aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorPierre Moreau <[email protected]>2018-01-28 11:38:46 +0100
committerKarol Herbst <[email protected]>2019-02-26 21:02:07 +0100
commitdddc5649bfa919713cfdae61fd3f4e756a40873c (patch)
treee9cdb27273f834c1724fb7c99542504eeab182f4
parent2ae07830e74549e3b5ff89ecb40feb595786aaa8 (diff)
include/CL: Update to the latest OpenCL 2.2 headers
Acked-by: Francisco Jerez <[email protected]> Reviewed-by: Karol Herbst <[email protected]>
Diffstat
-rw-r--r--include/CL/cl.h1796
-rw-r--r--include/CL/cl.hpp11639
-rw-r--r--include/CL/cl2.hpp9690
-rw-r--r--include/CL/cl_d3d10.h7
-rw-r--r--include/CL/cl_d3d11.h7
-rw-r--r--include/CL/cl_dx9_media_sharing.h9
-rw-r--r--include/CL/cl_dx9_media_sharing_intel.h182
-rw-r--r--include/CL/cl_egl.h101
-rw-r--r--include/CL/cl_ext.h672
-rw-r--r--include/CL/cl_ext_intel.h423
-rw-r--r--include/CL/cl_gl.h143
-rw-r--r--include/CL/cl_gl_ext.h39
-rw-r--r--include/CL/cl_platform.h605
-rw-r--r--include/CL/cl_va_api_media_sharing_intel.h172
-rw-r--r--include/CL/cl_version.h86
-rw-r--r--include/CL/opencl.h19
-rw-r--r--include/meson.build5
-rw-r--r--src/gallium/state_trackers/clover/Makefile.am5
18 files changed, 23136 insertions, 2464 deletions
diff --git a/include/CL/cl.h b/include/CL/cl.h
index 316565d6e45..54f6b29362a 100644
--- a/include/CL/cl.h
+++ b/include/CL/cl.h
@@ -1,5 +1,5 @@
/*******************************************************************************
- * Copyright (c) 2008 - 2012 The Khronos Group Inc.
+ * Copyright (c) 2008-2019 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -24,11 +29,8 @@
#ifndef __OPENCL_CL_H
#define __OPENCL_CL_H
-#ifdef __APPLE__
-#include <OpenCL/cl_platform.h>
-#else
+#include <CL/cl_version.h>
#include <CL/cl_platform.h>
-#endif
#ifdef __cplusplus
extern "C" {
@@ -46,7 +48,7 @@ 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_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;
@@ -55,45 +57,76 @@ 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;
+#ifdef CL_VERSION_2_0
+typedef cl_bitfield cl_device_svm_capabilities;
+#endif
typedef cl_bitfield cl_command_queue_properties;
+#ifdef CL_VERSION_1_2
typedef intptr_t cl_device_partition_property;
typedef cl_bitfield cl_device_affinity_domain;
+#endif
typedef intptr_t cl_context_properties;
typedef cl_uint cl_context_info;
+#ifdef CL_VERSION_2_0
+typedef cl_bitfield cl_queue_properties;
+#endif
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;
+#ifdef CL_VERSION_2_0
+typedef cl_bitfield cl_svm_mem_flags;
+#endif
typedef cl_uint cl_mem_object_type;
typedef cl_uint cl_mem_info;
+#ifdef CL_VERSION_1_2
typedef cl_bitfield cl_mem_migration_flags;
+#endif
typedef cl_uint cl_image_info;
+#ifdef CL_VERSION_1_1
typedef cl_uint cl_buffer_create_type;
+#endif
typedef cl_uint cl_addressing_mode;
typedef cl_uint cl_filter_mode;
typedef cl_uint cl_sampler_info;
typedef cl_bitfield cl_map_flags;
+#ifdef CL_VERSION_2_0
+typedef intptr_t cl_pipe_properties;
+typedef cl_uint cl_pipe_info;
+#endif
typedef cl_uint cl_program_info;
typedef cl_uint cl_program_build_info;
+#ifdef CL_VERSION_1_2
typedef cl_uint cl_program_binary_type;
+#endif
typedef cl_int cl_build_status;
typedef cl_uint cl_kernel_info;
+#ifdef CL_VERSION_1_2
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;
+#endif
typedef cl_uint cl_kernel_work_group_info;
+#ifdef CL_VERSION_2_1
+typedef cl_uint cl_kernel_sub_group_info;
+#endif
typedef cl_uint cl_event_info;
typedef cl_uint cl_command_type;
typedef cl_uint cl_profiling_info;
-
+#ifdef CL_VERSION_2_0
+typedef cl_bitfield cl_sampler_properties;
+typedef cl_uint cl_kernel_exec_info;
+#endif
typedef struct _cl_image_format {
cl_channel_order image_channel_order;
cl_channel_type image_channel_data_type;
} cl_image_format;
+#ifdef CL_VERSION_1_2
+
typedef struct _cl_image_desc {
cl_mem_object_type image_type;
size_t image_width;
@@ -104,14 +137,32 @@ typedef struct _cl_image_desc {
size_t image_slice_pitch;
cl_uint num_mip_levels;
cl_uint num_samples;
- cl_mem buffer;
+#ifdef __GNUC__
+ __extension__ /* Prevents warnings about anonymous union in -pedantic builds */
+#endif
+#ifdef _MSC_VER
+#pragma warning( push )
+#pragma warning( disable : 4201 ) /* Prevents warning about nameless struct/union in /W4 /Za builds */
+#endif
+ union {
+ cl_mem buffer;
+ cl_mem mem_object;
+ };
+#ifdef _MSC_VER
+#pragma warning( pop )
+#endif
} cl_image_desc;
+#endif
+
+#ifdef CL_VERSION_1_1
+
typedef struct _cl_buffer_region {
size_t origin;
size_t size;
} cl_buffer_region;
+#endif
/******************************************************************************/
@@ -129,13 +180,17 @@ typedef struct _cl_buffer_region {
#define CL_IMAGE_FORMAT_NOT_SUPPORTED -10
#define CL_BUILD_PROGRAM_FAILURE -11
#define CL_MAP_FAILURE -12
+#ifdef CL_VERSION_1_1
#define CL_MISALIGNED_SUB_BUFFER_OFFSET -13
#define CL_EXEC_STATUS_ERROR_FOR_EVENTS_IN_WAIT_LIST -14
+#endif
+#ifdef CL_VERSION_1_2
#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
+#endif
#define CL_INVALID_VALUE -30
#define CL_INVALID_DEVICE_TYPE -31
@@ -171,22 +226,32 @@ typedef struct _cl_buffer_region {
#define CL_INVALID_BUFFER_SIZE -61
#define CL_INVALID_MIP_LEVEL -62
#define CL_INVALID_GLOBAL_WORK_SIZE -63
+#ifdef CL_VERSION_1_1
#define CL_INVALID_PROPERTY -64
+#endif
+#ifdef CL_VERSION_1_2
#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
+#endif
+#ifdef CL_VERSION_2_0
+#define CL_INVALID_PIPE_SIZE -69
+#define CL_INVALID_DEVICE_QUEUE -70
+#endif
+#ifdef CL_VERSION_2_2
+#define CL_INVALID_SPEC_ID -71
+#define CL_MAX_SIZE_RESTRICTION_EXCEEDED -72
+#endif
-/* 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
+#ifdef CL_VERSION_1_2
#define CL_BLOCKING CL_TRUE
#define CL_NON_BLOCKING CL_FALSE
+#endif
/* cl_platform_info */
#define CL_PLATFORM_PROFILE 0x0900
@@ -194,92 +259,128 @@ typedef struct _cl_buffer_region {
#define CL_PLATFORM_NAME 0x0902
#define CL_PLATFORM_VENDOR 0x0903
#define CL_PLATFORM_EXTENSIONS 0x0904
+#ifdef CL_VERSION_2_1
+#define CL_PLATFORM_HOST_TIMER_RESOLUTION 0x0905
+#endif
/* 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)
+#ifdef CL_VERSION_1_2
#define CL_DEVICE_TYPE_CUSTOM (1 << 4)
+#endif
#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
+#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 /* deprecated */
+#ifdef CL_VERSION_2_0
+#define CL_DEVICE_QUEUE_ON_HOST_PROPERTIES 0x102A
+#endif
+#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
+#ifdef CL_VERSION_1_2
+#define CL_DEVICE_DOUBLE_FP_CONFIG 0x1032
+#endif
+/* 0x1033 reserved for CL_DEVICE_HALF_FP_CONFIG which is already defined in "cl_ext.h" */
+#ifdef CL_VERSION_1_1
+#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF 0x1034
+#define CL_DEVICE_HOST_UNIFIED_MEMORY 0x1035 /* deprecated */
+#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
+#endif
+#ifdef CL_VERSION_1_2
+#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
+#endif
+#ifdef CL_VERSION_2_0
+#define CL_DEVICE_MAX_READ_WRITE_IMAGE_ARGS 0x104C
+#define CL_DEVICE_MAX_GLOBAL_VARIABLE_SIZE 0x104D
+#define CL_DEVICE_QUEUE_ON_DEVICE_PROPERTIES 0x104E
+#define CL_DEVICE_QUEUE_ON_DEVICE_PREFERRED_SIZE 0x104F
+#define CL_DEVICE_QUEUE_ON_DEVICE_MAX_SIZE 0x1050
+#define CL_DEVICE_MAX_ON_DEVICE_QUEUES 0x1051
+#define CL_DEVICE_MAX_ON_DEVICE_EVENTS 0x1052
+#define CL_DEVICE_SVM_CAPABILITIES 0x1053
+#define CL_DEVICE_GLOBAL_VARIABLE_PREFERRED_TOTAL_SIZE 0x1054
+#define CL_DEVICE_MAX_PIPE_ARGS 0x1055
+#define CL_DEVICE_PIPE_MAX_ACTIVE_RESERVATIONS 0x1056
+#define CL_DEVICE_PIPE_MAX_PACKET_SIZE 0x1057
+#define CL_DEVICE_PREFERRED_PLATFORM_ATOMIC_ALIGNMENT 0x1058
+#define CL_DEVICE_PREFERRED_GLOBAL_ATOMIC_ALIGNMENT 0x1059
+#define CL_DEVICE_PREFERRED_LOCAL_ATOMIC_ALIGNMENT 0x105A
+#endif
+#ifdef CL_VERSION_2_1
+#define CL_DEVICE_IL_VERSION 0x105B
+#define CL_DEVICE_MAX_NUM_SUB_GROUPS 0x105C
+#define CL_DEVICE_SUB_GROUP_INDEPENDENT_FORWARD_PROGRESS 0x105D
+#endif
/* cl_device_fp_config - bitfield */
#define CL_FP_DENORM (1 << 0)
@@ -288,8 +389,12 @@ typedef struct _cl_buffer_region {
#define CL_FP_ROUND_TO_ZERO (1 << 3)
#define CL_FP_ROUND_TO_INF (1 << 4)
#define CL_FP_FMA (1 << 5)
+#ifdef CL_VERSION_1_1
#define CL_FP_SOFT_FLOAT (1 << 6)
+#endif
+#ifdef CL_VERSION_1_2
#define CL_FP_CORRECTLY_ROUNDED_DIVIDE_SQRT (1 << 7)
+#endif
/* cl_device_mem_cache_type */
#define CL_NONE 0x0
@@ -307,38 +412,70 @@ typedef struct _cl_buffer_region {
/* cl_command_queue_properties - bitfield */
#define CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE (1 << 0)
#define CL_QUEUE_PROFILING_ENABLE (1 << 1)
+#ifdef CL_VERSION_2_0
+#define CL_QUEUE_ON_DEVICE (1 << 2)
+#define CL_QUEUE_ON_DEVICE_DEFAULT (1 << 3)
+#endif
-/* cl_context_info */
+/* cl_context_info */
#define CL_CONTEXT_REFERENCE_COUNT 0x1080
#define CL_CONTEXT_DEVICES 0x1081
#define CL_CONTEXT_PROPERTIES 0x1082
+#ifdef CL_VERSION_1_1
#define CL_CONTEXT_NUM_DEVICES 0x1083
+#endif
/* cl_context_properties */
#define CL_CONTEXT_PLATFORM 0x1084
+#ifdef CL_VERSION_1_2
#define CL_CONTEXT_INTEROP_USER_SYNC 0x1085
-
+#endif
+
+#ifdef CL_VERSION_1_2
+
/* 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
-
+
+#endif
+
+#ifdef CL_VERSION_1_2
+
/* 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)
+#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)
+
+#endif
+
+#ifdef CL_VERSION_2_0
+
+/* cl_device_svm_capabilities */
+#define CL_DEVICE_SVM_COARSE_GRAIN_BUFFER (1 << 0)
+#define CL_DEVICE_SVM_FINE_GRAIN_BUFFER (1 << 1)
+#define CL_DEVICE_SVM_FINE_GRAIN_SYSTEM (1 << 2)
+#define CL_DEVICE_SVM_ATOMICS (1 << 3)
+
+#endif
/* 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
+#ifdef CL_VERSION_2_0
+#define CL_QUEUE_SIZE 0x1094
+#endif
+#ifdef CL_VERSION_2_1
+#define CL_QUEUE_DEVICE_DEFAULT 0x1095
+#endif
-/* cl_mem_flags - bitfield */
+/* cl_mem_flags and cl_svm_mem_flags - bitfield */
#define CL_MEM_READ_WRITE (1 << 0)
#define CL_MEM_WRITE_ONLY (1 << 1)
#define CL_MEM_READ_ONLY (1 << 2)
@@ -346,14 +483,25 @@ typedef struct _cl_buffer_region {
#define CL_MEM_ALLOC_HOST_PTR (1 << 4)
#define CL_MEM_COPY_HOST_PTR (1 << 5)
/* reserved (1 << 6) */
+#ifdef CL_VERSION_1_2
#define CL_MEM_HOST_WRITE_ONLY (1 << 7)
#define CL_MEM_HOST_READ_ONLY (1 << 8)
#define CL_MEM_HOST_NO_ACCESS (1 << 9)
+#endif
+#ifdef CL_VERSION_2_0
+#define CL_MEM_SVM_FINE_GRAIN_BUFFER (1 << 10) /* used by cl_svm_mem_flags only */
+#define CL_MEM_SVM_ATOMICS (1 << 11) /* used by cl_svm_mem_flags only */
+#define CL_MEM_KERNEL_READ_AND_WRITE (1 << 12)
+#endif
+
+#ifdef CL_VERSION_1_2
/* cl_mem_migration_flags - bitfield */
#define CL_MIGRATE_MEM_OBJECT_HOST (1 << 0)
#define CL_MIGRATE_MEM_OBJECT_CONTENT_UNDEFINED (1 << 1)
+#endif
+
/* cl_channel_order */
#define CL_R 0x10B0
#define CL_A 0x10B1
@@ -365,11 +513,22 @@ typedef struct _cl_buffer_region {
#define CL_ARGB 0x10B7
#define CL_INTENSITY 0x10B8
#define CL_LUMINANCE 0x10B9
+#ifdef CL_VERSION_1_1
#define CL_Rx 0x10BA
#define CL_RGx 0x10BB
#define CL_RGBx 0x10BC
+#endif
+#ifdef CL_VERSION_1_2
#define CL_DEPTH 0x10BD
#define CL_DEPTH_STENCIL 0x10BE
+#endif
+#ifdef CL_VERSION_2_0
+#define CL_sRGB 0x10BF
+#define CL_sRGBx 0x10C0
+#define CL_sRGBA 0x10C1
+#define CL_sBGRA 0x10C2
+#define CL_ABGR 0x10C3
+#endif
/* cl_channel_type */
#define CL_SNORM_INT8 0x10D0
@@ -387,16 +546,26 @@ typedef struct _cl_buffer_region {
#define CL_UNSIGNED_INT32 0x10DC
#define CL_HALF_FLOAT 0x10DD
#define CL_FLOAT 0x10DE
+#ifdef CL_VERSION_1_2
#define CL_UNORM_INT24 0x10DF
+#endif
+#ifdef CL_VERSION_2_1
+#define CL_UNORM_INT_101010_2 0x10E0
+#endif
/* cl_mem_object_type */
#define CL_MEM_OBJECT_BUFFER 0x10F0
#define CL_MEM_OBJECT_IMAGE2D 0x10F1
#define CL_MEM_OBJECT_IMAGE3D 0x10F2
+#ifdef CL_VERSION_1_2
#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
+#endif
+#ifdef CL_VERSION_2_0
+#define CL_MEM_OBJECT_PIPE 0x10F7
+#endif
/* cl_mem_info */
#define CL_MEM_TYPE 0x1100
@@ -406,8 +575,13 @@ typedef struct _cl_buffer_region {
#define CL_MEM_MAP_COUNT 0x1104
#define CL_MEM_REFERENCE_COUNT 0x1105
#define CL_MEM_CONTEXT 0x1106
+#ifdef CL_VERSION_1_1
#define CL_MEM_ASSOCIATED_MEMOBJECT 0x1107
#define CL_MEM_OFFSET 0x1108
+#endif
+#ifdef CL_VERSION_2_0
+#define CL_MEM_USES_SVM_POINTER 0x1109
+#endif
/* cl_image_info */
#define CL_IMAGE_FORMAT 0x1110
@@ -417,17 +591,29 @@ typedef struct _cl_buffer_region {
#define CL_IMAGE_WIDTH 0x1114
#define CL_IMAGE_HEIGHT 0x1115
#define CL_IMAGE_DEPTH 0x1116
+#ifdef CL_VERSION_1_2
#define CL_IMAGE_ARRAY_SIZE 0x1117
#define CL_IMAGE_BUFFER 0x1118
#define CL_IMAGE_NUM_MIP_LEVELS 0x1119
#define CL_IMAGE_NUM_SAMPLES 0x111A
+#endif
+
+#ifdef CL_VERSION_2_0
+
+/* cl_pipe_info */
+#define CL_PIPE_PACKET_SIZE 0x1120
+#define CL_PIPE_MAX_PACKETS 0x1121
+
+#endif
/* 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
+#ifdef CL_VERSION_1_1
#define CL_ADDRESS_MIRRORED_REPEAT 0x1134
+#endif
/* cl_filter_mode */
#define CL_FILTER_NEAREST 0x1140
@@ -439,11 +625,21 @@ typedef struct _cl_buffer_region {
#define CL_SAMPLER_NORMALIZED_COORDS 0x1152
#define CL_SAMPLER_ADDRESSING_MODE 0x1153
#define CL_SAMPLER_FILTER_MODE 0x1154
+#ifdef CL_VERSION_2_0
+/* These enumerants are for the cl_khr_mipmap_image extension.
+ They have since been added to cl_ext.h with an appropriate
+ KHR suffix, but are left here for backwards compatibility. */
+#define CL_SAMPLER_MIP_FILTER_MODE 0x1155
+#define CL_SAMPLER_LOD_MIN 0x1156
+#define CL_SAMPLER_LOD_MAX 0x1157
+#endif
/* cl_map_flags - bitfield */
#define CL_MAP_READ (1 << 0)
#define CL_MAP_WRITE (1 << 1)
+#ifdef CL_VERSION_1_2
#define CL_MAP_WRITE_INVALIDATE_REGION (1 << 2)
+#endif
/* cl_program_info */
#define CL_PROGRAM_REFERENCE_COUNT 0x1160
@@ -453,21 +649,39 @@ typedef struct _cl_buffer_region {
#define CL_PROGRAM_SOURCE 0x1164
#define CL_PROGRAM_BINARY_SIZES 0x1165
#define CL_PROGRAM_BINARIES 0x1166
+#ifdef CL_VERSION_1_2
#define CL_PROGRAM_NUM_KERNELS 0x1167
#define CL_PROGRAM_KERNEL_NAMES 0x1168
+#endif
+#ifdef CL_VERSION_2_1
+#define CL_PROGRAM_IL 0x1169
+#endif
+#ifdef CL_VERSION_2_2
+#define CL_PROGRAM_SCOPE_GLOBAL_CTORS_PRESENT 0x116A
+#define CL_PROGRAM_SCOPE_GLOBAL_DTORS_PRESENT 0x116B
+#endif
/* cl_program_build_info */
#define CL_PROGRAM_BUILD_STATUS 0x1181
#define CL_PROGRAM_BUILD_OPTIONS 0x1182
#define CL_PROGRAM_BUILD_LOG 0x1183
+#ifdef CL_VERSION_1_2
#define CL_PROGRAM_BINARY_TYPE 0x1184
-
+#endif
+#ifdef CL_VERSION_2_0
+#define CL_PROGRAM_BUILD_GLOBAL_VARIABLE_TOTAL_SIZE 0x1185
+#endif
+
+#ifdef CL_VERSION_1_2
+
/* 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
+#endif
+
/* cl_build_status */
#define CL_BUILD_SUCCESS 0
#define CL_BUILD_NONE -1
@@ -480,7 +694,15 @@ typedef struct _cl_buffer_region {
#define CL_KERNEL_REFERENCE_COUNT 0x1192
#define CL_KERNEL_CONTEXT 0x1193
#define CL_KERNEL_PROGRAM 0x1194
+#ifdef CL_VERSION_1_2
#define CL_KERNEL_ATTRIBUTES 0x1195
+#endif
+#ifdef CL_VERSION_2_1
+#define CL_KERNEL_MAX_NUM_SUB_GROUPS 0x11B9
+#define CL_KERNEL_COMPILE_NUM_SUB_GROUPS 0x11BA
+#endif
+
+#ifdef CL_VERSION_1_2
/* cl_kernel_arg_info */
#define CL_KERNEL_ARG_ADDRESS_QUALIFIER 0x1196
@@ -489,23 +711,40 @@ typedef struct _cl_buffer_region {
#define CL_KERNEL_ARG_TYPE_QUALIFIER 0x1199
#define CL_KERNEL_ARG_NAME 0x119A
+#endif
+
+#ifdef CL_VERSION_1_2
+
/* 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
+#endif
+
+#ifdef CL_VERSION_1_2
+
/* 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 */
+
+#endif
+
+#ifdef CL_VERSION_1_2
+
+/* cl_kernel_arg_type_qualifier */
#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)
+#ifdef CL_VERSION_2_0
+#define CL_KERNEL_ARG_TYPE_PIPE (1 << 3)
+#endif
+
+#endif
/* cl_kernel_work_group_info */
#define CL_KERNEL_WORK_GROUP_SIZE 0x11B0
@@ -513,14 +752,35 @@ typedef struct _cl_buffer_region {
#define CL_KERNEL_LOCAL_MEM_SIZE 0x11B2
#define CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE 0x11B3
#define CL_KERNEL_PRIVATE_MEM_SIZE 0x11B4
+#ifdef CL_VERSION_1_2
#define CL_KERNEL_GLOBAL_WORK_SIZE 0x11B5
+#endif
+
+#ifdef CL_VERSION_2_1
-/* cl_event_info */
+/* cl_kernel_sub_group_info */
+#define CL_KERNEL_MAX_SUB_GROUP_SIZE_FOR_NDRANGE 0x2033
+#define CL_KERNEL_SUB_GROUP_COUNT_FOR_NDRANGE 0x2034
+#define CL_KERNEL_LOCAL_SIZE_FOR_SUB_GROUP_COUNT 0x11B8
+
+#endif
+
+#ifdef CL_VERSION_2_0
+
+/* cl_kernel_exec_info */
+#define CL_KERNEL_EXEC_INFO_SVM_PTRS 0x11B6
+#define CL_KERNEL_EXEC_INFO_SVM_FINE_GRAIN_SYSTEM 0x11B7
+
+#endif
+
+/* 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
+#ifdef CL_VERSION_1_1
#define CL_EVENT_CONTEXT 0x11D4
+#endif
/* cl_command_type */
#define CL_COMMAND_NDRANGE_KERNEL 0x11F0
@@ -540,14 +800,25 @@ typedef struct _cl_buffer_region {
#define CL_COMMAND_MARKER 0x11FE
#define CL_COMMAND_ACQUIRE_GL_OBJECTS 0x11FF
#define CL_COMMAND_RELEASE_GL_OBJECTS 0x1200
+#ifdef CL_VERSION_1_1
#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
+#endif
+#ifdef CL_VERSION_1_2
#define CL_COMMAND_BARRIER 0x1205
#define CL_COMMAND_MIGRATE_MEM_OBJECTS 0x1206
#define CL_COMMAND_FILL_BUFFER 0x1207
#define CL_COMMAND_FILL_IMAGE 0x1208
+#endif
+#ifdef CL_VERSION_2_0
+#define CL_COMMAND_SVM_FREE 0x1209
+#define CL_COMMAND_SVM_MEMCPY 0x120A
+#define CL_COMMAND_SVM_MEMFILL 0x120B
+#define CL_COMMAND_SVM_MAP 0x120C
+#define CL_COMMAND_SVM_UNMAP 0x120D
+#endif
/* command execution status */
#define CL_COMPLETE 0x0
@@ -555,660 +826,975 @@ typedef struct _cl_buffer_region {
#define CL_SUBMITTED 0x2
#define CL_QUEUED 0x3
-/* cl_buffer_create_type */
+#ifdef CL_VERSION_1_1
+
+/* cl_buffer_create_type */
#define CL_BUFFER_CREATE_TYPE_REGION 0x1220
-/* cl_profiling_info */
+#endif
+
+/* 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
+#ifdef CL_VERSION_2_0
+#define CL_PROFILING_COMMAND_COMPLETE 0x1284
+#endif
/********************************************************************************************************/
/* 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;
+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;
+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;
+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;
+
+#ifdef CL_VERSION_1_2
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;
-
+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
-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;
+clRetainDevice(cl_device_id device) 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;
-
+clReleaseDevice(cl_device_id device) CL_API_SUFFIX__VERSION_1_2;
+
+#endif
+
+#ifdef CL_VERSION_2_1
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clSetDefaultDeviceCommandQueue(cl_context context,
+ cl_device_id device,
+ cl_command_queue command_queue) CL_API_SUFFIX__VERSION_2_1;
+
extern CL_API_ENTRY cl_int CL_API_CALL
-clReleaseDevice(cl_device_id /* device */) CL_API_SUFFIX__VERSION_1_2;
-
-/* Context APIs */
+clGetDeviceAndHostTimer(cl_device_id device,
+ cl_ulong* device_timestamp,
+ cl_ulong* host_timestamp) CL_API_SUFFIX__VERSION_2_1;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clGetHostTimer(cl_device_id device,
+ cl_ulong * host_timestamp) CL_API_SUFFIX__VERSION_2_1;
+
+#endif
+
+/* 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;
+clCreateContext(const cl_context_properties * properties,
+ cl_uint num_devices,
+ const cl_device_id * devices,
+ void (CL_CALLBACK * pfn_notify)(const char * errinfo,
+ const void * private_info,
+ size_t cb,
+ void * user_data),
+ 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;
+clCreateContextFromType(const cl_context_properties * properties,
+ cl_device_type device_type,
+ void (CL_CALLBACK * pfn_notify)(const char * errinfo,
+ const void * private_info,
+ size_t cb,
+ void * user_data),
+ 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;
+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;
+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;
+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 */
+
+#ifdef CL_VERSION_2_0
+
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;
+clCreateCommandQueueWithProperties(cl_context context,
+ cl_device_id device,
+ const cl_queue_properties * properties,
+ cl_int * errcode_ret) CL_API_SUFFIX__VERSION_2_0;
+
+#endif
extern CL_API_ENTRY cl_int CL_API_CALL
-clRetainCommandQueue(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
+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;
+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;
+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;
+clCreateBuffer(cl_context context,
+ cl_mem_flags flags,
+ size_t size,
+ void * host_ptr,
+ cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0;
+
+#ifdef CL_VERSION_1_1
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;
+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;
+
+#endif
+
+#ifdef CL_VERSION_1_2
+
+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;
+
+#endif
+
+#ifdef CL_VERSION_2_0
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;
-
+clCreatePipe(cl_context context,
+ cl_mem_flags flags,
+ cl_uint pipe_packet_size,
+ cl_uint pipe_max_packets,
+ const cl_pipe_properties * properties,
+ cl_int * errcode_ret) CL_API_SUFFIX__VERSION_2_0;
+
+#endif
+
extern CL_API_ENTRY cl_int CL_API_CALL
-clRetainMemObject(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0;
+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;
+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;
-
+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;
+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;
+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;
+
+#ifdef CL_VERSION_2_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;
+clGetPipeInfo(cl_mem pipe,
+ cl_pipe_info param_name,
+ size_t param_value_size,
+ void * param_value,
+ size_t * param_value_size_ret) CL_API_SUFFIX__VERSION_2_0;
+
+#endif
+
+#ifdef CL_VERSION_1_1
+
+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;
+
+#endif
+
+/* SVM Allocation APIs */
+
+#ifdef CL_VERSION_2_0
+
+extern CL_API_ENTRY void * CL_API_CALL
+clSVMAlloc(cl_context context,
+ cl_svm_mem_flags flags,
+ size_t size,
+ cl_uint alignment) CL_API_SUFFIX__VERSION_2_0;
+
+extern CL_API_ENTRY void CL_API_CALL
+clSVMFree(cl_context context,
+ void * svm_pointer) CL_API_SUFFIX__VERSION_2_0;
+
+#endif
/* Sampler APIs */
+
+#ifdef CL_VERSION_2_0
+
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;
+clCreateSamplerWithProperties(cl_context context,
+ const cl_sampler_properties * sampler_properties,
+ cl_int * errcode_ret) CL_API_SUFFIX__VERSION_2_0;
+
+#endif
extern CL_API_ENTRY cl_int CL_API_CALL
-clRetainSampler(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0;
+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;
+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 */
+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;
+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;
+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;
+
+#ifdef CL_VERSION_1_2
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;
+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;
+
+#endif
+
+#ifdef CL_VERSION_2_1
+
+extern CL_API_ENTRY cl_program CL_API_CALL
+clCreateProgramWithIL(cl_context context,
+ const void* il,
+ size_t length,
+ cl_int* errcode_ret) CL_API_SUFFIX__VERSION_2_1;
+
+#endif
extern CL_API_ENTRY cl_int CL_API_CALL
-clRetainProgram(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0;
+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;
+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;
+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;
+
+#ifdef CL_VERSION_1_2
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;
+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;
+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;
+
+#endif
+
+#ifdef CL_VERSION_2_2
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clSetProgramReleaseCallback(cl_program program,
+ void (CL_CALLBACK * pfn_notify)(cl_program program,
+ void * user_data),
+ void * user_data) CL_API_SUFFIX__VERSION_2_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clSetProgramSpecializationConstant(cl_program program,
+ cl_uint spec_id,
+ size_t spec_size,
+ const void* spec_value) CL_API_SUFFIX__VERSION_2_2;
+
+#endif
+#ifdef CL_VERSION_1_2
extern CL_API_ENTRY cl_int CL_API_CALL
-clUnloadPlatformCompiler(cl_platform_id /* platform */) CL_API_SUFFIX__VERSION_1_2;
+clUnloadPlatformCompiler(cl_platform_id platform) CL_API_SUFFIX__VERSION_1_2;
+
+#endif
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;
+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;
-
+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;
+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;
+clCreateKernelsInProgram(cl_program program,
+ cl_uint num_kernels,
+ cl_kernel * kernels,
+ cl_uint * num_kernels_ret) CL_API_SUFFIX__VERSION_1_0;
+
+#ifdef CL_VERSION_2_1
+
+extern CL_API_ENTRY cl_kernel CL_API_CALL
+clCloneKernel(cl_kernel source_kernel,
+ cl_int* errcode_ret) CL_API_SUFFIX__VERSION_2_1;
+
+#endif
extern CL_API_ENTRY cl_int CL_API_CALL
-clRetainKernel(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0;
+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;
+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;
+clSetKernelArg(cl_kernel kernel,
+ cl_uint arg_index,
+ size_t arg_size,
+ const void * arg_value) CL_API_SUFFIX__VERSION_1_0;
+
+#ifdef CL_VERSION_2_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;
+clSetKernelArgSVMPointer(cl_kernel kernel,
+ cl_uint arg_index,
+ const void * arg_value) CL_API_SUFFIX__VERSION_2_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;
+clSetKernelExecInfo(cl_kernel kernel,
+ cl_kernel_exec_info param_name,
+ size_t param_value_size,
+ const void * param_value) CL_API_SUFFIX__VERSION_2_0;
+
+#endif
+
+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;
+
+#ifdef CL_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;
+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;
+
+#endif
+
+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;
+
+#ifdef CL_VERSION_2_1
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clGetKernelSubGroupInfo(cl_kernel kernel,
+ cl_device_id device,
+ cl_kernel_sub_group_info param_name,
+ size_t input_value_size,
+ const void* input_value,
+ size_t param_value_size,
+ void* param_value,
+ size_t* param_value_size_ret) CL_API_SUFFIX__VERSION_2_1;
+
+#endif
/* 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;
+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;
-
+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;
+
+#ifdef CL_VERSION_1_1
+
extern CL_API_ENTRY cl_event CL_API_CALL
-clCreateUserEvent(cl_context /* context */,
- cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1;
-
+clCreateUserEvent(cl_context context,
+ cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_1;
+
+#endif
+
extern CL_API_ENTRY cl_int CL_API_CALL
-clRetainEvent(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0;
+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;
+clReleaseEvent(cl_event event) CL_API_SUFFIX__VERSION_1_0;
+
+#ifdef CL_VERSION_1_1
extern CL_API_ENTRY cl_int CL_API_CALL
-clSetUserEventStatus(cl_event /* event */,
- cl_int /* execution_status */) CL_API_SUFFIX__VERSION_1_1;
-
+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;
+clSetEventCallback(cl_event event,
+ cl_int command_exec_callback_type,
+ void (CL_CALLBACK * pfn_notify)(cl_event event,
+ cl_int event_command_status,
+ void * user_data),
+ void * user_data) CL_API_SUFFIX__VERSION_1_1;
+
+#endif
/* 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;
-
+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;
+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;
+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;
+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;
+
+#ifdef CL_VERSION_1_1
+
+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;
+
+#endif
+
+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;
+
+#ifdef CL_VERSION_1_1
+
+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;
+
+#endif
+
+#ifdef CL_VERSION_1_2
+
+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;
+
+#endif
+
+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;
+
+#ifdef CL_VERSION_1_1
+
+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;
+
+#endif
+
+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,
+ const size_t * region,
+ 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,
+ const size_t * region,
+ 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;
+
+#ifdef CL_VERSION_1_2
+
+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,
+ const size_t * region,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_API_SUFFIX__VERSION_1_2;
+
+#endif
+
+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,
+ const size_t * dst_origin,
+ const size_t * region,
+ 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,
+ const size_t * region,
+ 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,
+ const size_t * region,
+ 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;
+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;
+clEnqueueMapImage(cl_command_queue command_queue,
+ cl_mem image,
+ cl_bool blocking_map,
+ cl_map_flags map_flags,
+ const size_t * origin,
+ const size_t * region,
+ 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;
+
+#ifdef CL_VERSION_1_2
+
+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;
+
+#endif
+
+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
+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;
+
+#ifdef CL_VERSION_1_2
+
+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;
+
+#endif
+
+#ifdef CL_VERSION_2_0
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMFree(cl_command_queue command_queue,
+ cl_uint num_svm_pointers,
+ void * svm_pointers[],
+ void (CL_CALLBACK * pfn_free_func)(cl_command_queue queue,
+ cl_uint num_svm_pointers,
+ void * svm_pointers[],
+ void * user_data),
+ void * user_data,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_API_SUFFIX__VERSION_2_0;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMMemcpy(cl_command_queue command_queue,
+ cl_bool blocking_copy,
+ void * dst_ptr,
+ const void * src_ptr,
+ size_t size,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_API_SUFFIX__VERSION_2_0;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMMemFill(cl_command_queue command_queue,
+ void * svm_ptr,
+ const void * pattern,
+ size_t pattern_size,
+ size_t size,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_API_SUFFIX__VERSION_2_0;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMMap(cl_command_queue command_queue,
+ cl_bool blocking_map,
+ cl_map_flags flags,
+ void * svm_ptr,
+ size_t size,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_API_SUFFIX__VERSION_2_0;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMUnmap(cl_command_queue command_queue,
+ void * svm_ptr,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_API_SUFFIX__VERSION_2_0;
+#endif
+
+#ifdef CL_VERSION_2_1
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMMigrateMem(cl_command_queue command_queue,
+ cl_uint num_svm_pointers,
+ const void ** svm_pointers,
+ const size_t * sizes,
+ 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_2_1;
+
+#endif
+
+#ifdef CL_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
+ * 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;
-
+extern CL_API_ENTRY void * CL_API_CALL
+clGetExtensionFunctionAddressForPlatform(cl_platform_id platform,
+ const char * func_name) CL_API_SUFFIX__VERSION_1_2;
+
+#endif
+
+#ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS
+ /*
+ * WARNING:
+ * This API introduces mutable state into the OpenCL implementation. It has been REMOVED
+ * to better facilitate thread safety. The 1.0 API is not thread safe. It is not tested by the
+ * OpenCL 1.1 conformance test, and consequently may not work or may not work dependably.
