CLASS) \ F(cl_device_info, CL_DEVICE_VENDOR, STRING_CLASS) \ F(cl_device_info, CL_DRIVER_VERSION, STRING_CLASS) \ F(cl_device_info, CL_DEVICE_PROFILE, STRING_CLASS) \ F(cl_device_info, CL_DEVICE_VERSION, STRING_CLASS) \ F(cl_device_info, CL_DEVICE_EXTENSIONS, STRING_CLASS) \ \ F(cl_context_info, CL_CONTEXT_REFERENCE_COUNT, cl_uint) \ F(cl_context_info, CL_CONTEXT_DEVICES, VECTOR_CLASS) \ F(cl_context_info, CL_CONTEXT_PROPERTIES, VECTOR_CLASS) \ \ F(cl_event_info, CL_EVENT_COMMAND_QUEUE, cl::CommandQueue) \ F(cl_event_info, CL_EVENT_COMMAND_TYPE, cl_command_type) \ F(cl_event_info, CL_EVENT_REFERENCE_COUNT, cl_uint) \ F(cl_event_info, CL_EVENT_COMMAND_EXECUTION_STATUS, cl_uint) \ \ F(cl_profiling_info, CL_PROFILING_COMMAND_QUEUED, cl_ulong) \ F(cl_profiling_info, CL_PROFILING_COMMAND_SUBMIT, cl_ulong) \ F(cl_profiling_info, CL_PROFILING_COMMAND_START, cl_ulong) \ F(cl_profiling_info, CL_PROFILING_COMMAND_END, cl_ulong) \ \ F(cl_mem_info, CL_MEM_TYPE, cl_mem_object_type) \ F(cl_mem_info, CL_MEM_FLAGS, cl_mem_flags) \ F(cl_mem_info, CL_MEM_SIZE, ::size_t) \ F(cl_mem_info, CL_MEM_HOST_PTR, void*) \ F(cl_mem_info, CL_MEM_MAP_COUNT, cl_uint) \ F(cl_mem_info, CL_MEM_REFERENCE_COUNT, cl_uint) \ F(cl_mem_info, CL_MEM_CONTEXT, cl::Context) \ \ F(cl_image_info, CL_IMAGE_FORMAT, cl_image_format) \ F(cl_image_info, CL_IMAGE_ELEMENT_SIZE, ::size_t) \ F(cl_image_info, CL_IMAGE_ROW_PITCH, ::size_t) \ F(cl_image_info, CL_IMAGE_SLICE_PITCH, ::size_t) \ F(cl_image_info, CL_IMAGE_WIDTH, ::size_t) \ F(cl_image_info, CL_IMAGE_HEIGHT, ::size_t) \ F(cl_image_info, CL_IMAGE_DEPTH, ::size_t) \ \ F(cl_sampler_info, CL_SAMPLER_REFERENCE_COUNT, cl_uint) \ F(cl_sampler_info, CL_SAMPLER_CONTEXT, cl::Context) \ F(cl_sampler_info, CL_SAMPLER_NORMALIZED_COORDS, cl_addressing_mode) \ F(cl_sampler_info, CL_SAMPLER_ADDRESSING_MODE, cl_filter_mode) \ F(cl_sampler_info, CL_SAMPLER_FILTER_MODE, cl_bool) \ \ F(cl_program_info, CL_PROGRAM_REFERENCE_COUNT, cl_uint) \ F(cl_program_info, CL_PROGRAM_CONTEXT, cl::Context) \ F(cl_program_info, CL_PROGRAM_NUM_DEVICES, cl_uint) \ F(cl_program_info, CL_PROGRAM_DEVICES, VECTOR_CLASS) \ F(cl_program_info, CL_PROGRAM_SOURCE, STRING_CLASS) \ F(cl_program_info, CL_PROGRAM_BINARY_SIZES, VECTOR_CLASS< ::size_t>) \ F(cl_program_info, CL_PROGRAM_BINARIES, VECTOR_CLASS) \ \ F(cl_program_build_info, CL_PROGRAM_BUILD_STATUS, cl_build_status) \ F(cl_program_build_info, CL_PROGRAM_BUILD_OPTIONS, STRING_CLASS) \ F(cl_program_build_info, CL_PROGRAM_BUILD_LOG, STRING_CLASS) \ \ F(cl_kernel_info, CL_KERNEL_FUNCTION_NAME, STRING_CLASS) \ F(cl_kernel_info, CL_KERNEL_NUM_ARGS, cl_uint) \ F(cl_kernel_info, CL_KERNEL_REFERENCE_COUNT, cl_uint) \ F(cl_kernel_info, CL_KERNEL_CONTEXT, cl::Context) \ F(cl_kernel_info, CL_KERNEL_PROGRAM, cl::Program) \ \ F(cl_kernel_work_group_info, CL_KERNEL_WORK_GROUP_SIZE, ::size_t) \ F(cl_kernel_work_group_info, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, cl::size_t<3>) \ F(cl_kernel_work_group_info, CL_KERNEL_LOCAL_MEM_SIZE, cl_ulong) \ \ F(cl_command_queue_info, CL_QUEUE_CONTEXT, cl::Context) \ F(cl_command_queue_info, CL_QUEUE_DEVICE, cl::Device) \ F(cl_command_queue_info, CL_QUEUE_REFERENCE_COUNT, cl_uint) \ F(cl_command_queue_info, CL_QUEUE_PROPERTIES, cl_command_queue_properties) #if defined(CL_VERSION_1_1) #define __PARAM_NAME_INFO_1_1(F) \ F(cl_context_info, CL_CONTEXT_NUM_DEVICES, cl_uint)\ F(cl_device_info, CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF, cl_uint) \ F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR, cl_uint) \ F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT, cl_uint) \ F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_INT, cl_uint) \ F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG, cl_uint) \ F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT, cl_uint) \ F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE, cl_uint) \ F(cl_device_info, CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF, cl_uint) \ F(cl_device_info, CL_DEVICE_DOUBLE_FP_CONFIG, cl_device_fp_config) \ F(cl_device_info, CL_DEVICE_HALF_FP_CONFIG, cl_device_fp_config) \ F(cl_device_info, CL_DEVICE_HOST_UNIFIED_MEMORY, cl_bool) \ F(cl_device_info, CL_DEVICE_OPENCL_C_VERSION, STRING_CLASS) \ \ F(cl_mem_info, CL_MEM_ASSOCIATED_MEMOBJECT, cl::Memory) \ F(cl_mem_info, CL_MEM_OFFSET, ::size_t) \ \ F(cl_kernel_work_group_info, CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE, ::size_t) \ F(cl_kernel_work_group_info, CL_KERNEL_PRIVATE_MEM_SIZE, cl_ulong) \ \ F(cl_event_info, CL_EVENT_CONTEXT, cl::Context) #endif // CL_VERSION_1_1 #if defined(CL_VERSION_1_2) #define __PARAM_NAME_INFO_1_2(F) \ F(cl_image_info, CL_IMAGE_BUFFER, cl::Buffer) \ \ F(cl_program_info, CL_PROGRAM_NUM_KERNELS, ::size_t) \ F(cl_program_info, CL_PROGRAM_KERNEL_NAMES, STRING_CLASS) \ \ F(cl_program_build_info, CL_PROGRAM_BINARY_TYPE, cl_program_binary_type) \ \ F(cl_kernel_info, CL_KERNEL_ATTRIBUTES, STRING_CLASS) \ \ F(cl_kernel_arg_info, CL_KERNEL_ARG_ADDRESS_QUALIFIER, cl_kernel_arg_address_qualifier) \ F(cl_kernel_arg_info, CL_KERNEL_ARG_ACCESS_QUALIFIER, cl_kernel_arg_access_qualifier) \ F(cl_kernel_arg_info, CL_KERNEL_ARG_TYPE_NAME, STRING_CLASS) \ F(cl_kernel_arg_info, CL_KERNEL_ARG_NAME, STRING_CLASS) \ \ F(cl_device_info, CL_DEVICE_PARENT_DEVICE, cl_device_id) \ F(cl_device_info, CL_DEVICE_PARTITION_PROPERTIES, VECTOR_CLASS) \ F(cl_device_info, CL_DEVICE_PARTITION_TYPE, VECTOR_CLASS) \ F(cl_device_info, CL_DEVICE_REFERENCE_COUNT, cl_uint) \ F(cl_device_info, CL_DEVICE_PREFERRED_INTEROP_USER_SYNC, ::size_t) \ F(cl_device_info, CL_DEVICE_PARTITION_AFFINITY_DOMAIN, cl_device_affinity_domain) \ F(cl_device_info, CL_DEVICE_BUILT_IN_KERNELS, STRING_CLASS) #endif // #if defined(CL_VERSION_1_2) #if defined(USE_CL_DEVICE_FISSION) #define __PARAM_NAME_DEVICE_FISSION(F) \ F(cl_device_info, CL_DEVICE_PARENT_DEVICE_EXT, cl_device_id) \ F(cl_device_info, CL_DEVICE_PARTITION_TYPES_EXT, VECTOR_CLASS) \ F(cl_device_info, CL_DEVICE_AFFINITY_DOMAINS_EXT, VECTOR_CLASS) \ F(cl_device_info, CL_DEVICE_REFERENCE_COUNT_EXT , cl_uint) \ F(cl_device_info, CL_DEVICE_PARTITION_STYLE_EXT, VECTOR_CLASS) #endif // USE_CL_DEVICE_FISSION template struct param_traits {}; #define __CL_DECLARE_PARAM_TRAITS(token, param_name, T) \ struct token; \ template<> \ struct param_traits \ { \ enum { value = param_name }; \ typedef T param_type; \ }; __PARAM_NAME_INFO_1_0(__CL_DECLARE_PARAM_TRAITS) #if defined(CL_VERSION_1_1) __PARAM_NAME_INFO_1_1(__CL_DECLARE_PARAM_TRAITS) #endif // CL_VERSION_1_1 #if defined(CL_VERSION_1_2) __PARAM_NAME_INFO_1_2(__CL_DECLARE_PARAM_TRAITS) #endif // CL_VERSION_1_1 #if defined(USE_CL_DEVICE_FISSION) __PARAM_NAME_DEVICE_FISSION(__CL_DECLARE_PARAM_TRAITS); #endif // USE_CL_DEVICE_FISSION #ifdef CL_PLATFORM_ICD_SUFFIX_KHR __CL_DECLARE_PARAM_TRAITS(cl_platform_info, CL_PLATFORM_ICD_SUFFIX_KHR, STRING_CLASS) #endif #ifdef CL_DEVICE_PROFILING_TIMER_OFFSET_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_PROFILING_TIMER_OFFSET_AMD, cl_ulong) #endif #ifdef CL_DEVICE_GLOBAL_FREE_MEMORY_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_FREE_MEMORY_AMD, VECTOR_CLASS< ::size_t>) #endif #ifdef CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD, cl_uint) #endif #ifdef CL_DEVICE_SIMD_WIDTH_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_WIDTH_AMD, cl_uint) #endif #ifdef CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD, cl_uint) #endif #ifdef CL_DEVICE_WAVEFRONT_WIDTH_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_WAVEFRONT_WIDTH_AMD, cl_uint) #endif #ifdef CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD, cl_uint) #endif #ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD, cl_uint) #endif #ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNEL_BANK_WIDTH_AMD, cl_uint) #endif #ifdef CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_LOCAL_MEM_SIZE_PER_COMPUTE_UNIT_AMD, cl_uint) #endif #ifdef CL_DEVICE_LOCAL_MEM_BANKS_AMD __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_LOCAL_MEM_BANKS_AMD, cl_uint) #endif #ifdef CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV, cl_uint) #endif #ifdef CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV, cl_uint) #endif #ifdef CL_DEVICE_REGISTERS_PER_BLOCK_NV __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_REGISTERS_PER_BLOCK_NV, cl_uint) #endif #ifdef CL_DEVICE_WARP_SIZE_NV __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_WARP_SIZE_NV, cl_uint) #endif #ifdef CL_DEVICE_GPU_OVERLAP_NV __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_GPU_OVERLAP_NV, cl_bool) #endif #ifdef CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV, cl_bool) #endif #ifdef CL_DEVICE_INTEGRATED_MEMORY_NV __CL_DECLARE_PARAM_TRAITS(cl_device_info, CL_DEVICE_INTEGRATED_MEMORY_NV, cl_bool) #endif // Convenience functions template inline cl_int getInfo(Func f, cl_uint name, T* param) { return getInfoHelper(f, name, param, 0); } template struct GetInfoFunctor0 { Func f_; const Arg0& arg0_; cl_int operator ()( cl_uint param, ::size_t size, void* value, ::size_t* size_ret) { return f_(arg0_, param, size, value, size_ret); } }; template struct GetInfoFunctor1 { Func f_; const Arg0& arg0_; const Arg1& arg1_; cl_int operator ()( cl_uint param, ::size_t size, void* value, ::size_t* size_ret) { return f_(arg0_, arg1_, param, size, value, size_ret); } }; template inline cl_int getInfo(Func f, const Arg0& arg0, cl_uint name, T* param) { GetInfoFunctor0 f0 = { f, arg0 }; return getInfoHelper(f0, name, param, 0); } template inline cl_int getInfo(Func f, const Arg0& arg0, const Arg1& arg1, cl_uint name, T* param) { GetInfoFunctor1 f0 = { f, arg0, arg1 }; return getInfoHelper(f0, name, param, 0); } template struct ReferenceHandler { }; #if defined(CL_VERSION_1_2) /** * OpenCL 1.2 devices do have retain/release. */ template <> struct ReferenceHandler { /** * Retain the device. * \param device A valid device created using createSubDevices * \return * CL_SUCCESS if the function executed successfully. * CL_INVALID_DEVICE if device was not a valid subdevice * CL_OUT_OF_RESOURCES * CL_OUT_OF_HOST_MEMORY */ static cl_int retain(cl_device_id device) { return ::clRetainDevice(device); } /** * Retain the device. * \param device A valid device created using createSubDevices * \return * CL_SUCCESS if the function executed successfully. * CL_INVALID_DEVICE if device was not a valid subdevice * CL_OUT_OF_RESOURCES * CL_OUT_OF_HOST_MEMORY */ static cl_int release(cl_device_id device) { return ::clReleaseDevice(device); } }; #else // #if defined(CL_VERSION_1_2) /** * OpenCL 1.1 devices do not have retain/release. */ template <> struct ReferenceHandler { // cl_device_id does not have retain(). static cl_int retain(cl_device_id) { return CL_SUCCESS; } // cl_device_id does not have release(). static cl_int release(cl_device_id) { return CL_SUCCESS; } }; #endif // #if defined(CL_VERSION_1_2) template <> struct ReferenceHandler { // cl_platform_id does not have retain(). static cl_int retain(cl_platform_id) { return CL_SUCCESS; } // cl_platform_id does not have release(). static cl_int release(cl_platform_id) { return CL_SUCCESS; } }; template <> struct ReferenceHandler { static cl_int retain(cl_context context) { return ::clRetainContext(context); } static cl_int release(cl_context context) { return ::clReleaseContext(context); } }; template <> struct ReferenceHandler { static cl_int retain(cl_command_queue queue) { return ::clRetainCommandQueue(queue); } static cl_int release(cl_command_queue queue) { return ::clReleaseCommandQueue(queue); } }; template <> struct ReferenceHandler { static cl_int retain(cl_mem memory) { return ::clRetainMemObject(memory); } static cl_int release(cl_mem memory) { return ::clReleaseMemObject(memory); } }; template <> struct ReferenceHandler { static cl_int retain(cl_sampler sampler) { return ::clRetainSampler(sampler); } static cl_int release(cl_sampler sampler) { return ::clReleaseSampler(sampler); } }; template <> struct ReferenceHandler { static cl_int retain(cl_program program) { return ::clRetainProgram(program); } static cl_int release(cl_program program) { return ::clReleaseProgram(program); } }; template <> struct ReferenceHandler { static cl_int retain(cl_kernel kernel) { return ::clRetainKernel(kernel); } static cl_int release(cl_kernel kernel) { return ::clReleaseKernel(kernel); } }; template <> struct ReferenceHandler { static cl_int retain(cl_event event) { return ::clRetainEvent(event); } static cl_int release(cl_event event) { return ::clReleaseEvent(event); } }; // Extracts version number with major in the upper 16 bits, minor in the lower 16 static cl_uint getVersion(const char *versionInfo) { int highVersion = 0; int lowVersion = 0; int index = 7; while(versionInfo[index] != '.' ) { highVersion *= 10; highVersion += versionInfo[index]-'0'; ++index; } ++index; while(versionInfo[index] != ' ' ) { lowVersion *= 10; lowVersion += versionInfo[index]-'0'; ++index; } return (highVersion << 16) | lowVersion; } static cl_uint getPlatformVersion(cl_platform_id platform) { ::size_t size = 0; clGetPlatformInfo(platform, CL_PLATFORM_VERSION, 0, NULL, &size); char *versionInfo = (char *) alloca(size); clGetPlatformInfo(platform, CL_PLATFORM_VERSION, size, &versionInfo[0], &size); return getVersion(versionInfo); } static cl_uint getDevicePlatformVersion(cl_device_id device) { cl_platform_id platform; clGetDeviceInfo(device, CL_DEVICE_PLATFORM, sizeof(platform), &platform, NULL); return getPlatformVersion(platform); } #if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) static cl_uint getContextPlatformVersion(cl_context context) { // The platform cannot be queried directly, so we first have to grab a // device and obtain its context ::size_t size = 0; clGetContextInfo(context, CL_CONTEXT_DEVICES, 0, NULL, &size); if (size == 0) return 0; cl_device_id *devices = (cl_device_id *) alloca(size); clGetContextInfo(context, CL_CONTEXT_DEVICES, size, devices, NULL); return getDevicePlatformVersion(devices[0]); } #endif // #if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) template class Wrapper { public: typedef T cl_type; protected: cl_type object_; public: Wrapper() : object_(NULL) { } Wrapper(const cl_type &obj) : object_(obj) { } ~Wrapper() { if (object_ != NULL) { release(); } } Wrapper(const Wrapper& rhs) { object_ = rhs.object_; if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } } Wrapper& operator = (const Wrapper& rhs) { if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } object_ = rhs.object_; if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } return *this; } Wrapper& operator = (const cl_type &rhs) { if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } object_ = rhs; return *this; } cl_type operator ()() const { return object_; } cl_type& operator ()() { return object_; } protected: template friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type); cl_int retain() const { return ReferenceHandler::retain(object_); } cl_int release() const { return ReferenceHandler::release(object_); } }; template <> class Wrapper { public: typedef cl_device_id cl_type; protected: cl_type object_; bool referenceCountable_; static bool isReferenceCountable(cl_device_id device) { bool retVal = false; if (device != NULL) { int version = getDevicePlatformVersion(device); if(version > ((1 << 16) + 1)) { retVal = true; } } return retVal; } public: Wrapper() : object_(NULL), referenceCountable_(false) { } Wrapper(const cl_type &obj) : object_(obj), referenceCountable_(false) { referenceCountable_ = isReferenceCountable(obj); } ~Wrapper() { if (object_ != NULL) { release(); } } Wrapper(const Wrapper& rhs) { object_ = rhs.object_; referenceCountable_ = isReferenceCountable(object_); if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } } Wrapper& operator = (const Wrapper& rhs) { if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } object_ = rhs.object_; referenceCountable_ = rhs.referenceCountable_; if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); } return *this; } Wrapper& operator = (const cl_type &rhs) { if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); } object_ = rhs; referenceCountable_ = isReferenceCountable(object_); return *this; } cl_type operator ()() const { return object_; } cl_type& operator ()() { return object_; } protected: template friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type); template friend inline cl_int getInfoHelper(Func, cl_uint, VECTOR_CLASS*, int, typename U::cl_type); cl_int retain() const { if( referenceCountable_ ) { return ReferenceHandler::retain(object_); } else { return CL_SUCCESS; } } cl_int release() const { if( referenceCountable_ ) { return