+ * It is likely to be non-performant. Use of this API is not advised. Use at your own risk.
+ *
+ * Software developers previously relying on this API are instructed to set the command queue
+ * properties when creating the queue, instead.
+ */
+ extern CL_API_ENTRY cl_int CL_API_CALL
+ clSetCommandQueueProperty(cl_command_queue command_queue,
+ cl_command_queue_properties properties,
+ cl_bool enable,
+ cl_command_queue_properties * old_properties) CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED;
+#endif /* CL_USE_DEPRECATED_OPENCL_1_0_APIS */
/* 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;
-
+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;
-
+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;
-
+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;
-
+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;
+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;
+clGetExtensionFunctionAddress(const char * func_name) CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED;
+
+/* Deprecated OpenCL 2.0 APIs */
+extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_2_DEPRECATED cl_command_queue CL_API_CALL
+clCreateCommandQueue(cl_context context,
+ cl_device_id device,
+ cl_command_queue_properties properties,
+ cl_int * errcode_ret) CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED;
+
+extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_2_DEPRECATED 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_EXT_SUFFIX__VERSION_1_2_DEPRECATED;
+
+extern CL_API_ENTRY CL_EXT_PREFIX__VERSION_1_2_DEPRECATED 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_EXT_SUFFIX__VERSION_1_2_DEPRECATED;
#ifdef __cplusplus
}
#endif
#endif /* __OPENCL_CL_H */
-
diff --git a/include/CL/cl.hpp b/include/CL/cl.hpp
index 99b86a66563..838fa1ca30d 100644
--- a/include/CL/cl.hpp
+++ b/include/CL/cl.hpp
@@ -1,5 +1,5 @@
/*******************************************************************************
- * Copyright (c) 2008-2010 The Khronos Group Inc.
+ * Copyright (c) 2008-2015 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -23,13 +28,18 @@
/*! \file
*
- * \brief C++ bindings for OpenCL 1.0 (rev 48) and OpenCL 1.1 (rev 33)
- * \author Benedict R. Gaster and Laurent Morichetti
+ * \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.
+ * 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 1.1
- * \date June 2010
+ * \version 1.2.9
+ * \date December 2015
*
* Optional extension support
*
@@ -55,8 +65,8 @@
*
* For detail documentation on the bindings see:
*
- * The OpenCL C++ Wrapper API 1.1 (revision 04)
- * http://www.khronos.org/registry/cl/specs/opencl-cplusplus-1.1.pdf
+ * The OpenCL C++ Wrapper API 1.2 (revision 09)
+ * http://www.khronos.org/registry/cl/specs/opencl-cplusplus-1.2.pdf
*
* \section example Example
*
@@ -140,32 +150,77 @@
#ifndef CL_HPP_
#define CL_HPP_
+// The latest version of the OpenCL C++ bindings can be found on GitHub:
+// -> https://github.com/KhronosGroup/OpenCL-CLHPP
+#pragma message("This version of the OpenCL Host API C++ bindings is deprecated, please use cl2.hpp instead.")
+
#ifdef _WIN32
-#include <windows.h>
+
#include <malloc.h>
+
#if defined(USE_DX_INTEROP)
#include <CL/cl_d3d10.h>
+#include <CL/cl_dx9_media_sharing.h>
#endif
#endif // _WIN32
+#if defined(_MSC_VER)
+#include <intrin.h>
+#endif // _MSC_VER
+
//
#if defined(USE_CL_DEVICE_FISSION)
#include <CL/cl_ext.h>
#endif
#if defined(__APPLE__) || defined(__MACOSX)
-#include <OpenGL/OpenGL.h>
#include <OpenCL/opencl.h>
#else
-#include <GL/gl.h>
#include <CL/opencl.h>
#endif // !__APPLE__
+#if (_MSC_VER >= 1700) || (__cplusplus >= 201103L)
+#define CL_HPP_RVALUE_REFERENCES_SUPPORTED
+#define CL_HPP_CPP11_ATOMICS_SUPPORTED
+#include <atomic>
+#endif
+
+#if (__cplusplus >= 201103L)
+#define CL_HPP_NOEXCEPT noexcept
+#else
+#define CL_HPP_NOEXCEPT
+#endif
+
+
+// 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 <utility>
+#include <limits>
+#include <iterator>
+
+#if defined(__CL_ENABLE_EXCEPTIONS)
+#include <exception>
+#endif // #if defined(__CL_ENABLE_EXCEPTIONS)
#if !defined(__NO_STD_VECTOR)
#include <vector>
@@ -175,12 +230,32 @@
#include <string>
#endif
-#if defined(linux) || defined(__APPLE__) || defined(__MACOSX)
-# include <alloca.h>
+#if defined(__ANDROID__) || defined(linux) || defined(__APPLE__) || defined(__MACOSX)
+#include <alloca.h>
#endif // linux
#include <cstring>
+// Compiler specific weak linking
+#ifndef CL_WEAK_ATTRIB_PREFIX
+// C++17: use inline variables/functions
+#if __cplusplus >= 201703L
+#define CL_USE_INLINE
+#endif
+
+#ifdef CL_USE_INLINE
+#define CL_WEAK_ATTRIB_PREFIX inline
+#define CL_WEAK_ATTRIB_SUFFIX
+#elif _WIN32
+#define CL_WEAK_ATTRIB_PREFIX __declspec(selectany)
+#define CL_WEAK_ATTRIB_SUFFIX
+#else // GCC, CLANG, etc.
+#define CL_WEAK_ATTRIB_PREFIX
+#define CL_WEAK_ATTRIB_SUFFIX __attribute__((weak))
+#endif // CL_USE_INLINE
+
+#endif // CL_WEAK_ATTRIB_PREFIX
+
/*! \namespace cl
*
* \brief The OpenCL C++ bindings are defined within this namespace.
@@ -188,6 +263,12 @@
*/
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) \
@@ -195,17 +276,29 @@ namespace cl {
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)
-#include <exception>
-/*! \class Error
- * \brief Exception class
+/*! \brief Exception class
+ *
+ * This may be thrown by API functions when __CL_ENABLE_EXCEPTIONS is defined.
*/
class Error : public std::exception
{
@@ -213,8 +306,14 @@ private:
cl_int err_;
const char * errStr_;
public:
- /*! Create a new CL error exception for a given error code
+ /*! \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)
{}
@@ -239,7 +338,7 @@ public:
*
* \return The error code.
*/
- const cl_int err(void) const { return err_; }
+ cl_int err(void) const { return err_; }
};
#define __ERR_STR(x) #x
@@ -247,9 +346,33 @@ public:
#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_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)
@@ -260,20 +383,29 @@ public:
#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)
-#define __CREATE_IMAGE2D_ERR __ERR_STR(clCreateImage2D)
-#define __CREATE_IMAGE3D_ERR __ERR_STR(clCreateImage3D)
+#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)
@@ -286,7 +418,14 @@ public:
#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)
+#define __LINK_PROGRAM_ERR __ERR_STR(clLinkProgram)
+#endif // #if defined(CL_VERSION_1_2)
#define __CREATE_KERNELS_IN_PROGRAM_ERR __ERR_STR(clCreateKernelsInProgram)
#define __CREATE_COMMAND_QUEUE_ERR __ERR_STR(clCreateCommandQueue)
@@ -297,9 +436,11 @@ public:
#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)
@@ -308,76 +449,193 @@ public:
#define __ENQUEUE_NDRANGE_KERNEL_ERR __ERR_STR(clEnqueueNDRangeKernel)
#define __ENQUEUE_TASK_ERR __ERR_STR(clEnqueueTask)
#define __ENQUEUE_NATIVE_KERNEL __ERR_STR(clEnqueueNativeKernel)
-#define __ENQUEUE_MARKER_ERR __ERR_STR(clEnqueueMarker)
-#define __ENQUEUE_WAIT_FOR_EVENTS_ERR __ERR_STR(clEnqueueWaitForEvents)
-#define __ENQUEUE_BARRIER_ERR __ERR_STR(clEnqueueBarrier)
+#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 __UNLOAD_COMPILER_ERR __ERR_STR(clUnloadCompiler)
+#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 string
+class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED string
{
private:
::size_t size_;
char * str_;
public:
+ //! \brief Constructs an empty string, allocating no memory.
string(void) : size_(0), str_(NULL)
{
}
- string(char * str, ::size_t size) :
+ /*! \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)
{
- str_ = new char[size_+1];
- if (str_ != NULL) {
- memcpy(str_, str, size_ * sizeof(char));
- str_[size_] = '\0';
- }
- else {
- size_ = 0;
+ if( size > 0 ) {
+ str_ = new char[size_+1];
+ if (str_ != NULL) {
+ memcpy(str_, str, size_ * sizeof(char));
+ str_[size_] = '\0';
+ }
+ else {
+ size_ = 0;
+ }
}
}
- string(char * str) :
+ /*! \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)
{
- size_= ::strlen(str);
- str_ = new char[size_ + 1];
- if (str_ != NULL) {
- memcpy(str_, str, (size_ + 1) * sizeof(char));
+ if( str ) {
+ size_= ::strlen(str);
}
- else {
+ 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];
+ ::size_t 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;
- str_ = NULL;
}
else {
+ str_ = new char[rhs.size_ + 1];
size_ = rhs.size_;
- str_ = new char[size_ + 1];
+
if (str_ != NULL) {
memcpy(str_, rhs.str_, (size_ + 1) * sizeof(char));
}
@@ -389,37 +647,42 @@ public:
return *this;
}
- string(const string& rhs)
+ /*! \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()
{
- if (str_ != NULL) {
- delete[] str_;
- }
+ 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_ : "";}
-};
-
-#if !defined(__USE_DEV_STRING) && !defined(__NO_STD_STRING)
-#include <string>
-typedef std::string STRING_CLASS;
-#elif !defined(__USE_DEV_STRING)
+} CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED;
typedef cl::string STRING_CLASS;
-#endif
+#endif // #elif !defined(__USE_DEV_STRING)
#if !defined(__USE_DEV_VECTOR) && !defined(__NO_STD_VECTOR)
-#include <vector>
#define VECTOR_CLASS std::vector
#elif !defined(__USE_DEV_VECTOR)
#define VECTOR_CLASS cl::vector
-#endif
#if !defined(__MAX_DEFAULT_VECTOR_SIZE)
#define __MAX_DEFAULT_VECTOR_SIZE 10
@@ -427,189 +690,301 @@ typedef cl::string STRING_CLASS;
/*! \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 <typename T, unsigned int N = __MAX_DEFAULT_VECTOR_SIZE>
-class vector
+class CL_EXT_PREFIX__VERSION_1_1_DEPRECATED vector
{
private:
T data_[N];
unsigned int size_;
- bool empty_;
+
public:
- vector() :
- size_(-1),
- empty_(true)
+ //! \brief Constructs an empty vector with no memory allocated.
+ vector() :
+ size_(static_cast<unsigned int>(0))
{}
- ~vector() {}
+ //! \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_ + 1;
+ return size_;
}
-
+
+ /*! \brief Empties the vector of all elements.
+ * \note
+ * This does not deallocate memory but will invoke destructors
+ * on contained elements.
+ */
void clear()
{
- size_ = -1;
- empty_ = true;
+ 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) {
- size_++;
- data_[size_] = x;
- empty_ = false;
+ 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 (!empty_) {
+ if (size_ != 0) {
+ --size_;
data_[size_].~T();
- size_--;
- if (size_ == -1) {
- empty_ = true;
- }
+ } 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<T, N>& vec) :
- size_(vec.size_),
- empty_(vec.empty_)
+ size_(vec.size_)
{
- if (!empty_) {
- memcpy(&data_[0], &vec.data_[0], size() * sizeof(T));
+ 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_(-1),
- empty_(true)
+ 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<T, N>& operator=(const vector<T, N>& rhs)
{
if (this == &rhs) {
return *this;
}
- size_ = rhs.size_;
- empty_ = rhs.empty_;
-
- if (!empty_) {
- memcpy(&data_[0], &rhs.data_[0], size() * sizeof(T));
+ 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<T,N> &vec)
{
- if (empty_ && vec.empty_) {
- return true;
- }
-
if (size() != vec.size()) {
return false;
}
- return memcmp(&data_[0], &vec.data_[0], size() * sizeof(T)) == 0 ? true : 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 empty_;
+ 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 sizeof(T) * N;
+ return N;
+ }
+
+ //! \brief Resizes the vector to the given size
+ void resize(unsigned int newSize, T fill = T())
+ {
+ if (newSize > N)
+ {
+ detail::errHandler(CL_MEM_OBJECT_ALLOCATION_FAILURE, __VECTOR_CAPACITY_ERR);
+ }
+ else
+ {
+ while (size_ < newSize)
+ {
+ new (&data_[size_]) T(fill);
+ size_++;
+ }
+ while (size_ > newSize)
+ {
+ --size_;
+ data_[size_].~T();
+ }
+ }
}
+ /*! \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];
}
- T operator[](int index) const
+ /*! \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<class I>
void assign(I start, I end)
{
clear();
- while(start < end) {
+ while(start != end) {
push_back(*start);
start++;
}
}
/*! \class iterator
- * \brief Iterator class for vectors
+ * \brief Const iterator class for vectors
*/
class iterator
{
private:
- vector<T,N> vec_;
+ const vector<T,N> *vec_;
int index_;
- bool initialized_;
+
+ /**
+ * Internal iterator constructor to capture reference
+ * to the vector it iterates over rather than taking
+ * the vector by copy.
+ */
+ iterator (const vector<T,N> &vec, int index) :
+ vec_(&vec)
+ {
+ if( !vec.empty() ) {
+ index_ = index;
+ } else {
+ index_ = -1;
+ }
+ }
+
public:
iterator(void) :
index_(-1),
- initialized_(false)
+ vec_(NULL)
+ {
+ }
+
+ iterator(const iterator& rhs) :
+ vec_(rhs.vec_),
+ index_(rhs.index_)
{
- index_ = -1;
- initialized_ = false;
}
~iterator(void) {}
- static iterator begin(vector<T,N> &vec)
+ static iterator begin(const cl::vector<T,N> &vec)
{
- iterator i;
+ iterator i(vec, 0);
- if (!vec.empty()) {
- i.index_ = 0;
- }
-
- i.vec_ = vec;
- i.initialized_ = true;
return i;
}
- static iterator end(vector<T,N> &vec)
+ static iterator end(const cl::vector<T,N> &vec)
{
- iterator i;
+ iterator i(vec, vec.size());
- if (!vec.empty()) {
- i.index_ = vec.size();
- }
- i.vec_ = vec;
- i.initialized_ = true;
return i;
}
bool operator==(iterator i)
{
return ((vec_ == i.vec_) &&
- (index_ == i.index_) &&
- (initialized_ == i.initialized_));
+ (index_ == i.index_));
}
bool operator!=(iterator i)
@@ -617,29 +992,35 @@ public:
return (!(*this==i));
}
- void operator++()
+ iterator& operator++()
{
- index_++;
+ ++index_;
+ return *this;
}
- void operator++(int x)
+ iterator operator++(int)
{
- index_ += x;
+ iterator retVal(*this);
+ ++index_;
+ return retVal;
}
- void operator--()
+ iterator& operator--()
{
- index_--;
+ --index_;
+ return *this;
}
- void operator--(int x)
+ iterator operator--(int)
{
- index_ -= x;
+ iterator retVal(*this);
+ --index_;
+ return retVal;
}
- T operator *()
+ const T& operator *() const
{
- return vec_[index_];
+ return (*vec_)[index_];
}
};
@@ -648,11 +1029,21 @@ public:
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];
@@ -670,110 +1061,269 @@ public:
const T& back(void) const
{
- return data_[size_];
+ return data_[size_-1];
}
-};
+} CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED;
+#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
+ */
+
+#ifdef CL_HPP_CPP11_ATOMICS_SUPPORTED
+ inline int compare_exchange(std::atomic<int> * dest, int exchange, int comparand)
+#else // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ inline int compare_exchange(volatile int * dest, int exchange, int comparand)
+#endif // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ {
+#ifdef CL_HPP_CPP11_ATOMICS_SUPPORTED
+ std::atomic_compare_exchange_strong(dest, &comparand, exchange);
+ return comparand;
+#elif _MSC_VER
+ return (int)(_InterlockedCompareExchange(
+ (volatile long*)dest,
+ (long)exchange,
+ (long)comparand));
+#else // !_MSC_VER && !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ return (__sync_val_compare_and_swap(
+ dest,
+ comparand,
+ exchange));
+#endif // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ }
+
+ inline void fence() {
+#ifdef CL_HPP_CPP11_ATOMICS_SUPPORTED
+ std::atomic_thread_fence(std::memory_order_seq_cst);
+#elif _MSC_VER // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ _ReadWriteBarrier();
+#else // !_MSC_VER && !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ __sync_synchronize();
+#endif // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ }
+} // namespace detail
+
-/*!
- * \brief size_t class used to interface between C++ and
- * OpenCL C calls that require arrays of size_t values, who's
- * size is known statically.
+/*! \brief class used to interface between C++ and
+ * OpenCL C calls that require arrays of size_t values, whose
+ * size is known statically.
*/
template <int N>
-struct size_t : public cl::vector< ::size_t, N> { };
+class size_t
+{
+private:
+ ::size_t data_[N];
-namespace detail {
+public:
+ //! \brief Initialize size_t to all 0s
+ size_t()
+ {
+ for( int i = 0; i < N; ++i ) {
+ data_[i] = 0;
+ }
+ }
-// GetInfo help struct
-template <typename Functor, typename T>
-struct GetInfoHelper
-{
- static cl_int
- get(Functor f, cl_uint name, T* param)
+ ::size_t& operator[](int index)
{
- return f(name, sizeof(T), param, NULL);
+ 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_; }
};
-// Specialized GetInfoHelper for VECTOR_CLASS params
+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<typename Functor, typename T>
+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 <typename Func, typename T>
-struct GetInfoHelper<Func, VECTOR_CLASS<T> >
+inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS<T>* param, long)
{
- static cl_int get(Func f, cl_uint name, VECTOR_CLASS<T>* param)
- {
- ::size_t required;
- cl_int err = f(name, 0, NULL, &required);
- if (err != CL_SUCCESS) {
- return err;
- }
+ ::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;
- }
+ 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;
+ param->assign(&value[0], &value[required/sizeof(T)]);
+ return CL_SUCCESS;
+}
+
+/* Specialization for reference-counted types. This depends on the
+ * existence of Wrapper<T>::cl_type, and none of the other types having the
+ * cl_type member. Note that simplify specifying the parameter as Wrapper<T>
+ * does not work, because when using a derived type (e.g. Context) the generic
+ * template will provide a better match.
+ */
+template <typename Func, typename T>
+inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS<T>* 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<CL_PROGRAM_BINARIES>
template <typename Func>
-struct GetInfoHelper<Func, VECTOR_CLASS<char *> >
+inline cl_int getInfoHelper(Func f, cl_uint name, VECTOR_CLASS<char *>* param, int)
{
- static cl_int
- get(Func f, cl_uint name, VECTOR_CLASS<char *>* param)
- {
- cl_uint err = f(name, param->size() * sizeof(char *), &(*param)[0], NULL);
- if (err != CL_SUCCESS) {
+ cl_int err = f(name, param->size() * sizeof(char *), &(*param)[0], NULL);
+
+ if (err != CL_SUCCESS) {
return err;
- }
-
- return CL_SUCCESS;
}
-};
+
+ return CL_SUCCESS;
+}
// Specialized GetInfoHelper for STRING_CLASS params
template <typename Func>
-struct GetInfoHelper<Func, STRING_CLASS>
+inline cl_int getInfoHelper(Func f, cl_uint name, STRING_CLASS* param, long)
{
- static cl_int get(Func f, cl_uint name, STRING_CLASS* param)
- {
- ::size_t required;
- cl_int err = f(name, 0, NULL, &required);
- if (err != CL_SUCCESS) {
- return err;
- }
+#if defined(__NO_STD_VECTOR) || defined(__NO_STD_STRING)
+ ::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);
+ char* value = (char*)alloca(required);
+ err = f(name, required, value, NULL);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+
+ *param = value;
+ return CL_SUCCESS;
+#else
+ ::size_t required;
+ cl_int err = f(name, 0, NULL, &required);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+
+ if (required > 0) {
+ // std::string has a constant data member
+ // a char vector does not
+ VECTOR_CLASS<char> value(required);
+ err = f(name, required, value.data(), NULL);
if (err != CL_SUCCESS) {
return err;
}
+ if (param) {
+ param->assign(value.begin(), value.end() - 1u);
+ }
+ }
+ else if (param) {
+ param->assign("");
+ }
+#endif
+ return CL_SUCCESS;
+}
- *param = value;
- return CL_SUCCESS;
+// Specialized GetInfoHelper for cl::size_t params
+template <typename Func, ::size_t N>
+inline cl_int getInfoHelper(Func f, cl_uint name, size_t<N>* param, long)
+{
+ ::size_t required;
+ cl_int err = f(name, 0, NULL, &required);
+ if (err != CL_SUCCESS) {
+ return err;
}
-};
-#define __GET_INFO_HELPER_WITH_RETAIN(CPP_TYPE) \
-namespace detail { \
-template <typename Func> \
-struct GetInfoHelper<Func, CPP_TYPE> \
-{ \
- static cl_int get(Func f, cl_uint name, CPP_TYPE* param) \
- { \
- cl_uint err = f(name, sizeof(CPP_TYPE), param, NULL); \
- if (err != CL_SUCCESS) { \
- return err; \
- } \
- \
- return ReferenceHandler<CPP_TYPE::cl_type>::retain((*param)()); \
- } \
-}; \
-}
+ ::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<typename T> struct ReferenceHandler;
+
+/* Specialization for reference-counted types. This depends on the
+ * existence of Wrapper<T>::cl_type, and none of the other types having the
+ * cl_type member. Note that simplify specifying the parameter as Wrapper<T>
+ * does not work, because when using a derived type (e.g. Context) the generic
+ * template will provide a better match.
+ */
+template<typename Func, typename T>
+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) \
@@ -795,7 +1345,7 @@ struct GetInfoHelper<Func, CPP_TYPE> \
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_bitfield) \
+ 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) \
@@ -804,12 +1354,14 @@ struct GetInfoHelper<Func, CPP_TYPE> \
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_uint) \
+ 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_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_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) \
@@ -840,7 +1392,7 @@ struct GetInfoHelper<Func, CPP_TYPE> \
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_event_info, CL_EVENT_COMMAND_EXECUTION_STATUS, cl_int) \
\
F(cl_profiling_info, CL_PROFILING_COMMAND_QUEUED, cl_ulong) \
F(cl_profiling_info, CL_PROFILING_COMMAND_SUBMIT, cl_ulong) \
@@ -865,14 +1417,14 @@ struct GetInfoHelper<Func, CPP_TYPE> \
\
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_sampler_info, CL_SAMPLER_NORMALIZED_COORDS, cl_bool) \
+ F(cl_sampler_info, CL_SAMPLER_ADDRESSING_MODE, cl_addressing_mode) \
+ F(cl_sampler_info, CL_SAMPLER_FILTER_MODE, cl_filter_mode) \
\
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<cl_device_id>) \
+ F(cl_program_info, CL_PROGRAM_DEVICES, VECTOR_CLASS<Device>) \
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<char *>) \
@@ -907,9 +1459,8 @@ struct GetInfoHelper<Func, CPP_TYPE> \
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) \
@@ -920,19 +1471,54 @@ struct GetInfoHelper<Func, CPP_TYPE> \
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_ARRAY_SIZE, ::size_t) \
+ F(cl_image_info, CL_IMAGE_BUFFER, cl::Buffer) \
+ F(cl_image_info, CL_IMAGE_NUM_MIP_LEVELS, cl_uint) \
+ F(cl_image_info, CL_IMAGE_NUM_SAMPLES, cl_uint) \
+ \
+ 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_TYPE_QUALIFIER, cl_kernel_arg_type_qualifier) \
+ F(cl_kernel_arg_info, CL_KERNEL_ARG_NAME, STRING_CLASS) \
+ \
+ F(cl_device_info, CL_DEVICE_IMAGE_MAX_BUFFER_SIZE, ::size_t) \
+ F(cl_device_info, CL_DEVICE_IMAGE_MAX_ARRAY_SIZE, ::size_t) \
+ F(cl_device_info, CL_DEVICE_LINKER_AVAILABLE, cl_bool) \
+ F(cl_device_info, CL_DEVICE_BUILT_IN_KERNELS, STRING_CLASS) \
+ F(cl_device_info, CL_DEVICE_PRINTF_BUFFER_SIZE, ::size_t) \
+ F(cl_device_info, CL_DEVICE_PREFERRED_INTEROP_USER_SYNC, cl_bool) \
+ F(cl_device_info, CL_DEVICE_PARENT_DEVICE, cl_device_id) \
+ F(cl_device_info, CL_DEVICE_PARTITION_MAX_SUB_DEVICES, cl_uint) \
+ F(cl_device_info, CL_DEVICE_PARTITION_PROPERTIES, VECTOR_CLASS<cl_device_partition_property>) \
+ F(cl_device_info, CL_DEVICE_PARTITION_AFFINITY_DOMAIN, cl_device_affinity_domain) \
+ F(cl_device_info, CL_DEVICE_PARTITION_TYPE, VECTOR_CLASS<cl_device_partition_property>) \
+ F(cl_device_info, CL_DEVICE_REFERENCE_COUNT, cl_uint)
+#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<cl_device_partition_property_ext>) \
- F(cl_device_info, CL_DEVICE_AFFINITY_DOMAINS_EXT, VECTOR_CLASS<cl_device_partition_property_ext>) \
- F(cl_device_info, CL_DEVICE_REFERENCE_COUNT_EXT , cl_uint) \
- F(cl_device_info, CL_DEVICE_PARTITION_STYLE_EXT, VECTOR_CLASS<cl_device_partition_property_ext>)
+ F(cl_device_info, CL_DEVICE_PARTITION_TYPES_EXT, VECTOR_CLASS<cl_device_partition_property_ext>) \
+ F(cl_device_info, CL_DEVICE_AFFINITY_DOMAINS_EXT, VECTOR_CLASS<cl_device_partition_property_ext>) \
+ F(cl_device_info, CL_DEVICE_REFERENCE_COUNT_EXT , cl_uint) \
+ F(cl_device_info, CL_DEVICE_PARTITION_STYLE_EXT, VECTOR_CLASS<cl_device_partition_property_ext>)
#endif // USE_CL_DEVICE_FISSION
template <typename enum_type, cl_int Name>
struct param_traits {};
-#define __DECLARE_PARAM_TRAITS(token, param_name, T) \
+#define __CL_DECLARE_PARAM_TRAITS(token, param_name, T) \
struct token; \
template<> \
struct param_traits<detail:: token,param_name> \
@@ -941,16 +1527,78 @@ struct param_traits<detail:: token,param_name> \
typedef T param_type; \
};
-__PARAM_NAME_INFO_1_0(__DECLARE_PARAM_TRAITS);
+__PARAM_NAME_INFO_1_0(__CL_DECLARE_PARAM_TRAITS)
#if defined(CL_VERSION_1_1)
-__PARAM_NAME_INFO_1_1(__DECLARE_PARAM_TRAITS);
+__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(__DECLARE_PARAM_TRAITS);
+__PARAM_NAME_DEVICE_FISSION(__CL_DECLARE_PARAM_TRAITS);
#endif // USE_CL_DEVICE_FISSION
-#undef __DECLARE_PARAM_TRAITS
+#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
@@ -958,7 +1606,7 @@ template <typename Func, typename T>
inline cl_int
getInfo(Func f, cl_uint name, T* param)
{
- return GetInfoHelper<Func, T>::get(f, name, param);
+ return getInfoHelper(f, name, param, 0);
}
template <typename Func, typename Arg0>
@@ -984,8 +1632,7 @@ inline cl_int
getInfo(Func f, const Arg0& arg0, cl_uint name, T* param)
{
GetInfoFunctor0<Func, Arg0> f0 = { f, arg0 };
- return GetInfoHelper<GetInfoFunctor0<Func, Arg0>, T>
- ::get(f0, name, param);
+ return getInfoHelper(f0, name, param, 0);
}
template <typename Func, typename Arg0, typename Arg1, typename T>
@@ -993,34 +1640,68 @@ inline cl_int
getInfo(Func f, const Arg0& arg0, const Arg1& arg1, cl_uint name, T* param)
{
GetInfoFunctor1<Func, Arg0, Arg1> f0 = { f, arg0, arg1 };
- return GetInfoHelper<GetInfoFunctor1<Func, Arg0, Arg1>, T>
- ::get(f0, name, param);
+ return getInfoHelper(f0, name, param, 0);
}
template<typename T>
struct ReferenceHandler
{ };
+#if defined(CL_VERSION_1_2)
+/**
+ * OpenCL 1.2 devices do have retain/release.
+ */
+template <>
+struct ReferenceHandler<cl_device_id>
+{
+ /**
+ * 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>
{
// cl_device_id does not have retain().
static cl_int retain(cl_device_id)
- { return CL_INVALID_DEVICE; }
+ { return CL_SUCCESS; }
// cl_device_id does not have release().
static cl_int release(cl_device_id)
- { return CL_INVALID_DEVICE; }
+ { return CL_SUCCESS; }
};
+#endif // #if defined(CL_VERSION_1_2)
template <>
struct ReferenceHandler<cl_platform_id>
{
// cl_platform_id does not have retain().
static cl_int retain(cl_platform_id)
- { return CL_INVALID_PLATFORM; }
+ { return CL_SUCCESS; }
// cl_platform_id does not have release().
static cl_int release(cl_platform_id)
- { return CL_INVALID_PLATFORM; }
+ { return CL_SUCCESS; }
};
template <>
@@ -1086,6 +1767,58 @@ struct ReferenceHandler<cl_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] != ' ' && versionInfo[index] != '\0') {
+ 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 <typename T>
class Wrapper
{
@@ -1098,6 +1831,8 @@ protected:
public:
Wrapper() : object_(NULL) { }
+ Wrapper(const cl_type &obj) : object_(obj) { }
+
~Wrapper()
{
if (object_ != NULL) { release(); }
@@ -1106,14 +1841,43 @@ public:
Wrapper(const Wrapper<cl_type>& rhs)
{
object_ = rhs.object_;
- if (object_ != NULL) { retain(); }
+ if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); }
}
- Wrapper<cl_type>& operator = (const Wrapper<cl_type>& rhs)
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ Wrapper(Wrapper<cl_type>&& rhs) CL_HPP_NOEXCEPT
{
- if (object_ != NULL) { release(); }
object_ = rhs.object_;
- if (object_ != NULL) { retain(); }
+ rhs.object_ = NULL;
+ }
+#endif
+
+ Wrapper<cl_type>& operator = (const Wrapper<cl_type>& rhs)
+ {
+ if (this != &rhs) {
+ if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); }
+ object_ = rhs.object_;
+ if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); }
+ }
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ Wrapper<cl_type>& operator = (Wrapper<cl_type>&& rhs)
+ {
+ if (this != &rhs) {
+ if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); }
+ object_ = rhs.object_;
+ rhs.object_ = NULL;
+ }
+ return *this;
+ }
+#endif
+
+ Wrapper<cl_type>& operator = (const cl_type &rhs)
+ {
+ if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); }
+ object_ = rhs;
return *this;
}
@@ -1122,6 +1886,8 @@ public:
cl_type& operator ()() { return object_; }
protected:
+ template<typename Func, typename U>
+ friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type);
cl_int retain() const
{
@@ -1134,39 +1900,146 @@ protected:
}
};
-#if defined(__CL_ENABLE_EXCEPTIONS)
-static inline cl_int errHandler (
- cl_int err,
- const char * errStr = NULL) throw(Error)
+template <>
+class Wrapper<cl_device_id>
{
- if (err != CL_SUCCESS) {
- throw Error(err, errStr);
+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;
}
- return err;
-}
-#else
-static inline cl_int errHandler (cl_int err, const char * errStr = NULL)
-{
- return err;
-}
-#endif // __CL_ENABLE_EXCEPTIONS
+
+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<cl_type>& rhs)
+ {
+ object_ = rhs.object_;
+ referenceCountable_ = isReferenceCountable(object_);
+ if (object_ != NULL) { detail::errHandler(retain(), __RETAIN_ERR); }
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ Wrapper(Wrapper<cl_type>&& rhs) CL_HPP_NOEXCEPT
+ {
+ object_ = rhs.object_;
+ referenceCountable_ = rhs.referenceCountable_;
+ rhs.object_ = NULL;
+ rhs.referenceCountable_ = false;
+ }
+#endif
+
+ Wrapper<cl_type>& operator = (const Wrapper<cl_type>& rhs)
+ {
+ if (this != &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;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ Wrapper<cl_type>& operator = (Wrapper<cl_type>&& rhs)
+ {
+ if (this != &rhs) {
+ if (object_ != NULL) { detail::errHandler(release(), __RELEASE_ERR); }
+ object_ = rhs.object_;
+ referenceCountable_ = rhs.referenceCountable_;
+ rhs.object_ = NULL;
+ rhs.referenceCountable_ = false;
+ }
+ return *this;
+ }
+#endif
+
+ Wrapper<cl_type>& 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<typename Func, typename U>
+ friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type);
+
+ template<typename Func, typename U>
+ friend inline cl_int getInfoHelper(Func, cl_uint, VECTOR_CLASS<U>*, int, typename U::cl_type);
+
+ cl_int retain() const
+ {
+ if( referenceCountable_ ) {
+ return ReferenceHandler<cl_type>::retain(object_);
+ }
+ else {
+ return CL_SUCCESS;
+ }
+ }
+
+ cl_int release() const
+ {
+ if( referenceCountable_ ) {
+ return ReferenceHandler<cl_type>::release(object_);
+ }
+ else {
+ return CL_SUCCESS;
+ }
+ }
+};
} // namespace detail
//! \endcond
/*! \stuct ImageFormat
- * \brief ImageFormat interface fro cl_image_format.
+ * \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) {
@@ -1177,26 +2050,72 @@ struct ImageFormat : public cl_image_format
}
};
-/*! \class Device
- * \brief Device interface for cl_device_id.
+/*! \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<cl_device_id>
{
public:
- Device(cl_device_id device) { object_ = device; }
-
+ //! \brief Default constructor - initializes to NULL.
Device() : detail::Wrapper<cl_type>() { }
- Device(const Device& device) : detail::Wrapper<cl_type>(device) { }
+ /*! \brief Constructor from cl_device_id.
+ *
+ * This simply copies the device ID value, which is an inexpensive operation.
+ */
+ __CL_EXPLICIT_CONSTRUCTORS Device(const cl_device_id &device) : detail::Wrapper<cl_type>(device) { }
- Device& operator = (const Device& rhs)
+ /*! \brief Returns the first device on the default context.
+ *
+ * \see Context::getDefault()
+ */
+ static Device getDefault(cl_int * err = NULL);
+
+ /*! \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)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
- }
+ detail::Wrapper<cl_type>::operator=(rhs);
return *this;
}
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Device(const Device& dev) : detail::Wrapper<cl_type>(dev) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Device& operator = (const Device &dev)
+ {
+ detail::Wrapper<cl_type>::operator=(dev);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Device(Device&& dev) CL_HPP_NOEXCEPT : detail::Wrapper<cl_type>(std::move(dev)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Device& operator = (Device &&dev)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(dev));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+
+ //! \brief Wrapper for clGetDeviceInfo().
template <typename T>
cl_int getInfo(cl_device_info name, T* param) const
{
@@ -1205,6 +2124,7 @@ public:
__GET_DEVICE_INFO_ERR);
}
+ //! \brief Wrapper for clGetDeviceInfo() that returns by value.
template <cl_int name> typename
detail::param_traits<detail::cl_device_info, name>::param_type
getInfo(cl_int* err = NULL) const
@@ -1218,23 +2138,53 @@ public:
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<Device>* 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<Device>* devices)
- {
- typedef CL_API_ENTRY cl_int
- ( CL_API_CALL * PFN_clCreateSubDevicesEXT)(
- cl_device_id /*in_device*/,
+ cl_int createSubDevices(
+ const cl_device_partition_property_ext * properties,
+ VECTOR_CLASS<Device>* 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);
+ static PFN_clCreateSubDevicesEXT pfn_clCreateSubDevicesEXT = NULL;
+ __INIT_CL_EXT_FCN_PTR(clCreateSubDevicesEXT);
- cl_uint n = 0;
+ 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);
@@ -1248,32 +2198,41 @@ public:
devices->assign(&ids[0], &ids[n]);
return CL_SUCCESS;
- }
-#endif
+ }
+#endif // #if defined(USE_CL_DEVICE_FISSION)
+#endif // #if defined(CL_VERSION_1_1)
};
-/*! \class Platform
- * \brief Platform interface.
+/*! \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<cl_platform_id>
{
public:
- static const Platform null();
-
- Platform(cl_platform_id platform) { object_ = platform; }
-
+ //! \brief Default constructor - initializes to NULL.
Platform() : detail::Wrapper<cl_type>() { }
- Platform(const Platform& platform) : detail::Wrapper<cl_type>(platform) { }
+ /*! \brief Constructor from cl_platform_id.
+ *
+ * This simply copies the platform ID value, which is an inexpensive operation.
+ */
+ __CL_EXPLICIT_CONSTRUCTORS Platform(const cl_platform_id &platform) : detail::Wrapper<cl_type>(platform) { }
- Platform& operator = (const Platform& rhs)
+ /*! \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)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
- }
+ detail::Wrapper<cl_type>::operator=(rhs);
return *this;
}
+ //! \brief Wrapper for clGetPlatformInfo().
cl_int getInfo(cl_platform_info name, STRING_CLASS* param) const
{
return detail::errHandler(
@@ -1281,6 +2240,7 @@ public:
__GET_PLATFORM_INFO_ERR);
}
+ //! \brief Wrapper for clGetPlatformInfo() that returns by value.
template <cl_int name> typename
detail::param_traits<detail::cl_platform_info, name>::param_type
getInfo(cl_int* err = NULL) const
@@ -1294,11 +2254,18 @@ public:
return param;
}
+ /*! \brief Gets a list of devices for this platform.
+ *
+ * Wraps clGetDeviceIDs().
+ */
cl_int getDevices(
cl_device_type type,
VECTOR_CLASS<Device>* 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);
@@ -1353,8 +2320,12 @@ public:
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(clGetDeviceIDsFromD3D10KHR);
+ __INIT_CL_EXT_FCN_PTR_PLATFORM(object_, clGetDeviceIDsFromD3D10KHR);
cl_uint n = 0;
cl_int err = pfn_clGetDeviceIDsFromD3D10KHR(
@@ -1387,10 +2358,19 @@ public:
}
#endif
+ /*! \brief Gets a list of available platforms.
+ *
+ * Wraps clGetPlatformIDs().
+ */
static cl_int get(
VECTOR_CLASS<Platform>* 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);
@@ -1406,17 +2386,129 @@ public:
platforms->assign(&ids[0], &ids[n]);
return CL_SUCCESS;
}
-};
-static inline cl_int
+ /*! \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;
+ }
+ return Platform();
+ }
+
+ 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 Platform();
+ }
+
+
+ return Platform(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)
-class Context : public detail::Wrapper<cl_context>
+/*! \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<cl_context>
{
+private:
+
+#ifdef CL_HPP_CPP11_ATOMICS_SUPPORTED
+ static std::atomic<int> default_initialized_;
+#else // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ static volatile int default_initialized_;
+#endif // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ static Context default_;
+ static volatile cl_int default_error_;
public:
+ /*! \brief Constructs a context including a list of specified devices.
+ *
+ * Wraps clCreateContext().
+ */
Context(
const VECTOR_CLASS<Device>& devices,
cl_context_properties* properties = NULL,
@@ -1429,17 +2521,54 @@ public:
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) devices.size(),
- (cl_device_id*) &devices.front(),
+ properties, (cl_uint) numDevices,
+ deviceIDs,
notifyFptr, data, &error);
- detail::errHandler(error, __CREATE_CONTEXT_FROM_TYPE_ERR);
+ 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,
@@ -1452,6 +2581,66 @@ public:
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<Platform> 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<Device> 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);
@@ -1461,18 +2650,110 @@ public:
}
}
- Context() : detail::Wrapper<cl_type>() { }
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Context(const Context& ctx) : detail::Wrapper<cl_type>(ctx) {}
- Context(const Context& context) : detail::Wrapper<cl_type>(context) { }
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Context& operator = (const Context &ctx)
+ {
+ detail::Wrapper<cl_type>::operator=(ctx);
+ return *this;
+ }
- Context& operator = (const Context& rhs)
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Context(Context&& ctx) CL_HPP_NOEXCEPT : detail::Wrapper<cl_type>(std::move(ctx)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Context& operator = (Context &&ctx)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
+ detail::Wrapper<cl_type>::operator=(std::move(ctx));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+
+ /*! \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<cl_type>() { }
+
+ /*! \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<cl_type>(context) { }
+
+ /*! \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<cl_type>::operator=(rhs);
return *this;
}
+ //! \brief Wrapper for clGetContextInfo().
template <typename T>
cl_int getInfo(cl_context_info name, T* param) const
{
@@ -1481,6 +2762,7 @@ public:
__GET_CONTEXT_INFO_ERR);
}
+ //! \brief Wrapper for clGetContextInfo() that returns by value.
template <cl_int name> typename
detail::param_traits<detail::cl_context_info, name>::param_type
getInfo(cl_int* err = NULL) const
@@ -1494,61 +2776,120 @@ public:
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<ImageFormat>* 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);
+
+ if (!formats) {
+ return CL_SUCCESS;
}
- ImageFormat* value = (ImageFormat*)
- alloca(numEntries * sizeof(ImageFormat));
- err = ::clGetSupportedImageFormats(
- object_,
- flags,
- type,
- numEntries,
- (cl_image_format*) value,
- NULL);
+ cl_int err = ::clGetSupportedImageFormats(
+ object_,
+ flags,
+ type,
+ 0,
+ NULL,
+ &numEntries);
if (err != CL_SUCCESS) {
return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR);
}
- formats->assign(&value[0], &value[numEntries]);
+ if (numEntries > 0) {
+ 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]);
+ }
+ else {
+ formats->clear();
+ }
return CL_SUCCESS;
}
};
-__GET_INFO_HELPER_WITH_RETAIN(cl::Context)
+inline Device Device::getDefault(cl_int * err)
+{
+ cl_int error;
+ Device device;
+
+ Context context = Context::getDefault(&error);
+ detail::errHandler(error, __CREATE_CONTEXT_ERR);
+
+ if (error != CL_SUCCESS) {
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+ else {
+ device = context.getInfo<CL_CONTEXT_DEVICES>()[0];
+ if (err != NULL) {
+ *err = CL_SUCCESS;
+ }
+ }
+
+ return device;
+}
+
+#ifdef CL_HPP_CPP11_ATOMICS_SUPPORTED
+CL_WEAK_ATTRIB_PREFIX std::atomic<int> CL_WEAK_ATTRIB_SUFFIX Context::default_initialized_;
+#else // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+CL_WEAK_ATTRIB_PREFIX volatile int CL_WEAK_ATTRIB_SUFFIX Context::default_initialized_ = __DEFAULT_NOT_INITIALIZED;
+#endif // !CL_HPP_CPP11_ATOMICS_SUPPORTED
-/*! \class Event
- * \brief Event interface for cl_event.
+CL_WEAK_ATTRIB_PREFIX Context CL_WEAK_ATTRIB_SUFFIX Context::default_;
+CL_WEAK_ATTRIB_PREFIX volatile cl_int CL_WEAK_ATTRIB_SUFFIX Context::default_error_ = CL_SUCCESS;
+
+/*! \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<cl_event>
{
public:
+ //! \brief Default constructor - initializes to NULL.
Event() : detail::Wrapper<cl_type>() { }
- Event(const Event& event) : detail::Wrapper<cl_type>(event) { }
+ /*! \brief Constructor from cl_event - takes ownership.
+ *
+ * This effectively transfers ownership of a refcount on the cl_event
+ * into the new Event object.
+ */
+ __CL_EXPLICIT_CONSTRUCTORS Event(const cl_event& event) : detail::Wrapper<cl_type>(event) { }
- Event& operator = (const Event& rhs)
+ /*! \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 cl_event& rhs)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
- }
+ detail::Wrapper<cl_type>::operator=(rhs);
return *this;
}
+ //! \brief Wrapper for clGetEventInfo().
template <typename T>
cl_int getInfo(cl_event_info name, T* param) const
{
@@ -1557,6 +2898,7 @@ public:
__GET_EVENT_INFO_ERR);
}
+ //! \brief Wrapper for clGetEventInfo() that returns by value.
template <cl_int name> typename
detail::param_traits<detail::cl_event_info, name>::param_type
getInfo(cl_int* err = NULL) const
@@ -1570,6 +2912,7 @@ public:
return param;
}
+ //! \brief Wrapper for clGetEventProfilingInfo().
template <typename T>
cl_int getProfilingInfo(cl_profiling_info name, T* param) const
{
@@ -1578,6 +2921,7 @@ public:
__GET_EVENT_PROFILE_INFO_ERR);
}
+ //! \brief Wrapper for clGetEventProfilingInfo() that returns by value.
template <cl_int name> typename
detail::param_traits<detail::cl_profiling_info, name>::param_type
getProfilingInfo(cl_int* err = NULL) const
@@ -1591,6 +2935,10 @@ public:
return param;
}
+ /*! \brief Blocks the calling thread until this event completes.
+ *
+ * Wraps clWaitForEvents().
+ */
cl_int wait() const
{
return detail::errHandler(
@@ -1599,6 +2947,10 @@ public:
}
#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 *),
@@ -1614,25 +2966,32 @@ public:
}
#endif
+ /*! \brief Blocks the calling thread until every event specified is complete.
+ *
+ * Wraps clWaitForEvents().
+ */
static cl_int
waitForEvents(const VECTOR_CLASS<Event>& events)
{
return detail::errHandler(
::clWaitForEvents(
- (cl_uint) events.size(), (cl_event*)&events.front()),
+ (cl_uint) events.size(), (events.size() > 0) ? (cl_event*)&events.front() : NULL),
__WAIT_FOR_EVENTS_ERR);
}
};
-__GET_INFO_HELPER_WITH_RETAIN(cl::Event)
-
#if defined(CL_VERSION_1_1)
-/*! \class UserEvent
- * \brief User event interface for cl_event.
+/*! \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)
@@ -1648,18 +3007,13 @@ public:
}
}
+ //! \brief Default constructor - initializes to NULL.
UserEvent() : Event() { }
- UserEvent(const UserEvent& event) : Event(event) { }
-
- 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(
@@ -1669,33 +3023,82 @@ public:
};
#endif
+/*! \brief Blocks the calling thread until every event specified is complete.
+ *
+ * Wraps clWaitForEvents().
+ */
inline static cl_int
WaitForEvents(const VECTOR_CLASS<Event>& events)
{
return detail::errHandler(
::clWaitForEvents(
- (cl_uint) events.size(), (cl_event*)&events.front()),
+ (cl_uint) events.size(), (events.size() > 0) ? (cl_event*)&events.front() : NULL),
__WAIT_FOR_EVENTS_ERR);
}
-/*! \class Memory
- * \brief Memory interface for cl_mem.
+/*! \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<cl_mem>
{
public:
+ //! \brief Default constructor - initializes to NULL.
Memory() : detail::Wrapper<cl_type>() { }
- Memory(const Memory& memory) : detail::Wrapper<cl_type>(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<cl_type>(memory) { }
- Memory& operator = (const Memory& rhs)
+ /*! \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)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
- }
+ detail::Wrapper<cl_type>::operator=(rhs);
return *this;
}
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Memory(const Memory& mem) : detail::Wrapper<cl_type>(mem) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Memory& operator = (const Memory &mem)
+ {
+ detail::Wrapper<cl_type>::operator=(mem);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Memory(Memory&& mem) CL_HPP_NOEXCEPT : detail::Wrapper<cl_type>(std::move(mem)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Memory& operator = (Memory &&mem)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(mem));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+
+ //! \brief Wrapper for clGetMemObjectInfo().
template <typename T>
cl_int getInfo(cl_mem_info name, T* param) const
{
@@ -1704,6 +3107,7 @@ public:
__GET_MEM_OBJECT_INFO_ERR);
}
+ //! \brief Wrapper for clGetMemObjectInfo() that returns by value.
template <cl_int name> typename
detail::param_traits<detail::cl_mem_info, name>::param_type
getInfo(cl_int* err = NULL) const
@@ -1718,6 +3122,19 @@ public:
}
#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)
@@ -1733,14 +3150,35 @@ public:
};
-__GET_INFO_HELPER_WITH_RETAIN(cl::Memory)
+// 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 );
-/*! \class Buffer
- * \brief Memory buffer interface.
+
+/*! \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,
@@ -1757,19 +3195,155 @@ public:
}
}
- Buffer() : Memory() { }
+ /*! \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;
- Buffer(const Buffer& buffer) : Memory(buffer) { }
+ Context context = Context::getDefault(err);
- Buffer& operator = (const Buffer& rhs)
+ 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)
{
- if (this != &rhs) {
- Memory::operator=(rhs);
+ typedef typename std::iterator_traits<IteratorType>::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<DataType*>(&*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 Construct a Buffer from a host container via iterators using a specified queue.
+ * If useHostPtr is specified iterators must represent contiguous data.
+ */
+ template< typename IteratorType >
+ Buffer(const CommandQueue &queue, IteratorType startIterator, IteratorType endIterator,
+ bool readOnly, bool useHostPtr = false, cl_int* err = NULL);
+
+ //! \brief Default constructor - initializes to NULL.
+ Buffer() : Memory() { }
+
+ /*! \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 cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Buffer& operator = (const cl_mem& rhs)
+ {
+ Memory::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Buffer(const Buffer& buf) : Memory(buf) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Buffer& operator = (const Buffer &buf)
+ {
+ Memory::operator=(buf);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Buffer(Buffer&& buf) CL_HPP_NOEXCEPT : Memory(std::move(buf)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Buffer& operator = (Buffer &&buf)
+ {
+ Memory::operator=(std::move(buf));
return *this;
}
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
#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,
@@ -1791,11 +3365,19 @@ public:
}
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:
@@ -1803,6 +3385,11 @@ public:
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,
@@ -1810,7 +3397,20 @@ public:
cl_int * err = NULL)
{
static PFN_clCreateFromD3D10BufferKHR pfn_clCreateFromD3D10BufferKHR = NULL;
+
+#if defined(CL_VERSION_1_2)
+ vector<cl_context_properties> props = context.getInfo<CL_CONTEXT_PROPERTIES>();
+ 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(
@@ -1825,30 +3425,77 @@ public:
}
}
+ //! \brief Default constructor - initializes to NULL.
BufferD3D10() : Buffer() { }
- 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) { }
- BufferD3D10& operator = (const BufferD3D10& rhs)
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ BufferD3D10& operator = (const cl_mem& rhs)
{
- if (this != &rhs) {
- Buffer::operator=(rhs);
- }
+ Buffer::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferD3D10(const BufferD3D10& buf) : Buffer(buf) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferD3D10& operator = (const BufferD3D10 &buf)
+ {
+ Buffer::operator=(buf);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferD3D10(BufferD3D10&& buf) CL_HPP_NOEXCEPT : Buffer(std::move(buf)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferD3D10& operator = (BufferD3D10 &&buf)
+ {
+ Buffer::operator=(std::move(buf));
return *this;
}
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
};
#endif
-/*! \class BufferGL
- * \brief Memory buffer interface for GL interop.
+/*! \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_GLuint bufobj,
cl_int * err = NULL)
{
cl_int error;
@@ -1864,68 +3511,59 @@ public:
}
}
+ //! \brief Default constructor - initializes to NULL.
BufferGL() : Buffer() { }
- 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) { }
- BufferGL& operator = (const BufferGL& rhs)
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ BufferGL& operator = (const cl_mem& rhs)
{
- if (this != &rhs) {
- Buffer::operator=(rhs);
- }
+ Buffer::operator=(rhs);
return *this;
}
- 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 Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferGL(const BufferGL& buf) : Buffer(buf) {}
-/*! \class BufferRenderGL
- * \brief Memory buffer interface for GL interop with renderbuffer.
- */
-class BufferRenderGL : public Buffer
-{
-public:
- BufferRenderGL(
- const Context& context,
- cl_mem_flags flags,
- GLuint bufobj,
- cl_int * err = NULL)
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferGL& operator = (const BufferGL &buf)
{
- cl_int error;
- object_ = ::clCreateFromGLRenderbuffer(
- context(),
- flags,
- bufobj,
- &error);
-
- detail::errHandler(error, __CREATE_GL_BUFFER_ERR);
- if (err != NULL) {
- *err = error;
- }
+ Buffer::operator=(buf);
+ return *this;
}
- BufferRenderGL() : Buffer() { }
-
- BufferRenderGL(const BufferGL& buffer) : Buffer(buffer) { }
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferGL(BufferGL&& buf) CL_HPP_NOEXCEPT : Buffer(std::move(buf)) {}
- BufferRenderGL& operator = (const BufferRenderGL& rhs)
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferGL& operator = (BufferGL &&buf)
{
- if (this != &rhs) {
- Buffer::operator=(rhs);
- }
+ Buffer::operator=(std::move(buf));
return *this;
}
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ //! \brief Wrapper for clGetGLObjectInfo().
cl_int getObjectInfo(
cl_gl_object_type *type,
- GLuint * gl_object_name)
+ cl_GLuint * gl_object_name)
{
return detail::errHandler(
::clGetGLObjectInfo(object_,type,gl_object_name),
@@ -1933,24 +3571,66 @@ public:
}
};
-/*! \class Image
- * \brief Base class interface for all images.
+/*! \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() { }
- 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 cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image& operator = (const cl_mem& rhs)
+ {
+ Memory::operator=(rhs);
+ return *this;
+ }
- Image& operator = (const Image& rhs)
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image(const Image& img) : Memory(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image& operator = (const Image &img)
{
- if (this != &rhs) {
- Memory::operator=(rhs);
- }
+ Memory::operator=(img);
return *this;
}
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image(Image&& img) CL_HPP_NOEXCEPT : Memory(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image& operator = (Image &&img)
+ {
+ Memory::operator=(std::move(img));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+
public:
+ //! \brief Wrapper for clGetImageInfo().
template <typename T>
cl_int getImageInfo(cl_image_info name, T* param) const
{
@@ -1958,7 +3638,8 @@ public:
detail::getInfo(&::clGetImageInfo, object_, name, param),
__GET_IMAGE_INFO_ERR);
}
-
+
+ //! \brief Wrapper for clGetImageInfo() that returns by value.
template <cl_int name> typename
detail::param_traits<detail::cl_image_info, name>::param_type
getImageInfo(cl_int* err = NULL) const
@@ -1973,12 +3654,272 @@ public:
}
};
-/*! \class Image2D
- * \brief Image interface for 2D images.
+#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 Constructor from cl_mem - takes ownership.
+ *
+ * See Memory for further details.
+ */
+ __CL_EXPLICIT_CONSTRUCTORS Image1D(const cl_mem& image1D) : Image(image1D) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image1D& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1D(const Image1D& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1D& operator = (const Image1D &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1D(Image1D&& img) CL_HPP_NOEXCEPT : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1D& operator = (Image1D &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+};
+
+/*! \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() { }
+
+ __CL_EXPLICIT_CONSTRUCTORS Image1DBuffer(const cl_mem& image1D) : Image(image1D) { }
+
+ Image1DBuffer& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DBuffer(const Image1DBuffer& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DBuffer& operator = (const Image1DBuffer &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DBuffer(Image1DBuffer&& img) CL_HPP_NOEXCEPT : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DBuffer& operator = (Image1DBuffer &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+};
+
+/*! \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() { }
+
+ __CL_EXPLICIT_CONSTRUCTORS Image1DArray(const cl_mem& imageArray) : Image(imageArray) { }
+
+ Image1DArray& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DArray(const Image1DArray& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DArray& operator = (const Image1DArray &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DArray(Image1DArray&& img) CL_HPP_NOEXCEPT : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DArray& operator = (Image1DArray &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+};
+#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,
@@ -1990,40 +3931,135 @@ public:
cl_int* err = NULL)
{
cl_int error;
- object_ = ::clCreateImage2D(
- context(), flags,&format, width, height, row_pitch, host_ptr, &error);
+ bool useCreateImage;
- detail::errHandler(error, __CREATE_IMAGE2D_ERR);
- if (err != NULL) {
- *err = error;
+#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() { }
- 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) { }
- Image2D& operator = (const Image2D& rhs)
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image2D& operator = (const cl_mem& rhs)
{
- if (this != &rhs) {
- Image::operator=(rhs);
- }
+ Image::operator=(rhs);
return *this;
}
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2D(const Image2D& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2D& operator = (const Image2D &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2D(Image2D&& img) CL_HPP_NOEXCEPT : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2D& operator = (Image2D &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
};
-/*! \class Image2DGL
- * \brief 2D image interface for GL interop.
+
+#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 Image2DGL : public Image2D
+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_GLenum target,
+ cl_GLint miplevel,
+ cl_GLuint texobj,
cl_int * err = NULL)
{
cl_int error;
@@ -2035,31 +4071,164 @@ public:
texobj,
&error);
- detail::errHandler(error, __CREATE_GL_BUFFER_ERR);
+ detail::errHandler(error, __CREATE_GL_TEXTURE_2D_ERR);
if (err != NULL) {
*err = error;
}
- }
+ }
+
+ //! \brief Default constructor - initializes to NULL.
Image2DGL() : Image2D() { }
- 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) { }
- Image2DGL& operator = (const Image2DGL& rhs)
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image2DGL& operator = (const cl_mem& rhs)
{
- if (this != &rhs) {
- Image2D::operator=(rhs);
+ Image2D::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DGL(const Image2DGL& img) : Image2D(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DGL& operator = (const Image2DGL &img)
+ {
+ Image2D::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DGL(Image2DGL&& img) CL_HPP_NOEXCEPT : Image2D(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DGL& operator = (Image2DGL &&img)
+ {
+ Image2D::operator=(std::move(img));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+};
+#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() { }
+
+ __CL_EXPLICIT_CONSTRUCTORS Image2DArray(const cl_mem& imageArray) : Image(imageArray) { }
+
+ Image2DArray& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DArray(const Image2DArray& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DArray& operator = (const Image2DArray &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DArray(Image2DArray&& img) CL_HPP_NOEXCEPT : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DArray& operator = (Image2DArray &&img)
+ {
+ Image::operator=(std::move(img));
return *this;
}
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
};
+#endif // #if defined(CL_VERSION_1_2)
-/*! \class Image3D
- * \brief Image interface for 3D images.