ReferenceHandler::release(object_); } else { return CL_SUCCESS; } } }; } // namespace detail //! \endcond /*! \stuct ImageFormat * \brief Adds constructors and member functions for cl_image_format. * * \see cl_image_format */ struct ImageFormat : public cl_image_format { //! \brief Default constructor - performs no initialization. ImageFormat(){} //! \brief Initializing constructor. ImageFormat(cl_channel_order order, cl_channel_type type) { image_channel_order = order; image_channel_data_type = type; } //! \brief Assignment operator. ImageFormat& operator = (const ImageFormat& rhs) { if (this != &rhs) { this->image_channel_data_type = rhs.image_channel_data_type; this->image_channel_order = rhs.image_channel_order; } return *this; } }; /*! \brief Class interface for cl_device_id. * * \note Copies of these objects are inexpensive, since they don't 'own' * any underlying resources or data structures. * * \see cl_device_id */ class Device : public detail::Wrapper { public: //! \brief Default constructor - initializes to NULL. Device() : detail::Wrapper() { } /*! \brief Copy constructor. * * This simply copies the device ID value, which is an inexpensive operation. */ Device(const Device& device) : detail::Wrapper(device) { } /*! \brief Constructor from cl_device_id. * * This simply copies the device ID value, which is an inexpensive operation. */ Device(const cl_device_id &device) : detail::Wrapper(device) { } /*! \brief Returns the first device on the default context. * * \see Context::getDefault() */ static Device getDefault(cl_int * err = NULL); /*! \brief Assignment operator from Device. * * This simply copies the device ID value, which is an inexpensive operation. */ Device& operator = (const Device& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } /*! \brief Assignment operator from cl_device_id. * * This simply copies the device ID value, which is an inexpensive operation. */ Device& operator = (const cl_device_id& rhs) { detail::Wrapper::operator=(rhs); return *this; } //! \brief Wrapper for clGetDeviceInfo(). template cl_int getInfo(cl_device_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetDeviceInfo, object_, name, param), __GET_DEVICE_INFO_ERR); } //! \brief Wrapper for clGetDeviceInfo() that returns by value. template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_device_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } /** * CL 1.2 version */ #if defined(CL_VERSION_1_2) //! \brief Wrapper for clCreateSubDevicesEXT(). cl_int createSubDevices( const cl_device_partition_property * properties, VECTOR_CLASS* devices) { cl_uint n = 0; cl_int err = clCreateSubDevices(object_, properties, 0, NULL, &n); if (err != CL_SUCCESS) { return detail::errHandler(err, __CREATE_SUB_DEVICES); } cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); err = clCreateSubDevices(object_, properties, n, ids, NULL); if (err != CL_SUCCESS) { return detail::errHandler(err, __CREATE_SUB_DEVICES); } devices->assign(&ids[0], &ids[n]); return CL_SUCCESS; } #endif // #if defined(CL_VERSION_1_2) /** * CL 1.1 version that uses device fission. */ #if defined(CL_VERSION_1_1) #if defined(USE_CL_DEVICE_FISSION) cl_int createSubDevices( const cl_device_partition_property_ext * properties, VECTOR_CLASS* devices) { typedef CL_API_ENTRY cl_int ( CL_API_CALL * PFN_clCreateSubDevicesEXT)( cl_device_id /*in_device*/, const cl_device_partition_property_ext * /* properties */, cl_uint /*num_entries*/, cl_device_id * /*out_devices*/, cl_uint * /*num_devices*/ ) CL_EXT_SUFFIX__VERSION_1_1; static PFN_clCreateSubDevicesEXT pfn_clCreateSubDevicesEXT = NULL; __INIT_CL_EXT_FCN_PTR(clCreateSubDevicesEXT); cl_uint n = 0; cl_int err = pfn_clCreateSubDevicesEXT(object_, properties, 0, NULL, &n); if (err != CL_SUCCESS) { return detail::errHandler(err, __CREATE_SUB_DEVICES); } cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); err = pfn_clCreateSubDevicesEXT(object_, properties, n, ids, NULL); if (err != CL_SUCCESS) { return detail::errHandler(err, __CREATE_SUB_DEVICES); } devices->assign(&ids[0], &ids[n]); return CL_SUCCESS; } #endif // #if defined(USE_CL_DEVICE_FISSION) #endif // #if defined(CL_VERSION_1_1) }; /*! \brief Class interface for cl_platform_id. * * \note Copies of these objects are inexpensive, since they don't 'own' * any underlying resources or data structures. * * \see cl_platform_id */ class Platform : public detail::Wrapper { public: //! \brief Default constructor - initializes to NULL. Platform() : detail::Wrapper() { } /*! \brief Copy constructor. * * This simply copies the platform ID value, which is an inexpensive operation. */ Platform(const Platform& platform) : detail::Wrapper(platform) { } /*! \brief Constructor from cl_platform_id. * * This simply copies the platform ID value, which is an inexpensive operation. */ Platform(const cl_platform_id &platform) : detail::Wrapper(platform) { } /*! \brief Assignment operator from Platform. * * This simply copies the platform ID value, which is an inexpensive operation. */ Platform& operator = (const Platform& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } /*! \brief Assignment operator from cl_platform_id. * * This simply copies the platform ID value, which is an inexpensive operation. */ Platform& operator = (const cl_platform_id& rhs) { detail::Wrapper::operator=(rhs); return *this; } //! \brief Wrapper for clGetPlatformInfo(). cl_int getInfo(cl_platform_info name, STRING_CLASS* param) const { return detail::errHandler( detail::getInfo(&::clGetPlatformInfo, object_, name, param), __GET_PLATFORM_INFO_ERR); } //! \brief Wrapper for clGetPlatformInfo() that returns by value. template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_platform_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } /*! \brief Gets a list of devices for this platform. * * Wraps clGetDeviceIDs(). */ cl_int getDevices( cl_device_type type, VECTOR_CLASS* devices) const { cl_uint n = 0; if( devices == NULL ) { return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_DEVICE_IDS_ERR); } cl_int err = ::clGetDeviceIDs(object_, type, 0, NULL, &n); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_DEVICE_IDS_ERR); } cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); err = ::clGetDeviceIDs(object_, type, n, ids, NULL); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_DEVICE_IDS_ERR); } devices->assign(&ids[0], &ids[n]); return CL_SUCCESS; } #if defined(USE_DX_INTEROP) /*! \brief Get the list of available D3D10 devices. * * \param d3d_device_source. * * \param d3d_object. * * \param d3d_device_set. * * \param devices returns a vector of OpenCL D3D10 devices found. The cl::Device * values returned in devices can be used to identify a specific OpenCL * device. If \a devices argument is NULL, this argument is ignored. * * \return One of the following values: * - CL_SUCCESS if the function is executed successfully. * * The application can query specific capabilities of the OpenCL device(s) * returned by cl::getDevices. This can be used by the application to * determine which device(s) to use. * * \note In the case that exceptions are enabled and a return value * other than CL_SUCCESS is generated, then cl::Error exception is * generated. */ cl_int getDevices( cl_d3d10_device_source_khr d3d_device_source, void * d3d_object, cl_d3d10_device_set_khr d3d_device_set, VECTOR_CLASS* devices) const { typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clGetDeviceIDsFromD3D10KHR)( cl_platform_id platform, cl_d3d10_device_source_khr d3d_device_source, void * d3d_object, cl_d3d10_device_set_khr d3d_device_set, cl_uint num_entries, cl_device_id * devices, cl_uint* num_devices); if( devices == NULL ) { return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_DEVICE_IDS_ERR); } static PFN_clGetDeviceIDsFromD3D10KHR pfn_clGetDeviceIDsFromD3D10KHR = NULL; __INIT_CL_EXT_FCN_PTR_PLATFORM(object_, clGetDeviceIDsFromD3D10KHR); cl_uint n = 0; cl_int err = pfn_clGetDeviceIDsFromD3D10KHR( object_, d3d_device_source, d3d_object, d3d_device_set, 0, NULL, &n); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_DEVICE_IDS_ERR); } cl_device_id* ids = (cl_device_id*) alloca(n * sizeof(cl_device_id)); err = pfn_clGetDeviceIDsFromD3D10KHR( object_, d3d_device_source, d3d_object, d3d_device_set, n, ids, NULL); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_DEVICE_IDS_ERR); } devices->assign(&ids[0], &ids[n]); return CL_SUCCESS; } #endif /*! \brief Gets a list of available platforms. * * Wraps clGetPlatformIDs(). */ static cl_int get( VECTOR_CLASS* platforms) { cl_uint n = 0; if( platforms == NULL ) { return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_PLATFORM_IDS_ERR); } cl_int err = ::clGetPlatformIDs(0, NULL, &n); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); } cl_platform_id* ids = (cl_platform_id*) alloca( n * sizeof(cl_platform_id)); err = ::clGetPlatformIDs(n, ids, NULL); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); } platforms->assign(&ids[0], &ids[n]); return CL_SUCCESS; } /*! \brief Gets the first available platform. * * Wraps clGetPlatformIDs(), returning the first result. */ static cl_int get( Platform * platform) { cl_uint n = 0; if( platform == NULL ) { return detail::errHandler(CL_INVALID_ARG_VALUE, __GET_PLATFORM_IDS_ERR); } cl_int err = ::clGetPlatformIDs(0, NULL, &n); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); } cl_platform_id* ids = (cl_platform_id*) alloca( n * sizeof(cl_platform_id)); err = ::clGetPlatformIDs(n, ids, NULL); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_PLATFORM_IDS_ERR); } *platform = ids[0]; return CL_SUCCESS; } /*! \brief Gets the first available platform, returning it by value. * * Wraps clGetPlatformIDs(), returning the first result. */ static Platform get( cl_int * errResult = NULL) { Platform platform; cl_uint n = 0; cl_int err = ::clGetPlatformIDs(0, NULL, &n); if (err != CL_SUCCESS) { detail::errHandler(err, __GET_PLATFORM_IDS_ERR); if (errResult != NULL) { *errResult = err; } } cl_platform_id* ids = (cl_platform_id*) alloca( n * sizeof(cl_platform_id)); err = ::clGetPlatformIDs(n, ids, NULL); if (err != CL_SUCCESS) { detail::errHandler(err, __GET_PLATFORM_IDS_ERR); } if (errResult != NULL) { *errResult = err; } return ids[0]; } static Platform getDefault( cl_int *errResult = NULL ) { return get(errResult); } #if defined(CL_VERSION_1_2) //! \brief Wrapper for clUnloadCompiler(). cl_int unloadCompiler() { return ::clUnloadPlatformCompiler(object_); } #endif // #if defined(CL_VERSION_1_2) }; // class Platform /** * Deprecated APIs for 1.2 */ #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) /** * Unload the OpenCL compiler. * \note Deprecated for OpenCL 1.2. Use Platform::unloadCompiler instead. */ inline CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int UnloadCompiler() CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; inline cl_int UnloadCompiler() { return ::clUnloadCompiler(); } #endif // #if defined(CL_VERSION_1_1) /*! \brief Class interface for cl_context. * * \note Copies of these objects are shallow, meaning that the copy will refer * to the same underlying cl_context as the original. For details, see * clRetainContext() and clReleaseContext(). * * \see cl_context */ class Context : public detail::Wrapper { private: static volatile int default_initialized_; static Context default_; static volatile cl_int default_error_; public: /*! \brief Destructor. * * This calls clReleaseContext() on the value held by this instance. */ ~Context() { } /*! \brief Constructs a context including a list of specified devices. * * Wraps clCreateContext(). */ Context( const VECTOR_CLASS& devices, cl_context_properties* properties = NULL, void (CL_CALLBACK * notifyFptr)( const char *, const void *, ::size_t, void *) = NULL, void* data = NULL, cl_int* err = NULL) { cl_int error; ::size_t numDevices = devices.size(); cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { deviceIDs[deviceIndex] = (devices[deviceIndex])(); } object_ = ::clCreateContext( properties, (cl_uint) numDevices, deviceIDs, notifyFptr, data, &error); detail::errHandler(error, __CREATE_CONTEXT_ERR); if (err != NULL) { *err = error; } } Context( const Device& device, cl_context_properties* properties = NULL, void (CL_CALLBACK * notifyFptr)( const char *, const void *, ::size_t, void *) = NULL, void* data = NULL, cl_int* err = NULL) { cl_int error; cl_device_id deviceID = device(); object_ = ::clCreateContext( properties, 1, &deviceID, notifyFptr, data, &error); detail::errHandler(error, __CREATE_CONTEXT_ERR); if (err != NULL) { *err = error; } } /*! \brief Constructs a context including all or a subset of devices of a specified type. * * Wraps clCreateContextFromType(). */ Context( cl_device_type type, cl_context_properties* properties = NULL, void (CL_CALLBACK * notifyFptr)( const char *, const void *, ::size_t, void *) = NULL, void* data = NULL, cl_int* err = NULL) { cl_int error; #if !defined(__APPLE__) || !defined(__MACOS) cl_context_properties prop[4] = {CL_CONTEXT_PLATFORM, 0, 0, 0 }; if (properties == NULL) { // Get a valid platform ID as we cannot send in a blank one VECTOR_CLASS platforms; error = Platform::get(&platforms); if (error != CL_SUCCESS) { detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); if (err != NULL) { *err = error; } return; } // Check the platforms we found for a device of our specified type cl_context_properties platform_id = 0; for (unsigned int i = 0; i < platforms.size(); i++) { VECTOR_CLASS devices; #if defined(__CL_ENABLE_EXCEPTIONS) try { #endif error = platforms[i].getDevices(type, &devices); #if defined(__CL_ENABLE_EXCEPTIONS) } catch (Error) {} // Catch if exceptions are enabled as we don't want to exit if first platform has no devices of type // We do error checking next anyway, and can throw there if needed #endif // Only squash CL_SUCCESS and CL_DEVICE_NOT_FOUND if (error != CL_SUCCESS && error != CL_DEVICE_NOT_FOUND) { detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); if (err != NULL) { *err = error; } } if (devices.size() > 0) { platform_id = (cl_context_properties)platforms[i](); break; } } if (platform_id == 0) { detail::errHandler(CL_DEVICE_NOT_FOUND, __CREATE_CONTEXT_FROM_TYPE_ERR); if (err != NULL) { *err = CL_DEVICE_NOT_FOUND; } return; } prop[1] = platform_id; properties = &prop[0]; } #endif object_ = ::clCreateContextFromType( properties, type, notifyFptr, data, &error); detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR); if (err != NULL) { *err = error; } } /*! \brief Returns a singleton context including all devices of CL_DEVICE_TYPE_DEFAULT. * * \note All calls to this function return the same cl_context as the first. */ static Context getDefault(cl_int * err = NULL) { int state = detail::compare_exchange( &default_initialized_, __DEFAULT_BEING_INITIALIZED, __DEFAULT_NOT_INITIALIZED); if (state & __DEFAULT_INITIALIZED) { if (err != NULL) { *err = default_error_; } return default_; } if (state & __DEFAULT_BEING_INITIALIZED) { // Assume writes will propagate eventually... while(default_initialized_ != __DEFAULT_INITIALIZED) { detail::fence(); } if (err != NULL) { *err = default_error_; } return default_; } cl_int error; default_ = Context( CL_DEVICE_TYPE_DEFAULT, NULL, NULL, NULL, &error); detail::fence(); default_error_ = error; // Assume writes will propagate eventually... default_initialized_ = __DEFAULT_INITIALIZED; detail::fence(); if (err != NULL) { *err = default_error_; } return default_; } //! \brief Default constructor - initializes to NULL. Context() : detail::Wrapper() { } /*! \brief Copy constructor. * * This calls clRetainContext() on the parameter's cl_context. */ Context(const Context& context) : detail::Wrapper(context) { } /*! \brief Constructor from cl_context - takes ownership. * * This effectively transfers ownership of a refcount on the cl_context * into the new Context object. */ __CL_EXPLICIT_CONSTRUCTORS Context(const cl_context& context) : detail::Wrapper(context) { } /*! \brief Assignment operator from Context. * * This calls clRetainContext() on the parameter and clReleaseContext() on * the previous value held by this instance. */ Context& operator = (const Context& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } /*! \brief Assignment operator from cl_context - takes ownership. * * This effectively transfers ownership of a refcount on the rhs and calls * clReleaseContext() on the value previously held by this instance. */ Context& operator = (const cl_context& rhs) { detail::Wrapper::operator=(rhs); return *this; } //! \brief Wrapper for clGetContextInfo(). template cl_int