+/*! \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,
@@ -2073,41 +4242,136 @@ public:
cl_int* err = NULL)
{
cl_int error;
- object_ = ::clCreateImage3D(
- context(), flags, &format, width, height, depth, row_pitch,
- slice_pitch, host_ptr, &error);
+ bool useCreateImage;
- detail::errHandler(error, __CREATE_IMAGE3D_ERR);
- if (err != NULL) {
- *err = error;
+#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)
}
- Image3D() { }
+ //! \brief Default constructor - initializes to NULL.
+ Image3D() : Image() { }
- 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) { }
- Image3D& operator = (const Image3D& rhs)
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image3D& operator = (const cl_mem& rhs)
{
- if (this != &rhs) {
- Image::operator=(rhs);
- }
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3D(const Image3D& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3D& operator = (const Image3D &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3D(Image3D&& img) CL_HPP_NOEXCEPT : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3D& operator = (Image3D &&img)
+ {
+ Image::operator=(std::move(img));
return *this;
}
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
};
-/*! \class Image2DGL
- * \brief 2D image interface for GL interop.
+#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_GLenum target,
+ cl_GLint miplevel,
+ cl_GLuint texobj,
cl_int * err = NULL)
{
cl_int error;
@@ -2119,33 +4383,288 @@ public:
texobj,
&error);
- detail::errHandler(error, __CREATE_GL_BUFFER_ERR);
+ detail::errHandler(error, __CREATE_GL_TEXTURE_3D_ERR);
if (err != NULL) {
*err = error;
}
}
+ //! \brief Default constructor - initializes to NULL.
Image3DGL() : Image3D() { }
- 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) { }
- Image3DGL& operator = (const Image3DGL& rhs)
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image3DGL& operator = (const cl_mem& rhs)
{
- if (this != &rhs) {
- Image3D::operator=(rhs);
+ Image3D::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3DGL(const Image3DGL& img) : Image3D(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3DGL& operator = (const Image3DGL &img)
+ {
+ Image3D::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3DGL(Image3DGL&& img) CL_HPP_NOEXCEPT : Image3D(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3DGL& operator = (Image3DGL &&img)
+ {
+ Image3D::operator=(std::move(img));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+};
+#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,
+ cl_GLenum target,
+ cl_GLint miplevel,
+ cl_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() { }
+
+ __CL_EXPLICIT_CONSTRUCTORS ImageGL(const cl_mem& image) : Image(image) { }
+
+ ImageGL& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ ImageGL(const ImageGL& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ ImageGL& operator = (const ImageGL &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ ImageGL(ImageGL&& img) CL_HPP_NOEXCEPT : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ ImageGL& operator = (ImageGL &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+};
+#endif // #if defined(CL_VERSION_1_2)
+
+/*! \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 :
+#if defined(CL_VERSION_1_2)
+ public ImageGL
+#else // #if defined(CL_VERSION_1_2)
+ public Image2DGL
+#endif //#if defined(CL_VERSION_1_2)
+{
+public:
+ /*! \brief Constructs a BufferRenderGL in a specified context, from a given
+ * GL Renderbuffer.
+ *
+ * Wraps clCreateFromGLRenderbuffer().
+ */
+ BufferRenderGL(
+ const Context& context,
+ cl_mem_flags flags,
+ cl_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.
+#if defined(CL_VERSION_1_2)
+ BufferRenderGL() : ImageGL() {};
+#else // #if defined(CL_VERSION_1_2)
+ BufferRenderGL() : Image2DGL() {};
+#endif //#if defined(CL_VERSION_1_2)
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * See Memory for further details.
+ */
+#if defined(CL_VERSION_1_2)
+ __CL_EXPLICIT_CONSTRUCTORS BufferRenderGL(const cl_mem& buffer) : ImageGL(buffer) { }
+#else // #if defined(CL_VERSION_1_2)
+ __CL_EXPLICIT_CONSTRUCTORS BufferRenderGL(const cl_mem& buffer) : Image2DGL(buffer) { }
+#endif //#if defined(CL_VERSION_1_2)
+
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ BufferRenderGL& operator = (const cl_mem& rhs)
+ {
+#if defined(CL_VERSION_1_2)
+ ImageGL::operator=(rhs);
+#else // #if defined(CL_VERSION_1_2)
+ Image2DGL::operator=(rhs);
+#endif //#if defined(CL_VERSION_1_2)
+
return *this;
}
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+#if defined(CL_VERSION_1_2)
+ BufferRenderGL(const BufferRenderGL& buf) : ImageGL(buf) {}
+#else // #if defined(CL_VERSION_1_2)
+ BufferRenderGL(const BufferRenderGL& buf) : Image2DGL(buf) {}
+#endif //#if defined(CL_VERSION_1_2)
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferRenderGL& operator = (const BufferRenderGL &rhs)
+ {
+#if defined(CL_VERSION_1_2)
+ ImageGL::operator=(rhs);
+#else // #if defined(CL_VERSION_1_2)
+ Image2DGL::operator=(rhs);
+#endif //#if defined(CL_VERSION_1_2)
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+#if defined(CL_VERSION_1_2)
+ BufferRenderGL(BufferRenderGL&& buf) CL_HPP_NOEXCEPT : ImageGL(std::move(buf)) {}
+#else // #if defined(CL_VERSION_1_2)
+ BufferRenderGL(BufferRenderGL&& buf) CL_HPP_NOEXCEPT : Image2DGL(std::move(buf)) {}
+#endif //#if defined(CL_VERSION_1_2)
+
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferRenderGL& operator = (BufferRenderGL &&buf)
+ {
+#if defined(CL_VERSION_1_2)
+ ImageGL::operator=(std::move(buf));
+#else // #if defined(CL_VERSION_1_2)
+ Image2DGL::operator=(std::move(buf));
+#endif //#if defined(CL_VERSION_1_2)
+
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+
+ //! \brief Wrapper for clGetGLObjectInfo().
+ cl_int getObjectInfo(
+ cl_gl_object_type *type,
+ cl_GLuint * gl_object_name)
+ {
+ return detail::errHandler(
+ ::clGetGLObjectInfo(object_, type, gl_object_name),
+ __GET_GL_OBJECT_INFO_ERR);
+ }
};
-/*! \class Sampler
- * \brief Sampler interface for cl_sampler.
+/*! \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<cl_sampler>
{
public:
+ //! \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,
@@ -2167,16 +4686,55 @@ public:
}
}
- Sampler(const Sampler& sampler) : detail::Wrapper<cl_type>(sampler) { }
+ /*! \brief Constructor from cl_sampler - takes ownership.
+ *
+ * This effectively transfers ownership of a refcount on the cl_sampler
+ * into the new Sampler object.
+ */
+ __CL_EXPLICIT_CONSTRUCTORS Sampler(const cl_sampler& sampler) : detail::Wrapper<cl_type>(sampler) { }
- Sampler& operator = (const Sampler& rhs)
+ /*! \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)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
- }
+ detail::Wrapper<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Sampler(const Sampler& sam) : detail::Wrapper<cl_type>(sam) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Sampler& operator = (const Sampler &sam)
+ {
+ detail::Wrapper<cl_type>::operator=(sam);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Sampler(Sampler&& sam) CL_HPP_NOEXCEPT : detail::Wrapper<cl_type>(std::move(sam)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Sampler& operator = (Sampler &&sam)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(sam));
return *this;
}
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ //! \brief Wrapper for clGetSamplerInfo().
template <typename T>
cl_int getInfo(cl_sampler_info name, T* param) const
{
@@ -2185,6 +4743,7 @@ public:
__GET_SAMPLER_INFO_ERR);
}
+ //! \brief Wrapper for clGetSamplerInfo() that returns by value.
template <cl_int name> typename
detail::param_traits<detail::cl_sampler_info, name>::param_type
getInfo(cl_int* err = NULL) const
@@ -2199,15 +4758,11 @@ public:
}
};
-__GET_INFO_HELPER_WITH_RETAIN(cl::Sampler)
-
class Program;
class CommandQueue;
class Kernel;
-/*! \class NDRange
- * \brief NDRange interface
- */
+//! \brief Class interface for specifying NDRange values.
class NDRange
{
private:
@@ -2215,41 +4770,51 @@ private:
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_.push_back(size0);
+ sizes_[0] = size0;
}
+ //! \brief Constructs two-dimensional range.
NDRange(::size_t size0, ::size_t size1)
: dimensions_(2)
{
- sizes_.push_back(size0);
- sizes_.push_back(size1);
+ sizes_[0] = size0;
+ sizes_[1] = size1;
}
+ //! \brief Constructs three-dimensional range.
NDRange(::size_t size0, ::size_t size1, ::size_t size2)
: dimensions_(3)
{
- sizes_.push_back(size0);
- sizes_.push_back(size1);
- sizes_.push_back(size2);
+ 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_;
}
- 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;
-/*!
- * \struct LocalSpaceArg
- * \brief Local address raper for use with Kernel::setArg
- */
+//! \brief Local address wrapper for use with Kernel::setArg
struct LocalSpaceArg
{
::size_t size_;
@@ -2261,19 +4826,25 @@ template <typename T>
struct KernelArgumentHandler
{
static ::size_t size(const T&) { return sizeof(T); }
- static T* ptr(T& value) { return &value; }
+ static const T* ptr(const T& value) { return &value; }
};
template <>
struct KernelArgumentHandler<LocalSpaceArg>
{
static ::size_t size(const LocalSpaceArg& value) { return value.size_; }
- static void* ptr(LocalSpaceArg&) { return NULL; }
+ static const void* ptr(const 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)
{
@@ -2281,28 +4852,82 @@ __local(::size_t size)
return ret;
}
-class KernelFunctor;
+/*! Local
+ * \brief Helper function for generating LocalSpaceArg objects.
+ */
+inline LocalSpaceArg
+Local(::size_t size)
+{
+ LocalSpaceArg ret = { size };
+ return ret;
+}
+
+//class KernelFunctor;
-/*! \class Kernel
- * \brief Kernel interface that implements cl_kernel
+/*! \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<cl_kernel>
{
public:
inline Kernel(const Program& program, const char* name, cl_int* err = NULL);
+ //! \brief Default constructor - initializes to NULL.
Kernel() { }
- Kernel(const Kernel& kernel) : detail::Wrapper<cl_type>(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<cl_type>(kernel) { }
+
+ /*! \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<cl_type>::operator=(rhs);
+ return *this;
+ }
- Kernel& operator = (const Kernel& rhs)
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Kernel(const Kernel& kernel) : detail::Wrapper<cl_type>(kernel) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Kernel& operator = (const Kernel &kernel)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
- }
+ detail::Wrapper<cl_type>::operator=(kernel);
return *this;
}
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Kernel(Kernel&& kernel) CL_HPP_NOEXCEPT : detail::Wrapper<cl_type>(std::move(kernel)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Kernel& operator = (Kernel &&kernel)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(kernel));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+
template <typename T>
cl_int getInfo(cl_kernel_info name, T* param) const
{
@@ -2324,6 +4949,29 @@ public:
return param;
}
+#if defined(CL_VERSION_1_2)
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_kernel_arg_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+#endif // #if defined(CL_VERSION_1_2)
+
template <typename T>
cl_int getWorkGroupInfo(
const Device& device, cl_kernel_work_group_info name, T* param) const
@@ -2348,7 +4996,7 @@ public:
}
template <typename T>
- cl_int setArg(cl_uint index, T value)
+ cl_int setArg(cl_uint index, const T &value)
{
return detail::errHandler(
::clSetKernelArg(
@@ -2359,27 +5007,14 @@ public:
__SET_KERNEL_ARGS_ERR);
}
- cl_int setArg(cl_uint index, ::size_t size, void* argPtr)
+ cl_int setArg(cl_uint index, ::size_t size, const void* argPtr)
{
return detail::errHandler(
::clSetKernelArg(object_, index, size, argPtr),
__SET_KERNEL_ARGS_ERR);
}
-
- KernelFunctor bind(
- const CommandQueue& queue,
- const NDRange& offset,
- const NDRange& global,
- const NDRange& local);
-
- KernelFunctor bind(
- const CommandQueue& queue,
- const NDRange& global,
- const NDRange& local);
};
-__GET_INFO_HELPER_WITH_RETAIN(cl::Kernel)
-
/*! \class Program
* \brief Program interface that implements cl_program.
*/
@@ -2390,6 +5025,75 @@ public:
typedef VECTOR_CLASS<std::pair<const char*, ::size_t> > 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)
@@ -2414,6 +5118,25 @@ public:
}
}
+ /**
+ * 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<Device>& devices,
@@ -2422,20 +5145,41 @@ public:
cl_int* err = NULL)
{
cl_int error;
- const ::size_t n = binaries.size();
- ::size_t* lengths = (::size_t*) alloca(n * sizeof(::size_t));
- const unsigned char** images = (const unsigned char**) alloca(n * sizeof(const void*));
+
+ 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;
+ }
- for (::size_t i = 0; i < n; ++i) {
- images[i] = (const unsigned char*)binaries[(int)i].first;
+ ::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(),
- (cl_device_id*)&devices.front(),
- lengths, images, binaryStatus != NULL
- ? (cl_int*) &binaryStatus->front()
+ deviceIDs,
+ lengths, images, (binaryStatus != NULL && numDevices > 0)
+ ? &binaryStatus->front()
: NULL, &error);
detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR);
@@ -2444,36 +5188,142 @@ public:
}
}
+
+#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<Device>& 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<cl_type>(program) { }
+ __CL_EXPLICIT_CONSTRUCTORS Program(const cl_program& program) : detail::Wrapper<cl_type>(program) { }
- Program& operator = (const Program& rhs)
+ Program& operator = (const cl_program& rhs)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
- }
+ detail::Wrapper<cl_type>::operator=(rhs);
return *this;
}
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Program(const Program& program) : detail::Wrapper<cl_type>(program) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Program& operator = (const Program &program)
+ {
+ detail::Wrapper<cl_type>::operator=(program);
+ return *this;
+ }
+
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Program(Program&& program) CL_HPP_NOEXCEPT : detail::Wrapper<cl_type>(std::move(program)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Program& operator = (Program &&program)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(program));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+
cl_int build(
const VECTOR_CLASS<Device>& 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(),
- (cl_device_id*)&devices.front(),
+ 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 <typename T>
cl_int getInfo(cl_program_info name, T* param) const
{
@@ -2538,7 +5388,106 @@ public:
}
};
-__GET_INFO_HELPER_WITH_RETAIN(cl::Program)
+#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 error_local = CL_SUCCESS;
+
+ cl_program programs[2] = { input1(), input2() };
+
+ Context ctx = input1.getInfo<CL_PROGRAM_CONTEXT>(&error_local);
+ if(error_local!=CL_SUCCESS) {
+ detail::errHandler(error_local, __LINK_PROGRAM_ERR);
+ }
+
+ cl_program prog = ::clLinkProgram(
+ ctx(),
+ 0,
+ NULL,
+ options,
+ 2,
+ programs,
+ notifyFptr,
+ data,
+ &error_local);
+
+ detail::errHandler(error_local,__COMPILE_PROGRAM_ERR);
+ if (err != NULL) {
+ *err = error_local;
+ }
+
+ return Program(prog);
+}
+
+inline Program linkProgram(
+ VECTOR_CLASS<Program> inputPrograms,
+ const char* options = NULL,
+ void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL,
+ void* data = NULL,
+ cl_int* err = NULL)
+{
+ cl_int error_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]();
+ }
+ }
+
+ Context ctx;
+ if(inputPrograms.size() > 0) {
+ ctx = inputPrograms[0].getInfo<CL_PROGRAM_CONTEXT>(&error_local);
+ if(error_local!=CL_SUCCESS) {
+ detail::errHandler(error_local, __LINK_PROGRAM_ERR);
+ }
+ }
+ cl_program prog = ::clLinkProgram(
+ ctx(),
+ 0,
+ NULL,
+ options,
+ (cl_uint)inputPrograms.size(),
+ programs,
+ notifyFptr,
+ data,
+ &error_local);
+
+ detail::errHandler(error_local,__COMPILE_PROGRAM_ERR);
+ if (err != NULL) {
+ *err = error_local;
+ }
+
+ return Program(prog);
+}
+#endif
+
+template<>
+inline VECTOR_CLASS<char *> cl::Program::getInfo<CL_PROGRAM_BINARIES>(cl_int* err) const
+{
+ VECTOR_CLASS< ::size_t> sizes = getInfo<CL_PROGRAM_BINARY_SIZES>();
+ VECTOR_CLASS<char *> 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)
{
@@ -2558,7 +5507,73 @@ inline Kernel::Kernel(const Program& program, const char* name, cl_int* err)
*/
class CommandQueue : public detail::Wrapper<cl_command_queue>
{
+private:
+#ifdef CL_HPP_CPP11_ATOMICS_SUPPORTED
+ static std::atomic<int> default_initialized_;
+#else // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ static volatile int default_initialized_;
+#endif // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+ 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_CONTEXT_ERR);
+
+ if (error != CL_SUCCESS) {
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+ else {
+ Device device = context.getInfo<CL_CONTEXT_DEVICES>()[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<cl::Device> devices;
+ error = context.getInfo(CL_CONTEXT_DEVICES, &devices);
+
+ detail::errHandler(error, __CREATE_CONTEXT_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,
@@ -2575,15 +5590,104 @@ public:
}
}
- CommandQueue() { }
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ CommandQueue(const CommandQueue& queue) : detail::Wrapper<cl_type>(queue) {}
- CommandQueue(const CommandQueue& commandQueue) : detail::Wrapper<cl_type>(commandQueue) { }
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ CommandQueue& operator = (const CommandQueue &queue)
+ {
+ detail::Wrapper<cl_type>::operator=(queue);
+ return *this;
+ }
- CommandQueue& operator = (const CommandQueue& rhs)
+#if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ CommandQueue(CommandQueue&& queue) CL_HPP_NOEXCEPT : detail::Wrapper<cl_type>(std::move(queue)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ CommandQueue& operator = (CommandQueue &&queue)
{
- if (this != &rhs) {
- detail::Wrapper<cl_type>::operator=(rhs);
+ detail::Wrapper<cl_type>::operator=(std::move(queue));
+ return *this;
+ }
+#endif // #if defined(CL_HPP_RVALUE_REFERENCES_SUPPORTED)
+
+ 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<CL_CONTEXT_DEVICES>()[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() { }
+
+ __CL_EXPLICIT_CONSTRUCTORS CommandQueue(const cl_command_queue& commandQueue) : detail::Wrapper<cl_type>(commandQueue) { }
+
+ CommandQueue& operator = (const cl_command_queue& rhs)
+ {
+ detail::Wrapper<cl_type>::operator=(rhs);
return *this;
}
@@ -2618,14 +5722,20 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_READ_BUFFER_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueWriteBuffer(
@@ -2637,14 +5747,20 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_WRITE_BUFFER_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueCopyBuffer(
@@ -2656,16 +5772,22 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQEUE_COPY_BUFFER_ERR);
- }
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
+ }
#if defined(CL_VERSION_1_1)
+
cl_int enqueueReadBufferRect(
const Buffer& buffer,
cl_bool blocking,
@@ -2680,7 +5802,8 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ cl_event tmp;
+ cl_int err = detail::errHandler(
::clEnqueueReadBufferRect(
object_,
buffer(),
@@ -2695,10 +5818,14 @@ public:
ptr,
(events != NULL) ? (cl_uint) events->size() : 0,
(events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event),
+ (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,
@@ -2710,11 +5837,12 @@ public:
::size_t buffer_slice_pitch,
::size_t host_row_pitch,
::size_t host_slice_pitch,
- void *ptr,
+ const void *ptr,
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ cl_event tmp;
+ cl_int err = detail::errHandler(
::clEnqueueWriteBufferRect(
object_,
buffer(),
@@ -2729,8 +5857,13 @@ public:
ptr,
(events != NULL) ? (cl_uint) events->size() : 0,
(events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_WRITE_BUFFER_RECT_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueCopyBufferRect(
@@ -2746,7 +5879,8 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ cl_event tmp;
+ cl_int err = detail::errHandler(
::clEnqueueCopyBufferRect(
object_,
src(),
@@ -2760,10 +5894,52 @@ public:
dst_slice_pitch,
(events != NULL) ? (cl_uint) events->size() : 0,
(events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQEUE_COPY_BUFFER_RECT_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
-#endif
+#endif //if defined(CL_VERSION_1_1)
+
+#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<typename PatternType>
+ cl_int enqueueFillBuffer(
+ const Buffer& buffer,
+ PatternType pattern,
+ ::size_t offset,
+ ::size_t size,
+ const VECTOR_CLASS<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueFillBuffer(
+ object_,
+ buffer(),
+ static_cast<void*>(&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,
@@ -2776,14 +5952,20 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_READ_IMAGE_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueWriteImage(
@@ -2793,18 +5975,24 @@ public:
const size_t<3>& region,
::size_t row_pitch,
::size_t slice_pitch,
- void* ptr,
+ const void* ptr,
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_WRITE_IMAGE_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueCopyImage(
@@ -2816,16 +6004,126 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (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<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueFillImage(
+ object_,
+ image(),
+ static_cast<void*>(&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<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueFillImage(
+ object_,
+ image(),
+ static_cast<void*>(&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<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueFillImage(
+ object_,
+ image(),
+ static_cast<void*>(&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,
@@ -2835,14 +6133,20 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueCopyBufferToImage(
@@ -2854,14 +6158,20 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
void* enqueueMapBuffer(
@@ -2874,18 +6184,22 @@ public:
Event* event = NULL,
cl_int* err = NULL) const
{
+ cl_event tmp;
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,
+ (event != NULL) ? &tmp : NULL,
&error);
detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR);
if (err != NULL) {
*err = error;
}
+ if (event != NULL && error == CL_SUCCESS)
+ *event = tmp;
+
return result;
}
@@ -2901,6 +6215,7 @@ public:
Event* event = NULL,
cl_int* err = NULL) const
{
+ cl_event tmp;
cl_int error;
void * result = ::clEnqueueMapImage(
object_, buffer(), blocking, flags,
@@ -2908,13 +6223,15 @@ public:
row_pitch, slice_pitch,
(events != NULL) ? (cl_uint) events->size() : 0,
(events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event,
+ (event != NULL) ? &tmp : NULL,
&error);
detail::errHandler(error, __ENQUEUE_MAP_IMAGE_ERR);
if (err != NULL) {
*err = error;
}
+ if (event != NULL && error == CL_SUCCESS)
+ *event = tmp;
return result;
}
@@ -2924,24 +6241,129 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (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<Event> *events = 0,
+ Event *event = 0) const
+ {
+ 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<Event> *events = 0,
+ Event *event = 0) const
+ {
+ 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<Memory> &memObjects,
+ cl_mem_migration_flags flags,
+ const VECTOR_CLASS<Event>* events = NULL,
+ Event* event = NULL
+ ) const
+ {
+ cl_event tmp;
+
+ cl_mem* localMemObjects = static_cast<cl_mem*>(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<const cl_mem*>(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,
+ const NDRange& local = NullRange,
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ cl_event tmp;
+ cl_int err = detail::errHandler(
::clEnqueueNDRangeKernel(
object_, kernel(), (cl_uint) global.dimensions(),
offset.dimensions() != 0 ? (const ::size_t*) offset : NULL,
@@ -2949,8 +6371,13 @@ public:
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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_NDRANGE_KERNEL_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueTask(
@@ -2958,17 +6385,23 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_TASK_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueNativeKernel(
- void (*userFptr)(void *),
+ void (CL_CALLBACK *userFptr)(void *),
std::pair<void*, ::size_t> args,
const VECTOR_CLASS<Memory>* mem_objects = NULL,
const VECTOR_CLASS<const void*>* mem_locs = NULL,
@@ -2985,49 +6418,76 @@ public:
}
}
- return detail::errHandler(
+ 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,
+ (mem_locs != NULL && mem_locs->size() > 0) ? (const void **) &mem_locs->front() : NULL,
(events != NULL) ? (cl_uint) events->size() : 0,
(events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_NATIVE_KERNEL);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
- cl_int enqueueMarker(Event* event = NULL) const
+/**
+ * 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),
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueMarker(
+ object_,
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_MARKER_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
- cl_int enqueueWaitForEvents(const VECTOR_CLASS<Event>& events) const
+ CL_EXT_PREFIX__VERSION_1_1_DEPRECATED
+ cl_int enqueueWaitForEvents(const VECTOR_CLASS<Event>& events) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED
{
return detail::errHandler(
::clEnqueueWaitForEvents(
object_,
(cl_uint) events.size(),
- (const cl_event*) &events.front()),
+ events.size() > 0 ? (const cl_event*) &events.front() : NULL),
__ENQUEUE_WAIT_FOR_EVENTS_ERR);
}
+#endif // #if defined(CL_VERSION_1_1)
cl_int enqueueAcquireGLObjects(
const VECTOR_CLASS<Memory>* mem_objects = NULL,
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
+ (mem_objects != NULL && mem_objects->size() > 0) ? (const cl_mem *) &mem_objects->front(): NULL,
(events != NULL) ? (cl_uint) events->size() : 0,
(events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_ACQUIRE_GL_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueReleaseGLObjects(
@@ -3035,15 +6495,21 @@ public:
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
{
- return detail::errHandler(
+ 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,
+ (mem_objects != NULL && mem_objects->size() > 0) ? (const cl_mem *) &mem_objects->front(): NULL,
(events != NULL) ? (cl_uint) events->size() : 0,
(events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_RELEASE_GL_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
#if defined (USE_DX_INTEROP)
@@ -3060,19 +6526,33 @@ typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueReleaseD3D10ObjectsKHR)(
const VECTOR_CLASS<Memory>* mem_objects = NULL,
const VECTOR_CLASS<Event>* events = NULL,
Event* event = NULL) const
- {
- static PFN_clEnqueueAcquireD3D10ObjectsKHR pfn_clEnqueueAcquireD3D10ObjectsKHR = NULL;
- __INIT_CL_EXT_FCN_PTR(clEnqueueAcquireD3D10ObjectsKHR);
-
- return detail::errHandler(
+ {
+ static PFN_clEnqueueAcquireD3D10ObjectsKHR pfn_clEnqueueAcquireD3D10ObjectsKHR = NULL;
+#if defined(CL_VERSION_1_2)
+ cl_context context = getInfo<CL_QUEUE_CONTEXT>();
+ cl::Device device(getInfo<CL_QUEUE_DEVICE>());
+ cl_platform_id platform = device.getInfo<CL_DEVICE_PLATFORM>();
+ __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,
+ (mem_objects != NULL && mem_objects->size() > 0) ? (const cl_mem *) &mem_objects->front(): NULL,
(events != NULL) ? (cl_uint) events->size() : 0,
(events != NULL) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event),
+ (event != NULL) ? &tmp : NULL),
__ENQUEUE_ACQUIRE_GL_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
}
cl_int enqueueReleaseD3D10Objects(
@@ -3081,26 +6561,46 @@ typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueReleaseD3D10ObjectsKHR)(
Event* event = NULL) const
{
static PFN_clEnqueueReleaseD3D10ObjectsKHR pfn_clEnqueueReleaseD3D10ObjectsKHR = NULL;
+#if defined(CL_VERSION_1_2)
+ cl_context context = getInfo<CL_QUEUE_CONTEXT>();
+ cl::Device device(getInfo<CL_QUEUE_DEVICE>());
+ cl_platform_id platform = device.getInfo<CL_DEVICE_PLATFORM>();
+ __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)
- return detail::errHandler(
+ 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,
+ (mem_objects != NULL && mem_objects->size() > 0) ? (const cl_mem *) &mem_objects->front(): NULL,
(events != NULL) ? (cl_uint) events->size() : 0,
- (events != NULL) ? (cl_event*) &events->front() : NULL,
- (cl_event*) event),
+ (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;
}
#endif
- cl_int enqueueBarrier() const
+/**
+ * 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
{
@@ -3113,825 +6613,6254 @@ typedef CL_API_ENTRY cl_int (CL_API_CALL *PFN_clEnqueueReleaseD3D10ObjectsKHR)(
}
};
-__GET_INFO_HELPER_WITH_RETAIN(cl::CommandQueue)
-
-/*! \class KernelFunctor
- * \brief Kernel functor interface
- *
- * \note Currently only functors of zero to ten arguments are supported. It
- * is straightforward to add more and a more general solution, similar to
- * Boost.Lambda could be followed if required in the future.