getInfo(cl_context_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetContextInfo, object_, name, param), __GET_CONTEXT_INFO_ERR); } //! \brief Wrapper for clGetContextInfo() that returns by value. template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_context_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } /*! \brief Gets a list of supported image formats. * * Wraps clGetSupportedImageFormats(). */ cl_int getSupportedImageFormats( cl_mem_flags flags, cl_mem_object_type type, VECTOR_CLASS* formats) const { cl_uint numEntries; cl_int err = ::clGetSupportedImageFormats( object_, flags, type, 0, NULL, &numEntries); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR); } ImageFormat* value = (ImageFormat*) alloca(numEntries * sizeof(ImageFormat)); err = ::clGetSupportedImageFormats( object_, flags, type, numEntries, (cl_image_format*) value, NULL); if (err != CL_SUCCESS) { return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR); } formats->assign(&value[0], &value[numEntries]); return CL_SUCCESS; } }; inline Device Device::getDefault(cl_int * err) { cl_int error; Device device; Context context = Context::getDefault(&error); detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); if (error != CL_SUCCESS) { if (err != NULL) { *err = error; } } else { device = context.getInfo()[0]; if (err != NULL) { *err = CL_SUCCESS; } } return device; } #ifdef _WIN32 __declspec(selectany) volatile int Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED; __declspec(selectany) Context Context::default_; __declspec(selectany) volatile cl_int Context::default_error_ = CL_SUCCESS; #else __attribute__((weak)) volatile int Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED; __attribute__((weak)) Context Context::default_; __attribute__((weak)) volatile cl_int Context::default_error_ = CL_SUCCESS; #endif /*! \brief Class interface for cl_event. * * \note Copies of these objects are shallow, meaning that the copy will refer * to the same underlying cl_event as the original. For details, see * clRetainEvent() and clReleaseEvent(). * * \see cl_event */ class Event : public detail::Wrapper { public: /*! \brief Destructor. * * This calls clReleaseEvent() on the value held by this instance. */ ~Event() { } //! \brief Default constructor - initializes to NULL. Event() : detail::Wrapper() { } /*! \brief Copy constructor. * * This calls clRetainEvent() on the parameter's cl_event. */ Event(const Event& event) : detail::Wrapper(event) { } /*! \brief Constructor from cl_event - takes ownership. * * This effectively transfers ownership of a refcount on the cl_event * into the new Event object. */ Event(const cl_event& event) : detail::Wrapper(event) { } /*! \brief Assignment operator from cl_event - takes ownership. * * This effectively transfers ownership of a refcount on the rhs and calls * clReleaseEvent() on the value previously held by this instance. */ Event& operator = (const Event& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } /*! \brief Assignment operator from cl_event. * * This calls clRetainEvent() on the parameter and clReleaseEvent() on * the previous value held by this instance. */ Event& operator = (const cl_event& rhs) { detail::Wrapper::operator=(rhs); return *this; } //! \brief Wrapper for clGetEventInfo(). template cl_int getInfo(cl_event_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetEventInfo, object_, name, param), __GET_EVENT_INFO_ERR); } //! \brief Wrapper for clGetEventInfo() that returns by value. template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_event_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } //! \brief Wrapper for clGetEventProfilingInfo(). template cl_int getProfilingInfo(cl_profiling_info name, T* param) const { return detail::errHandler(detail::getInfo( &::clGetEventProfilingInfo, object_, name, param), __GET_EVENT_PROFILE_INFO_ERR); } //! \brief Wrapper for clGetEventProfilingInfo() that returns by value. template typename detail::param_traits::param_type getProfilingInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_profiling_info, name>::param_type param; cl_int result = getProfilingInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } /*! \brief Blocks the calling thread until this event completes. * * Wraps clWaitForEvents(). */ cl_int wait() const { return detail::errHandler( ::clWaitForEvents(1, &object_), __WAIT_FOR_EVENTS_ERR); } #if defined(CL_VERSION_1_1) /*! \brief Registers a user callback function for a specific command execution status. * * Wraps clSetEventCallback(). */ cl_int setCallback( cl_int type, void (CL_CALLBACK * pfn_notify)(cl_event, cl_int, void *), void * user_data = NULL) { return detail::errHandler( ::clSetEventCallback( object_, type, pfn_notify, user_data), __SET_EVENT_CALLBACK_ERR); } #endif /*! \brief Blocks the calling thread until every event specified is complete. * * Wraps clWaitForEvents(). */ static cl_int waitForEvents(const VECTOR_CLASS& events) { return detail::errHandler( ::clWaitForEvents( (cl_uint) events.size(), (cl_event*)&events.front()), __WAIT_FOR_EVENTS_ERR); } }; #if defined(CL_VERSION_1_1) /*! \brief Class interface for user events (a subset of cl_event's). * * See Event for details about copy semantics, etc. */ class UserEvent : public Event { public: /*! \brief Constructs a user event on a given context. * * Wraps clCreateUserEvent(). */ UserEvent( const Context& context, cl_int * err = NULL) { cl_int error; object_ = ::clCreateUserEvent( context(), &error); detail::errHandler(error, __CREATE_USER_EVENT_ERR); if (err != NULL) { *err = error; } } //! \brief Default constructor - initializes to NULL. UserEvent() : Event() { } //! \brief Copy constructor - performs shallow copy. UserEvent(const UserEvent& event) : Event(event) { } //! \brief Assignment Operator - performs shallow copy. UserEvent& operator = (const UserEvent& rhs) { if (this != &rhs) { Event::operator=(rhs); } return *this; } /*! \brief Sets the execution status of a user event object. * * Wraps clSetUserEventStatus(). */ cl_int setStatus(cl_int status) { return detail::errHandler( ::clSetUserEventStatus(object_,status), __SET_USER_EVENT_STATUS_ERR); } }; #endif /*! \brief Blocks the calling thread until every event specified is complete. * * Wraps clWaitForEvents(). */ inline static cl_int WaitForEvents(const VECTOR_CLASS& events) { return detail::errHandler( ::clWaitForEvents( (cl_uint) events.size(), (cl_event*)&events.front()), __WAIT_FOR_EVENTS_ERR); } /*! \brief Class interface for cl_mem. * * \note Copies of these objects are shallow, meaning that the copy will refer * to the same underlying cl_mem as the original. For details, see * clRetainMemObject() and clReleaseMemObject(). * * \see cl_mem */ class Memory : public detail::Wrapper { public: /*! \brief Destructor. * * This calls clReleaseMemObject() on the value held by this instance. */ ~Memory() {} //! \brief Default constructor - initializes to NULL. Memory() : detail::Wrapper() { } /*! \brief Copy constructor - performs shallow copy. * * This calls clRetainMemObject() on the parameter's cl_mem. */ Memory(const Memory& memory) : detail::Wrapper(memory) { } /*! \brief Constructor from cl_mem - takes ownership. * * This effectively transfers ownership of a refcount on the cl_mem * into the new Memory object. */ __CL_EXPLICIT_CONSTRUCTORS Memory(const cl_mem& memory) : detail::Wrapper(memory) { } /*! \brief Assignment operator from Memory. * * This calls clRetainMemObject() on the parameter and clReleaseMemObject() * on the previous value held by this instance. */ Memory& operator = (const Memory& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } /*! \brief Assignment operator from cl_mem - takes ownership. * * This effectively transfers ownership of a refcount on the rhs and calls * clReleaseMemObject() on the value previously held by this instance. */ Memory& operator = (const cl_mem& rhs) { detail::Wrapper::operator=(rhs); return *this; } //! \brief Wrapper for clGetMemObjectInfo(). template cl_int getInfo(cl_mem_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetMemObjectInfo, object_, name, param), __GET_MEM_OBJECT_INFO_ERR); } //! \brief Wrapper for clGetMemObjectInfo() that returns by value. template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_mem_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } #if defined(CL_VERSION_1_1) /*! \brief Registers a callback function to be called when the memory object * is no longer needed. * * Wraps clSetMemObjectDestructorCallback(). * * Repeated calls to this function, for a given cl_mem value, will append * to the list of functions called (in reverse order) when memory object's * resources are freed and the memory object is deleted. * * \note * The registered callbacks are associated with the underlying cl_mem * value - not the Memory class instance. */ cl_int setDestructorCallback( void (CL_CALLBACK * pfn_notify)(cl_mem, void *), void * user_data = NULL) { return detail::errHandler( ::clSetMemObjectDestructorCallback( object_, pfn_notify, user_data), __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR); } #endif }; // Pre-declare copy functions class Buffer; template< typename IteratorType > cl_int copy( IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ); template< typename IteratorType > cl_int copy( const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ); template< typename IteratorType > cl_int copy( const CommandQueue &queue, IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ); template< typename IteratorType > cl_int copy( const CommandQueue &queue, const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ); /*! \brief Class interface for Buffer Memory Objects. * * See Memory for details about copy semantics, etc. * * \see Memory */ class Buffer : public Memory { public: /*! \brief Constructs a Buffer in a specified context. * * Wraps clCreateBuffer(). * * \param host_ptr Storage to be used if the CL_MEM_USE_HOST_PTR flag was * specified. Note alignment & exclusivity requirements. */ Buffer( const Context& context, cl_mem_flags flags, ::size_t size, void* host_ptr = NULL, cl_int* err = NULL) { cl_int error; object_ = ::clCreateBuffer(context(), flags, size, host_ptr, &error); detail::errHandler(error, __CREATE_BUFFER_ERR); if (err != NULL) { *err = error; } } /*! \brief Constructs a Buffer in the default context. * * Wraps clCreateBuffer(). * * \param host_ptr Storage to be used if the CL_MEM_USE_HOST_PTR flag was * specified. Note alignment & exclusivity requirements. * * \see Context::getDefault() */ Buffer( cl_mem_flags flags, ::size_t size, void* host_ptr = NULL, cl_int* err = NULL) { cl_int error; Context context = Context::getDefault(err); object_ = ::clCreateBuffer(context(), flags, size, host_ptr, &error); detail::errHandler(error, __CREATE_BUFFER_ERR); if (err != NULL) { *err = error; } } /*! * \brief Construct a Buffer from a host container via iterators. * IteratorType must be random access. * If useHostPtr is specified iterators must represent contiguous data. */ template< typename IteratorType > Buffer( IteratorType startIterator, IteratorType endIterator, bool readOnly, bool useHostPtr = false, cl_int* err = NULL) { typedef typename std::iterator_traits::value_type DataType; cl_int error; cl_mem_flags flags = 0; if( readOnly ) { flags |= CL_MEM_READ_ONLY; } else { flags |= CL_MEM_READ_WRITE; } if( useHostPtr ) { flags |= CL_MEM_USE_HOST_PTR; } ::size_t size = sizeof(DataType)*(endIterator - startIterator); Context context = Context::getDefault(err); if( useHostPtr ) { object_ = ::clCreateBuffer(context(), flags, size, static_cast(&*startIterator), &error); } else { object_ = ::clCreateBuffer(context(), flags, size, 0, &error); } detail::errHandler(error, __CREATE_BUFFER_ERR); if (err != NULL) { *err = error; } if( !useHostPtr ) { error = cl::copy(startIterator, endIterator, *this); detail::errHandler(error, __CREATE_BUFFER_ERR); if (err != NULL) { *err = error; } } } /*! * \brief Construct a Buffer from a host container via iterators using a specified context. * IteratorType must be random access. * If useHostPtr is specified iterators must represent contiguous data. */ template< typename IteratorType > Buffer(const Context &context, IteratorType startIterator, IteratorType endIterator, bool readOnly, bool useHostPtr = false, cl_int* err = NULL); //! \brief Default constructor - initializes to NULL. Buffer() : Memory() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ Buffer(const Buffer& buffer) : Memory(buffer) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS Buffer(const cl_mem& buffer) : Memory(buffer) { } /*! \brief Assignment from Buffer - performs shallow copy. * * See Memory for further details. */ Buffer& operator = (const Buffer& rhs) { if (this != &rhs) { Memory::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ Buffer& operator = (const cl_mem& rhs) { Memory::operator=(rhs); return *this; } #if defined(CL_VERSION_1_1) /*! \brief Creates a new buffer object from this. * * Wraps clCreateSubBuffer(). */ Buffer createSubBuffer( cl_mem_flags flags, cl_buffer_create_type buffer_create_type, const void * buffer_create_info, cl_int * err = NULL) { Buffer result; cl_int error; result.object_ = ::clCreateSubBuffer( object_, flags, buffer_create_type, buffer_create_info, &error); detail::errHandler(error, __CREATE_SUBBUFFER_ERR); if (err != NULL) { *err = error; } return result; } #endif }; #if defined (USE_DX_INTEROP) /*! \brief Class interface for creating OpenCL buffers from ID3D10Buffer's. * * This is provided to facilitate interoperability with Direct3D. * * See Memory for details about copy semantics, etc. * * \see Memory */ class BufferD3D10 : public Buffer { public: typedef CL_API_ENTRY cl_mem (CL_API_CALL *PFN_clCreateFromD3D10BufferKHR)( cl_context context, cl_mem_flags flags, ID3D10Buffer* buffer, cl_int* errcode_ret); /*! \brief Constructs a BufferD3D10, in a specified context, from a * given ID3D10Buffer. * * Wraps clCreateFromD3D10BufferKHR(). */ BufferD3D10( const Context& context, cl_mem_flags flags, ID3D10Buffer* bufobj, cl_int * err = NULL) { static PFN_clCreateFromD3D10BufferKHR pfn_clCreateFromD3D10BufferKHR = NULL; #if defined(CL_VERSION_1_2) vector props = context.getInfo(); cl_platform platform = -1; for( int i = 0; i < props.size(); ++i ) { if( props[i] == CL_CONTEXT_PLATFORM ) { platform = props[i+1]; } } __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clCreateFromD3D10BufferKHR); #endif #if defined(CL_VERSION_1_1) __INIT_CL_EXT_FCN_PTR(clCreateFromD3D10BufferKHR); #endif cl_int error; object_ = pfn_clCreateFromD3D10BufferKHR( context(), flags, bufobj, &error); detail::errHandler(error, __CREATE_GL_BUFFER_ERR); if (err != NULL) { *err = error; } } //! \brief Default constructor - initializes to NULL. BufferD3D10() : Buffer() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ BufferD3D10(const BufferD3D10& buffer) : Buffer(buffer) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS BufferD3D10(const cl_mem& buffer) : Buffer(buffer) { } /*! \brief Assignment from BufferD3D10 - performs shallow copy. * * See Memory for further details. */ BufferD3D10& operator = (const BufferD3D10& rhs) { if (this != &rhs) { Buffer::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ BufferD3D10& operator = (const cl_mem& rhs) { Buffer::operator=(rhs); return *this; } }; #endif /*! \brief Class interface for GL Buffer Memory Objects. * * This is provided to facilitate interoperability with OpenGL. * * See Memory for details about copy semantics, etc. * * \see Memory */ class BufferGL : public Buffer { public: /*! \brief Constructs a BufferGL in a specified context, from a given * GL buffer. * * Wraps clCreateFromGLBuffer(). */ BufferGL( const Context& context, cl_mem_flags flags, GLuint bufobj, cl_int * err = NULL) { cl_int error; object_ = ::clCreateFromGLBuffer( context(), flags, bufobj, &error); detail::errHandler(error, __CREATE_GL_BUFFER_ERR); if (err != NULL) { *err = error; } } //! \brief Default constructor - initializes to NULL. BufferGL() : Buffer() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ BufferGL(const BufferGL& buffer) : Buffer(buffer) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS BufferGL(const cl_mem& buffer) : Buffer(buffer) { } /*! \brief Assignment from BufferGL - performs shallow copy. * * See Memory for further details. */ BufferGL& operator = (const BufferGL& rhs) { if (this != &rhs) { Buffer::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ BufferGL& operator = (const cl_mem& rhs) { Buffer::operator=(rhs); return *this; } //! \brief Wrapper for clGetGLObjectInfo(). cl_int getObjectInfo( cl_gl_object_type *type, GLuint * gl_object_name) { return