- */
-class KernelFunctor
+#ifdef CL_HPP_CPP11_ATOMICS_SUPPORTED
+CL_WEAK_ATTRIB_PREFIX std::atomic<int> CL_WEAK_ATTRIB_SUFFIX CommandQueue::default_initialized_;
+#else // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+CL_WEAK_ATTRIB_PREFIX volatile int CL_WEAK_ATTRIB_SUFFIX CommandQueue::default_initialized_ = __DEFAULT_NOT_INITIALIZED;
+#endif // !CL_HPP_CPP11_ATOMICS_SUPPORTED
+
+CL_WEAK_ATTRIB_PREFIX CommandQueue CL_WEAK_ATTRIB_SUFFIX CommandQueue::default_;
+CL_WEAK_ATTRIB_PREFIX volatile cl_int CL_WEAK_ATTRIB_SUFFIX CommandQueue::default_error_ = CL_SUCCESS;
+
+template< typename IteratorType >
+Buffer::Buffer(
+ const Context &context,
+ IteratorType startIterator,
+ IteratorType endIterator,
+ bool readOnly,
+ bool useHostPtr,
+ cl_int* err)
{
-private:
- Kernel kernel_;
- CommandQueue queue_;
- NDRange offset_;
- NDRange global_;
- NDRange local_;
+ typedef typename std::iterator_traits<IteratorType>::value_type DataType;
+ cl_int error;
- cl_int err_;
-public:
- KernelFunctor() { }
+ 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);
- KernelFunctor(
- const Kernel& kernel,
- const CommandQueue& queue,
- const NDRange& offset,
- const NDRange& global,
- const NDRange& local) :
- kernel_(kernel),
- queue_(queue),
- offset_(offset),
- global_(global),
- local_(local),
- err_(CL_SUCCESS)
- {}
+ if( useHostPtr ) {
+ object_ = ::clCreateBuffer(context(), flags, size, static_cast<DataType*>(&*startIterator), &error);
+ } else {
+ object_ = ::clCreateBuffer(context(), flags, size, 0, &error);
+ }
- KernelFunctor& operator=(const KernelFunctor& rhs);
-
- KernelFunctor(const KernelFunctor& rhs);
-
- cl_int getError() { return err_; }
-
- inline Event operator()(const VECTOR_CLASS<Event>* events = NULL);
-
- template<typename A1>
- inline Event operator()(
- const A1& a1,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5, class A6>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4,
- class A5, class A6, class A7>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11, class A12>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const A12& a12,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11, class A12, class A13>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const A12& a12,
- const A13& a13,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11, class A12, class A13, class A14>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const A12& a12,
- const A13& a13,
- const A14& a14,
- const VECTOR_CLASS<Event>* events = NULL);
-
- template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11, class A12, class A13, class A14, class A15>
- inline Event operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const A12& a12,
- const A13& a13,
- const A14& a14,
- const A15& a15,
- const VECTOR_CLASS<Event>* events = NULL);
-};
-
-inline KernelFunctor Kernel::bind(
- const CommandQueue& queue,
- const NDRange& offset,
- const NDRange& global,
- const NDRange& local)
-{
- return KernelFunctor(*this,queue,offset,global,local);
+ 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 KernelFunctor Kernel::bind(
- const CommandQueue& queue,
- const NDRange& global,
- const NDRange& local)
+template< typename IteratorType >
+Buffer::Buffer(
+ const CommandQueue &queue,
+ IteratorType startIterator,
+ IteratorType endIterator,
+ bool readOnly,
+ bool useHostPtr,
+ cl_int* err)
{
- return KernelFunctor(*this,queue,NullRange,global,local);
+ typedef typename std::iterator_traits<IteratorType>::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 = queue.getInfo<CL_QUEUE_CONTEXT>();
+
+ if (useHostPtr) {
+ object_ = ::clCreateBuffer(context(), flags, size, static_cast<DataType*>(&*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(queue, startIterator, endIterator, *this);
+ detail::errHandler(error, __CREATE_BUFFER_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
}
-inline KernelFunctor& KernelFunctor::operator=(const KernelFunctor& rhs)
+inline cl_int enqueueReadBuffer(
+ const Buffer& buffer,
+ cl_bool blocking,
+ ::size_t offset,
+ ::size_t size,
+ void* ptr,
+ const VECTOR_CLASS<Event>* events = NULL,
+ Event* event = NULL)
{
- if (this == &rhs) {
- return *this;
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+
+ if (error != CL_SUCCESS) {
+ return error;
}
-
- kernel_ = rhs.kernel_;
- queue_ = rhs.queue_;
- offset_ = rhs.offset_;
- global_ = rhs.global_;
- local_ = rhs.local_;
-
- return *this;
+
+ return queue.enqueueReadBuffer(buffer, blocking, offset, size, ptr, events, event);
}
-inline KernelFunctor::KernelFunctor(const KernelFunctor& rhs) :
- kernel_(rhs.kernel_),
- queue_(rhs.queue_),
- offset_(rhs.offset_),
- global_(rhs.global_),
- local_(rhs.local_)
+inline cl_int enqueueWriteBuffer(
+ const Buffer& buffer,
+ cl_bool blocking,
+ ::size_t offset,
+ ::size_t size,
+ const void* ptr,
+ const VECTOR_CLASS<Event>* 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);
}
-Event KernelFunctor::operator()(const VECTOR_CLASS<Event>* events)
+inline void* enqueueMapBuffer(
+ const Buffer& buffer,
+ cl_bool blocking,
+ cl_map_flags flags,
+ ::size_t offset,
+ ::size_t size,
+ const VECTOR_CLASS<Event>* events = NULL,
+ Event* event = NULL,
+ cl_int* err = NULL)
{
- Event event;
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
+ 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);
- return event;
+ detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ return result;
}
-template<typename A1>
-Event KernelFunctor::operator()(
- const A1& a1,
- const VECTOR_CLASS<Event>* events)
+inline cl_int enqueueUnmapMemObject(
+ const Memory& memory,
+ void* mapped_ptr,
+ const VECTOR_CLASS<Event>* events = NULL,
+ Event* event = NULL)
{
- Event event;
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR);
+ if (error != CL_SUCCESS) {
+ return error;
+ }
- kernel_.setArg(0,a1);
+ 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);
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
- return event;
+ return err;
}
-template<typename A1, typename A2>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const VECTOR_CLASS<Event>* events)
+inline cl_int enqueueCopyBuffer(
+ const Buffer& src,
+ const Buffer& dst,
+ ::size_t src_offset,
+ ::size_t dst_offset,
+ ::size_t size,
+ const VECTOR_CLASS<Event>* events = NULL,
+ Event* event = NULL)
{
- Event event;
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
+ if (error != CL_SUCCESS) {
+ return error;
+ }
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
+ return queue.enqueueCopyBuffer(src, dst, src_offset, dst_offset, size, events, event);
+}
- return 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);
}
-template<typename A1, typename A2, typename A3>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const VECTOR_CLASS<Event>* events)
+/**
+ * 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 )
{
- Event event;
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ if (error != CL_SUCCESS)
+ return error;
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
+ return cl::copy(queue, buffer, startIterator, endIterator);
+}
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
+/**
+ * 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<IteratorType>::value_type DataType;
+ cl_int error;
+
+ ::size_t length = endIterator-startIterator;
+ ::size_t byteLength = length*sizeof(DataType);
+
+ DataType *pointer =
+ static_cast<DataType*>(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<DataType*>(
+ 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;
+}
- return event;
+/**
+ * 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<IteratorType>::value_type DataType;
+ cl_int error;
+
+ ::size_t length = endIterator-startIterator;
+ ::size_t byteLength = length*sizeof(DataType);
+
+ DataType *pointer =
+ static_cast<DataType*>(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;
}
-template<typename A1, typename A2, typename A3, typename A4>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return event;
+#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<Event>* 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);
}
-template<typename A1, typename A2, typename A3, typename A4, typename A5>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return 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,
+ const void *ptr,
+ const VECTOR_CLASS<Event>* 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);
}
-template<typename A1, typename A2, typename A3, typename A4, typename A5,
- typename A6>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return 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<Event>* 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
-template<typename A1, typename A2, typename A3, typename A4,
- typename A5, typename A6, typename A7>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return event;
+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<Event>* 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);
}
-template<typename A1, typename A2, typename A3, typename A4, typename A5,
- typename A6, typename A7, typename A8>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
- kernel_.setArg(7,a8);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return 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,
+ const void* ptr,
+ const VECTOR_CLASS<Event>* 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);
}
-template<typename A1, typename A2, typename A3, typename A4, typename A5,
- typename A6, typename A7, typename A8, typename A9>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
- kernel_.setArg(7,a8);
- kernel_.setArg(8,a9);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return 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<Event>* 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);
}
-template<typename A1, typename A2, typename A3, typename A4, typename A5,
- typename A6, typename A7, typename A8, typename A9, typename A10>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
- kernel_.setArg(7,a8);
- kernel_.setArg(8,a9);
- kernel_.setArg(9,a10);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return 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<Event>* 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);
}
-template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
- kernel_.setArg(7,a8);
- kernel_.setArg(8,a9);
- kernel_.setArg(9,a10);
- kernel_.setArg(10,a11);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return 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<Event>* 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);
}
-template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11, class A12>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const A12& a12,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
- kernel_.setArg(7,a8);
- kernel_.setArg(8,a9);
- kernel_.setArg(9,a10);
- kernel_.setArg(10,a11);
- kernel_.setArg(11,a12);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return event;
+
+inline cl_int flush(void)
+{
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+
+ if (error != CL_SUCCESS) {
+ return error;
+ }
+
+ return queue.flush();
}
-template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11, class A12, class A13>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const A12& a12,
- const A13& a13,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
-
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
- kernel_.setArg(7,a8);
- kernel_.setArg(8,a9);
- kernel_.setArg(9,a10);
- kernel_.setArg(10,a11);
- kernel_.setArg(11,a12);
- kernel_.setArg(12,a13);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return event;
+inline cl_int finish(void)
+{
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+
+ if (error != CL_SUCCESS) {
+ return error;
+ }
+
+
+ return queue.finish();
}
-template<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11, class A12, class A13, class A14>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const A12& a12,
- const A13& a13,
- const A14& a14,
- const VECTOR_CLASS<Event>* events)
-{
- Event event;
+// 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<Event> 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<Event> &events, NDRange global) :
+ queue_(CommandQueue::getDefault()),
+ offset_(NullRange),
+ global_(global),
+ local_(NullRange),
+ events_(events)
+ {
+
+ }
+
+ EnqueueArgs(const VECTOR_CLASS<Event> &events, NDRange global, NDRange local) :
+ queue_(CommandQueue::getDefault()),
+ offset_(NullRange),
+ global_(global),
+ local_(local),
+ events_(events)
+ {
+
+ }
+
+ EnqueueArgs(const VECTOR_CLASS<Event> &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<Event> &events, NDRange global) :
+ queue_(queue),
+ offset_(NullRange),
+ global_(global),
+ local_(NullRange),
+ events_(events)
+ {
+
+ }
+
+ EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS<Event> &events, NDRange global, NDRange local) :
+ queue_(queue),
+ offset_(NullRange),
+ global_(global),
+ local_(local),
+ events_(events)
+ {
+
+ }
+
+ EnqueueArgs(CommandQueue &queue, const VECTOR_CLASS<Event> &events, NDRange offset, NDRange global, NDRange local) :
+ queue_(queue),
+ offset_(offset),
+ global_(global),
+ local_(local),
+ events_(events)
+ {
+
+ }
+};
+
+namespace detail {
+
+class NullType {};
+
+template<int index, typename T0>
+struct SetArg
+{
+ static void set (Kernel kernel, T0 arg)
+ {
+ kernel.setArg(index, arg);
+ }
+};
+
+template<int index>
+struct SetArg<index, NullType>
+{
+ 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)
+ {
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
- kernel_.setArg(7,a8);
- kernel_.setArg(8,a9);
- kernel_.setArg(9,a10);
- kernel_.setArg(10,a11);
- kernel_.setArg(11,a12);
- kernel_.setArg(12,a13);
- kernel_.setArg(13,a14);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return event;
-}
+ #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<class A1, class A2, class A3, class A4, class A5,
- class A6, class A7, class A8, class A9, class A10,
- class A11, class A12, class A13, class A14, class A15>
-Event KernelFunctor::operator()(
- const A1& a1,
- const A2& a2,
- const A3& a3,
- const A4& a4,
- const A5& a5,
- const A6& a6,
- const A7& a7,
- const A8& a8,
- const A9& a9,
- const A10& a10,
- const A11& a11,
- const A12& a12,
- const A13& a13,
- const A14& a14,
- const A15& a15,
- const VECTOR_CLASS<Event>* events)
-{
- Event 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>
+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)
+ {
- kernel_.setArg(0,a1);
- kernel_.setArg(1,a2);
- kernel_.setArg(2,a3);
- kernel_.setArg(3,a4);
- kernel_.setArg(4,a5);
- kernel_.setArg(5,a6);
- kernel_.setArg(6,a7);
- kernel_.setArg(7,a8);
- kernel_.setArg(8,a9);
- kernel_.setArg(9,a10);
- kernel_.setArg(10,a11);
- kernel_.setArg(11,a12);
- kernel_.setArg(12,a13);
- kernel_.setArg(13,a14);
- kernel_.setArg(14,a15);
-
- err_ = queue_.enqueueNDRangeKernel(
- kernel_,
- offset_,
- global_,
- local_,
- NULL, // bgaster_fixme - do we want to allow wait event lists?
- &event);
-
- return event;
-}
+ #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)
@@ -3945,11 +12874,13 @@ Event KernelFunctor::operator()(
#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
@@ -3963,6 +12894,7 @@ Event KernelFunctor::operator()(
#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
@@ -3970,6 +12902,7 @@ Event KernelFunctor::operator()(
#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
@@ -3993,19 +12926,33 @@ Event KernelFunctor::operator()(
#undef __ENQUEUE_TASK_ERR
#undef __ENQUEUE_NATIVE_KERNEL
+#undef __CL_EXPLICIT_CONSTRUCTORS
+
#undef __UNLOAD_COMPILER_ERR
#endif //__CL_USER_OVERRIDE_ERROR_STRINGS
-#undef __GET_INFO_HELPER_WITH_RETAIN
+#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
+
+#undef CL_HPP_RVALUE_REFERENCES_SUPPORTED
+#undef CL_HPP_NOEXCEPT
+
} // namespace cl
#endif // CL_HPP_
diff --git a/include/CL/cl2.hpp b/include/CL/cl2.hpp
new file mode 100644
index 00000000000..4ead85cb4ad
--- /dev/null
+++ b/include/CL/cl2.hpp
@@ -0,0 +1,9690 @@
+/*******************************************************************************
+ * Copyright (c) 2008-2016 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.
+ *
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
+ * 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),
+ * OpenCL 1.2 (rev 15) and OpenCL 2.0 (rev 29)
+ * \author Lee Howes and Bruce Merry
+ *
+ * Derived from the OpenCL 1.x C++ bindings written by
+ * Benedict R. Gaster, Laurent Morichetti and Lee Howes
+ * With 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
+ * James Price, 2015-
+ *
+ * \version 2.0.10
+ * \date 2016-07-20
+ *
+ * Optional extension support
+ *
+ * cl_ext_device_fission
+ * #define CL_HPP_USE_CL_DEVICE_FISSION
+ * cl_khr_d3d10_sharing
+ * #define CL_HPP_USE_DX_INTEROP
+ * cl_khr_sub_groups
+ * #define CL_HPP_USE_CL_SUB_GROUPS_KHR
+ * cl_khr_image2d_from_buffer
+ * #define CL_HPP_USE_CL_IMAGE2D_FROM_BUFFER_KHR
+ *
+ * Doxygen documentation for this header is available here:
+ *
+ * http://khronosgroup.github.io/OpenCL-CLHPP/
+ *
+ * The latest version of this header can be found on the GitHub releases page:
+ *
+ * https://github.com/KhronosGroup/OpenCL-CLHPP/releases
+ *
+ * Bugs and patches can be submitted to the GitHub repository:
+ *
+ * https://github.com/KhronosGroup/OpenCL-CLHPP
+ */
+
+/*! \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 cl2.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 cl2.hpp.
+ *
+ * The bindings themselves are lightweight and correspond closely to the
+ * underlying C API. Using the C++ bindings introduces no additional execution
+ * overhead.
+ *
+ * There are numerous compatibility, portability and memory management
+ * fixes in the new header as well as additional OpenCL 2.0 features.
+ * As a result the header is not directly backward compatible and for this
+ * reason we release it as cl2.hpp rather than a new version of cl.hpp.
+ *
+ *
+ * \section compatibility Compatibility
+ * Due to the evolution of the underlying OpenCL API the 2.0 C++ bindings
+ * include an updated approach to defining supported feature versions
+ * and the range of valid underlying OpenCL runtime versions supported.
+ *
+ * The combination of preprocessor macros CL_HPP_TARGET_OPENCL_VERSION and
+ * CL_HPP_MINIMUM_OPENCL_VERSION control this range. These are three digit
+ * decimal values representing OpenCL runime versions. The default for
+ * the target is 200, representing OpenCL 2.0 and the minimum is also
+ * defined as 200. These settings would use 2.0 API calls only.
+ * If backward compatibility with a 1.2 runtime is required, the minimum
+ * version may be set to 120.
+ *
+ * Note that this is a compile-time setting, and so affects linking against
+ * a particular SDK version rather than the versioning of the loaded runtime.
+ *
+ * The earlier versions of the header included basic vector and string
+ * classes based loosely on STL versions. These were difficult to
+ * maintain and very rarely used. For the 2.0 header we now assume
+ * the presence of the standard library unless requested otherwise.
+ * We use std::array, std::vector, std::shared_ptr and std::string
+ * throughout to safely manage memory and reduce the chance of a
+ * recurrance of earlier memory management bugs.
+ *
+ * These classes are used through typedefs in the cl namespace:
+ * cl::array, cl::vector, cl::pointer and cl::string.
+ * In addition cl::allocate_pointer forwards to std::allocate_shared
+ * by default.
+ * In all cases these standard library classes can be replaced with
+ * custom interface-compatible versions using the CL_HPP_NO_STD_ARRAY,
+ * CL_HPP_NO_STD_VECTOR, CL_HPP_NO_STD_UNIQUE_PTR and
+ * CL_HPP_NO_STD_STRING macros.
+ *
+ * The OpenCL 1.x versions of the C++ bindings included a size_t wrapper
+ * class to interface with kernel enqueue. This caused unpleasant interactions
+ * with the standard size_t declaration and led to namespacing bugs.
+ * In the 2.0 version we have replaced this with a std::array-based interface.
+ * However, the old behaviour can be regained for backward compatibility
+ * using the CL_HPP_ENABLE_SIZE_T_COMPATIBILITY macro.
+ *
+ * Finally, the program construction interface used a clumsy vector-of-pairs
+ * design in the earlier versions. We have replaced that with a cleaner
+ * vector-of-vectors and vector-of-strings design. However, for backward
+ * compatibility old behaviour can be regained with the
+ * CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY macro.
+ *
+ * In OpenCL 2.0 OpenCL C is not entirely backward compatibility with
+ * earlier versions. As a result a flag must be passed to the OpenCL C
+ * compiled to request OpenCL 2.0 compilation of kernels with 1.2 as
+ * the default in the absence of the flag.
+ * In some cases the C++ bindings automatically compile code for ease.
+ * For those cases the compilation defaults to OpenCL C 2.0.
+ * If this is not wanted, the CL_HPP_CL_1_2_DEFAULT_BUILD macro may
+ * be specified to assume 1.2 compilation.
+ * If more fine-grained decisions on a per-kernel bases are required
+ * then explicit build operations that take the flag should be used.
+ *
+ *
+ * \section parameterization Parameters
+ * This header may be parameterized by a set of preprocessor macros.
+ *
+ * - CL_HPP_TARGET_OPENCL_VERSION
+ *
+ * Defines the target OpenCL runtime version to build the header
+ * against. Defaults to 200, representing OpenCL 2.0.
+ *
+ * - CL_HPP_NO_STD_STRING
+ *
+ * Do not use the standard library string class. cl::string is not
+ * defined and may be defined by the user before cl2.hpp is
+ * included.
+ *
+ * - CL_HPP_NO_STD_VECTOR
+ *
+ * Do not use the standard library vector class. cl::vector is not
+ * defined and may be defined by the user before cl2.hpp is
+ * included.
+ *
+ * - CL_HPP_NO_STD_ARRAY
+ *
+ * Do not use the standard library array class. cl::array is not
+ * defined and may be defined by the user before cl2.hpp is
+ * included.
+ *
+ * - CL_HPP_NO_STD_UNIQUE_PTR
+ *
+ * Do not use the standard library unique_ptr class. cl::pointer and
+ * the cl::allocate_pointer functions are not defined and may be
+ * defined by the user before cl2.hpp is included.
+ *
+ * - CL_HPP_ENABLE_DEVICE_FISSION
+ *
+ * Enables device fission for OpenCL 1.2 platforms.
+ *
+ * - CL_HPP_ENABLE_EXCEPTIONS
+ *
+ * Enable exceptions for use in the C++ bindings header. This is the
+ * preferred error handling mechanism but is not required.
+ *
+ * - CL_HPP_ENABLE_SIZE_T_COMPATIBILITY
+ *
+ * Backward compatibility option to support cl.hpp-style size_t
+ * class. Replaces the updated std::array derived version and
+ * removal of size_t from the namespace. Note that in this case the
+ * new size_t class is placed in the cl::compatibility namespace and
+ * thus requires an additional using declaration for direct backward
+ * compatibility.
+ *
+ * - CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY
+ *
+ * Enable older vector of pairs interface for construction of
+ * programs.
+ *
+ * - CL_HPP_CL_1_2_DEFAULT_BUILD
+ *
+ * Default to OpenCL C 1.2 compilation rather than OpenCL C 2.0
+ * applies to use of cl::Program construction and other program
+ * build variants.
+ *
+ *
+ * \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_HPP_ENABLE_EXCEPTIONS
+ #define CL_HPP_TARGET_OPENCL_VERSION 200
+
+ #include <CL/cl2.hpp>
+ #include <iostream>
+ #include <vector>
+ #include <memory>
+ #include <algorithm>
+
+ const int numElements = 32;
+
+ int main(void)
+ {
+ // Filter for a 2.0 platform and set it as the default
+ std::vector<cl::Platform> platforms;
+ cl::Platform::get(&platforms);
+ cl::Platform plat;
+ for (auto &p : platforms) {
+ std::string platver = p.getInfo<CL_PLATFORM_VERSION>();
+ if (platver.find("OpenCL 2.") != std::string::npos) {
+ plat = p;
+ }
+ }
+ if (plat() == 0) {
+ std::cout << "No OpenCL 2.0 platform found.";
+ return -1;
+ }
+
+ cl::Platform newP = cl::Platform::setDefault(plat);
+ if (newP != plat) {
+ std::cout << "Error setting default platform.";
+ return -1;
+ }
+
+ // Use C++11 raw string literals for kernel source code
+ std::string kernel1{R"CLC(
+ global int globalA;
+ kernel void updateGlobal()
+ {
+ globalA = 75;
+ }
+ )CLC"};
+ std::string kernel2{R"CLC(
+ typedef struct { global int *bar; } Foo;
+ kernel void vectorAdd(global const Foo* aNum, global const int *inputA, global const int *inputB,
+ global int *output, int val, write_only pipe int outPipe, queue_t childQueue)
+ {
+ output[get_global_id(0)] = inputA[get_global_id(0)] + inputB[get_global_id(0)] + val + *(aNum->bar);
+ write_pipe(outPipe, &val);
+ queue_t default_queue = get_default_queue();
+ ndrange_t ndrange = ndrange_1D(get_global_size(0)/2, get_global_size(0)/2);
+
+ // Have a child kernel write into third quarter of output
+ enqueue_kernel(default_queue, CLK_ENQUEUE_FLAGS_WAIT_KERNEL, ndrange,
+ ^{
+ output[get_global_size(0)*2 + get_global_id(0)] =
+ inputA[get_global_size(0)*2 + get_global_id(0)] + inputB[get_global_size(0)*2 + get_global_id(0)] + globalA;
+ });
+
+ // Have a child kernel write into last quarter of output
+ enqueue_kernel(childQueue, CLK_ENQUEUE_FLAGS_WAIT_KERNEL, ndrange,
+ ^{
+ output[get_global_size(0)*3 + get_global_id(0)] =
+ inputA[get_global_size(0)*3 + get_global_id(0)] + inputB[get_global_size(0)*3 + get_global_id(0)] + globalA + 2;
+ });
+ }
+ )CLC"};
+
+ // New simpler string interface style
+ std::vector<std::string> programStrings {kernel1, kernel2};
+
+ cl::Program vectorAddProgram(programStrings);
+ try {
+ vectorAddProgram.build("-cl-std=CL2.0");
+ }
+ catch (...) {
+ // Print build info for all devices
+ cl_int buildErr = CL_SUCCESS;
+ auto buildInfo = vectorAddProgram.getBuildInfo<CL_PROGRAM_BUILD_LOG>(&buildErr);
+ for (auto &pair : buildInfo) {
+ std::cerr << pair.second << std::endl << std::endl;
+ }
+
+ return 1;
+ }
+
+ typedef struct { int *bar; } Foo;
+
+ // Get and run kernel that initializes the program-scope global
+ // A test for kernels that take no arguments
+ auto program2Kernel =
+ cl::KernelFunctor<>(vectorAddProgram, "updateGlobal");
+ program2Kernel(
+ cl::EnqueueArgs(
+ cl::NDRange(1)));
+
+ //////////////////
+ // SVM allocations
+
+ auto anSVMInt = cl::allocate_svm<int, cl::SVMTraitCoarse<>>();
+ *anSVMInt = 5;
+ cl::SVMAllocator<Foo, cl::SVMTraitCoarse<cl::SVMTraitReadOnly<>>> svmAllocReadOnly;
+ auto fooPointer = cl::allocate_pointer<Foo>(svmAllocReadOnly);
+ fooPointer->bar = anSVMInt.get();
+ cl::SVMAllocator<int, cl::SVMTraitCoarse<>> svmAlloc;
+ std::vector<int, cl::SVMAllocator<int, cl::SVMTraitCoarse<>>> inputA(numElements, 1, svmAlloc);
+ cl::coarse_svm_vector<int> inputB(numElements, 2, svmAlloc);
+
+ //
+ //////////////
+
+ // Traditional cl_mem allocations
+ std::vector<int> output(numElements, 0xdeadbeef);
+ cl::Buffer outputBuffer(begin(output), end(output), false);
+ cl::Pipe aPipe(sizeof(cl_int), numElements / 2);
+
+ // Default command queue, also passed in as a parameter
+ cl::DeviceCommandQueue defaultDeviceQueue = cl::DeviceCommandQueue::makeDefault(
+ cl::Context::getDefault(), cl::Device::getDefault());
+
+ auto vectorAddKernel =
+ cl::KernelFunctor<
+ decltype(fooPointer)&,
+ int*,
+ cl::coarse_svm_vector<int>&,
+ cl::Buffer,
+ int,
+ cl::Pipe&,
+ cl::DeviceCommandQueue
+ >(vectorAddProgram, "vectorAdd");
+
+ // Ensure that the additional SVM pointer is available to the kernel
+ // This one was not passed as a parameter
+ vectorAddKernel.setSVMPointers(anSVMInt);
+
+ // Hand control of coarse allocations to runtime
+ cl::enqueueUnmapSVM(anSVMInt);
+ cl::enqueueUnmapSVM(fooPointer);
+ cl::unmapSVM(inputB);
+ cl::unmapSVM(output2);
+
+ cl_int error;
+ vectorAddKernel(
+ cl::EnqueueArgs(
+ cl::NDRange(numElements/2),
+ cl::NDRange(numElements/2)),
+ fooPointer,
+ inputA.data(),
+ inputB,
+ outputBuffer,
+ 3,
+ aPipe,
+ defaultDeviceQueue,
+ error
+ );
+
+ cl::copy(outputBuffer, begin(output), end(output));
+ // Grab the SVM output vector using a map
+ cl::mapSVM(output2);
+
+ cl::Device d = cl::Device::getDefault();
+
+ std::cout << "Output:\n";
+ for (int i = 1; i < numElements; ++i) {
+ std::cout << "\t" << output[i] << "\n";
+ }
+ std::cout << "\n\n";
+
+ return 0;
+ }
+ *
+ * \endcode
+ *
+ */
+#ifndef CL_HPP_
+#define CL_HPP_
+
+/* Handle deprecated preprocessor definitions. In each case, we only check for
+ * the old name if the new name is not defined, so that user code can define
+ * both and hence work with either version of the bindings.
+ */
+#if !defined(CL_HPP_USE_DX_INTEROP) && defined(USE_DX_INTEROP)
+# pragma message("cl2.hpp: USE_DX_INTEROP is deprecated. Define CL_HPP_USE_DX_INTEROP instead")
+# define CL_HPP_USE_DX_INTEROP
+#endif
+#if !defined(CL_HPP_USE_CL_DEVICE_FISSION) && defined(USE_CL_DEVICE_FISSION)
+# pragma message("cl2.hpp: USE_CL_DEVICE_FISSION is deprecated. Define CL_HPP_USE_CL_DEVICE_FISSION instead")
+# define CL_HPP_USE_CL_DEVICE_FISSION
+#endif
+#if !defined(CL_HPP_ENABLE_EXCEPTIONS) && defined(__CL_ENABLE_EXCEPTIONS)
+# pragma message("cl2.hpp: __CL_ENABLE_EXCEPTIONS is deprecated. Define CL_HPP_ENABLE_EXCEPTIONS instead")
+# define CL_HPP_ENABLE_EXCEPTIONS
+#endif
+#if !defined(CL_HPP_NO_STD_VECTOR) && defined(__NO_STD_VECTOR)
+# pragma message("cl2.hpp: __NO_STD_VECTOR is deprecated. Define CL_HPP_NO_STD_VECTOR instead")
+# define CL_HPP_NO_STD_VECTOR
+#endif
+#if !defined(CL_HPP_NO_STD_STRING) && defined(__NO_STD_STRING)
+# pragma message("cl2.hpp: __NO_STD_STRING is deprecated. Define CL_HPP_NO_STD_STRING instead")
+# define CL_HPP_NO_STD_STRING
+#endif
+#if defined(VECTOR_CLASS)
+# pragma message("cl2.hpp: VECTOR_CLASS is deprecated. Alias cl::vector instead")
+#endif
+#if defined(STRING_CLASS)
+# pragma message("cl2.hpp: STRING_CLASS is deprecated. Alias cl::string instead.")
+#endif
+#if !defined(CL_HPP_USER_OVERRIDE_ERROR_STRINGS) && defined(__CL_USER_OVERRIDE_ERROR_STRINGS)
+# pragma message("cl2.hpp: __CL_USER_OVERRIDE_ERROR_STRINGS is deprecated. Define CL_HPP_USER_OVERRIDE_ERROR_STRINGS instead")
+# define CL_HPP_USER_OVERRIDE_ERROR_STRINGS
+#endif
+
+/* Warn about features that are no longer supported
+ */
+#if defined(__USE_DEV_VECTOR)
+# pragma message("cl2.hpp: __USE_DEV_VECTOR is no longer supported. Expect compilation errors")
+#endif
+#if defined(__USE_DEV_STRING)
+# pragma message("cl2.hpp: __USE_DEV_STRING is no longer supported. Expect compilation errors")
+#endif
+
+/* Detect which version to target */
+#if !defined(CL_HPP_TARGET_OPENCL_VERSION)
+# pragma message("cl2.hpp: CL_HPP_TARGET_OPENCL_VERSION is not defined. It will default to 200 (OpenCL 2.0)")
+# define CL_HPP_TARGET_OPENCL_VERSION 200
+#endif
+#if CL_HPP_TARGET_OPENCL_VERSION != 100 && CL_HPP_TARGET_OPENCL_VERSION != 110 && CL_HPP_TARGET_OPENCL_VERSION != 120 && CL_HPP_TARGET_OPENCL_VERSION != 200
+# pragma message("cl2.hpp: CL_HPP_TARGET_OPENCL_VERSION is not a valid value (100, 110, 120 or 200). It will be set to 200")
+# undef CL_HPP_TARGET_OPENCL_VERSION
+# define CL_HPP_TARGET_OPENCL_VERSION 200
+#endif
+
+/* Forward target OpenCL version to C headers if necessary */
+#if defined(CL_TARGET_OPENCL_VERSION)
+/* Warn if prior definition of CL_TARGET_OPENCL_VERSION is lower than
+ * requested C++ bindings version */
+#if CL_TARGET_OPENCL_VERSION < CL_HPP_TARGET_OPENCL_VERSION
+# pragma message("CL_TARGET_OPENCL_VERSION is already defined as is lower than CL_HPP_TARGET_OPENCL_VERSION")
+#endif
+#else
+# define CL_TARGET_OPENCL_VERSION CL_HPP_TARGET_OPENCL_VERSION
+#endif
+
+#if !defined(CL_HPP_MINIMUM_OPENCL_VERSION)
+# define CL_HPP_MINIMUM_OPENCL_VERSION 200
+#endif
+#if CL_HPP_MINIMUM_OPENCL_VERSION != 100 && CL_HPP_MINIMUM_OPENCL_VERSION != 110 && CL_HPP_MINIMUM_OPENCL_VERSION != 120 && CL_HPP_MINIMUM_OPENCL_VERSION != 200
+# pragma message("cl2.hpp: CL_HPP_MINIMUM_OPENCL_VERSION is not a valid value (100, 110, 120 or 200). It will be set to 100")
+# undef CL_HPP_MINIMUM_OPENCL_VERSION
+# define CL_HPP_MINIMUM_OPENCL_VERSION 100
+#endif
+#if CL_HPP_MINIMUM_OPENCL_VERSION > CL_HPP_TARGET_OPENCL_VERSION
+# error "CL_HPP_MINIMUM_OPENCL_VERSION must not be greater than CL_HPP_TARGET_OPENCL_VERSION"
+#endif
+
+#if CL_HPP_MINIMUM_OPENCL_VERSION <= 100 && !defined(CL_USE_DEPRECATED_OPENCL_1_0_APIS)
+# define CL_USE_DEPRECATED_OPENCL_1_0_APIS
+#endif
+#if CL_HPP_MINIMUM_OPENCL_VERSION <= 110 && !defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+# define CL_USE_DEPRECATED_OPENCL_1_1_APIS
+#endif
+#if CL_HPP_MINIMUM_OPENCL_VERSION <= 120 && !defined(CL_USE_DEPRECATED_OPENCL_1_2_APIS)
+# define CL_USE_DEPRECATED_OPENCL_1_2_APIS
+#endif
+#if CL_HPP_MINIMUM_OPENCL_VERSION <= 200 && !defined(CL_USE_DEPRECATED_OPENCL_2_0_APIS)
+# define CL_USE_DEPRECATED_OPENCL_2_0_APIS
+#endif
+
+#ifdef _WIN32
+
+#include <malloc.h>
+
+#if defined(CL_HPP_USE_DX_INTEROP)
+#include <CL/cl_d3d10.h>
+#include <CL/cl_dx9_media_sharing.h>
+#endif
+#endif // _WIN32
+
+#if defined(_MSC_VER)
+#include <intrin.h>
+#endif // _MSC_VER
+
+ // Check for a valid C++ version
+
+// Need to do both tests here because for some reason __cplusplus is not
+// updated in visual studio
+#if (!defined(_MSC_VER) && __cplusplus < 201103L) || (defined(_MSC_VER) && _MSC_VER < 1700)
+#error Visual studio 2013 or another C++11-supporting compiler required
+#endif
+
+//
+#if defined(CL_HPP_USE_CL_DEVICE_FISSION) || defined(CL_HPP_USE_CL_SUB_GROUPS_KHR)
+#include <CL/cl_ext.h>
+#endif
+
+#if defined(__APPLE__) || defined(__MACOSX)
+#include <OpenCL/opencl.h>
+#else
+#include <CL/opencl.h>
+#endif // !__APPLE__
+
+#if (__cplusplus >= 201103L)
+#define CL_HPP_NOEXCEPT_ noexcept
+#else
+#define CL_HPP_NOEXCEPT_
+#endif
+
+#if defined(_MSC_VER)
+# define CL_HPP_DEFINE_STATIC_MEMBER_ __declspec(selectany)
+#else
+# define CL_HPP_DEFINE_STATIC_MEMBER_ __attribute__((weak))
+#endif // !_MSC_VER
+
+// 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_EXT_PREFIX__VERSION_1_2_DEPRECATED)
+#define CL_EXT_PREFIX__VERSION_1_2_DEPRECATED
+#endif // #if !defined(CL_EXT_PREFIX__VERSION_1_2_DEPRECATED)
+#if !defined(CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED)
+#define CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED
+#endif // #if !defined(CL_EXT_PREFIX__VERSION_1_2_DEPRECATED)
+
+#if !defined(CL_CALLBACK)
+#define CL_CALLBACK
+#endif //CL_CALLBACK
+
+#include <utility>
+#include <limits>
+#include <iterator>
+#include <mutex>
+#include <cstring>
+#include <functional>
+
+
+// Define a size_type to represent a correctly resolved size_t
+#if defined(CL_HPP_ENABLE_SIZE_T_COMPATIBILITY)
+namespace cl {
+ using size_type = ::size_t;
+} // namespace cl
+#else // #if defined(CL_HPP_ENABLE_SIZE_T_COMPATIBILITY)
+namespace cl {
+ using size_type = size_t;
+} // namespace cl
+#endif // #if defined(CL_HPP_ENABLE_SIZE_T_COMPATIBILITY)
+
+
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+#include <exception>
+#endif // #if defined(CL_HPP_ENABLE_EXCEPTIONS)
+
+#if !defined(CL_HPP_NO_STD_VECTOR)
+#include <vector>
+namespace cl {
+ template < class T, class Alloc = std::allocator<T> >
+ using vector = std::vector<T, Alloc>;
+} // namespace cl
+#endif // #if !defined(CL_HPP_NO_STD_VECTOR)
+
+#if !defined(CL_HPP_NO_STD_STRING)
+#include <string>
+namespace cl {
+ using string = std::string;
+} // namespace cl
+#endif // #if !defined(CL_HPP_NO_STD_STRING)
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+#if !defined(CL_HPP_NO_STD_UNIQUE_PTR)
+#include <memory>
+namespace cl {
+ // Replace unique_ptr and allocate_pointer for internal use
+ // to allow user to replace them
+ template<class T, class D>
+ using pointer = std::unique_ptr<T, D>;
+} // namespace cl
+#endif
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+#if !defined(CL_HPP_NO_STD_ARRAY)
+#include <array>
+namespace cl {
+ template < class T, size_type N >
+ using array = std::array<T, N>;
+} // namespace cl
+#endif // #if !defined(CL_HPP_NO_STD_ARRAY)
+
+// Define size_type appropriately to allow backward-compatibility
+// use of the old size_t interface class
+#if defined(CL_HPP_ENABLE_SIZE_T_COMPATIBILITY)
+namespace cl {
+ namespace compatibility {
+ /*! \brief class used to interface between C++ and
+ * OpenCL C calls that require arrays of size_t values, whose
+ * size is known statically.