detail::errHandler( ::clGetGLObjectInfo(object_,type,gl_object_name), __GET_GL_OBJECT_INFO_ERR); } }; /*! \brief Class interface for GL Render Buffer Memory Objects. * * This is provided to facilitate interoperability with OpenGL. * * See Memory for details about copy semantics, etc. * * \see Memory */ class BufferRenderGL : public Buffer { public: /*! \brief Constructs a BufferRenderGL in a specified context, from a given * GL Renderbuffer. * * Wraps clCreateFromGLRenderbuffer(). */ BufferRenderGL( const Context& context, cl_mem_flags flags, GLuint bufobj, cl_int * err = NULL) { cl_int error; object_ = ::clCreateFromGLRenderbuffer( context(), flags, bufobj, &error); detail::errHandler(error, __CREATE_GL_RENDER_BUFFER_ERR); if (err != NULL) { *err = error; } } //! \brief Default constructor - initializes to NULL. BufferRenderGL() : Buffer() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ BufferRenderGL(const BufferGL& buffer) : Buffer(buffer) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS BufferRenderGL(const cl_mem& buffer) : Buffer(buffer) { } /*! \brief Assignment from BufferGL - performs shallow copy. * * See Memory for further details. */ BufferRenderGL& operator = (const BufferRenderGL& rhs) { if (this != &rhs) { Buffer::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ BufferRenderGL& operator = (const cl_mem& rhs) { Buffer::operator=(rhs); return *this; } //! \brief Wrapper for clGetGLObjectInfo(). cl_int getObjectInfo( cl_gl_object_type *type, GLuint * gl_object_name) { return detail::errHandler( ::clGetGLObjectInfo(object_,type,gl_object_name), __GET_GL_OBJECT_INFO_ERR); } }; /*! \brief C++ base class for Image Memory objects. * * See Memory for details about copy semantics, etc. * * \see Memory */ class Image : public Memory { protected: //! \brief Default constructor - initializes to NULL. Image() : Memory() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ Image(const Image& image) : Memory(image) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS Image(const cl_mem& image) : Memory(image) { } /*! \brief Assignment from Image - performs shallow copy. * * See Memory for further details. */ Image& operator = (const Image& rhs) { if (this != &rhs) { Memory::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ Image& operator = (const cl_mem& rhs) { Memory::operator=(rhs); return *this; } public: //! \brief Wrapper for clGetImageInfo(). template cl_int getImageInfo(cl_image_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetImageInfo, object_, name, param), __GET_IMAGE_INFO_ERR); } //! \brief Wrapper for clGetImageInfo() that returns by value. template typename detail::param_traits::param_type getImageInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_image_info, name>::param_type param; cl_int result = getImageInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } }; #if defined(CL_VERSION_1_2) /*! \brief Class interface for 1D Image Memory objects. * * See Memory for details about copy semantics, etc. * * \see Memory */ class Image1D : public Image { public: /*! \brief Constructs a 1D Image in a specified context. * * Wraps clCreateImage(). */ Image1D( const Context& context, cl_mem_flags flags, ImageFormat format, ::size_t width, void* host_ptr = NULL, cl_int* err = NULL) { cl_int error; cl_image_desc desc = { CL_MEM_OBJECT_IMAGE1D, width, 0, 0, 0, 0, 0, 0, 0, 0 }; object_ = ::clCreateImage( context(), flags, &format, &desc, host_ptr, &error); detail::errHandler(error, __CREATE_IMAGE_ERR); if (err != NULL) { *err = error; } } //! \brief Default constructor - initializes to NULL. Image1D() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ Image1D(const Image1D& image1D) : Image(image1D) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS Image1D(const cl_mem& image1D) : Image(image1D) { } /*! \brief Assignment from Image1D - performs shallow copy. * * See Memory for further details. */ Image1D& operator = (const Image1D& rhs) { if (this != &rhs) { Image::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ Image1D& operator = (const cl_mem& rhs) { Image::operator=(rhs); return *this; } }; /*! \class Image1DBuffer * \brief Image interface for 1D buffer images. */ class Image1DBuffer : public Image { public: Image1DBuffer( const Context& context, cl_mem_flags flags, ImageFormat format, ::size_t width, const Buffer &buffer, cl_int* err = NULL) { cl_int error; cl_image_desc desc = { CL_MEM_OBJECT_IMAGE1D_BUFFER, width, 0, 0, 0, 0, 0, 0, 0, buffer() }; object_ = ::clCreateImage( context(), flags, &format, &desc, NULL, &error); detail::errHandler(error, __CREATE_IMAGE_ERR); if (err != NULL) { *err = error; } } Image1DBuffer() { } Image1DBuffer(const Image1DBuffer& image1D) : Image(image1D) { } __CL_EXPLICIT_CONSTRUCTORS Image1DBuffer(const cl_mem& image1D) : Image(image1D) { } Image1DBuffer& operator = (const Image1DBuffer& rhs) { if (this != &rhs) { Image::operator=(rhs); } return *this; } Image1DBuffer& operator = (const cl_mem& rhs) { Image::operator=(rhs); return *this; } }; /*! \class Image1DArray * \brief Image interface for arrays of 1D images. */ class Image1DArray : public Image { public: Image1DArray( const Context& context, cl_mem_flags flags, ImageFormat format, ::size_t arraySize, ::size_t width, ::size_t rowPitch, void* host_ptr = NULL, cl_int* err = NULL) { cl_int error; cl_image_desc desc = { CL_MEM_OBJECT_IMAGE1D_ARRAY, width, 0, 0, // height, depth (unused) arraySize, rowPitch, 0, 0, 0, 0 }; object_ = ::clCreateImage( context(), flags, &format, &desc, host_ptr, &error); detail::errHandler(error, __CREATE_IMAGE_ERR); if (err != NULL) { *err = error; } } Image1DArray() { } Image1DArray(const Image1DArray& imageArray) : Image(imageArray) { } __CL_EXPLICIT_CONSTRUCTORS Image1DArray(const cl_mem& imageArray) : Image(imageArray) { } Image1DArray& operator = (const Image1DArray& rhs) { if (this != &rhs) { Image::operator=(rhs); } return *this; } Image1DArray& operator = (const cl_mem& rhs) { Image::operator=(rhs); return *this; } }; #endif // #if defined(CL_VERSION_1_2) /*! \brief Class interface for 2D Image Memory objects. * * See Memory for details about copy semantics, etc. * * \see Memory */ class Image2D : public Image { public: /*! \brief Constructs a 1D Image in a specified context. * * Wraps clCreateImage(). */ Image2D( const Context& context, cl_mem_flags flags, ImageFormat format, ::size_t width, ::size_t height, ::size_t row_pitch = 0, void* host_ptr = NULL, cl_int* err = NULL) { cl_int error; bool useCreateImage; #if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) // Run-time decision based on the actual platform { cl_uint version = detail::getContextPlatformVersion(context()); useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above } #elif defined(CL_VERSION_1_2) useCreateImage = true; #else useCreateImage = false; #endif #if defined(CL_VERSION_1_2) if (useCreateImage) { cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, // depth, array size (unused) row_pitch, 0, 0, 0, 0 }; object_ = ::clCreateImage( context(), flags, &format, &desc, host_ptr, &error); detail::errHandler(error, __CREATE_IMAGE_ERR); if (err != NULL) { *err = error; } } #endif // #if defined(CL_VERSION_1_2) #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) if (!useCreateImage) { object_ = ::clCreateImage2D( context(), flags,&format, width, height, row_pitch, host_ptr, &error); detail::errHandler(error, __CREATE_IMAGE2D_ERR); if (err != NULL) { *err = error; } } #endif // #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) } //! \brief Default constructor - initializes to NULL. Image2D() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ Image2D(const Image2D& image2D) : Image(image2D) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS Image2D(const cl_mem& image2D) : Image(image2D) { } /*! \brief Assignment from Image2D - performs shallow copy. * * See Memory for further details. */ Image2D& operator = (const Image2D& rhs) { if (this != &rhs) { Image::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ Image2D& operator = (const cl_mem& rhs) { Image::operator=(rhs); return *this; } }; #if !defined(CL_VERSION_1_2) /*! \brief Class interface for GL 2D Image Memory objects. * * This is provided to facilitate interoperability with OpenGL. * * See Memory for details about copy semantics, etc. * * \see Memory * \note Deprecated for OpenCL 1.2. Please use ImageGL instead. */ class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED Image2DGL CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED : public Image2D { public: /*! \brief Constructs an Image2DGL in a specified context, from a given * GL Texture. * * Wraps clCreateFromGLTexture2D(). */ Image2DGL( const Context& context, cl_mem_flags flags, GLenum target, GLint miplevel, GLuint texobj, cl_int * err = NULL) { cl_int error; object_ = ::clCreateFromGLTexture2D( context(), flags, target, miplevel, texobj, &error); detail::errHandler(error, __CREATE_GL_TEXTURE_2D_ERR); if (err != NULL) { *err = error; } } //! \brief Default constructor - initializes to NULL. Image2DGL() : Image2D() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ Image2DGL(const Image2DGL& image) : Image2D(image) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS Image2DGL(const cl_mem& image) : Image2D(image) { } /*! \brief Assignment from Image2DGL - performs shallow copy. * * See Memory for further details. */ Image2DGL& operator = (const Image2DGL& rhs) { if (this != &rhs) { Image2D::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ Image2DGL& operator = (const cl_mem& rhs) { Image2D::operator=(rhs); return *this; } }; #endif // #if !defined(CL_VERSION_1_2) #if defined(CL_VERSION_1_2) /*! \class Image2DArray * \brief Image interface for arrays of 2D images. */ class Image2DArray : public Image { public: Image2DArray( const Context& context, cl_mem_flags flags, ImageFormat format, ::size_t arraySize, ::size_t width, ::size_t height, ::size_t rowPitch, ::size_t slicePitch, void* host_ptr = NULL, cl_int* err = NULL) { cl_int error; cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D_ARRAY, width, height, 0, // depth (unused) arraySize, rowPitch, slicePitch, 0, 0, 0 }; object_ = ::clCreateImage( context(), flags, &format, &desc, host_ptr, &error); detail::errHandler(error, __CREATE_IMAGE_ERR); if (err != NULL) { *err = error; } } Image2DArray() { } Image2DArray(const Image2DArray& imageArray) : Image(imageArray) { } __CL_EXPLICIT_CONSTRUCTORS Image2DArray(const cl_mem& imageArray) : Image(imageArray) { } Image2DArray& operator = (const Image2DArray& rhs) { if (this != &rhs) { Image::operator=(rhs); } return *this; } Image2DArray& operator = (const cl_mem& rhs) { Image::operator=(rhs); return *this; } }; #endif // #if defined(CL_VERSION_1_2) /*! \brief Class interface for 3D Image Memory objects. * * See Memory for details about copy semantics, etc. * * \see Memory */ class Image3D : public Image { public: /*! \brief Constructs a 3D Image in a specified context. * * Wraps clCreateImage(). */ Image3D( const Context& context, cl_mem_flags flags, ImageFormat format, ::size_t width, ::size_t height, ::size_t depth, ::size_t row_pitch = 0, ::size_t slice_pitch = 0, void* host_ptr = NULL, cl_int* err = NULL) { cl_int error; bool useCreateImage; #if defined(CL_VERSION_1_2) && defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) // Run-time decision based on the actual platform { cl_uint version = detail::getContextPlatformVersion(context()); useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above } #elif defined(CL_VERSION_1_2) useCreateImage = true; #else useCreateImage = false; #endif #if defined(CL_VERSION_1_2) if (useCreateImage) { cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, // array size (unused) row_pitch, slice_pitch, 0, 0, 0 }; object_ = ::clCreateImage( context(), flags, &format, &desc, host_ptr, &error); detail::errHandler(error, __CREATE_IMAGE_ERR); if (err != NULL) { *err = error; } } #endif // #if defined(CL_VERSION_1_2) #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) if (!useCreateImage) { object_ = ::clCreateImage3D( context(), flags, &format, width, height, depth, row_pitch, slice_pitch, host_ptr, &error); detail::errHandler(error, __CREATE_IMAGE3D_ERR); if (err != NULL) { *err = error; } } #endif // #if !defined(CL_VERSION_1_2) || defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) } //! \brief Default constructor - initializes to NULL. Image3D() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ Image3D(const Image3D& image3D) : Image(image3D) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS Image3D(const cl_mem& image3D) : Image(image3D) { } /*! \brief Assignment from Image3D - performs shallow copy. * * See Memory for further details. */ Image3D& operator = (const Image3D& rhs) { if (this != &rhs) { Image::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ Image3D& operator = (const cl_mem& rhs) { Image::operator=(rhs); return *this; } }; #if !defined(CL_VERSION_1_2) /*! \brief Class interface for GL 3D Image Memory objects. * * This is provided to facilitate interoperability with OpenGL. * * See Memory for details about copy semantics, etc. * * \see Memory */ class Image3DGL : public Image3D { public: /*! \brief Constructs an Image3DGL in a specified context, from a given * GL Texture. * * Wraps clCreateFromGLTexture3D(). */ Image3DGL( const Context& context, cl_mem_flags flags, GLenum target, GLint miplevel, GLuint texobj, cl_int * err = NULL) { cl_int error; object_ = ::clCreateFromGLTexture3D( context(), flags, target, miplevel, texobj, &error); detail::errHandler(error, __CREATE_GL_TEXTURE_3D_ERR); if (err != NULL) { *err = error; } } //! \brief Default constructor - initializes to NULL. Image3DGL() : Image3D() { } /*! \brief Copy constructor - performs shallow copy. * * See Memory for further details. */ Image3DGL(const Image3DGL& image) : Image3D(image) { } /*! \brief Constructor from cl_mem - takes ownership. * * See Memory for further details. */ __CL_EXPLICIT_CONSTRUCTORS Image3DGL(const cl_mem& image) : Image3D(image) { } /*! \brief Assignment from Image3DGL - performs shallow copy. * * See Memory for further details. */ Image3DGL& operator = (const Image3DGL& rhs) { if (this != &rhs) { Image3D::operator=(rhs); } return *this; } /*! \brief Assignment from cl_mem - performs shallow copy. * * See Memory for further details. */ Image3DGL& operator = (const cl_mem& rhs) { Image3D::operator=(rhs); return *this; } }; #endif // #if !defined(CL_VERSION_1_2) #if defined(CL_VERSION_1_2) /*! \class ImageGL * \brief general image interface for GL interop. * We abstract the 2D and 3D GL images into a single instance here * that wraps all GL sourced images on the grounds that setup information * was performed by OpenCL anyway. */ class ImageGL : public Image { public: ImageGL( const Context& context, cl_mem_flags flags, GLenum target, GLint miplevel, GLuint texobj, cl_int * err = NULL) { cl_int error; object_ = ::clCreateFromGLTexture( context(), flags, target, miplevel, texobj, &error); detail::errHandler(error, __CREATE_GL_TEXTURE_ERR); if (err != NULL) { *err = error; } } ImageGL() : Image() { } ImageGL(const ImageGL& image) : Image(image) { } __CL_EXPLICIT_CONSTRUCTORS ImageGL(const cl_mem& image) : Image(image) { } ImageGL& operator = (const ImageGL& rhs) { if (this != &rhs) { Image::operator=(rhs); } return *this; } ImageGL& operator = (const cl_mem& rhs) { Image::operator=(rhs); return *this; } }; #endif // #if defined(CL_VERSION_1_2) /*! \brief Class interface for cl_sampler. * * \note Copies of these objects are shallow, meaning that the copy will refer * to the same underlying cl_sampler as the original. For details, see * clRetainSampler() and clReleaseSampler(). * * \see cl_sampler */ class Sampler : public detail::Wrapper { public: /*! \brief Destructor. * * This calls clReleaseSampler() on the value held by this instance. */ ~Sampler() { } //! \brief Default constructor - initializes to NULL. Sampler() { } /*! \brief Constructs a Sampler in a specified context. * * Wraps clCreateSampler(). */ Sampler( const Context& context, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode, cl_int* err = NULL) { cl_int error; object_ = ::clCreateSampler( context(), normalized_coords, addressing_mode, filter_mode, &error); detail::errHandler(error, __CREATE_SAMPLER_ERR); if (err != NULL) { *err = error; } } /*! \brief Copy constructor - performs shallow copy. * * This calls clRetainSampler() on the parameter's cl_sampler. */ Sampler(const Sampler& sampler) : detail::Wrapper(sampler) { } /*! \brief Constructor from cl_sampler - takes ownership. * * This effectively transfers