+ */
+ template <int N>
+ class size_t
+ {
+ private:
+ size_type data_[N];
+
+ public:
+ //! \brief Initialize size_t to all 0s
+ size_t()
+ {
+ for (int i = 0; i < N; ++i) {
+ data_[i] = 0;
+ }
+ }
+
+ size_t(const array<size_type, N> &rhs)
+ {
+ for (int i = 0; i < N; ++i) {
+ data_[i] = rhs[i];
+ }
+ }
+
+ size_type& operator[](int index)
+ {
+ return data_[index];
+ }
+
+ const size_type& operator[](int index) const
+ {
+ return data_[index];
+ }
+
+ //! \brief Conversion operator to T*.
+ operator size_type* () { return data_; }
+
+ //! \brief Conversion operator to const T*.
+ operator const size_type* () const { return data_; }
+
+ operator array<size_type, N>() const
+ {
+ array<size_type, N> ret;
+
+ for (int i = 0; i < N; ++i) {
+ ret[i] = data_[i];
+ }
+ return ret;
+ }
+ };
+ } // namespace compatibility
+
+ template<int N>
+ using size_t = compatibility::size_t<N>;
+} // namespace cl
+#endif // #if defined(CL_HPP_ENABLE_SIZE_T_COMPATIBILITY)
+
+// Helper alias to avoid confusing the macros
+namespace cl {
+ namespace detail {
+ using size_t_array = array<size_type, 3>;
+ } // namespace detail
+} // namespace cl
+
+
+/*! \namespace cl
+ *
+ * \brief The OpenCL C++ bindings are defined within this namespace.
+ *
+ */
+namespace cl {
+ class Memory;
+
+#define CL_HPP_INIT_CL_EXT_FCN_PTR_(name) \
+ if (!pfn_##name) { \
+ pfn_##name = (PFN_##name) \
+ clGetExtensionFunctionAddress(#name); \
+ if (!pfn_##name) { \
+ } \
+ }
+
+#define CL_HPP_INIT_CL_EXT_FCN_PTR_PLATFORM_(platform, name) \
+ if (!pfn_##name) { \
+ pfn_##name = (PFN_##name) \
+ clGetExtensionFunctionAddressForPlatform(platform, #name); \
+ if (!pfn_##name) { \
+ } \
+ }
+
+ class Program;
+ class Device;
+ class Context;
+ class CommandQueue;
+ class DeviceCommandQueue;
+ class Memory;
+ class Buffer;
+ class Pipe;
+
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ /*! \brief Exception class
+ *
+ * This may be thrown by API functions when CL_HPP_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 CL_HPP_ERR_STR_(x) #x
+#else
+#define CL_HPP_ERR_STR_(x) NULL
+#endif // CL_HPP_ENABLE_EXCEPTIONS
+
+
+namespace detail
+{
+#if defined(CL_HPP_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_HPP_ENABLE_EXCEPTIONS
+}
+
+
+
+//! \cond DOXYGEN_DETAIL
+#if !defined(CL_HPP_USER_OVERRIDE_ERROR_STRINGS)
+#define __GET_DEVICE_INFO_ERR CL_HPP_ERR_STR_(clGetDeviceInfo)
+#define __GET_PLATFORM_INFO_ERR CL_HPP_ERR_STR_(clGetPlatformInfo)
+#define __GET_DEVICE_IDS_ERR CL_HPP_ERR_STR_(clGetDeviceIDs)
+#define __GET_PLATFORM_IDS_ERR CL_HPP_ERR_STR_(clGetPlatformIDs)
+#define __GET_CONTEXT_INFO_ERR CL_HPP_ERR_STR_(clGetContextInfo)
+#define __GET_EVENT_INFO_ERR CL_HPP_ERR_STR_(clGetEventInfo)
+#define __GET_EVENT_PROFILE_INFO_ERR CL_HPP_ERR_STR_(clGetEventProfileInfo)
+#define __GET_MEM_OBJECT_INFO_ERR CL_HPP_ERR_STR_(clGetMemObjectInfo)
+#define __GET_IMAGE_INFO_ERR CL_HPP_ERR_STR_(clGetImageInfo)
+#define __GET_SAMPLER_INFO_ERR CL_HPP_ERR_STR_(clGetSamplerInfo)
+#define __GET_KERNEL_INFO_ERR CL_HPP_ERR_STR_(clGetKernelInfo)
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __GET_KERNEL_ARG_INFO_ERR CL_HPP_ERR_STR_(clGetKernelArgInfo)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __GET_KERNEL_WORK_GROUP_INFO_ERR CL_HPP_ERR_STR_(clGetKernelWorkGroupInfo)
+#define __GET_PROGRAM_INFO_ERR CL_HPP_ERR_STR_(clGetProgramInfo)
+#define __GET_PROGRAM_BUILD_INFO_ERR CL_HPP_ERR_STR_(clGetProgramBuildInfo)
+#define __GET_COMMAND_QUEUE_INFO_ERR CL_HPP_ERR_STR_(clGetCommandQueueInfo)
+
+#define __CREATE_CONTEXT_ERR CL_HPP_ERR_STR_(clCreateContext)
+#define __CREATE_CONTEXT_FROM_TYPE_ERR CL_HPP_ERR_STR_(clCreateContextFromType)
+#define __GET_SUPPORTED_IMAGE_FORMATS_ERR CL_HPP_ERR_STR_(clGetSupportedImageFormats)
+
+#define __CREATE_BUFFER_ERR CL_HPP_ERR_STR_(clCreateBuffer)
+#define __COPY_ERR CL_HPP_ERR_STR_(cl::copy)
+#define __CREATE_SUBBUFFER_ERR CL_HPP_ERR_STR_(clCreateSubBuffer)
+#define __CREATE_GL_BUFFER_ERR CL_HPP_ERR_STR_(clCreateFromGLBuffer)
+#define __CREATE_GL_RENDER_BUFFER_ERR CL_HPP_ERR_STR_(clCreateFromGLBuffer)
+#define __GET_GL_OBJECT_INFO_ERR CL_HPP_ERR_STR_(clGetGLObjectInfo)
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __CREATE_IMAGE_ERR CL_HPP_ERR_STR_(clCreateImage)
+#define __CREATE_GL_TEXTURE_ERR CL_HPP_ERR_STR_(clCreateFromGLTexture)
+#define __IMAGE_DIMENSION_ERR CL_HPP_ERR_STR_(Incorrect image dimensions)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __SET_MEM_OBJECT_DESTRUCTOR_CALLBACK_ERR CL_HPP_ERR_STR_(clSetMemObjectDestructorCallback)
+
+#define __CREATE_USER_EVENT_ERR CL_HPP_ERR_STR_(clCreateUserEvent)
+#define __SET_USER_EVENT_STATUS_ERR CL_HPP_ERR_STR_(clSetUserEventStatus)
+#define __SET_EVENT_CALLBACK_ERR CL_HPP_ERR_STR_(clSetEventCallback)
+#define __WAIT_FOR_EVENTS_ERR CL_HPP_ERR_STR_(clWaitForEvents)
+
+#define __CREATE_KERNEL_ERR CL_HPP_ERR_STR_(clCreateKernel)
+#define __SET_KERNEL_ARGS_ERR CL_HPP_ERR_STR_(clSetKernelArg)
+#define __CREATE_PROGRAM_WITH_SOURCE_ERR CL_HPP_ERR_STR_(clCreateProgramWithSource)
+#define __CREATE_PROGRAM_WITH_BINARY_ERR CL_HPP_ERR_STR_(clCreateProgramWithBinary)
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR CL_HPP_ERR_STR_(clCreateProgramWithBuiltInKernels)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __BUILD_PROGRAM_ERR CL_HPP_ERR_STR_(clBuildProgram)
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __COMPILE_PROGRAM_ERR CL_HPP_ERR_STR_(clCompileProgram)
+#define __LINK_PROGRAM_ERR CL_HPP_ERR_STR_(clLinkProgram)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __CREATE_KERNELS_IN_PROGRAM_ERR CL_HPP_ERR_STR_(clCreateKernelsInProgram)
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+#define __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR CL_HPP_ERR_STR_(clCreateCommandQueueWithProperties)
+#define __CREATE_SAMPLER_WITH_PROPERTIES_ERR CL_HPP_ERR_STR_(clCreateSamplerWithProperties)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+#define __SET_COMMAND_QUEUE_PROPERTY_ERR CL_HPP_ERR_STR_(clSetCommandQueueProperty)
+#define __ENQUEUE_READ_BUFFER_ERR CL_HPP_ERR_STR_(clEnqueueReadBuffer)
+#define __ENQUEUE_READ_BUFFER_RECT_ERR CL_HPP_ERR_STR_(clEnqueueReadBufferRect)
+#define __ENQUEUE_WRITE_BUFFER_ERR CL_HPP_ERR_STR_(clEnqueueWriteBuffer)
+#define __ENQUEUE_WRITE_BUFFER_RECT_ERR CL_HPP_ERR_STR_(clEnqueueWriteBufferRect)
+#define __ENQEUE_COPY_BUFFER_ERR CL_HPP_ERR_STR_(clEnqueueCopyBuffer)
+#define __ENQEUE_COPY_BUFFER_RECT_ERR CL_HPP_ERR_STR_(clEnqueueCopyBufferRect)
+#define __ENQUEUE_FILL_BUFFER_ERR CL_HPP_ERR_STR_(clEnqueueFillBuffer)
+#define __ENQUEUE_READ_IMAGE_ERR CL_HPP_ERR_STR_(clEnqueueReadImage)
+#define __ENQUEUE_WRITE_IMAGE_ERR CL_HPP_ERR_STR_(clEnqueueWriteImage)
+#define __ENQUEUE_COPY_IMAGE_ERR CL_HPP_ERR_STR_(clEnqueueCopyImage)
+#define __ENQUEUE_FILL_IMAGE_ERR CL_HPP_ERR_STR_(clEnqueueFillImage)
+#define __ENQUEUE_COPY_IMAGE_TO_BUFFER_ERR CL_HPP_ERR_STR_(clEnqueueCopyImageToBuffer)
+#define __ENQUEUE_COPY_BUFFER_TO_IMAGE_ERR CL_HPP_ERR_STR_(clEnqueueCopyBufferToImage)
+#define __ENQUEUE_MAP_BUFFER_ERR CL_HPP_ERR_STR_(clEnqueueMapBuffer)
+#define __ENQUEUE_MAP_IMAGE_ERR CL_HPP_ERR_STR_(clEnqueueMapImage)
+#define __ENQUEUE_UNMAP_MEM_OBJECT_ERR CL_HPP_ERR_STR_(clEnqueueUnMapMemObject)
+#define __ENQUEUE_NDRANGE_KERNEL_ERR CL_HPP_ERR_STR_(clEnqueueNDRangeKernel)
+#define __ENQUEUE_NATIVE_KERNEL CL_HPP_ERR_STR_(clEnqueueNativeKernel)
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __ENQUEUE_MIGRATE_MEM_OBJECTS_ERR CL_HPP_ERR_STR_(clEnqueueMigrateMemObjects)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+#define __ENQUEUE_ACQUIRE_GL_ERR CL_HPP_ERR_STR_(clEnqueueAcquireGLObjects)
+#define __ENQUEUE_RELEASE_GL_ERR CL_HPP_ERR_STR_(clEnqueueReleaseGLObjects)
+
+#define __CREATE_PIPE_ERR CL_HPP_ERR_STR_(clCreatePipe)
+#define __GET_PIPE_INFO_ERR CL_HPP_ERR_STR_(clGetPipeInfo)
+
+
+#define __RETAIN_ERR CL_HPP_ERR_STR_(Retain Object)
+#define __RELEASE_ERR CL_HPP_ERR_STR_(Release Object)
+#define __FLUSH_ERR CL_HPP_ERR_STR_(clFlush)
+#define __FINISH_ERR CL_HPP_ERR_STR_(clFinish)
+#define __VECTOR_CAPACITY_ERR CL_HPP_ERR_STR_(Vector capacity error)
+
+/**
+ * CL 1.2 version that uses device fission.
+ */
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __CREATE_SUB_DEVICES_ERR CL_HPP_ERR_STR_(clCreateSubDevices)
+#else
+#define __CREATE_SUB_DEVICES_ERR CL_HPP_ERR_STR_(clCreateSubDevicesEXT)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+/**
+ * Deprecated APIs for 1.2
+ */
+#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+#define __ENQUEUE_MARKER_ERR CL_HPP_ERR_STR_(clEnqueueMarker)
+#define __ENQUEUE_WAIT_FOR_EVENTS_ERR CL_HPP_ERR_STR_(clEnqueueWaitForEvents)
+#define __ENQUEUE_BARRIER_ERR CL_HPP_ERR_STR_(clEnqueueBarrier)
+#define __UNLOAD_COMPILER_ERR CL_HPP_ERR_STR_(clUnloadCompiler)
+#define __CREATE_GL_TEXTURE_2D_ERR CL_HPP_ERR_STR_(clCreateFromGLTexture2D)
+#define __CREATE_GL_TEXTURE_3D_ERR CL_HPP_ERR_STR_(clCreateFromGLTexture3D)
+#define __CREATE_IMAGE2D_ERR CL_HPP_ERR_STR_(clCreateImage2D)
+#define __CREATE_IMAGE3D_ERR CL_HPP_ERR_STR_(clCreateImage3D)
+#endif // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+
+/**
+ * Deprecated APIs for 2.0
+ */
+#if defined(CL_USE_DEPRECATED_OPENCL_1_2_APIS)
+#define __CREATE_COMMAND_QUEUE_ERR CL_HPP_ERR_STR_(clCreateCommandQueue)
+#define __ENQUEUE_TASK_ERR CL_HPP_ERR_STR_(clEnqueueTask)
+#define __CREATE_SAMPLER_ERR CL_HPP_ERR_STR_(clCreateSampler)
+#endif // #if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+
+/**
+ * CL 1.2 marker and barrier commands
+ */
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+#define __ENQUEUE_MARKER_WAIT_LIST_ERR CL_HPP_ERR_STR_(clEnqueueMarkerWithWaitList)
+#define __ENQUEUE_BARRIER_WAIT_LIST_ERR CL_HPP_ERR_STR_(clEnqueueBarrierWithWaitList)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+#endif // CL_HPP_USER_OVERRIDE_ERROR_STRINGS
+//! \endcond
+
+
+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<typename Functor, typename T>
+inline cl_int getInfoHelper(Functor f, cl_uint name, T* param, long)
+{
+ return f(name, sizeof(T), param, NULL);
+}
+
+// Specialized for getInfo<CL_PROGRAM_BINARIES>
+// Assumes that the output vector was correctly resized on the way in
+template <typename Func>
+inline cl_int getInfoHelper(Func f, cl_uint name, vector<vector<unsigned char>>* param, int)
+{
+ if (name != CL_PROGRAM_BINARIES) {
+ return CL_INVALID_VALUE;
+ }
+ if (param) {
+ // Create array of pointers, calculate total size and pass pointer array in
+ size_type numBinaries = param->size();
+ vector<unsigned char*> binariesPointers(numBinaries);
+
+ for (size_type i = 0; i < numBinaries; ++i)
+ {
+ binariesPointers[i] = (*param)[i].data();
+ }
+
+ cl_int err = f(name, numBinaries * sizeof(unsigned char*), binariesPointers.data(), NULL);
+
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+ }
+
+
+ return CL_SUCCESS;
+}
+
+// Specialized getInfoHelper for vector params
+template <typename Func, typename T>
+inline cl_int getInfoHelper(Func f, cl_uint name, vector<T>* param, long)
+{
+ size_type required;
+ cl_int err = f(name, 0, NULL, &required);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+ const size_type elements = required / sizeof(T);
+
+ // Temporary to avoid changing param on an error
+ vector<T> localData(elements);
+ err = f(name, required, localData.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+ if (param) {
+ *param = std::move(localData);
+ }
+
+ return CL_SUCCESS;
+}
+
+/* Specialization for reference-counted types. This depends on the
+ * existence of Wrapper<T>::cl_type, and none of the other types having the
+ * cl_type member. Note that simplify specifying the parameter as Wrapper<T>
+ * does not work, because when using a derived type (e.g. Context) the generic
+ * template will provide a better match.
+ */
+template <typename Func, typename T>
+inline cl_int getInfoHelper(
+ Func f, cl_uint name, vector<T>* param, int, typename T::cl_type = 0)
+{
+ size_type required;
+ cl_int err = f(name, 0, NULL, &required);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+
+ const size_type elements = required / sizeof(typename T::cl_type);
+
+ vector<typename T::cl_type> value(elements);
+ err = f(name, required, value.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+
+ if (param) {
+ // Assign to convert CL type to T for each element
+ param->resize(elements);
+
+ // Assign to param, constructing with retain behaviour
+ // to correctly capture each underlying CL object
+ for (size_type i = 0; i < elements; i++) {
+ (*param)[i] = T(value[i], true);
+ }
+ }
+ return CL_SUCCESS;
+}
+
+// Specialized GetInfoHelper for string params
+template <typename Func>
+inline cl_int getInfoHelper(Func f, cl_uint name, string* param, long)
+{
+ size_type required;
+ cl_int err = f(name, 0, NULL, &required);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+
+ // std::string has a constant data member
+ // a char vector does not
+ if (required > 0) {
+ vector<char> value(required);
+ err = f(name, required, value.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+ if (param) {
+ param->assign(begin(value), prev(end(value)));
+ }
+ }
+ else if (param) {
+ param->assign("");
+ }
+ return CL_SUCCESS;
+}
+
+// Specialized GetInfoHelper for clsize_t params
+template <typename Func, size_type N>
+inline cl_int getInfoHelper(Func f, cl_uint name, array<size_type, N>* param, long)
+{
+ size_type required;
+ cl_int err = f(name, 0, NULL, &required);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+
+ size_type elements = required / sizeof(size_type);
+ vector<size_type> value(elements, 0);
+
+ err = f(name, required, value.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return err;
+ }
+
+ // Bound the copy with N to prevent overruns
+ // if passed N > than the amount copied
+ if (elements > N) {
+ elements = N;
+ }
+ for (size_type i = 0; i < elements; ++i) {
+ (*param)[i] = value[i];
+ }
+
+ return CL_SUCCESS;
+}
+
+template<typename T> struct ReferenceHandler;
+
+/* Specialization for reference-counted types. This depends on the
+ * existence of Wrapper<T>::cl_type, and none of the other types having the
+ * cl_type member. Note that simplify specifying the parameter as Wrapper<T>
+ * does not work, because when using a derived type (e.g. Context) the generic
+ * template will provide a better match.
+ */
+template<typename Func, typename T>
+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 CL_HPP_PARAM_NAME_INFO_1_0_(F) \
+ F(cl_platform_info, CL_PLATFORM_PROFILE, string) \
+ F(cl_platform_info, CL_PLATFORM_VERSION, string) \
+ F(cl_platform_info, CL_PLATFORM_NAME, string) \
+ F(cl_platform_info, CL_PLATFORM_VENDOR, string) \
+ F(cl_platform_info, CL_PLATFORM_EXTENSIONS, string) \
+ \
+ 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_type) \
+ F(cl_device_info, CL_DEVICE_MAX_WORK_ITEM_SIZES, cl::vector<size_type>) \
+ 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_type) \
+ F(cl_device_info, CL_DEVICE_IMAGE2D_MAX_HEIGHT, size_type) \
+ F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_WIDTH, size_type) \
+ F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_HEIGHT, size_type) \
+ F(cl_device_info, CL_DEVICE_IMAGE3D_MAX_DEPTH, size_type) \
+ F(cl_device_info, CL_DEVICE_IMAGE_SUPPORT, cl_bool) \
+ F(cl_device_info, CL_DEVICE_MAX_PARAMETER_SIZE, size_type) \
+ 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_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_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_type) \
+ 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_PLATFORM, cl_platform_id) \
+ F(cl_device_info, CL_DEVICE_NAME, string) \
+ F(cl_device_info, CL_DEVICE_VENDOR, string) \
+ F(cl_device_info, CL_DRIVER_VERSION, string) \
+ F(cl_device_info, CL_DEVICE_PROFILE, string) \
+ F(cl_device_info, CL_DEVICE_VERSION, string) \
+ F(cl_device_info, CL_DEVICE_EXTENSIONS, string) \
+ \
+ F(cl_context_info, CL_CONTEXT_REFERENCE_COUNT, cl_uint) \
+ F(cl_context_info, CL_CONTEXT_DEVICES, cl::vector<Device>) \
+ F(cl_context_info, CL_CONTEXT_PROPERTIES, cl::vector<cl_context_properties>) \
+ \
+ 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_int) \
+ \
+ 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_type) \
+ 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_type) \
+ F(cl_image_info, CL_IMAGE_ROW_PITCH, size_type) \
+ F(cl_image_info, CL_IMAGE_SLICE_PITCH, size_type) \
+ F(cl_image_info, CL_IMAGE_WIDTH, size_type) \
+ F(cl_image_info, CL_IMAGE_HEIGHT, size_type) \
+ F(cl_image_info, CL_IMAGE_DEPTH, size_type) \
+ \
+ 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_bool) \
+ F(cl_sampler_info, CL_SAMPLER_ADDRESSING_MODE, cl_addressing_mode) \
+ F(cl_sampler_info, CL_SAMPLER_FILTER_MODE, cl_filter_mode) \
+ \
+ 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, cl::vector<Device>) \
+ F(cl_program_info, CL_PROGRAM_SOURCE, string) \
+ F(cl_program_info, CL_PROGRAM_BINARY_SIZES, cl::vector<size_type>) \
+ F(cl_program_info, CL_PROGRAM_BINARIES, cl::vector<cl::vector<unsigned char>>) \
+ \
+ F(cl_program_build_info, CL_PROGRAM_BUILD_STATUS, cl_build_status) \
+ F(cl_program_build_info, CL_PROGRAM_BUILD_OPTIONS, string) \
+ F(cl_program_build_info, CL_PROGRAM_BUILD_LOG, string) \
+ \
+ F(cl_kernel_info, CL_KERNEL_FUNCTION_NAME, string) \
+ 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_type) \
+ F(cl_kernel_work_group_info, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, cl::detail::size_t_array) \
+ 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)
+
+
+#define CL_HPP_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_OPENCL_C_VERSION, string) \
+ \
+ F(cl_mem_info, CL_MEM_ASSOCIATED_MEMOBJECT, cl::Memory) \
+ F(cl_mem_info, CL_MEM_OFFSET, size_type) \
+ \
+ F(cl_kernel_work_group_info, CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE, size_type) \
+ F(cl_kernel_work_group_info, CL_KERNEL_PRIVATE_MEM_SIZE, cl_ulong) \
+ \
+ F(cl_event_info, CL_EVENT_CONTEXT, cl::Context)
+
+#define CL_HPP_PARAM_NAME_INFO_1_2_(F) \
+ F(cl_program_info, CL_PROGRAM_NUM_KERNELS, size_type) \
+ F(cl_program_info, CL_PROGRAM_KERNEL_NAMES, string) \
+ \
+ F(cl_program_build_info, CL_PROGRAM_BINARY_TYPE, cl_program_binary_type) \
+ \
+ F(cl_kernel_info, CL_KERNEL_ATTRIBUTES, string) \
+ \
+ 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) \
+ F(cl_kernel_arg_info, CL_KERNEL_ARG_NAME, string) \
+ F(cl_kernel_arg_info, CL_KERNEL_ARG_TYPE_QUALIFIER, cl_kernel_arg_type_qualifier) \
+ \
+ F(cl_device_info, CL_DEVICE_PARENT_DEVICE, cl::Device) \
+ F(cl_device_info, CL_DEVICE_PARTITION_PROPERTIES, cl::vector<cl_device_partition_property>) \
+ F(cl_device_info, CL_DEVICE_PARTITION_TYPE, cl::vector<cl_device_partition_property>) \
+ F(cl_device_info, CL_DEVICE_REFERENCE_COUNT, cl_uint) \
+ F(cl_device_info, CL_DEVICE_PREFERRED_INTEROP_USER_SYNC, size_type) \
+ F(cl_device_info, CL_DEVICE_PARTITION_AFFINITY_DOMAIN, cl_device_affinity_domain) \
+ F(cl_device_info, CL_DEVICE_BUILT_IN_KERNELS, string) \
+ \
+ F(cl_image_info, CL_IMAGE_ARRAY_SIZE, size_type) \
+ F(cl_image_info, CL_IMAGE_NUM_MIP_LEVELS, cl_uint) \
+ F(cl_image_info, CL_IMAGE_NUM_SAMPLES, cl_uint)
+
+#define CL_HPP_PARAM_NAME_INFO_2_0_(F) \
+ F(cl_device_info, CL_DEVICE_QUEUE_ON_HOST_PROPERTIES, cl_command_queue_properties) \
+ F(cl_device_info, CL_DEVICE_QUEUE_ON_DEVICE_PROPERTIES, cl_command_queue_properties) \
+ F(cl_device_info, CL_DEVICE_QUEUE_ON_DEVICE_PREFERRED_SIZE, cl_uint) \
+ F(cl_device_info, CL_DEVICE_QUEUE_ON_DEVICE_MAX_SIZE, cl_uint) \
+ F(cl_device_info, CL_DEVICE_MAX_ON_DEVICE_QUEUES, cl_uint) \
+ F(cl_device_info, CL_DEVICE_MAX_ON_DEVICE_EVENTS, cl_uint) \
+ F(cl_device_info, CL_DEVICE_MAX_PIPE_ARGS, cl_uint) \
+ F(cl_device_info, CL_DEVICE_PIPE_MAX_ACTIVE_RESERVATIONS, cl_uint) \
+ F(cl_device_info, CL_DEVICE_PIPE_MAX_PACKET_SIZE, cl_uint) \
+ F(cl_device_info, CL_DEVICE_SVM_CAPABILITIES, cl_device_svm_capabilities) \
+ F(cl_device_info, CL_DEVICE_PREFERRED_PLATFORM_ATOMIC_ALIGNMENT, cl_uint) \
+ F(cl_device_info, CL_DEVICE_PREFERRED_GLOBAL_ATOMIC_ALIGNMENT, cl_uint) \
+ F(cl_device_info, CL_DEVICE_PREFERRED_LOCAL_ATOMIC_ALIGNMENT, cl_uint) \
+ F(cl_command_queue_info, CL_QUEUE_SIZE, cl_uint) \
+ F(cl_mem_info, CL_MEM_USES_SVM_POINTER, cl_bool) \
+ F(cl_program_build_info, CL_PROGRAM_BUILD_GLOBAL_VARIABLE_TOTAL_SIZE, size_type) \
+ F(cl_pipe_info, CL_PIPE_PACKET_SIZE, cl_uint) \
+ F(cl_pipe_info, CL_PIPE_MAX_PACKETS, cl_uint)
+
+#define CL_HPP_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, cl::vector<cl_device_partition_property_ext>) \
+ F(cl_device_info, CL_DEVICE_AFFINITY_DOMAINS_EXT, cl::vector<cl_device_partition_property_ext>) \
+ F(cl_device_info, CL_DEVICE_REFERENCE_COUNT_EXT , cl_uint) \
+ F(cl_device_info, CL_DEVICE_PARTITION_STYLE_EXT, cl::vector<cl_device_partition_property_ext>)
+
+template <typename enum_type, cl_int Name>
+struct param_traits {};
+
+#define CL_HPP_DECLARE_PARAM_TRAITS_(token, param_name, T) \
+struct token; \
+template<> \
+struct param_traits<detail:: token,param_name> \
+{ \
+ enum { value = param_name }; \
+ typedef T param_type; \
+};
+
+CL_HPP_PARAM_NAME_INFO_1_0_(CL_HPP_DECLARE_PARAM_TRAITS_)
+#if CL_HPP_TARGET_OPENCL_VERSION >= 110
+CL_HPP_PARAM_NAME_INFO_1_1_(CL_HPP_DECLARE_PARAM_TRAITS_)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 110
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+CL_HPP_PARAM_NAME_INFO_1_2_(CL_HPP_DECLARE_PARAM_TRAITS_)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 110
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+CL_HPP_PARAM_NAME_INFO_2_0_(CL_HPP_DECLARE_PARAM_TRAITS_)
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 110
+
+
+// Flags deprecated in OpenCL 2.0
+#define CL_HPP_PARAM_NAME_INFO_1_0_DEPRECATED_IN_2_0_(F) \
+ F(cl_device_info, CL_DEVICE_QUEUE_PROPERTIES, cl_command_queue_properties)
+
+#define CL_HPP_PARAM_NAME_INFO_1_1_DEPRECATED_IN_2_0_(F) \
+ F(cl_device_info, CL_DEVICE_HOST_UNIFIED_MEMORY, cl_bool)
+
+#define CL_HPP_PARAM_NAME_INFO_1_2_DEPRECATED_IN_2_0_(F) \
+ F(cl_image_info, CL_IMAGE_BUFFER, cl::Buffer)
+
+// Include deprecated query flags based on versions
+// Only include deprecated 1.0 flags if 2.0 not active as there is an enum clash
+#if CL_HPP_TARGET_OPENCL_VERSION > 100 && CL_HPP_MINIMUM_OPENCL_VERSION < 200 && CL_HPP_TARGET_OPENCL_VERSION < 200
+CL_HPP_PARAM_NAME_INFO_1_0_DEPRECATED_IN_2_0_(CL_HPP_DECLARE_PARAM_TRAITS_)
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 110
+#if CL_HPP_TARGET_OPENCL_VERSION > 110 && CL_HPP_MINIMUM_OPENCL_VERSION < 200
+CL_HPP_PARAM_NAME_INFO_1_1_DEPRECATED_IN_2_0_(CL_HPP_DECLARE_PARAM_TRAITS_)
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 120
+#if CL_HPP_TARGET_OPENCL_VERSION > 120 && CL_HPP_MINIMUM_OPENCL_VERSION < 200
+CL_HPP_PARAM_NAME_INFO_1_2_DEPRECATED_IN_2_0_(CL_HPP_DECLARE_PARAM_TRAITS_)
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 200
+
+#if defined(CL_HPP_USE_CL_DEVICE_FISSION)
+CL_HPP_PARAM_NAME_DEVICE_FISSION_(CL_HPP_DECLARE_PARAM_TRAITS_);
+#endif // CL_HPP_USE_CL_DEVICE_FISSION
+
+#ifdef CL_PLATFORM_ICD_SUFFIX_KHR
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_platform_info, CL_PLATFORM_ICD_SUFFIX_KHR, string)
+#endif
+
+#ifdef CL_DEVICE_PROFILING_TIMER_OFFSET_AMD
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_PROFILING_TIMER_OFFSET_AMD, cl_ulong)
+#endif
+
+#ifdef CL_DEVICE_GLOBAL_FREE_MEMORY_AMD
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_GLOBAL_FREE_MEMORY_AMD, vector<size_type>)
+#endif
+#ifdef CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_SIMD_PER_COMPUTE_UNIT_AMD, cl_uint)
+#endif
+#ifdef CL_DEVICE_SIMD_WIDTH_AMD
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_SIMD_WIDTH_AMD, cl_uint)
+#endif
+#ifdef CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_SIMD_INSTRUCTION_WIDTH_AMD, cl_uint)
+#endif
+#ifdef CL_DEVICE_WAVEFRONT_WIDTH_AMD
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_WAVEFRONT_WIDTH_AMD, cl_uint)
+#endif
+#ifdef CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_GLOBAL_MEM_CHANNELS_AMD, cl_uint)
+#endif
+#ifdef CL_DEVICE_GLOBAL_MEM_CHANNEL_BANKS_AMD
+CL_HPP_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_HPP_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_HPP_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_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_LOCAL_MEM_BANKS_AMD, cl_uint)
+#endif
+
+#ifdef CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MAJOR_NV, cl_uint)
+#endif
+#ifdef CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_COMPUTE_CAPABILITY_MINOR_NV, cl_uint)
+#endif
+#ifdef CL_DEVICE_REGISTERS_PER_BLOCK_NV
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_REGISTERS_PER_BLOCK_NV, cl_uint)
+#endif
+#ifdef CL_DEVICE_WARP_SIZE_NV
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_WARP_SIZE_NV, cl_uint)
+#endif
+#ifdef CL_DEVICE_GPU_OVERLAP_NV
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_GPU_OVERLAP_NV, cl_bool)
+#endif
+#ifdef CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV, cl_bool)
+#endif
+#ifdef CL_DEVICE_INTEGRATED_MEMORY_NV
+CL_HPP_DECLARE_PARAM_TRAITS_(cl_device_info, CL_DEVICE_INTEGRATED_MEMORY_NV, cl_bool)
+#endif
+
+// Convenience functions
+
+template <typename Func, typename T>
+inline cl_int
+getInfo(Func f, cl_uint name, T* param)
+{
+ return getInfoHelper(f, name, param, 0);
+}
+
+template <typename Func, typename Arg0>
+struct GetInfoFunctor0
+{
+ Func f_; const Arg0& arg0_;
+ cl_int operator ()(
+ cl_uint param, size_type size, void* value, size_type* size_ret)
+ { return f_(arg0_, param, size, value, size_ret); }
+};
+
+template <typename Func, typename Arg0, typename Arg1>
+struct GetInfoFunctor1
+{
+ Func f_; const Arg0& arg0_; const Arg1& arg1_;
+ cl_int operator ()(
+ cl_uint param, size_type size, void* value, size_type* size_ret)
+ { return f_(arg0_, arg1_, param, size, value, size_ret); }
+};
+
+template <typename Func, typename Arg0, typename T>
+inline cl_int
+getInfo(Func f, const Arg0& arg0, cl_uint name, T* param)
+{
+ GetInfoFunctor0<Func, Arg0> f0 = { f, arg0 };
+ return getInfoHelper(f0, name, param, 0);
+}
+
+template <typename Func, typename Arg0, typename Arg1, typename T>
+inline cl_int
+getInfo(Func f, const Arg0& arg0, const Arg1& arg1, cl_uint name, T* param)
+{
+ GetInfoFunctor1<Func, Arg0, Arg1> f0 = { f, arg0, arg1 };
+ return getInfoHelper(f0, name, param, 0);
+}
+
+
+template<typename T>
+struct ReferenceHandler
+{ };
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+/**
+ * OpenCL 1.2 devices do have retain/release.
+ */
+template <>
+struct ReferenceHandler<cl_device_id>
+{
+ /**
+ * 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 // CL_HPP_TARGET_OPENCL_VERSION >= 120
+/**
+ * OpenCL 1.1 devices do not have retain/release.
+ */
+template <>
+struct ReferenceHandler<cl_device_id>
+{
+ // 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 // ! (CL_HPP_TARGET_OPENCL_VERSION >= 120)
+
+template <>
+struct ReferenceHandler<cl_platform_id>
+{
+ // 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<cl_context>
+{
+ static cl_int retain(cl_context context)
+ { return ::clRetainContext(context); }
+ static cl_int release(cl_context context)
+ { return ::clReleaseContext(context); }
+};
+
+template <>
+struct ReferenceHandler<cl_command_queue>
+{
+ 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<cl_mem>
+{
+ static cl_int retain(cl_mem memory)
+ { return ::clRetainMemObject(memory); }
+ static cl_int release(cl_mem memory)
+ { return ::clReleaseMemObject(memory); }
+};
+
+template <>
+struct ReferenceHandler<cl_sampler>
+{
+ static cl_int retain(cl_sampler sampler)
+ { return ::clRetainSampler(sampler); }
+ static cl_int release(cl_sampler sampler)
+ { return ::clReleaseSampler(sampler); }
+};
+
+template <>
+struct ReferenceHandler<cl_program>
+{
+ static cl_int retain(cl_program program)
+ { return ::clRetainProgram(program); }
+ static cl_int release(cl_program program)
+ { return ::clReleaseProgram(program); }
+};
+
+template <>
+struct ReferenceHandler<cl_kernel>
+{
+ static cl_int retain(cl_kernel kernel)
+ { return ::clRetainKernel(kernel); }
+ static cl_int release(cl_kernel kernel)
+ { return ::clReleaseKernel(kernel); }
+};
+
+template <>
+struct ReferenceHandler<cl_event>
+{
+ static cl_int retain(cl_event event)
+ { return ::clRetainEvent(event); }
+ static cl_int release(cl_event event)
+ { return ::clReleaseEvent(event); }
+};
+
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120 && CL_HPP_MINIMUM_OPENCL_VERSION < 120
+// Extracts version number with major in the upper 16 bits, minor in the lower 16
+static cl_uint getVersion(const vector<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] != ' ' && versionInfo[index] != '\0') {
+ lowVersion *= 10;
+ lowVersion += versionInfo[index]-'0';
+ ++index;
+ }
+ return (highVersion << 16) | lowVersion;
+}
+
+static cl_uint getPlatformVersion(cl_platform_id platform)
+{
+ size_type size = 0;
+ clGetPlatformInfo(platform, CL_PLATFORM_VERSION, 0, NULL, &size);
+
+ vector<char> versionInfo(size);
+ clGetPlatformInfo(platform, CL_PLATFORM_VERSION, size, versionInfo.data(), &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);
+}
+
+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_type size = 0;
+ clGetContextInfo(context, CL_CONTEXT_DEVICES, 0, NULL, &size);
+ if (size == 0)
+ return 0;
+ vector<cl_device_id> devices(size/sizeof(cl_device_id));
+ clGetContextInfo(context, CL_CONTEXT_DEVICES, size, devices.data(), NULL);
+ return getDevicePlatformVersion(devices[0]);
+}
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120 && CL_HPP_MINIMUM_OPENCL_VERSION < 120
+
+template <typename T>
+class Wrapper
+{
+public:
+ typedef T cl_type;
+
+protected:
+ cl_type object_;
+
+public:
+ Wrapper() : object_(NULL) { }
+
+ Wrapper(const cl_type &obj, bool retainObject) : object_(obj)
+ {
+ if (retainObject) {
+ detail::errHandler(retain(), __RETAIN_ERR);
+ }
+ }
+
+ ~Wrapper()
+ {
+ if (object_ != NULL) { release(); }
+ }
+
+ Wrapper(const Wrapper<cl_type>& rhs)
+ {
+ object_ = rhs.object_;
+ detail::errHandler(retain(), __RETAIN_ERR);
+ }
+
+ Wrapper(Wrapper<cl_type>&& rhs) CL_HPP_NOEXCEPT_
+ {
+ object_ = rhs.object_;
+ rhs.object_ = NULL;
+ }
+
+ Wrapper<cl_type>& operator = (const Wrapper<cl_type>& rhs)
+ {
+ if (this != &rhs) {
+ detail::errHandler(release(), __RELEASE_ERR);
+ object_ = rhs.object_;
+ detail::errHandler(retain(), __RETAIN_ERR);
+ }
+ return *this;
+ }
+
+ Wrapper<cl_type>& operator = (Wrapper<cl_type>&& rhs)
+ {
+ if (this != &rhs) {
+ detail::errHandler(release(), __RELEASE_ERR);
+ object_ = rhs.object_;
+ rhs.object_ = NULL;
+ }
+ return *this;
+ }
+
+ Wrapper<cl_type>& operator = (const cl_type &rhs)
+ {
+ detail::errHandler(release(), __RELEASE_ERR);
+ object_ = rhs;
+ return *this;
+ }
+
+ const cl_type& operator ()() const { return object_; }
+
+ cl_type& operator ()() { return object_; }
+
+ const cl_type get() const { return object_; }
+
+ cl_type get() { return object_; }
+
+
+protected:
+ template<typename Func, typename U>
+ friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type);
+
+ cl_int retain() const
+ {
+ if (object_ != nullptr) {
+ return ReferenceHandler<cl_type>::retain(object_);
+ }
+ else {
+ return CL_SUCCESS;
+ }
+ }
+
+ cl_int release() const
+ {
+ if (object_ != nullptr) {
+ return ReferenceHandler<cl_type>::release(object_);
+ }
+ else {
+ return CL_SUCCESS;
+ }
+ }
+};
+
+template <>
+class Wrapper<cl_device_id>
+{
+public:
+ typedef cl_device_id cl_type;
+
+protected:
+ cl_type object_;
+ bool referenceCountable_;
+
+ static bool isReferenceCountable(cl_device_id device)
+ {
+ bool retVal = false;
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 120
+ if (device != NULL) {
+ int version = getDevicePlatformVersion(device);
+ if(version > ((1 << 16) + 1)) {
+ retVal = true;
+ }
+ }
+#else // CL_HPP_MINIMUM_OPENCL_VERSION < 120
+ retVal = true;
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 120
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+ return retVal;
+ }
+
+public:
+ Wrapper() : object_(NULL), referenceCountable_(false)
+ {
+ }
+
+ Wrapper(const cl_type &obj, bool retainObject) :
+ object_(obj),
+ referenceCountable_(false)
+ {
+ referenceCountable_ = isReferenceCountable(obj);
+
+ if (retainObject) {
+ detail::errHandler(retain(), __RETAIN_ERR);
+ }
+ }
+
+ ~Wrapper()
+ {
+ release();
+ }
+
+ Wrapper(const Wrapper<cl_type>& rhs)
+ {
+ object_ = rhs.object_;
+ referenceCountable_ = isReferenceCountable(object_);
+ detail::errHandler(retain(), __RETAIN_ERR);
+ }
+
+ Wrapper(Wrapper<cl_type>&& rhs) CL_HPP_NOEXCEPT_
+ {
+ object_ = rhs.object_;
+ referenceCountable_ = rhs.referenceCountable_;
+ rhs.object_ = NULL;
+ rhs.referenceCountable_ = false;
+ }
+
+ Wrapper<cl_type>& operator = (const Wrapper<cl_type>& rhs)
+ {
+ if (this != &rhs) {
+ detail::errHandler(release(), __RELEASE_ERR);
+ object_ = rhs.object_;
+ referenceCountable_ = rhs.referenceCountable_;
+ detail::errHandler(retain(), __RETAIN_ERR);
+ }
+ return *this;
+ }
+
+ Wrapper<cl_type>& operator = (Wrapper<cl_type>&& rhs)
+ {
+ if (this != &rhs) {
+ detail::errHandler(release(), __RELEASE_ERR);
+ object_ = rhs.object_;
+ referenceCountable_ = rhs.referenceCountable_;
+ rhs.object_ = NULL;
+ rhs.referenceCountable_ = false;
+ }
+ return *this;
+ }
+
+ Wrapper<cl_type>& operator = (const cl_type &rhs)
+ {
+ detail::errHandler(release(), __RELEASE_ERR);
+ object_ = rhs;
+ referenceCountable_ = isReferenceCountable(object_);
+ return *this;
+ }
+
+ const cl_type& operator ()() const { return object_; }
+
+ cl_type& operator ()() { return object_; }
+
+ cl_type get() const { return object_; }
+
+protected:
+ template<typename Func, typename U>
+ friend inline cl_int getInfoHelper(Func, cl_uint, U*, int, typename U::cl_type);
+
+ template<typename Func, typename U>
+ friend inline cl_int getInfoHelper(Func, cl_uint, vector<U>*, int, typename U::cl_type);
+
+ cl_int retain() const
+ {
+ if( object_ != nullptr && referenceCountable_ ) {
+ return ReferenceHandler<cl_type>::retain(object_);
+ }
+ else {
+ return CL_SUCCESS;
+ }
+ }
+
+ cl_int release() const
+ {
+ if (object_ != nullptr && referenceCountable_) {
+ return ReferenceHandler<cl_type>::release(object_);
+ }
+ else {
+ return CL_SUCCESS;
+ }
+ }
+};
+
+template <typename T>
+inline bool operator==(const Wrapper<T> &lhs, const Wrapper<T> &rhs)
+{
+ return lhs() == rhs();
+}
+
+template <typename T>
+inline bool operator!=(const Wrapper<T> &lhs, const Wrapper<T> &rhs)
+{
+ return !operator==(lhs, rhs);
+}
+
+} // namespace detail
+//! \endcond
+
+
+using BuildLogType = vector<std::pair<cl::Device, typename detail::param_traits<detail::cl_program_build_info, CL_PROGRAM_BUILD_LOG>::param_type>>;
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+/**
+* Exception class for build errors to carry build info
+*/
+class BuildError : public Error
+{
+private:
+ BuildLogType buildLogs;
+public:
+ BuildError(cl_int err, const char * errStr, const BuildLogType &vec) : Error(err, errStr), buildLogs(vec)
+ {
+ }
+
+ BuildLogType getBuildLog() const
+ {
+ return buildLogs;
+ }
+};
+namespace detail {
+ static inline cl_int buildErrHandler(
+ cl_int err,
+ const char * errStr,
+ const BuildLogType &buildLogs)
+ {
+ if (err != CL_SUCCESS) {
+ throw BuildError(err, errStr, buildLogs);
+ }
+ return err;
+ }
+} // namespace detail
+
+#else
+namespace detail {
+ static inline cl_int buildErrHandler(
+ cl_int err,
+ const char * errStr,
+ const BuildLogType &buildLogs)
+ {
+ (void)buildLogs; // suppress unused variable warning
+ (void)errStr;
+ return err;
+ }
+} // namespace detail
+#endif // #if defined(CL_HPP_ENABLE_EXCEPTIONS)
+
+
+/*! \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<cl_device_id>
+{
+private:
+ static std::once_flag default_initialized_;
+ static Device default_;
+ static cl_int default_error_;
+
+ /*! \brief Create the default context.
+ *
+ * This sets @c default_ and @c default_error_. It does not throw
+ * @c cl::Error.
+ */
+ static void makeDefault();
+
+ /*! \brief Create the default platform from a provided platform.
+ *
+ * This sets @c default_. It does not throw
+ * @c cl::Error.
+ */
+ static void makeDefaultProvided(const Device &p) {
+ default_ = p;
+ }
+
+public:
+#ifdef CL_HPP_UNIT_TEST_ENABLE
+ /*! \brief Reset the default.
+ *
+ * This sets @c default_ to an empty value to support cleanup in
+ * the unit test framework.
+ * This function is not thread safe.
+ */
+ static void unitTestClearDefault() {
+ default_ = Device();
+ }
+#endif // #ifdef CL_HPP_UNIT_TEST_ENABLE
+
+ //! \brief Default constructor - initializes to NULL.
+ Device() : detail::Wrapper<cl_type>() { }
+
+ /*! \brief Constructor from cl_device_id.
+ *
+ * This simply copies the device ID value, which is an inexpensive operation.
+ */
+ explicit Device(const cl_device_id &device, bool retainObject = false) :
+ detail::Wrapper<cl_type>(device, retainObject) { }
+
+ /*! \brief Returns the first device on the default context.
+ *
+ * \see Context::getDefault()
+ */
+ static Device getDefault(
+ cl_int *errResult = NULL)
+ {
+ std::call_once(default_initialized_, makeDefault);
+ detail::errHandler(default_error_);
+ if (errResult != NULL) {
+ *errResult = default_error_;
+ }
+ return default_;
+ }
+
+ /**
+ * Modify the default device to be used by
+ * subsequent operations.
+ * Will only set the default if no default was previously created.
+ * @return updated default device.
+ * Should be compared to the passed value to ensure that it was updated.
+ */
+ static Device setDefault(const Device &default_device)
+ {
+ std::call_once(default_initialized_, makeDefaultProvided, std::cref(default_device));
+ detail::errHandler(default_error_);
+ return default_;
+ }
+
+ /*! \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<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Device(const Device& dev) : detail::Wrapper<cl_type>(dev) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Device& operator = (const Device &dev)
+ {
+ detail::Wrapper<cl_type>::operator=(dev);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Device(Device&& dev) CL_HPP_NOEXCEPT_ : detail::Wrapper<cl_type>(std::move(dev)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Device& operator = (Device &&dev)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(dev));
+ return *this;
+ }
+
+ //! \brief Wrapper for clGetDeviceInfo().
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_device_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+ /**
+ * CL 1.2 version
+ */
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ //! \brief Wrapper for clCreateSubDevices().
+ cl_int createSubDevices(
+ const cl_device_partition_property * properties,
+ vector<Device>* 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_ERR);
+ }
+
+ vector<cl_device_id> ids(n);
+ err = clCreateSubDevices(object_, properties, n, ids.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return detail::errHandler(err, __CREATE_SUB_DEVICES_ERR);
+ }
+
+ // Cannot trivially assign because we need to capture intermediates
+ // with safe construction
+ if (devices) {
+ devices->resize(ids.size());
+
+ // Assign to param, constructing with retain behaviour
+ // to correctly capture each underlying CL object
+ for (size_type i = 0; i < ids.size(); i++) {
+ // We do not need to retain because this device is being created
+ // by the runtime
+ (*devices)[i] = Device(ids[i], false);
+ }
+ }
+
+ return CL_SUCCESS;
+ }
+#elif defined(CL_HPP_USE_CL_DEVICE_FISSION)
+
+/**
+ * CL 1.1 version that uses device fission extension.
+ */
+ cl_int createSubDevices(
+ const cl_device_partition_property_ext * properties,
+ vector<Device>* 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;
+ CL_HPP_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_ERR);
+ }
+
+ vector<cl_device_id> ids(n);
+ err = pfn_clCreateSubDevicesEXT(object_, properties, n, ids.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return detail::errHandler(err, __CREATE_SUB_DEVICES_ERR);
+ }
+ // Cannot trivially assign because we need to capture intermediates
+ // with safe construction
+ if (devices) {
+ devices->resize(ids.size());
+
+ // Assign to param, constructing with retain behaviour
+ // to correctly capture each underlying CL object
+ for (size_type i = 0; i < ids.size(); i++) {
+ // We do not need to retain because this device is being created
+ // by the runtime
+ (*devices)[i] = Device(ids[i], false);
+ }
+ }
+ return CL_SUCCESS;
+ }
+#endif // defined(CL_HPP_USE_CL_DEVICE_FISSION)
+};
+
+CL_HPP_DEFINE_STATIC_MEMBER_ std::once_flag Device::default_initialized_;
+CL_HPP_DEFINE_STATIC_MEMBER_ Device Device::default_;
+CL_HPP_DEFINE_STATIC_MEMBER_ cl_int Device::default_error_ = CL_SUCCESS;
+
+/*! \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<cl_platform_id>
+{
+private:
+ static std::once_flag default_initialized_;
+ static Platform default_;
+ static cl_int default_error_;
+
+ /*! \brief Create the default context.
+ *
+ * This sets @c default_ and @c default_error_. It does not throw
+ * @c cl::Error.
+ */
+ static void makeDefault() {
+ /* Throwing an exception from a call_once invocation does not do
+ * what we wish, so we catch it and save the error.
+ */
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ try
+#endif
+ {
+ // If default wasn't passed ,generate one
+ // Otherwise set it
+ cl_uint n = 0;
+
+ cl_int err = ::clGetPlatformIDs(0, NULL, &n);
+ if (err != CL_SUCCESS) {
+ default_error_ = err;
+ return;
+ }
+ if (n == 0) {
+ default_error_ = CL_INVALID_PLATFORM;
+ return;
+ }
+
+ vector<cl_platform_id> ids(n);
+ err = ::clGetPlatformIDs(n, ids.data(), NULL);
+ if (err != CL_SUCCESS) {
+ default_error_ = err;
+ return;
+ }
+
+ default_ = Platform(ids[0]);
+ }
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ catch (cl::Error &e) {
+ default_error_ = e.err();
+ }
+#endif
+ }
+
+ /*! \brief Create the default platform from a provided platform.
+ *
+ * This sets @c default_. It does not throw
+ * @c cl::Error.
+ */
+ static void makeDefaultProvided(const Platform &p) {
+ default_ = p;
+ }
+
+public:
+#ifdef CL_HPP_UNIT_TEST_ENABLE
+ /*! \brief Reset the default.
+ *
+ * This sets @c default_ to an empty value to support cleanup in
+ * the unit test framework.
+ * This function is not thread safe.
+ */
+ static void unitTestClearDefault() {
+ default_ = Platform();
+ }
+#endif // #ifdef CL_HPP_UNIT_TEST_ENABLE
+
+ //! \brief Default constructor - initializes to NULL.
+ Platform() : detail::Wrapper<cl_type>() { }
+
+ /*! \brief Constructor from cl_platform_id.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * This simply copies the platform ID value, which is an inexpensive operation.
+ */
+ explicit Platform(const cl_platform_id &platform, bool retainObject = false) :
+ detail::Wrapper<cl_type>(platform, retainObject) { }
+
+ /*! \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<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ static Platform getDefault(
+ cl_int *errResult = NULL)
+ {
+ std::call_once(default_initialized_, makeDefault);
+ detail::errHandler(default_error_);
+ if (errResult != NULL) {
+ *errResult = default_error_;
+ }
+ return default_;
+ }
+
+ /**
+ * Modify the default platform to be used by
+ * subsequent operations.
+ * Will only set the default if no default was previously created.
+ * @return updated default platform.
+ * Should be compared to the passed value to ensure that it was updated.
+ */
+ static Platform setDefault(const Platform &default_platform)
+ {
+ std::call_once(default_initialized_, makeDefaultProvided, std::cref(default_platform));
+ detail::errHandler(default_error_);
+ return default_;
+ }
+
+ //! \brief Wrapper for clGetPlatformInfo().
+ cl_int getInfo(cl_platform_info name, string* param) const
+ {
+ return detail::errHandler(
+ detail::getInfo(&::clGetPlatformInfo, object_, name, param),
+ __GET_PLATFORM_INFO_ERR);
+ }
+
+ //! \brief Wrapper for clGetPlatformInfo() that returns by value.
+ template <cl_int name> typename
+ detail::param_traits<detail::cl_platform_info, name>::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, &param);
+ 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<Device>* 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);
+ }
+
+ vector<cl_device_id> ids(n);
+ err = ::clGetDeviceIDs(object_, type, n, ids.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return detail::errHandler(err, __GET_DEVICE_IDS_ERR);
+ }
+
+ // Cannot trivially assign because we need to capture intermediates
+ // with safe construction
+ // We must retain things we obtain from the API to avoid releasing
+ // API-owned objects.
+ if (devices) {
+ devices->resize(ids.size());
+
+ // Assign to param, constructing with retain behaviour
+ // to correctly capture each underlying CL object
+ for (size_type i = 0; i < ids.size(); i++) {
+ (*devices)[i] = Device(ids[i], true);
+ }
+ }
+ return CL_SUCCESS;
+ }
+
+#if defined(CL_HPP_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<Device>* 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;
+ CL_HPP_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);
+ }
+
+ vector<cl_device_id> ids(n);
+ err = pfn_clGetDeviceIDsFromD3D10KHR(
+ object_,
+ d3d_device_source,
+ d3d_object,
+ d3d_device_set,
+ n,
+ ids.data(),
+ NULL);
+ if (err != CL_SUCCESS) {
+ return detail::errHandler(err, __GET_DEVICE_IDS_ERR);
+ }
+
+ // Cannot trivially assign because we need to capture intermediates
+ // with safe construction
+ // We must retain things we obtain from the API to avoid releasing
+ // API-owned objects.
+ if (devices) {
+ devices->resize(ids.size());
+
+ // Assign to param, constructing with retain behaviour
+ // to correctly capture each underlying CL object
+ for (size_type i = 0; i < ids.size(); i++) {
+ (*devices)[i] = Device(ids[i], true);
+ }
+ }
+ return CL_SUCCESS;
+ }
+#endif
+
+ /*! \brief Gets a list of available platforms.
+ *
+ * Wraps clGetPlatformIDs().
+ */
+ static cl_int get(
+ vector<Platform>* 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);
+ }
+
+ vector<cl_platform_id> ids(n);
+ err = ::clGetPlatformIDs(n, ids.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return detail::errHandler(err, __GET_PLATFORM_IDS_ERR);
+ }
+
+ if (platforms) {
+ platforms->resize(ids.size());
+
+ // Platforms don't reference count
+ for (size_type i = 0; i < ids.size(); i++) {
+ (*platforms)[i] = Platform(ids[i]);
+ }
+ }
+ return CL_SUCCESS;
+ }
+
+ /*! \brief Gets the first available platform.
+ *
+ * Wraps clGetPlatformIDs(), returning the first result.
+ */
+ static cl_int get(
+ Platform * platform)
+ {
+ cl_int err;
+ Platform default_platform = Platform::getDefault(&err);
+ if (platform) {
+ *platform = default_platform;
+ }
+ return err;
+ }
+
+ /*! \brief Gets the first available platform, returning it by value.
+ *
+ * \return Returns a valid platform if one is available.
+ * If no platform is available will return a null platform.
+ * Throws an exception if no platforms are available
+ * or an error condition occurs.
+ * Wraps clGetPlatformIDs(), returning the first result.
+ */
+ static Platform get(
+ cl_int * errResult = NULL)
+ {
+ cl_int err;
+ Platform default_platform = Platform::getDefault(&err);
+ if (errResult) {
+ *errResult = err;
+ }
+ return default_platform;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ //! \brief Wrapper for clUnloadCompiler().
+ cl_int
+ unloadCompiler()
+ {
+ return ::clUnloadPlatformCompiler(object_);
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+}; // class Platform
+
+CL_HPP_DEFINE_STATIC_MEMBER_ std::once_flag Platform::default_initialized_;
+CL_HPP_DEFINE_STATIC_MEMBER_ Platform Platform::default_;
+CL_HPP_DEFINE_STATIC_MEMBER_ cl_int Platform::default_error_ = CL_SUCCESS;
+
+
+/**
+ * Deprecated APIs for 1.2
+ */
+#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+/**
+ * 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_USE_DEPRECATED_OPENCL_1_1_APIS)
+
+/*! \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<cl_context>
+{
+private:
+ static std::once_flag default_initialized_;
+ static Context default_;
+ static cl_int default_error_;
+
+ /*! \brief Create the default context from the default device type in the default platform.
+ *
+ * This sets @c default_ and @c default_error_. It does not throw
+ * @c cl::Error.
+ */
+ static void makeDefault() {
+ /* Throwing an exception from a call_once invocation does not do
+ * what we wish, so we catch it and save the error.
+ */
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ try
+#endif
+ {
+#if !defined(__APPLE__) && !defined(__MACOS)
+ const Platform &p = Platform::getDefault();
+ cl_platform_id defaultPlatform = p();
+ cl_context_properties properties[3] = {
+ CL_CONTEXT_PLATFORM, (cl_context_properties)defaultPlatform, 0
+ };
+#else // #if !defined(__APPLE__) && !defined(__MACOS)
+ cl_context_properties *properties = nullptr;
+#endif // #if !defined(__APPLE__) && !defined(__MACOS)
+
+ default_ = Context(
+ CL_DEVICE_TYPE_DEFAULT,
+ properties,
+ NULL,
+ NULL,
+ &default_error_);
+ }
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ catch (cl::Error &e) {
+ default_error_ = e.err();
+ }
+#endif
+ }
+
+
+ /*! \brief Create the default context from a provided Context.
+ *
+ * This sets @c default_. It does not throw
+ * @c cl::Error.
+ */
+ static void makeDefaultProvided(const Context &c) {
+ default_ = c;
+ }
+
+public:
+#ifdef CL_HPP_UNIT_TEST_ENABLE
+ /*! \brief Reset the default.
+ *
+ * This sets @c default_ to an empty value to support cleanup in
+ * the unit test framework.
+ * This function is not thread safe.
+ */
+ static void unitTestClearDefault() {
+ default_ = Context();
+ }
+#endif // #ifdef CL_HPP_UNIT_TEST_ENABLE
+
+ /*! \brief Constructs a context including a list of specified devices.
+ *
+ * Wraps clCreateContext().