ownership of a refcount on the cl_sampler * into the new Sampler object. */ Sampler(const cl_sampler& sampler) : detail::Wrapper(sampler) { } /*! \brief Assignment operator from Sampler. * * This calls clRetainSampler() on the parameter and clReleaseSampler() * on the previous value held by this instance. */ Sampler& operator = (const Sampler& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } /*! \brief Assignment operator from cl_sampler - takes ownership. * * This effectively transfers ownership of a refcount on the rhs and calls * clReleaseSampler() on the value previously held by this instance. */ Sampler& operator = (const cl_sampler& rhs) { detail::Wrapper::operator=(rhs); return *this; } //! \brief Wrapper for clGetSamplerInfo(). template cl_int getInfo(cl_sampler_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetSamplerInfo, object_, name, param), __GET_SAMPLER_INFO_ERR); } //! \brief Wrapper for clGetSamplerInfo() that returns by value. template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_sampler_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } }; class Program; class CommandQueue; class Kernel; //! \brief Class interface for specifying NDRange values. class NDRange { private: size_t<3> sizes_; cl_uint dimensions_; public: //! \brief Default constructor - resulting range has zero dimensions. NDRange() : dimensions_(0) { } //! \brief Constructs one-dimensional range. NDRange(::size_t size0) : dimensions_(1) { sizes_[0] = size0; } //! \brief Constructs two-dimensional range. NDRange(::size_t size0, ::size_t size1) : dimensions_(2) { sizes_[0] = size0; sizes_[1] = size1; } //! \brief Constructs three-dimensional range. NDRange(::size_t size0, ::size_t size1, ::size_t size2) : dimensions_(3) { sizes_[0] = size0; sizes_[1] = size1; sizes_[2] = size2; } /*! \brief Conversion operator to const ::size_t *. * * \returns a pointer to the size of the first dimension. */ operator const ::size_t*() const { return (const ::size_t*) sizes_; } //! \brief Queries the number of dimensions in the range. ::size_t dimensions() const { return dimensions_; } }; //! \brief A zero-dimensional range. static const NDRange NullRange; //! \brief Local address wrapper for use with Kernel::setArg struct LocalSpaceArg { ::size_t size_; }; namespace detail { template struct KernelArgumentHandler { static ::size_t size(const T&) { return sizeof(T); } static T* ptr(T& value) { return &value; } }; template <> struct KernelArgumentHandler { static ::size_t size(const LocalSpaceArg& value) { return value.size_; } static void* ptr(LocalSpaceArg&) { return NULL; } }; } //! \endcond /*! __local * \brief Helper function for generating LocalSpaceArg objects. * Deprecated. Replaced with Local. */ inline CL_EXT_PREFIX__VERSION_1_1_DEPRECATED LocalSpaceArg __local(::size_t size) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED; inline LocalSpaceArg __local(::size_t size) { LocalSpaceArg ret = { size }; return ret; } /*! Local * \brief Helper function for generating LocalSpaceArg objects. */ inline LocalSpaceArg Local(::size_t size) { LocalSpaceArg ret = { size }; return ret; } //class KernelFunctor; /*! \brief Class interface for cl_kernel. * * \note Copies of these objects are shallow, meaning that the copy will refer * to the same underlying cl_kernel as the original. For details, see * clRetainKernel() and clReleaseKernel(). * * \see cl_kernel */ class Kernel : public detail::Wrapper { public: inline Kernel(const Program& program, const char* name, cl_int* err = NULL); /*! \brief Destructor. * * This calls clReleaseKernel() on the value held by this instance. */ ~Kernel() { } //! \brief Default constructor - initializes to NULL. Kernel() { } /*! \brief Copy constructor - performs shallow copy. * * This calls clRetainKernel() on the parameter's cl_kernel. */ Kernel(const Kernel& kernel) : detail::Wrapper(kernel) { } /*! \brief Constructor from cl_kernel - takes ownership. * * This effectively transfers ownership of a refcount on the cl_kernel * into the new Kernel object. */ __CL_EXPLICIT_CONSTRUCTORS Kernel(const cl_kernel& kernel) : detail::Wrapper(kernel) { } /*! \brief Assignment operator from Kernel. * * This calls clRetainKernel() on the parameter and clReleaseKernel() * on the previous value held by this instance. */ Kernel& operator = (const Kernel& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } /*! \brief Assignment operator from cl_kernel - takes ownership. * * This effectively transfers ownership of a refcount on the rhs and calls * clReleaseKernel() on the value previously held by this instance. */ Kernel& operator = (const cl_kernel& rhs) { detail::Wrapper::operator=(rhs); return *this; } template cl_int getInfo(cl_kernel_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetKernelInfo, object_, name, param), __GET_KERNEL_INFO_ERR); } template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_kernel_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } #if defined(CL_VERSION_1_2) template cl_int getArgInfo(cl_uint argIndex, cl_kernel_arg_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetKernelArgInfo, object_, argIndex, name, param), __GET_KERNEL_ARG_INFO_ERR); } template typename detail::param_traits::param_type getArgInfo(cl_uint argIndex, cl_int* err = NULL) const { typename detail::param_traits< detail::cl_kernel_arg_info, name>::param_type param; cl_int result = getArgInfo(argIndex, name, ¶m); if (err != NULL) { *err = result; } return param; } #endif // #if defined(CL_VERSION_1_2) template cl_int getWorkGroupInfo( const Device& device, cl_kernel_work_group_info name, T* param) const { return detail::errHandler( detail::getInfo( &::clGetKernelWorkGroupInfo, object_, device(), name, param), __GET_KERNEL_WORK_GROUP_INFO_ERR); } template typename detail::param_traits::param_type getWorkGroupInfo(const Device& device, cl_int* err = NULL) const { typename detail::param_traits< detail::cl_kernel_work_group_info, name>::param_type param; cl_int result = getWorkGroupInfo(device, name, ¶m); if (err != NULL) { *err = result; } return param; } template cl_int setArg(cl_uint index, T value) { return detail::errHandler( ::clSetKernelArg( object_, index, detail::KernelArgumentHandler::size(value), detail::KernelArgumentHandler::ptr(value)), __SET_KERNEL_ARGS_ERR); } cl_int setArg(cl_uint index, ::size_t size, void* argPtr) { return detail::errHandler( ::clSetKernelArg(object_, index, size, argPtr), __SET_KERNEL_ARGS_ERR); } }; /*! \class Program * \brief Program interface that implements cl_program. */ class Program : public detail::Wrapper { public: typedef VECTOR_CLASS > Binaries; typedef VECTOR_CLASS > Sources; Program( const STRING_CLASS& source, bool build = false, cl_int* err = NULL) { cl_int error; const char * strings = source.c_str(); const ::size_t length = source.size(); Context context = Context::getDefault(err); object_ = ::clCreateProgramWithSource( context(), (cl_uint)1, &strings, &length, &error); detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); if (error == CL_SUCCESS && build) { error = ::clBuildProgram( object_, 0, NULL, "", NULL, NULL); detail::errHandler(error, __BUILD_PROGRAM_ERR); } if (err != NULL) { *err = error; } } Program( const Context& context, const STRING_CLASS& source, bool build = false, cl_int* err = NULL) { cl_int error; const char * strings = source.c_str(); const ::size_t length = source.size(); object_ = ::clCreateProgramWithSource( context(), (cl_uint)1, &strings, &length, &error); detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); if (error == CL_SUCCESS && build) { error = ::clBuildProgram( object_, 0, NULL, "", NULL, NULL); detail::errHandler(error, __BUILD_PROGRAM_ERR); } if (err != NULL) { *err = error; } } Program( const Context& context, const Sources& sources, cl_int* err = NULL) { cl_int error; const ::size_t n = (::size_t)sources.size(); ::size_t* lengths = (::size_t*) alloca(n * sizeof(::size_t)); const char** strings = (const char**) alloca(n * sizeof(const char*)); for (::size_t i = 0; i < n; ++i) { strings[i] = sources[(int)i].first; lengths[i] = sources[(int)i].second; } object_ = ::clCreateProgramWithSource( context(), (cl_uint)n, strings, lengths, &error); detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR); if (err != NULL) { *err = error; } } /** * Construct a program object from a list of devices and a per-device list of binaries. * \param context A valid OpenCL context in which to construct the program. * \param devices A vector of OpenCL device objects for which the program will be created. * \param binaries A vector of pairs of a pointer to a binary object and its length. * \param binaryStatus An optional vector that on completion will be resized to * match the size of binaries and filled with values to specify if each binary * was successfully loaded. * Set to CL_SUCCESS if the binary was successfully loaded. * Set to CL_INVALID_VALUE if the length is 0 or the binary pointer is NULL. * Set to CL_INVALID_BINARY if the binary provided is not valid for the matching device. * \param err if non-NULL will be set to CL_SUCCESS on successful operation or one of the following errors: * CL_INVALID_CONTEXT if context is not a valid context. * CL_INVALID_VALUE if the length of devices is zero; or if the length of binaries does not match the length of devices; * or if any entry in binaries is NULL or has length 0. * CL_INVALID_DEVICE if OpenCL devices listed in devices are not in the list of devices associated with context. * CL_INVALID_BINARY if an invalid program binary was encountered for any device. binaryStatus will return specific status for each device. * CL_OUT_OF_HOST_MEMORY if there is a failure to allocate resources required by the OpenCL implementation on the host. */ Program( const Context& context, const VECTOR_CLASS& devices, const Binaries& binaries, VECTOR_CLASS* binaryStatus = NULL, cl_int* err = NULL) { cl_int error; const ::size_t numDevices = devices.size(); // Catch size mismatch early and return if(binaries.size() != numDevices) { error = CL_INVALID_VALUE; detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR); if (err != NULL) { *err = error; } return; } ::size_t* lengths = (::size_t*) alloca(numDevices * sizeof(::size_t)); const unsigned char** images = (const unsigned char**) alloca(numDevices * sizeof(const unsigned char**)); for (::size_t i = 0; i < numDevices; ++i) { images[i] = (const unsigned char*)binaries[i].first; lengths[i] = binaries[(int)i].second; } cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { deviceIDs[deviceIndex] = (devices[deviceIndex])(); } if(binaryStatus) { binaryStatus->resize(numDevices); } object_ = ::clCreateProgramWithBinary( context(), (cl_uint) devices.size(), deviceIDs, lengths, images, binaryStatus != NULL ? &binaryStatus->front() : NULL, &error); detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR); if (err != NULL) { *err = error; } } #if defined(CL_VERSION_1_2) /** * Create program using builtin kernels. * \param kernelNames Semi-colon separated list of builtin kernel names */ Program( const Context& context, const VECTOR_CLASS& devices, const STRING_CLASS& kernelNames, cl_int* err = NULL) { cl_int error; ::size_t numDevices = devices.size(); cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { deviceIDs[deviceIndex] = (devices[deviceIndex])(); } object_ = ::clCreateProgramWithBuiltInKernels( context(), (cl_uint) devices.size(), deviceIDs, kernelNames.c_str(), &error); detail::errHandler(error, __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR); if (err != NULL) { *err = error; } } #endif // #if defined(CL_VERSION_1_2) Program() { } Program(const Program& program) : detail::Wrapper(program) { } __CL_EXPLICIT_CONSTRUCTORS Program(const cl_program& program) : detail::Wrapper(program) { } Program& operator = (const Program& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } Program& operator = (const cl_program& rhs) { detail::Wrapper::operator=(rhs); return *this; } cl_int build( const VECTOR_CLASS& devices, const char* options = NULL, void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, void* data = NULL) const { ::size_t numDevices = devices.size(); cl_device_id* deviceIDs = (cl_device_id*) alloca(numDevices * sizeof(cl_device_id)); for( ::size_t deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) { deviceIDs[deviceIndex] = (devices[deviceIndex])(); } return detail::errHandler( ::clBuildProgram( object_, (cl_uint) devices.size(), deviceIDs, options, notifyFptr, data), __BUILD_PROGRAM_ERR); } cl_int build( const char* options = NULL, void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, void* data = NULL) const { return detail::errHandler( ::clBuildProgram( object_, 0, NULL, options, notifyFptr, data), __BUILD_PROGRAM_ERR); } #if defined(CL_VERSION_1_2) cl_int compile( const char* options = NULL, void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, void* data = NULL) const { return detail::errHandler( ::clCompileProgram( object_, 0, NULL, options, 0, NULL, NULL, notifyFptr, data), __COMPILE_PROGRAM_ERR); } #endif template cl_int getInfo(cl_program_info name, T* param) const { return detail::errHandler( detail::getInfo(&::clGetProgramInfo, object_, name, param), __GET_PROGRAM_INFO_ERR); } template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_program_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } template cl_int getBuildInfo( const Device& device, cl_program_build_info name, T* param) const { return detail::errHandler( detail::getInfo( &::clGetProgramBuildInfo, object_, device(), name, param), __GET_PROGRAM_BUILD_INFO_ERR); } template typename detail::param_traits::param_type getBuildInfo(const Device& device, cl_int* err = NULL) const { typename detail::param_traits< detail::cl_program_build_info, name>::param_type param; cl_int result = getBuildInfo(device, name, ¶m); if (err != NULL) { *err = result; } return param; } cl_int createKernels(VECTOR_CLASS* kernels) { cl_uint numKernels; cl_int err = ::clCreateKernelsInProgram(object_, 0, NULL, &numKernels); if (err != CL_SUCCESS) { return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR); } Kernel* value = (Kernel*) alloca(numKernels * sizeof(Kernel)); err = ::clCreateKernelsInProgram( object_, numKernels, (cl_kernel*) value, NULL); if (err != CL_SUCCESS) { return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR); } kernels->assign(&value[0], &value[numKernels]); return CL_SUCCESS; } }; #if defined(CL_VERSION_1_2) inline Program linkProgram( Program input1, Program input2, const char* options = NULL, void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, void* data = NULL, cl_int* err = NULL) { cl_int err_local = CL_SUCCESS; cl_program programs[2] = { input1(), input2() }; Context ctx = input1.getInfo(); cl_program prog = ::clLinkProgram( ctx(), 0, NULL, options, 2, programs, notifyFptr, data, &err_local); detail::errHandler(err_local,__COMPILE_PROGRAM_ERR); if (err != NULL) { *err = err_local; } return Program(prog); } inline Program linkProgram( VECTOR_CLASS inputPrograms, const char* options = NULL, void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL, void* data = NULL, cl_int* err = NULL) { cl_int err_local = CL_SUCCESS; cl_program * programs = (cl_program*) alloca(inputPrograms.size() * sizeof(cl_program)); if (programs != NULL) { for (unsigned int i = 0; i < inputPrograms.size(); i++) { programs[i] = inputPrograms[i](); } } cl_program prog = ::clLinkProgram( Context::getDefault()(), 0, NULL, options, (cl_uint)inputPrograms.size(), programs, notifyFptr, data, &err_local); detail::errHandler(err_local,__COMPILE_PROGRAM_ERR); if (err != NULL) { *err = err_local; } return Program(prog); } #endif template<> inline VECTOR_CLASS cl::Program::getInfo(cl_int* err) const { VECTOR_CLASS< ::size_t> sizes = getInfo(); VECTOR_CLASS binaries; for (VECTOR_CLASS< ::size_t>::iterator s = sizes.begin(); s != sizes.end(); ++s) { char *ptr = NULL; if (*s != 0) ptr = new char[*s]; binaries.push_back(ptr); } cl_int result = getInfo(CL_PROGRAM_BINARIES, &binaries); if (err != NULL) { *err = result; } return binaries; } inline Kernel::Kernel(const Program& program, const char* name, cl_int* err) { cl_int error; object_ = ::clCreateKernel(program(), name, &error); detail::errHandler(error, __CREATE_KERNEL_ERR); if (err != NULL) { *err = error; } } /*! \class CommandQueue * \brief CommandQueue interface for cl_command_queue. */ class CommandQueue : public detail::Wrapper { private: static volatile int default_initialized_; static CommandQueue default_; static volatile cl_int default_error_; public: CommandQueue( cl_command_queue_properties properties, cl_int* err = NULL) { cl_int error; Context context = Context::getDefault(&error); detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); if (error != CL_SUCCESS) { if (err != NULL) { *err = error; } } else { Device device = context.getInfo()[0]; object_ = ::clCreateCommandQueue( context(), device(), properties, &error); detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); if (err != NULL) { *err = error; } } } /*! * \brief Constructs a CommandQueue for an implementation defined device in the given context */ explicit CommandQueue( const Context& context, cl_command_queue_properties properties = 0, cl_int* err = NULL) { cl_int error; VECTOR_CLASS devices; error = context.getInfo(CL_CONTEXT_DEVICES, &devices); detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); if (error != CL_SUCCESS) { if (err != NULL) { *err = error; } return; } object_ = ::clCreateCommandQueue(context(), devices[0](), properties, &error); detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); if (err != NULL) { *err = error; } } CommandQueue( const Context& context, const Device& device, cl_command_queue_properties properties = 0, cl_int* err = NULL) { cl_int error; object_ = ::clCreateCommandQueue( context(), device(), properties, &error); detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); if (err != NULL) { *err = error; } } static CommandQueue getDefault(cl_int * err = NULL) { int state = detail::compare_exchange( &default_initialized_, __DEFAULT_BEING_INITIALIZED, __DEFAULT_NOT_INITIALIZED); if (state & __DEFAULT_INITIALIZED) { if (err != NULL) { *err = default_error_; } return default_; } if (state & __DEFAULT_BEING_INITIALIZED) { // Assume writes will propagate eventually... while(default_initialized_ != __DEFAULT_INITIALIZED) { detail::fence(); } if (err != NULL) { *err = default_error_; } return default_; } cl_int error; Context context = Context::getDefault(&error); detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); if (error != CL_SUCCESS) { if (err != NULL) { *err = error; } } else { Device device = context.getInfo()[0]; default_ = CommandQueue(context, device, 0, &error); detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR); if (err != NULL) { *err = error; } } detail::fence(); default_error_ = error; // Assume writes will propagate eventually... default_initialized_ = __DEFAULT_INITIALIZED; detail::fence(); if (err != NULL) { *err = default_error_; } return default_; } CommandQueue() { } CommandQueue(const CommandQueue& commandQueue) : detail::Wrapper(commandQueue) { } CommandQueue(const cl_command_queue& commandQueue) : detail::Wrapper(commandQueue) { } CommandQueue& operator = (const CommandQueue& rhs) { if (this != &rhs) { detail::Wrapper::operator=(rhs); } return *this; } CommandQueue& operator = (const cl_command_queue& rhs) { detail::Wrapper::operator=(rhs); return *this; } template cl_int getInfo(cl_command_queue_info name, T* param) const { return detail::errHandler( detail::getInfo( &::clGetCommandQueueInfo, object_, name, param), __GET_COMMAND_QUEUE_INFO_ERR); } template typename detail::param_traits::param_type getInfo(cl_int* err = NULL) const { typename detail::param_traits< detail::cl_command_queue_info, name>::param_type param; cl_int result = getInfo(name, ¶m); if (err != NULL) { *err = result; } return param; } cl_int enqueueReadBuffer( const Buffer& buffer, cl_bool blocking, ::size_t offset, ::size_t size, void* ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueReadBuffer( object_, buffer(), blocking, offset, size, ptr, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_READ_BUFFER_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueWriteBuffer( const Buffer& buffer, cl_bool blocking, ::size_t offset, ::size_t size, const void* ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueWriteBuffer( object_, buffer(), blocking, offset, size, ptr, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_WRITE_BUFFER_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueCopyBuffer( const Buffer& src, const Buffer& dst, ::size_t src_offset, ::size_t dst_offset, ::size_t size, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueCopyBuffer( object_, src(), dst(), src_offset, dst_offset, size, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQEUE_COPY_BUFFER_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueReadBufferRect( const Buffer& buffer, cl_bool blocking, const size_t<3>& buffer_offset, const size_t<3>& host_offset, const size_t<3>& region, ::size_t buffer_row_pitch, ::size_t buffer_slice_pitch, ::size_t host_row_pitch, ::size_t host_slice_pitch, void *ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueReadBufferRect( object_, buffer(), blocking, (const ::size_t *)buffer_offset, (const ::size_t *)host_offset, (const ::size_t *)region, buffer_row_pitch, buffer_slice_pitch, host_row_pitch, host_slice_pitch, ptr, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_READ_BUFFER_RECT_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueWriteBufferRect( const Buffer& buffer, cl_bool blocking, const size_t<3>& buffer_offset, const size_t<3>& host_offset, const size_t<3>& region, ::size_t buffer_row_pitch, ::size_t buffer_slice_pitch, ::size_t host_row_pitch, ::size_t host_slice_pitch, void *ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueWriteBufferRect( object_, buffer(), blocking, (const ::size_t *)buffer_offset, (const ::size_t *)host_offset, (const ::size_t *)region, buffer_row_pitch, buffer_slice_pitch, host_row_pitch, host_slice_pitch, ptr, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_WRITE_BUFFER_RECT_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueCopyBufferRect( const Buffer& src, const Buffer& dst, const size_t<3>& src_origin, const size_t<3>& dst_origin, const size_t<3>& region, ::size_t src_row_pitch, ::size_t src_slice_pitch, ::size_t dst_row_pitch, ::size_t dst_slice_pitch, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueCopyBufferRect( object_, src(), dst(), (const ::size_t *)src_origin, (const ::size_t *)dst_origin, (const ::size_t *)region, src_row_pitch, src_slice_pitch, dst_row_pitch, dst_slice_pitch, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQEUE_COPY_BUFFER_RECT_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } #if defined(CL_VERSION_1_2) /** * Enqueue a command to fill a buffer object with a pattern * of a given size. The pattern is specified a as vector. * \tparam PatternType The datatype of the pattern field. * The pattern type must be an accepted OpenCL data type. */ template cl_int enqueueFillBuffer( const Buffer& buffer, PatternType pattern, ::size_t offset, ::size_t size, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueFillBuffer( object_, buffer(), static_cast(&pattern), sizeof(PatternType), offset, size, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_FILL_BUFFER_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } #endif // #if defined(CL_VERSION_1_2) cl_int enqueueReadImage( const Image& image, cl_bool blocking, const size_t<3>& origin, const size_t<3>& region, ::size_t row_pitch, ::size_t slice_pitch, void* ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueReadImage( object_, image(), blocking, (const ::size_t *) origin, (const ::size_t *) region, row_pitch, slice_pitch, ptr, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_READ_IMAGE_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueWriteImage( const Image& image, cl_bool blocking, const size_t<3>& origin, const size_t<3>& region, ::size_t row_pitch, ::size_t slice_pitch, void* ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueWriteImage( object_, image(), blocking, (const ::size_t *) origin, (const ::size_t *) region, row_pitch, slice_pitch, ptr, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_WRITE_IMAGE_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueCopyImage( const Image& src, const Image& dst, const size_t<3>& src_origin, const size_t<3>& dst_origin, const size_t<3>& region, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueCopyImage( object_, src(), dst(), (const ::size_t *) src_origin, (const ::size_t *)dst_origin, (const ::size_t *) region, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_COPY_IMAGE_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } #if defined(CL_VERSION_1_2) /** * Enqueue a command to fill an image object with a specified color. * \param fillColor is the color to use to fill the image. * This is a four component RGBA floating-point color value if * the image channel data type is not an unnormalized signed or * unsigned data type. */ cl_int enqueueFillImage( const Image& image, cl_float4 fillColor, const size_t<3>& origin, const size_t<3>& region, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueFillImage( object_, image(), static_cast(&fillColor), (const ::size_t *) origin, (const ::size_t *) region, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_FILL_IMAGE_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } /** * Enqueue a command to fill an image object with a specified color. * \param fillColor is the color to use to fill the image. * This is a four component RGBA signed integer color value if * the image channel data type is an unnormalized signed integer * type. */ cl_int enqueueFillImage( const Image& image, cl_int4 fillColor, const size_t<3>& origin, const size_t<3>& region, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueFillImage( object_, image(), static_cast(&fillColor), (const ::size_t *) origin, (const ::size_t *) region, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_FILL_IMAGE_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } /** * Enqueue a command to fill an image object with a specified color. * \param fillColor is the color to use to fill the image. * This is a four component RGBA unsigned integer color value if * the image channel data type is an unnormalized unsigned integer * type. */ cl_int enqueueFillImage( const Image& image, cl_uint4 fillColor, const size_t<3>& origin, const size_t<3>& region, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueFillImage( object_, image(), static_cast(&fillColor), (const ::size_t *) origin, (const ::size_t *) region, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_FILL_IMAGE_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } #endif // #if defined(CL_VERSION_1_2) cl_int enqueueCopyImageToBuffer( const Image& src, const Buffer& dst, const size_t<3>& src_origin, const size_t<3>& region, ::size_t dst_offset, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueCopyImageToBuffer( object_, src(), dst(), (const ::size_t *) src_origin, (const ::size_t *) region, dst_offset, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueCopyBufferToImage( const Buffer& src, const Image& dst, ::size_t src_offset, const size_t<3>& dst_origin, const size_t<3>& region, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueCopyBufferToImage( object_, src(), dst(), src_offset, (const ::size_t *) dst_origin, (const ::size_t *) region, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } void* enqueueMapBuffer( const Buffer& buffer, cl_bool blocking, cl_map_flags flags, ::size_t offset, ::size_t size, const VECTOR_CLASS* events = NULL, Event* event = NULL, cl_int* err = NULL) const { cl_int error; void * result = ::clEnqueueMapBuffer( object_, buffer(), blocking, flags, offset, size, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (cl_event*) event, &error); detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); if (err != NULL) { *err = error; } return result; } void* enqueueMapImage( const Image& buffer, cl_bool blocking, cl_map_flags flags, const size_t<3>& origin, const size_t<3>& region, ::size_t * row_pitch, ::size_t * slice_pitch, const VECTOR_CLASS* events = NULL, Event* event = NULL, cl_int* err = NULL) const { cl_int error; void * result = ::clEnqueueMapImage( object_, buffer(), blocking, flags, (const ::size_t *) origin, (const ::size_t *) region, row_pitch, slice_pitch, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (cl_event*) event, &error); detail::errHandler(error, __ENQUEUE_MAP_IMAGE_ERR); if (err != NULL) { *err = error; } return result; } cl_int enqueueUnmapMemObject( const Memory& memory, void* mapped_ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueUnmapMemObject( object_, memory(), mapped_ptr, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_UNMAP_MEM_OBJECT_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } #if defined(CL_VERSION_1_2) /** * Enqueues a marker command which waits for either a list of events to complete, * or all previously enqueued commands to complete. * * Enqueues a marker command which waits for either a list of events to complete, * or if the list is empty it waits for all commands previously enqueued in command_queue * to complete before it completes. This command returns an event which can be waited on, * i.e. this event can be waited on to insure that all events either in the event_wait_list * or all previously enqueued commands, queued before this command to command_queue, * have completed. */ cl_int enqueueMarkerWithWaitList( const VECTOR_CLASS *events = 0, Event *event = 0) { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueMarkerWithWaitList( object_, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_MARKER_WAIT_LIST_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } /** * A synchronization point that enqueues a barrier operation. * * Enqueues a barrier command which waits for either a list of events to complete, * or if the list is empty it waits for all commands previously enqueued in command_queue * to complete before it completes. This command blocks command execution, that is, any * following commands enqueued after it do not execute until it completes. This command * returns an event which can be waited on, i.e. this event can be waited on to insure that * all events either in the event_wait_list or all previously enqueued commands, queued * before this command to command_queue, have completed. */ cl_int enqueueBarrierWithWaitList( const VECTOR_CLASS *events = 0, Event *event = 0) { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueBarrierWithWaitList( object_, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_BARRIER_WAIT_LIST_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } /** * Enqueues a command to indicate with which device a set of memory objects * should be associated. */ cl_int enqueueMigrateMemObjects( const VECTOR_CLASS &memObjects, cl_mem_migration_flags flags, const VECTOR_CLASS* events = NULL, Event* event = NULL ) { cl_event tmp; cl_mem* localMemObjects = static_cast(alloca(memObjects.size() * sizeof(cl_mem))); for( int i = 0; i < (int)memObjects.size(); ++i ) { localMemObjects[i] = memObjects[i](); } cl_int err = detail::errHandler( ::clEnqueueMigrateMemObjects( object_, (cl_uint)memObjects.size(), static_cast(localMemObjects), flags, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_UNMAP_MEM_OBJECT_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } #endif // #if defined(CL_VERSION_1_2) cl_int enqueueNDRangeKernel( const Kernel& kernel, const NDRange& offset, const NDRange& global, const NDRange& local = NullRange, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueNDRangeKernel( object_, kernel(), (cl_uint) global.dimensions(), offset.dimensions() != 0 ? (const ::size_t*) offset : NULL, (const ::size_t*) global, local.dimensions() != 0 ? (const ::size_t*) local : NULL, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_NDRANGE_KERNEL_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueTask( const Kernel& kernel, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueTask( object_, kernel(), (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_TASK_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueNativeKernel( void (CL_CALLBACK *userFptr)(void *), std::pair args, const VECTOR_CLASS* mem_objects = NULL, const VECTOR_CLASS* mem_locs = NULL, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_mem * mems = (mem_objects != NULL && mem_objects->size() > 0) ? (cl_mem*) alloca(mem_objects->size() * sizeof(cl_mem)) : NULL; if (mems != NULL) { for (unsigned int i = 0; i < mem_objects->size(); i++) { mems[i] = ((*mem_objects)[i])(); } } cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueNativeKernel( object_, userFptr, args.first, args.second, (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, mems, (mem_locs != NULL) ? (const void **) &mem_locs->front() : NULL, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_NATIVE_KERNEL); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } /** * Deprecated APIs for 1.2 */ #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int enqueueMarker(Event* event = NULL) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED { return detail::errHandler( ::clEnqueueMarker(object_, (cl_event*) event), __ENQUEUE_MARKER_ERR); } CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int enqueueWaitForEvents(const VECTOR_CLASS& events) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED { return detail::errHandler( ::clEnqueueWaitForEvents( object_, (cl_uint) events.size(), (const cl_event*) &events.front()), __ENQUEUE_WAIT_FOR_EVENTS_ERR); } #endif // #if defined(CL_VERSION_1_1) cl_int enqueueAcquireGLObjects( const VECTOR_CLASS* mem_objects = NULL, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueAcquireGLObjects( object_, (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_ACQUIRE_GL_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueReleaseGLObjects( const VECTOR_CLASS* mem_objects = NULL, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueReleaseGLObjects( object_, (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_RELEASE_GL_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } #if defined (USE_DX_INTEROP) typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueAcquireD3D10ObjectsKHR)( cl_command_queue command_queue, cl_uint num_objects, const cl_mem* mem_objects, cl_uint num_events_in_wait_list, const cl_event* event_wait_list, cl_event* event); typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueReleaseD3D10ObjectsKHR)( cl_command_queue command_queue, cl_uint num_objects, const cl_mem* mem_objects, cl_uint num_events_in_wait_list, const cl_event* event_wait_list, cl_event* event); cl_int enqueueAcquireD3D10Objects( const VECTOR_CLASS* mem_objects = NULL, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { static PFN_clEnqueueAcquireD3D10ObjectsKHR pfn_clEnqueueAcquireD3D10ObjectsKHR = NULL; #if defined(CL_VERSION_1_2) cl_context context = getInfo(); cl::Device device(getInfo()); cl_platform_id platform = device.getInfo(); __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clEnqueueAcquireD3D10ObjectsKHR); #endif #if defined(CL_VERSION_1_1) __INIT_CL_EXT_FCN_PTR(clEnqueueAcquireD3D10ObjectsKHR); #endif cl_event tmp; cl_int err = detail::errHandler( pfn_clEnqueueAcquireD3D10ObjectsKHR( object_, (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_ACQUIRE_GL_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } cl_int enqueueReleaseD3D10Objects( const VECTOR_CLASS* mem_objects = NULL, const VECTOR_CLASS* events = NULL, Event* event = NULL) const { static PFN_clEnqueueReleaseD3D10ObjectsKHR pfn_clEnqueueReleaseD3D10ObjectsKHR = NULL; #if defined(CL_VERSION_1_2) cl_context context = getInfo(); cl::Device device(getInfo()); cl_platform_id platform = device.getInfo(); __INIT_CL_EXT_FCN_PTR_PLATFORM(platform, clEnqueueReleaseD3D10ObjectsKHR); #endif // #if defined(CL_VERSION_1_2) #if defined(CL_VERSION_1_1) __INIT_CL_EXT_FCN_PTR(clEnqueueReleaseD3D10ObjectsKHR); #endif // #if defined(CL_VERSION_1_1) cl_event tmp; cl_int err = detail::errHandler( pfn_clEnqueueReleaseD3D10ObjectsKHR( object_, (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0, (mem_objects != NULL) ? (const cl_mem *) &mem_objects->front(): NULL, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_RELEASE_GL_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } #endif /** * Deprecated APIs for 1.2 */ #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS) || (defined(CL_VERSION_1_1) && !defined(CL_VERSION_1_2)) CL_EXT_PREFIX__VERSION_1_1_DEPRECATED cl_int enqueueBarrier() const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED { return detail::errHandler( ::clEnqueueBarrier(object_), __ENQUEUE_BARRIER_ERR); } #endif // #if defined(CL_VERSION_1_1) cl_int flush() const { return detail::errHandler(::clFlush(object_), __FLUSH_ERR); } cl_int finish() const { return detail::errHandler(::clFinish(object_), __FINISH_ERR); } }; #ifdef _WIN32 __declspec(selectany) volatile int CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED; __declspec(selectany) CommandQueue CommandQueue::default_; __declspec(selectany) volatile cl_int CommandQueue::default_error_ = CL_SUCCESS; #else __attribute__((weak)) volatile int CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED; __attribute__((weak)) CommandQueue CommandQueue::default_; __attribute__((weak)) volatile cl_int CommandQueue::default_error_ = CL_SUCCESS; #endif template< typename IteratorType > Buffer::Buffer( const Context &context, IteratorType startIterator, IteratorType endIterator, bool readOnly, bool useHostPtr, cl_int* err) { typedef typename std::iterator_traits::value_type DataType; cl_int error; cl_mem_flags flags = 0; if( readOnly ) { flags |= CL_MEM_READ_ONLY; } else { flags |= CL_MEM_READ_WRITE; } if( useHostPtr ) { flags |= CL_MEM_USE_HOST_PTR; } ::size_t size = sizeof(DataType)*(endIterator - startIterator); if( useHostPtr ) { object_ = ::clCreateBuffer(context(), flags, size, static_cast(&*startIterator), &error); } else { object_ = ::clCreateBuffer(context(), flags, size, 0, &error); } detail::errHandler(error, __CREATE_BUFFER_ERR); if (err != NULL) { *err = error; } if( !useHostPtr ) { CommandQueue queue(context, 0, &error); detail::errHandler(error, __CREATE_BUFFER_ERR); if (err != NULL) { *err = error; } error = cl::copy(queue, startIterator, endIterator, *this); detail::errHandler(error, __CREATE_BUFFER_ERR); if (err != NULL) { *err = error; } } } inline cl_int enqueueReadBuffer( const Buffer& buffer, cl_bool blocking, ::size_t offset, ::size_t size, void* ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueReadBuffer(buffer, blocking, offset, size, ptr, events, event); } inline cl_int enqueueWriteBuffer( const Buffer& buffer, cl_bool blocking, ::size_t offset, ::size_t size, const void* ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueWriteBuffer(buffer, blocking, offset, size, ptr, events, event); } inline void* enqueueMapBuffer( const Buffer& buffer, cl_bool blocking, cl_map_flags flags, ::size_t offset, ::size_t size, const VECTOR_CLASS* events = NULL, Event* event = NULL, cl_int* err = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); if (err != NULL) { *err = error; } void * result = ::clEnqueueMapBuffer( queue(), buffer(), blocking, flags, offset, size, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (cl_event*) event, &error); detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); if (err != NULL) { *err = error; } return result; } inline cl_int enqueueUnmapMemObject( const Memory& memory, void* mapped_ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR); if (error != CL_SUCCESS) { return error; } cl_event tmp; cl_int err = detail::errHandler( ::clEnqueueUnmapMemObject( queue(), memory(), mapped_ptr, (events != NULL) ? (cl_uint) events->size() : 0, (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL, (event != NULL) ? &tmp : NULL), __ENQUEUE_UNMAP_MEM_OBJECT_ERR); if (event != NULL && err == CL_SUCCESS) *event = tmp; return err; } inline cl_int enqueueCopyBuffer( const Buffer& src, const Buffer& dst, ::size_t src_offset, ::size_t dst_offset, ::size_t size, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueCopyBuffer(src, dst, src_offset, dst_offset, size, events, event); } /** * Blocking copy operation between iterators and a buffer. * Host to Device. * Uses default command queue. */ template< typename IteratorType > inline cl_int copy( IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) return error; return cl::copy(queue, startIterator, endIterator, buffer); } /** * Blocking copy operation between iterators and a buffer. * Device to Host. * Uses default command queue. */ template< typename IteratorType > inline cl_int copy( const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) return error; return cl::copy(queue, buffer, startIterator, endIterator); } /** * Blocking copy operation between iterators and a buffer. * Host to Device. * Uses specified queue. */ template< typename IteratorType > inline cl_int copy( const CommandQueue &queue, IteratorType startIterator, IteratorType endIterator, cl::Buffer &buffer ) { typedef typename std::iterator_traits::value_type DataType; cl_int error; ::size_t length = endIterator-startIterator; ::size_t byteLength = length*sizeof(DataType); DataType *pointer = static_cast(queue.enqueueMapBuffer(buffer, CL_TRUE, CL_MAP_WRITE, 0, byteLength, 0, 0, &error)); // if exceptions enabled, enqueueMapBuffer will throw if( error != CL_SUCCESS ) { return error; } #if defined(_MSC_VER) std::copy( startIterator, endIterator, stdext::checked_array_iterator( pointer, length)); #else std::copy(startIterator, endIterator, pointer); #endif Event endEvent; error = queue.enqueueUnmapMemObject(buffer, pointer, 0, &endEvent); // if exceptions enabled, enqueueUnmapMemObject will throw if( error != CL_SUCCESS ) { return error; } endEvent.wait(); return CL_SUCCESS; } /** * Blocking copy operation between iterators and a buffer. * Device to Host. * Uses specified queue. */ template< typename IteratorType > inline cl_int copy( const CommandQueue &queue, const cl::Buffer &buffer, IteratorType startIterator, IteratorType endIterator ) { typedef typename std::iterator_traits::value_type DataType; cl_int error; ::size_t length = endIterator-startIterator; ::size_t byteLength = length*sizeof(DataType); DataType *pointer = static_cast(queue.enqueueMapBuffer(buffer, CL_TRUE, CL_MAP_READ, 0, byteLength, 0, 0, &error)); // if exceptions enabled, enqueueMapBuffer will throw if( error != CL_SUCCESS ) { return error; } std::copy(pointer, pointer + length, startIterator); Event endEvent; error = queue.enqueueUnmapMemObject(buffer, pointer, 0, &endEvent); // if exceptions enabled, enqueueUnmapMemObject will throw if( error != CL_SUCCESS ) { return error; } endEvent.wait(); return CL_SUCCESS; } #if defined(CL_VERSION_1_1) inline cl_int enqueueReadBufferRect( const Buffer& buffer, cl_bool blocking, const size_t<3>& buffer_offset, const size_t<3>& host_offset, const size_t<3>& region, ::size_t buffer_row_pitch, ::size_t buffer_slice_pitch, ::size_t host_row_pitch, ::size_t host_slice_pitch, void *ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueReadBufferRect( buffer, blocking, buffer_offset, host_offset, region, buffer_row_pitch, buffer_slice_pitch, host_row_pitch, host_slice_pitch, ptr, events, event); } inline cl_int enqueueWriteBufferRect( const Buffer& buffer, cl_bool blocking, const size_t<3>& buffer_offset, const size_t<3>& host_offset, const size_t<3>& region, ::size_t buffer_row_pitch, ::size_t buffer_slice_pitch, ::size_t host_row_pitch, ::size_t host_slice_pitch, void *ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueWriteBufferRect( buffer, blocking, buffer_offset, host_offset, region, buffer_row_pitch, buffer_slice_pitch, host_row_pitch, host_slice_pitch, ptr, events, event); } inline cl_int enqueueCopyBufferRect( const Buffer& src, const Buffer& dst, const size_t<3>& src_origin, const size_t<3>& dst_origin, const size_t<3>& region, ::size_t src_row_pitch, ::size_t src_slice_pitch, ::size_t dst_row_pitch, ::size_t dst_slice_pitch, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueCopyBufferRect( src, dst, src_origin, dst_origin, region, src_row_pitch, src_slice_pitch, dst_row_pitch, dst_slice_pitch, events, event); } #endif inline cl_int enqueueReadImage( const Image& image, cl_bool blocking, const size_t<3>& origin, const size_t<3>& region, ::size_t row_pitch, ::size_t slice_pitch, void* ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueReadImage( image, blocking, origin, region, row_pitch, slice_pitch, ptr, events, event); } inline cl_int enqueueWriteImage( const Image& image, cl_bool blocking, const size_t<3>& origin, const size_t<3>& region, ::size_t row_pitch, ::size_t slice_pitch, void* ptr, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueWriteImage( image, blocking, origin, region, row_pitch, slice_pitch, ptr, events, event); } inline cl_int enqueueCopyImage( const Image& src, const Image& dst, const size_t<3>& src_origin, const size_t<3>& dst_origin, const size_t<3>& region, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueCopyImage( src, dst, src_origin, dst_origin, region, events, event); } inline cl_int enqueueCopyImageToBuffer( const Image& src, const Buffer& dst, const size_t<3>& src_origin, const size_t<3>& region, ::size_t dst_offset, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueCopyImageToBuffer( src, dst, src_origin, region, dst_offset, events, event); } inline cl_int enqueueCopyBufferToImage( const Buffer& src, const Image& dst, ::size_t src_offset, const size_t<3>& dst_origin, const size_t<3>& region, const VECTOR_CLASS* events = NULL, Event* event = NULL) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.enqueueCopyBufferToImage( src, dst, src_offset, dst_origin, region, events, event); } inline cl_int flush(void) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.flush(); } inline cl_int finish(void) { cl_int error; CommandQueue queue = CommandQueue::getDefault(&error); if (error != CL_SUCCESS) { return error; } return queue.finish(); } // Kernel Functor support // New interface as of September 2011 // Requires the C++11 std::tr1::function (note do not support TR1) // Visual Studio 2010 and GCC 4.2 struct EnqueueArgs { CommandQueue queue_; const NDRange offset_; const NDRange global_; const NDRange local_; VECTOR_CLASS events_; EnqueueArgs(NDRange global) : queue_(CommandQueue::getDefault()), offset_(NullRange), global_(global), local_(NullRange) { } EnqueueArgs(NDRange global, NDRange local) : queue_(CommandQueue::getDefault()), offset_(NullRange), global_(global), local_(local) { } EnqueueArgs(NDRange offset, NDRange global, NDRange local) : queue_(CommandQueue::getDefault()), offset_(offset), global_(global), local_(local) { } EnqueueArgs(Event e, NDRange global) : queue_(CommandQueue::getDefault()), offset_(NullRange), global_(global), local_(NullRange) { events_.push_back(e); } EnqueueArgs(Event e, NDRange global, NDRange local) : queue_(CommandQueue::getDefault()), offset_(NullRange), global_(global), local_(local) { events_.push_back(e); } EnqueueArgs(Event e, NDRange offset, NDRange global, NDRange local) : queue_(CommandQueue::getDefault()), offset_(offset), global_(global), local_(local) { events_.push_back(e); } EnqueueArgs(const VECTOR_CLASS &events, NDRange global) : queue_(CommandQueue::getDefault()), offset_(NullRange), global_(global), local_(NullRange), events_(events) { } EnqueueArgs(const VECTOR_CLASS &events, NDRange global, NDRange local) : queue_(CommandQueue::getDefault()), offset_(NullRange), global_(global), local_(local), events_(events) { } EnqueueArgs(const VECTOR_CLASS &events, NDRange offset, NDRange global, NDRange local) : queue_(CommandQueue::getDefault()), offset_(offset), global_(global), local_(local), events_(events) { } EnqueueArgs(CommandQueue &queue, NDRange global) : queue_(queue), offset_(NullRange), global_(global), local_(NullRange) { } EnqueueArgs(CommandQueue &queue, NDRange global, NDRange local) : queue_(queue), offset_(NullRange), global_(global), local_(local) { } EnqueueArgs(CommandQueue &queue, NDRange offset, NDRange global, NDRange local) : queue_(queue), offset_(offset), global_(global), local_(local) { } EnqueueArgs(CommandQueue &queue, Event e, NDRange global) : queue_(queue), offset_(NullRange), global_(global), local_(NullRange) { events_.push_back(e); } EnqueueArgs(CommandQueue &queue, Event e, NDRange global, NDRange local) : queue_(queue), offset_(NullRange), global_(global), local_(local) { events_.push_back(e); } EnqueueArgs(CommandQueue &queue, Event e, NDRange offset, NDRange global, NDRange local) : queue_(queue), offset_(offset), global_(global), local_(local) { events_.push_back(e); } EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange global) : queue_(queue), offset_(NullRange), global_(global), local_(NullRange), events_(events) { } EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange global, NDRange local) : queue_(queue), offset_(NullRange), global_(global), local_(local), events_(events) { } EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS &events, NDRange offset, NDRange global, NDRange local) : queue_(queue), offset_(offset), global_(global), local_(local), events_(events) { } }; namespace detail { class NullType {}; template struct SetArg { static void set (Kernel kernel, T0 arg) { kernel.setArg(index, arg); } }; template struct SetArg { static void set (Kernel, NullType) { } }; template < typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24, typename T25, typename T26, typename T27, typename T28, typename T29, typename T30, typename T31 > class KernelFunctorGlobal { private: Kernel kernel_; public: KernelFunctorGlobal( Kernel kernel) : kernel_(kernel) {} KernelFunctorGlobal( const Program& program, const STRING_CLASS name, cl_int * err = NULL) : kernel_(program, name.c_str(), err) {} Event operator() ( const EnqueueArgs& args, T0 t0, T1 t1 = NullType(), T2 t2 = NullType(), T3 t3 = NullType(), T4 t4 = NullType(), T5 t5 = NullType(), T6 t6 = NullType(), T7 t7 = NullType(), T8 t8 = NullType(), T9 t9 = NullType(), T10 t10 = NullType(), T11 t11 = NullType(), T12 t12 = NullType(), T13 t13 = NullType(), T14 t14 = NullType(), T15 t15 = NullType(), T16 t16 = NullType(), T17 t17 = NullType(), T18 t18 = NullType(), T19 t19 = NullType(), T20 t20 = NullType(), T21 t21 = NullType(), T22 t22 = NullType(), T23 t23 = NullType(), T24 t24 = NullType(), T25 t25 = NullType(), T26 t26 = NullType(), T27 t27 = NullType(), T28 t28 = NullType(), T29 t29 = NullType(), T30 t30 = NullType(), T31 t31 = NullType() ) { Event event; SetArg<0, T0>::set(kernel_, t0); SetArg<1, T1>::set(kernel_, t1); SetArg<2, T2>::set(kernel_, t2); SetArg<3, T3>::set(kernel_, t3); SetArg<4, T4>::set(kernel_, t4); SetArg<5, T5>::set(kernel_, t5); SetArg<6, T6>::set(kernel_, t6); SetArg<7, T7>::set(kernel_, t7); SetArg<8, T8>::set(kernel_, t8); SetArg<9, T9>::set(kernel_, t9); SetArg<10, T10>::set(kernel_, t10); SetArg<11, T11>::set(kernel_, t11); SetArg<12, T12>::set(kernel_, t12); SetArg<13, T13>::set(kernel_, t13); SetArg<14, T14>::set(kernel_, t14); SetArg<15, T15>::set(kernel_, t15); SetArg<16, T16>::set(kernel_, t16); SetArg<17, T17>::set(kernel_, t17); SetArg<18, T18>::set(kernel_, t18); SetArg<19, T19>::set(kernel_, t19); SetArg<20, T20>::set(kernel_, t20); SetArg<21, T21>::set(kernel_, t21); SetArg<22, T22>::set(kernel_, t22); SetArg<23, T23>::set(kernel_, t23); SetArg<24, T24>::set(kernel_, t24); SetArg<25, T25>::set(kernel_, t25); SetArg<26, T26>::set(kernel_, t26); SetArg<27, T27>::set(kernel_, t27); SetArg<28, T28>::set(kernel_, t28); SetArg<29, T29>::set(kernel_, t29); SetArg<30, T30>::set(kernel_, t30); SetArg<31, T31>::set(kernel_, t31); args.queue_.enqueueNDRangeKernel( kernel_, args.offset_, args.global_, args.local_, &args.events_, &event); return event; } }; //------------------------------------------------------------------------------------------------------ template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24, typename T25, typename T26, typename T27, typename T28, typename T29, typename T30, typename T31> struct functionImplementation_ { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30, T31> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 32)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30, T31); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23, T24 arg24, T25 arg25, T26 arg26, T27 arg27, T28 arg28, T29 arg29, T30 arg30, T31 arg31) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24, arg25, arg26, arg27, arg28, arg29, arg30, arg31); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24, typename T25, typename T26, typename T27, typename T28, typename T29, typename T30> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 31)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23, T24 arg24, T25 arg25, T26 arg26, T27 arg27, T28 arg28, T29 arg29, T30 arg30) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24, arg25, arg26, arg27, arg28, arg29, arg30); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24, typename T25, typename T26, typename T27, typename T28, typename T29> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 30)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23, T24 arg24, T25 arg25, T26 arg26, T27 arg27, T28 arg28, T29 arg29) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24, arg25, arg26, arg27, arg28, arg29); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24, typename T25, typename T26, typename T27, typename T28> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 29)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23, T24 arg24, T25 arg25, T26 arg26, T27 arg27, T28 arg28) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24, arg25, arg26, arg27, arg28); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24, typename T25, typename T26, typename T27> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 28)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23, T24 arg24, T25 arg25, T26 arg26, T27 arg27) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24, arg25, arg26, arg27); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24, typename T25, typename T26> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 27)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23, T24 arg24, T25 arg25, T26 arg26) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24, arg25, arg26); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24, typename T25> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 26)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23, T24 arg24, T25 arg25) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24, arg25); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23, typename T24> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 25)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23, T24 arg24) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23, arg24); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22, typename T23> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 24)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22, T23 arg23) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22, arg23); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21, typename T22> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 23)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21, T22 arg22) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21, arg22); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20, typename T21> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 22)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20, T21 arg21) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20, arg21); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19, typename T20> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 21)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19, T20 arg20) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19, arg20); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18, typename T19> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 20)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18, T19 arg19) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18, arg19); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17, typename T18> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 19)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17, T18 arg18) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, arg18); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16, typename T17> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 18)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16, T17 arg17) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15, typename T16> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 17)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15, T16 arg16) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14, typename T15> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 16)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14, T15 arg15) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13, typename T14> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 15)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13, T14 arg14) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12, typename T13> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 14)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12, T13 arg13) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11, typename T12> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 13)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11, T12 arg12) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10, typename T11> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 12)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10, T11 arg11) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9, typename T10> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 11)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9, T10 arg10) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 10)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8, T9); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8, T9 arg9) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, T8, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 9)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7, T8); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7, T8 arg8) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, T7, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 8)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6, T7); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6, T7 arg7) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, T6, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 7)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5, T6); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5, T6 arg6) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5, arg6); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4, typename T5> struct functionImplementation_ < T0, T1, T2, T3, T4, T5, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 6)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4, T5); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4, T5 arg5) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4, arg5); } }; template< typename T0, typename T1, typename T2, typename T3, typename T4> struct functionImplementation_ < T0, T1, T2, T3, T4, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 5)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3, T4); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3, T4 arg4) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3, arg4); } }; template< typename T0, typename T1, typename T2, typename T3> struct functionImplementation_ < T0, T1, T2, T3, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 4)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2, T3); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2, T3 arg3) { return functor_( enqueueArgs, arg0, arg1, arg2, arg3); } }; template< typename T0, typename T1, typename T2> struct functionImplementation_ < T0, T1, T2, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, T2, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 3)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1, T2); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1, T2 arg2) { return functor_( enqueueArgs, arg0, arg1, arg2); } }; template< typename T0, typename T1> struct functionImplementation_ < T0, T1, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, T1, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 2)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0, T1); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0, T1 arg1) { return functor_( enqueueArgs, arg0, arg1); } }; template< typename T0> struct functionImplementation_ < T0, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> { typedef detail::KernelFunctorGlobal< T0, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType, NullType> FunctorType; FunctorType functor_; functionImplementation_(const FunctorType &functor) : functor_(functor) { #if (defined(_WIN32) && defined(_VARIADIC_MAX) && (_VARIADIC_MAX < 1)) // Fail variadic expansion for dev11 static_assert(0, "Visual Studio has a hard limit of argument count for a std::function expansion. Please define _VARIADIC_MAX to be 10. If you need more arguments than that VC12 and below cannot support it."); #endif } //! \brief Return type of the functor typedef Event result_type; //! \brief Function signature of kernel functor with no event dependency. typedef Event type_( const EnqueueArgs&, T0); Event operator()( const EnqueueArgs& enqueueArgs, T0 arg0) { return functor_( enqueueArgs, arg0); } }; } // namespace detail //---------------------------------------------------------------------------------------------- template < typename T0, typename T1 = detail::NullType, typename T2 = detail::NullType, typename T3 = detail::NullType, typename T4 = detail::NullType, typename T5 = detail::NullType, typename T6 = detail::NullType, typename T7 = detail::NullType, typename T8 = detail::NullType, typename T9 = detail::NullType, typename T10 = detail::NullType, typename T11 = detail::NullType, typename T12 = detail::NullType, typename T13 = detail::NullType, typename T14 = detail::NullType, typename T15 = detail::NullType, typename T16 = detail::NullType, typename T17 = detail::NullType, typename T18 = detail::NullType, typename T19 = detail::NullType, typename T20 = detail::NullType, typename T21 = detail::NullType, typename T22 = detail::NullType, typename T23 = detail::NullType, typename T24 = detail::NullType, typename T25 = detail::NullType, typename T26 = detail::NullType, typename T27 = detail::NullType, typename T28 = detail::NullType, typename T29 = detail::NullType, typename T30 = detail::NullType, typename T31 = detail::NullType > struct make_kernel : public detail::functionImplementation_< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30, T31 > { public: typedef detail::KernelFunctorGlobal< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30, T31 > FunctorType; make_kernel( const Program& program, const STRING_CLASS name, cl_int * err = NULL) : detail::functionImplementation_< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30, T31 >( FunctorType(program, name, err)) {} make_kernel( const Kernel kernel) : detail::functionImplementation_< T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30, T31 >( FunctorType(kernel)) {} }; //---------------------------------------------------------------------------------------------------------------------- #undef __ERR_STR #if !defined(__CL_USER_OVERRIDE_ERROR_STRINGS) #undef __GET_DEVICE_INFO_ERR #undef __GET_PLATFORM_INFO_ERR #undef __GET_DEVICE_IDS_ERR #undef __GET_CONTEXT_INFO_ERR #undef __GET_EVENT_INFO_ERR #undef __GET_EVENT_PROFILE_INFO_ERR #undef __GET_MEM_OBJECT_INFO_ERR #undef __GET_IMAGE_INFO_ERR #undef __GET_SAMPLER_INFO_ERR #undef __GET_KERNEL_INFO_ERR #undef __GET_KERNEL_ARG_INFO_ERR #undef __GET_KERNEL_WORK_GROUP_INFO_ERR #undef __GET_PROGRAM_INFO_ERR #undef __GET_PROGRAM_BUILD_INFO_ERR #undef __GET_COMMAND_QUEUE_INFO_ERR #undef __CREATE_CONTEXT_ERR #undef __CREATE_CONTEXT_FROM_TYPE_ERR #undef __GET_SUPPORTED_IMAGE_FORMATS_ERR #undef __CREATE_BUFFER_ERR #undef __CREATE_SUBBUFFER_ERR #undef __CREATE_IMAGE2D_ERR #undef __CREATE_IMAGE3D_ERR #undef __CREATE_SAMPLER_ERR #undef __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR #undef __CREATE_USER_EVENT_ERR #undef __SET_USER_EVENT_STATUS_ERR #undef __SET_EVENT_CALLBACK_ERR #undef __SET_PRINTF_CALLBACK_ERR #undef __WAIT_FOR_EVENTS_ERR #undef __CREATE_KERNEL_ERR #undef __SET_KERNEL_ARGS_ERR #undef __CREATE_PROGRAM_WITH_SOURCE_ERR #undef __CREATE_PROGRAM_WITH_BINARY_ERR #undef __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR #undef __BUILD_PROGRAM_ERR #undef __CREATE_KERNELS_IN_PROGRAM_ERR #undef __CREATE_COMMAND_QUEUE_ERR #undef __SET_COMMAND_QUEUE_PROPERTY_ERR #undef __ENQUEUE_READ_BUFFER_ERR #undef __ENQUEUE_WRITE_BUFFER_ERR #undef __ENQUEUE_READ_BUFFER_RECT_ERR #undef __ENQUEUE_WRITE_BUFFER_RECT_ERR #undef __ENQEUE_COPY_BUFFER_ERR #undef __ENQEUE_COPY_BUFFER_RECT_ERR #undef __ENQUEUE_READ_IMAGE_ERR #undef __ENQUEUE_WRITE_IMAGE_ERR #undef __ENQUEUE_COPY_IMAGE_ERR #undef __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR #undef __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR #undef __ENQUEUE_MAP_BUFFER_ERR #undef __ENQUEUE_MAP_IMAGE_ERR #undef __ENQUEUE_UNMAP_MEM_OBJECT_ERR #undef __ENQUEUE_NDRANGE_KERNEL_ERR #undef __ENQUEUE_TASK_ERR #undef __ENQUEUE_NATIVE_KERNEL #undef __CL_EXPLICIT_CONSTRUCTORS #undef __UNLOAD_COMPILER_ERR #endif //__CL_USER_OVERRIDE_ERROR_STRINGS #undef __CL_FUNCTION_TYPE // Extensions /** * Deprecated APIs for 1.2 */ #if defined(CL_VERSION_1_1) #undef __INIT_CL_EXT_FCN_PTR #endif // #if defined(CL_VERSION_1_1) #undef __CREATE_SUB_DEVICES #if defined(USE_CL_DEVICE_FISSION) #undef __PARAM_NAME_DEVICE_FISSION #endif // USE_CL_DEVICE_FISSION #undef __DEFAULT_NOT_INITIALIZED #undef __DEFAULT_BEING_INITIALIZED #undef __DEFAULT_INITIALIZED } // namespace cl #ifdef _WIN32 #pragma pop_macro("max") #endif // _WIN32 #endif // CL_HPP_