+ */
+ Context(
+ const vector<Device>& devices,
+ cl_context_properties* properties = NULL,
+ void (CL_CALLBACK * notifyFptr)(
+ const char *,
+ const void *,
+ size_type,
+ void *) = NULL,
+ void* data = NULL,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ size_type numDevices = devices.size();
+ vector<cl_device_id> deviceIDs(numDevices);
+
+ for( size_type deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) {
+ deviceIDs[deviceIndex] = (devices[deviceIndex])();
+ }
+
+ object_ = ::clCreateContext(
+ properties, (cl_uint) numDevices,
+ deviceIDs.data(),
+ 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_type,
+ 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_type,
+ 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<Platform> 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<Device> devices;
+
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ try {
+#endif
+
+ error = platforms[i].getDevices(type, &devices);
+
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ } catch (cl::Error& e) {
+ error = e.err();
+ }
+ // 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 Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Context(const Context& ctx) : detail::Wrapper<cl_type>(ctx) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Context& operator = (const Context &ctx)
+ {
+ detail::Wrapper<cl_type>::operator=(ctx);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Context(Context&& ctx) CL_HPP_NOEXCEPT_ : detail::Wrapper<cl_type>(std::move(ctx)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Context& operator = (Context &&ctx)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(ctx));
+ return *this;
+ }
+
+
+ /*! \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)
+ {
+ std::call_once(default_initialized_, makeDefault);
+ detail::errHandler(default_error_);
+ if (err != NULL) {
+ *err = default_error_;
+ }
+ return default_;
+ }
+
+ /**
+ * Modify the default context to be used by
+ * subsequent operations.
+ * Will only set the default if no default was previously created.
+ * @return updated default context.
+ * Should be compared to the passed value to ensure that it was updated.
+ */
+ static Context setDefault(const Context &default_context)
+ {
+ std::call_once(default_initialized_, makeDefaultProvided, std::cref(default_context));
+ detail::errHandler(default_error_);
+ return default_;
+ }
+
+ //! \brief Default constructor - initializes to NULL.
+ Context() : detail::Wrapper<cl_type>() { }
+
+ /*! \brief Constructor from cl_context - takes ownership.
+ *
+ * This effectively transfers ownership of a refcount on the cl_context
+ * into the new Context object.
+ */
+ explicit Context(const cl_context& context, bool retainObject = false) :
+ detail::Wrapper<cl_type>(context, retainObject) { }
+
+ /*! \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<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ //! \brief Wrapper for clGetContextInfo().
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_context_info, name>::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, &param);
+ 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<ImageFormat>* formats) const
+ {
+ cl_uint numEntries;
+
+ if (!formats) {
+ return CL_SUCCESS;
+ }
+
+ cl_int err = ::clGetSupportedImageFormats(
+ object_,
+ flags,
+ type,
+ 0,
+ NULL,
+ &numEntries);
+ if (err != CL_SUCCESS) {
+ return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR);
+ }
+
+ if (numEntries > 0) {
+ vector<ImageFormat> value(numEntries);
+ err = ::clGetSupportedImageFormats(
+ object_,
+ flags,
+ type,
+ numEntries,
+ (cl_image_format*)value.data(),
+ NULL);
+ if (err != CL_SUCCESS) {
+ return detail::errHandler(err, __GET_SUPPORTED_IMAGE_FORMATS_ERR);
+ }
+
+ formats->assign(begin(value), end(value));
+ }
+ else {
+ // If no values are being returned, ensure an empty vector comes back
+ formats->clear();
+ }
+
+ return CL_SUCCESS;
+ }
+};
+
+inline void Device::makeDefault()
+{
+ /* Throwing an exception from a call_once invocation does not do
+ * what we wish, so we catch it and save the error.
+ */
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ try
+#endif
+ {
+ cl_int error = 0;
+
+ Context context = Context::getDefault(&error);
+ detail::errHandler(error, __CREATE_CONTEXT_ERR);
+
+ if (error != CL_SUCCESS) {
+ default_error_ = error;
+ }
+ else {
+ default_ = context.getInfo<CL_CONTEXT_DEVICES>()[0];
+ default_error_ = CL_SUCCESS;
+ }
+ }
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ catch (cl::Error &e) {
+ default_error_ = e.err();
+ }
+#endif
+}
+
+CL_HPP_DEFINE_STATIC_MEMBER_ std::once_flag Context::default_initialized_;
+CL_HPP_DEFINE_STATIC_MEMBER_ Context Context::default_;
+CL_HPP_DEFINE_STATIC_MEMBER_ cl_int Context::default_error_ = CL_SUCCESS;
+
+/*! \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<cl_event>
+{
+public:
+ //! \brief Default constructor - initializes to NULL.
+ Event() : detail::Wrapper<cl_type>() { }
+
+ /*! \brief Constructor from cl_event - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * This effectively transfers ownership of a refcount on the cl_event
+ * into the new Event object.
+ */
+ explicit Event(const cl_event& event, bool retainObject = false) :
+ detail::Wrapper<cl_type>(event, retainObject) { }
+
+ /*! \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 cl_event& rhs)
+ {
+ detail::Wrapper<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ //! \brief Wrapper for clGetEventInfo().
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_event_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+ //! \brief Wrapper for clGetEventProfilingInfo().
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_profiling_info, name>::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, &param);
+ 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 CL_HPP_TARGET_OPENCL_VERSION >= 110
+ /*! \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 // CL_HPP_TARGET_OPENCL_VERSION >= 110
+
+ /*! \brief Blocks the calling thread until every event specified is complete.
+ *
+ * Wraps clWaitForEvents().
+ */
+ static cl_int
+ waitForEvents(const vector<Event>& events)
+ {
+ return detail::errHandler(
+ ::clWaitForEvents(
+ (cl_uint) events.size(), (events.size() > 0) ? (cl_event*)&events.front() : NULL),
+ __WAIT_FOR_EVENTS_ERR);
+ }
+};
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 110
+/*! \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 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 // CL_HPP_TARGET_OPENCL_VERSION >= 110
+
+/*! \brief Blocks the calling thread until every event specified is complete.
+ *
+ * Wraps clWaitForEvents().
+ */
+inline static cl_int
+WaitForEvents(const vector<Event>& events)
+{
+ return detail::errHandler(
+ ::clWaitForEvents(
+ (cl_uint) events.size(), (events.size() > 0) ? (cl_event*)&events.front() : NULL),
+ __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<cl_mem>
+{
+public:
+ //! \brief Default constructor - initializes to NULL.
+ Memory() : detail::Wrapper<cl_type>() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * Optionally transfer ownership of a refcount on the cl_mem
+ * into the new Memory object.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ *
+ * See Memory for further details.
+ */
+ explicit Memory(const cl_mem& memory, bool retainObject) :
+ detail::Wrapper<cl_type>(memory, retainObject) { }
+
+ /*! \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<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Memory(const Memory& mem) : detail::Wrapper<cl_type>(mem) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Memory& operator = (const Memory &mem)
+ {
+ detail::Wrapper<cl_type>::operator=(mem);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Memory(Memory&& mem) CL_HPP_NOEXCEPT_ : detail::Wrapper<cl_type>(std::move(mem)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Memory& operator = (Memory &&mem)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(mem));
+ return *this;
+ }
+
+
+ //! \brief Wrapper for clGetMemObjectInfo().
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_mem_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 110
+ /*! \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 // CL_HPP_TARGET_OPENCL_VERSION >= 110
+
+};
+
+// 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 );
+
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+namespace detail
+{
+ class SVMTraitNull
+ {
+ public:
+ static cl_svm_mem_flags getSVMMemFlags()
+ {
+ return 0;
+ }
+ };
+} // namespace detail
+
+template<class Trait = detail::SVMTraitNull>
+class SVMTraitReadWrite
+{
+public:
+ static cl_svm_mem_flags getSVMMemFlags()
+ {
+ return CL_MEM_READ_WRITE |
+ Trait::getSVMMemFlags();
+ }
+};
+
+template<class Trait = detail::SVMTraitNull>
+class SVMTraitReadOnly
+{
+public:
+ static cl_svm_mem_flags getSVMMemFlags()
+ {
+ return CL_MEM_READ_ONLY |
+ Trait::getSVMMemFlags();
+ }
+};
+
+template<class Trait = detail::SVMTraitNull>
+class SVMTraitWriteOnly
+{
+public:
+ static cl_svm_mem_flags getSVMMemFlags()
+ {
+ return CL_MEM_WRITE_ONLY |
+ Trait::getSVMMemFlags();
+ }
+};
+
+template<class Trait = SVMTraitReadWrite<>>
+class SVMTraitCoarse
+{
+public:
+ static cl_svm_mem_flags getSVMMemFlags()
+ {
+ return Trait::getSVMMemFlags();
+ }
+};
+
+template<class Trait = SVMTraitReadWrite<>>
+class SVMTraitFine
+{
+public:
+ static cl_svm_mem_flags getSVMMemFlags()
+ {
+ return CL_MEM_SVM_FINE_GRAIN_BUFFER |
+ Trait::getSVMMemFlags();
+ }
+};
+
+template<class Trait = SVMTraitReadWrite<>>
+class SVMTraitAtomic
+{
+public:
+ static cl_svm_mem_flags getSVMMemFlags()
+ {
+ return
+ CL_MEM_SVM_FINE_GRAIN_BUFFER |
+ CL_MEM_SVM_ATOMICS |
+ Trait::getSVMMemFlags();
+ }
+};
+
+// Pre-declare SVM map function
+template<typename T>
+inline cl_int enqueueMapSVM(
+ T* ptr,
+ cl_bool blocking,
+ cl_map_flags flags,
+ size_type size,
+ const vector<Event>* events = NULL,
+ Event* event = NULL);
+
+/**
+ * STL-like allocator class for managing SVM objects provided for convenience.
+ *
+ * Note that while this behaves like an allocator for the purposes of constructing vectors and similar objects,
+ * care must be taken when using with smart pointers.
+ * The allocator should not be used to construct a unique_ptr if we are using coarse-grained SVM mode because
+ * the coarse-grained management behaviour would behave incorrectly with respect to reference counting.
+ *
+ * Instead the allocator embeds a Deleter which may be used with unique_ptr and is used
+ * with the allocate_shared and allocate_ptr supplied operations.
+ */
+template<typename T, class SVMTrait>
+class SVMAllocator {
+private:
+ Context context_;
+
+public:
+ typedef T value_type;
+ typedef value_type* pointer;
+ typedef const value_type* const_pointer;
+ typedef value_type& reference;
+ typedef const value_type& const_reference;
+ typedef std::size_t size_type;
+ typedef std::ptrdiff_t difference_type;
+
+ template<typename U>
+ struct rebind
+ {
+ typedef SVMAllocator<U, SVMTrait> other;
+ };
+
+ template<typename U, typename V>
+ friend class SVMAllocator;
+
+ SVMAllocator() :
+ context_(Context::getDefault())
+ {
+ }
+
+ explicit SVMAllocator(cl::Context context) :
+ context_(context)
+ {
+ }
+
+
+ SVMAllocator(const SVMAllocator &other) :
+ context_(other.context_)
+ {
+ }
+
+ template<typename U>
+ SVMAllocator(const SVMAllocator<U, SVMTrait> &other) :
+ context_(other.context_)
+ {
+ }
+
+ ~SVMAllocator()
+ {
+ }
+
+ pointer address(reference r) CL_HPP_NOEXCEPT_
+ {
+ return std::addressof(r);
+ }
+
+ const_pointer address(const_reference r) CL_HPP_NOEXCEPT_
+ {
+ return std::addressof(r);
+ }
+
+ /**
+ * Allocate an SVM pointer.
+ *
+ * If the allocator is coarse-grained, this will take ownership to allow
+ * containers to correctly construct data in place.
+ */
+ pointer allocate(
+ size_type size,
+ typename cl::SVMAllocator<void, SVMTrait>::const_pointer = 0)
+ {
+ // Allocate memory with default alignment matching the size of the type
+ void* voidPointer =
+ clSVMAlloc(
+ context_(),
+ SVMTrait::getSVMMemFlags(),
+ size*sizeof(T),
+ 0);
+ pointer retValue = reinterpret_cast<pointer>(
+ voidPointer);
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ if (!retValue) {
+ std::bad_alloc excep;
+ throw excep;
+ }
+#endif // #if defined(CL_HPP_ENABLE_EXCEPTIONS)
+
+ // If allocation was coarse-grained then map it
+ if (!(SVMTrait::getSVMMemFlags() & CL_MEM_SVM_FINE_GRAIN_BUFFER)) {
+ cl_int err = enqueueMapSVM(retValue, CL_TRUE, CL_MAP_READ | CL_MAP_WRITE, size*sizeof(T));
+ if (err != CL_SUCCESS) {
+ std::bad_alloc excep;
+ throw excep;
+ }
+ }
+
+ // If exceptions disabled, return null pointer from allocator
+ return retValue;
+ }
+
+ void deallocate(pointer p, size_type)
+ {
+ clSVMFree(context_(), p);
+ }
+
+ /**
+ * Return the maximum possible allocation size.
+ * This is the minimum of the maximum sizes of all devices in the context.
+ */
+ size_type max_size() const CL_HPP_NOEXCEPT_
+ {
+ size_type maxSize = std::numeric_limits<size_type>::max() / sizeof(T);
+
+ for (const Device &d : context_.getInfo<CL_CONTEXT_DEVICES>()) {
+ maxSize = std::min(
+ maxSize,
+ static_cast<size_type>(d.getInfo<CL_DEVICE_MAX_MEM_ALLOC_SIZE>()));
+ }
+
+ return maxSize;
+ }
+
+ template< class U, class... Args >
+ void construct(U* p, Args&&... args)
+ {
+ new(p)T(args...);
+ }
+
+ template< class U >
+ void destroy(U* p)
+ {
+ p->~U();
+ }
+
+ /**
+ * Returns true if the contexts match.
+ */
+ inline bool operator==(SVMAllocator const& rhs)
+ {
+ return (context_==rhs.context_);
+ }
+
+ inline bool operator!=(SVMAllocator const& a)
+ {
+ return !operator==(a);
+ }
+}; // class SVMAllocator return cl::pointer<T>(tmp, detail::Deleter<T, Alloc>{alloc, copies});
+
+
+template<class SVMTrait>
+class SVMAllocator<void, SVMTrait> {
+public:
+ typedef void value_type;
+ typedef value_type* pointer;
+ typedef const value_type* const_pointer;
+
+ template<typename U>
+ struct rebind
+ {
+ typedef SVMAllocator<U, SVMTrait> other;
+ };
+
+ template<typename U, typename V>
+ friend class SVMAllocator;
+};
+
+#if !defined(CL_HPP_NO_STD_UNIQUE_PTR)
+namespace detail
+{
+ template<class Alloc>
+ class Deleter {
+ private:
+ Alloc alloc_;
+ size_type copies_;
+
+ public:
+ typedef typename std::allocator_traits<Alloc>::pointer pointer;
+
+ Deleter(const Alloc &alloc, size_type copies) : alloc_{ alloc }, copies_{ copies }
+ {
+ }
+
+ void operator()(pointer ptr) const {
+ Alloc tmpAlloc{ alloc_ };
+ std::allocator_traits<Alloc>::destroy(tmpAlloc, std::addressof(*ptr));
+ std::allocator_traits<Alloc>::deallocate(tmpAlloc, ptr, copies_);
+ }
+ };
+} // namespace detail
+
+/**
+ * Allocation operation compatible with std::allocate_ptr.
+ * Creates a unique_ptr<T> by default.
+ * This requirement is to ensure that the control block is not
+ * allocated in memory inaccessible to the host.
+ */
+template <class T, class Alloc, class... Args>
+cl::pointer<T, detail::Deleter<Alloc>> allocate_pointer(const Alloc &alloc_, Args&&... args)
+{
+ Alloc alloc(alloc_);
+ static const size_type copies = 1;
+
+ // Ensure that creation of the management block and the
+ // object are dealt with separately such that we only provide a deleter
+
+ T* tmp = std::allocator_traits<Alloc>::allocate(alloc, copies);
+ if (!tmp) {
+ std::bad_alloc excep;
+ throw excep;
+ }
+ try {
+ std::allocator_traits<Alloc>::construct(
+ alloc,
+ std::addressof(*tmp),
+ std::forward<Args>(args)...);
+
+ return cl::pointer<T, detail::Deleter<Alloc>>(tmp, detail::Deleter<Alloc>{alloc, copies});
+ }
+ catch (std::bad_alloc b)
+ {
+ std::allocator_traits<Alloc>::deallocate(alloc, tmp, copies);
+ throw;
+ }
+}
+
+template< class T, class SVMTrait, class... Args >
+cl::pointer<T, detail::Deleter<SVMAllocator<T, SVMTrait>>> allocate_svm(Args... args)
+{
+ SVMAllocator<T, SVMTrait> alloc;
+ return cl::allocate_pointer<T>(alloc, args...);
+}
+
+template< class T, class SVMTrait, class... Args >
+cl::pointer<T, detail::Deleter<SVMAllocator<T, SVMTrait>>> allocate_svm(const cl::Context &c, Args... args)
+{
+ SVMAllocator<T, SVMTrait> alloc(c);
+ return cl::allocate_pointer<T>(alloc, args...);
+}
+#endif // #if !defined(CL_HPP_NO_STD_UNIQUE_PTR)
+
+/*! \brief Vector alias to simplify contruction of coarse-grained SVM containers.
+ *
+ */
+template < class T >
+using coarse_svm_vector = vector<T, cl::SVMAllocator<int, cl::SVMTraitCoarse<>>>;
+
+/*! \brief Vector alias to simplify contruction of fine-grained SVM containers.
+*
+*/
+template < class T >
+using fine_svm_vector = vector<T, cl::SVMAllocator<int, cl::SVMTraitFine<>>>;
+
+/*! \brief Vector alias to simplify contruction of fine-grained SVM containers that support platform atomics.
+*
+*/
+template < class T >
+using atomic_svm_vector = vector<T, cl::SVMAllocator<int, cl::SVMTraitAtomic<>>>;
+
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+
+/*! \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_type 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_type 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<IteratorType>::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_type size = sizeof(DataType)*(endIterator - startIterator);
+
+ Context context = Context::getDefault(err);
+
+ if( useHostPtr ) {
+ object_ = ::clCreateBuffer(context(), flags, size, static_cast<DataType*>(&*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 Construct a Buffer from a host container via iterators using a specified queue.
+ * If useHostPtr is specified iterators must be random access.
+ */
+ template< typename IteratorType >
+ Buffer(const CommandQueue &queue, IteratorType startIterator, IteratorType endIterator,
+ bool readOnly, bool useHostPtr = false, cl_int* err = NULL);
+
+ //! \brief Default constructor - initializes to NULL.
+ Buffer() : Memory() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with earlier versions.
+ *
+ * See Memory for further details.
+ */
+ explicit Buffer(const cl_mem& buffer, bool retainObject = false) :
+ Memory(buffer, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Buffer& operator = (const cl_mem& rhs)
+ {
+ Memory::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Buffer(const Buffer& buf) : Memory(buf) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Buffer& operator = (const Buffer &buf)
+ {
+ Memory::operator=(buf);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Buffer(Buffer&& buf) CL_HPP_NOEXCEPT_ : Memory(std::move(buf)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Buffer& operator = (Buffer &&buf)
+ {
+ Memory::operator=(std::move(buf));
+ return *this;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 110
+ /*! \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 // CL_HPP_TARGET_OPENCL_VERSION >= 110
+};
+
+#if defined (CL_HPP_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:
+
+
+ /*! \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) : pfn_clCreateFromD3D10BufferKHR(nullptr)
+ {
+ typedef CL_API_ENTRY cl_mem (CL_API_CALL *PFN_clCreateFromD3D10BufferKHR)(
+ cl_context context, cl_mem_flags flags, ID3D10Buffer* buffer,
+ cl_int* errcode_ret);
+ PFN_clCreateFromD3D10BufferKHR pfn_clCreateFromD3D10BufferKHR;
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ vector<cl_context_properties> props = context.getInfo<CL_CONTEXT_PROPERTIES>();
+ cl_platform platform = -1;
+ for( int i = 0; i < props.size(); ++i ) {
+ if( props[i] == CL_CONTEXT_PLATFORM ) {
+ platform = props[i+1];
+ }
+ }
+ CL_HPP_INIT_CL_EXT_FCN_PTR_PLATFORM_(platform, clCreateFromD3D10BufferKHR);
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 110
+ CL_HPP_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 Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit BufferD3D10(const cl_mem& buffer, bool retainObject = false) :
+ Buffer(buffer, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ BufferD3D10& operator = (const cl_mem& rhs)
+ {
+ Buffer::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferD3D10(const BufferD3D10& buf) :
+ Buffer(buf) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferD3D10& operator = (const BufferD3D10 &buf)
+ {
+ Buffer::operator=(buf);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferD3D10(BufferD3D10&& buf) CL_HPP_NOEXCEPT_ : Buffer(std::move(buf)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferD3D10& operator = (BufferD3D10 &&buf)
+ {
+ Buffer::operator=(std::move(buf));
+ 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,
+ cl_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 Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit BufferGL(const cl_mem& buffer, bool retainObject = false) :
+ Buffer(buffer, retainObject) { }
+
+ /*! \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 Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferGL(const BufferGL& buf) : Buffer(buf) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferGL& operator = (const BufferGL &buf)
+ {
+ Buffer::operator=(buf);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferGL(BufferGL&& buf) CL_HPP_NOEXCEPT_ : Buffer(std::move(buf)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferGL& operator = (BufferGL &&buf)
+ {
+ Buffer::operator=(std::move(buf));
+ return *this;
+ }
+
+ //! \brief Wrapper for clGetGLObjectInfo().
+ cl_int getObjectInfo(
+ cl_gl_object_type *type,
+ cl_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,
+ cl_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 Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit BufferRenderGL(const cl_mem& buffer, bool retainObject = false) :
+ Buffer(buffer, retainObject) { }
+
+ /*! \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 Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferRenderGL(const BufferRenderGL& buf) : Buffer(buf) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferRenderGL& operator = (const BufferRenderGL &buf)
+ {
+ Buffer::operator=(buf);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferRenderGL(BufferRenderGL&& buf) CL_HPP_NOEXCEPT_ : Buffer(std::move(buf)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ BufferRenderGL& operator = (BufferRenderGL &&buf)
+ {
+ Buffer::operator=(std::move(buf));
+ return *this;
+ }
+
+ //! \brief Wrapper for clGetGLObjectInfo().
+ cl_int getObjectInfo(
+ cl_gl_object_type *type,
+ cl_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 Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image(const cl_mem& image, bool retainObject = false) :
+ Memory(image, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image& operator = (const cl_mem& rhs)
+ {
+ Memory::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image(const Image& img) : Memory(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image& operator = (const Image &img)
+ {
+ Memory::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image(Image&& img) CL_HPP_NOEXCEPT_ : Memory(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image& operator = (Image &&img)
+ {
+ Memory::operator=(std::move(img));
+ return *this;
+ }
+
+
+public:
+ //! \brief Wrapper for clGetImageInfo().
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_image_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+};
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+/*! \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_type 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 Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image1D(const cl_mem& image1D, bool retainObject = false) :
+ Image(image1D, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image1D& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1D(const Image1D& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1D& operator = (const Image1D &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1D(Image1D&& img) CL_HPP_NOEXCEPT_ : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1D& operator = (Image1D &&img)
+ {
+ Image::operator=(std::move(img));
+ 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_type 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() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image1DBuffer(const cl_mem& image1D, bool retainObject = false) :
+ Image(image1D, retainObject) { }
+
+ Image1DBuffer& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DBuffer(const Image1DBuffer& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DBuffer& operator = (const Image1DBuffer &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DBuffer(Image1DBuffer&& img) CL_HPP_NOEXCEPT_ : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DBuffer& operator = (Image1DBuffer &&img)
+ {
+ Image::operator=(std::move(img));
+ 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_type arraySize,
+ size_type width,
+ size_type 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() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image1DArray(const cl_mem& imageArray, bool retainObject = false) :
+ Image(imageArray, retainObject) { }
+
+
+ Image1DArray& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DArray(const Image1DArray& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DArray& operator = (const Image1DArray &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DArray(Image1DArray&& img) CL_HPP_NOEXCEPT_ : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image1DArray& operator = (Image1DArray &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+
+};
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+
+/*! \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 2D Image in a specified context.
+ *
+ * Wraps clCreateImage().
+ */
+ Image2D(
+ const Context& context,
+ cl_mem_flags flags,
+ ImageFormat format,
+ size_type width,
+ size_type height,
+ size_type row_pitch = 0,
+ void* host_ptr = NULL,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+ bool useCreateImage;
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120 && CL_HPP_MINIMUM_OPENCL_VERSION < 120
+ // Run-time decision based on the actual platform
+ {
+ cl_uint version = detail::getContextPlatformVersion(context());
+ useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above
+ }
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 120
+ useCreateImage = true;
+#else
+ useCreateImage = false;
+#endif
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ 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 // CL_HPP_TARGET_OPENCL_VERSION >= 120
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 120
+ 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 // CL_HPP_MINIMUM_OPENCL_VERSION < 120
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200 || defined(CL_HPP_USE_CL_IMAGE2D_FROM_BUFFER_KHR)
+ /*! \brief Constructs a 2D Image from a buffer.
+ * \note This will share storage with the underlying buffer.
+ *
+ * Wraps clCreateImage().
+ */
+ Image2D(
+ const Context& context,
+ ImageFormat format,
+ const Buffer &sourceBuffer,
+ size_type width,
+ size_type height,
+ size_type row_pitch = 0,
+ cl_int* err = nullptr)
+ {
+ cl_int error;
+
+ cl_image_desc desc =
+ {
+ CL_MEM_OBJECT_IMAGE2D,
+ width,
+ height,
+ 0, 0, // depth, array size (unused)
+ row_pitch,
+ 0, 0, 0,
+ // Use buffer as input to image
+ sourceBuffer()
+ };
+ object_ = ::clCreateImage(
+ context(),
+ 0, // flags inherited from buffer
+ &format,
+ &desc,
+ nullptr,
+ &error);
+
+ detail::errHandler(error, __CREATE_IMAGE_ERR);
+ if (err != nullptr) {
+ *err = error;
+ }
+ }
+#endif //#if CL_HPP_TARGET_OPENCL_VERSION >= 200 || defined(CL_HPP_USE_CL_IMAGE2D_FROM_BUFFER_KHR)
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ /*! \brief Constructs a 2D Image from an image.
+ * \note This will share storage with the underlying image but may
+ * reinterpret the channel order and type.
+ *
+ * The image will be created matching with a descriptor matching the source.
+ *
+ * \param order is the channel order to reinterpret the image data as.
+ * The channel order may differ as described in the OpenCL
+ * 2.0 API specification.
+ *
+ * Wraps clCreateImage().
+ */
+ Image2D(
+ const Context& context,
+ cl_channel_order order,
+ const Image &sourceImage,
+ cl_int* err = nullptr)
+ {
+ cl_int error;
+
+ // Descriptor fields have to match source image
+ size_type sourceWidth =
+ sourceImage.getImageInfo<CL_IMAGE_WIDTH>();
+ size_type sourceHeight =
+ sourceImage.getImageInfo<CL_IMAGE_HEIGHT>();
+ size_type sourceRowPitch =
+ sourceImage.getImageInfo<CL_IMAGE_ROW_PITCH>();
+ cl_uint sourceNumMIPLevels =
+ sourceImage.getImageInfo<CL_IMAGE_NUM_MIP_LEVELS>();
+ cl_uint sourceNumSamples =
+ sourceImage.getImageInfo<CL_IMAGE_NUM_SAMPLES>();
+ cl_image_format sourceFormat =
+ sourceImage.getImageInfo<CL_IMAGE_FORMAT>();
+
+ // Update only the channel order.
+ // Channel format inherited from source.
+ sourceFormat.image_channel_order = order;
+ cl_image_desc desc =
+ {
+ CL_MEM_OBJECT_IMAGE2D,
+ sourceWidth,
+ sourceHeight,
+ 0, 0, // depth (unused), array size (unused)
+ sourceRowPitch,
+ 0, // slice pitch (unused)
+ sourceNumMIPLevels,
+ sourceNumSamples,
+ // Use buffer as input to image
+ sourceImage()
+ };
+ object_ = ::clCreateImage(
+ context(),
+ 0, // flags should be inherited from mem_object
+ &sourceFormat,
+ &desc,
+ nullptr,
+ &error);
+
+ detail::errHandler(error, __CREATE_IMAGE_ERR);
+ if (err != nullptr) {
+ *err = error;
+ }
+ }
+#endif //#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+ //! \brief Default constructor - initializes to NULL.
+ Image2D() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image2D(const cl_mem& image2D, bool retainObject = false) :
+ Image(image2D, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image2D& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2D(const Image2D& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2D& operator = (const Image2D &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2D(Image2D&& img) CL_HPP_NOEXCEPT_ : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2D& operator = (Image2D &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+
+};
+
+
+#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+/*! \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 : 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,
+ cl_GLenum target,
+ cl_GLint miplevel,
+ cl_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 Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image2DGL(const cl_mem& image, bool retainObject = false) :
+ Image2D(image, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *c
+ * See Memory for further details.
+ */
+ Image2DGL& operator = (const cl_mem& rhs)
+ {
+ Image2D::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DGL(const Image2DGL& img) : Image2D(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DGL& operator = (const Image2DGL &img)
+ {
+ Image2D::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DGL(Image2DGL&& img) CL_HPP_NOEXCEPT_ : Image2D(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DGL& operator = (Image2DGL &&img)
+ {
+ Image2D::operator=(std::move(img));
+ return *this;
+ }
+
+} CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED;
+#endif // CL_USE_DEPRECATED_OPENCL_1_1_APIS
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+/*! \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_type arraySize,
+ size_type width,
+ size_type height,
+ size_type rowPitch,
+ size_type 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() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image2DArray(const cl_mem& imageArray, bool retainObject = false) : Image(imageArray, retainObject) { }
+
+ Image2DArray& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DArray(const Image2DArray& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DArray& operator = (const Image2DArray &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DArray(Image2DArray&& img) CL_HPP_NOEXCEPT_ : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image2DArray& operator = (Image2DArray &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+};
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+/*! \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_type width,
+ size_type height,
+ size_type depth,
+ size_type row_pitch = 0,
+ size_type slice_pitch = 0,
+ void* host_ptr = NULL,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+ bool useCreateImage;
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120 && CL_HPP_MINIMUM_OPENCL_VERSION < 120
+ // Run-time decision based on the actual platform
+ {
+ cl_uint version = detail::getContextPlatformVersion(context());
+ useCreateImage = (version >= 0x10002); // OpenCL 1.2 or above
+ }
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 120
+ useCreateImage = true;
+#else
+ useCreateImage = false;
+#endif
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ 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 // CL_HPP_TARGET_OPENCL_VERSION >= 120
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 120
+ 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 // CL_HPP_MINIMUM_OPENCL_VERSION < 120
+ }
+
+ //! \brief Default constructor - initializes to NULL.
+ Image3D() : Image() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image3D(const cl_mem& image3D, bool retainObject = false) :
+ Image(image3D, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image3D& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3D(const Image3D& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3D& operator = (const Image3D &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3D(Image3D&& img) CL_HPP_NOEXCEPT_ : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3D& operator = (Image3D &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+};
+
+#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+/*! \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,
+ cl_GLenum target,
+ cl_GLint miplevel,
+ cl_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 Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit Image3DGL(const cl_mem& image, bool retainObject = false) :
+ Image3D(image, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Image3DGL& operator = (const cl_mem& rhs)
+ {
+ Image3D::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3DGL(const Image3DGL& img) : Image3D(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3DGL& operator = (const Image3DGL &img)
+ {
+ Image3D::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3DGL(Image3DGL&& img) CL_HPP_NOEXCEPT_ : Image3D(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Image3DGL& operator = (Image3DGL &&img)
+ {
+ Image3D::operator=(std::move(img));
+ return *this;
+ }
+};
+#endif // CL_USE_DEPRECATED_OPENCL_1_1_APIS
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+/*! \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,
+ cl_GLenum target,
+ cl_GLint miplevel,
+ cl_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() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * See Memory for further details.
+ */
+ explicit ImageGL(const cl_mem& image, bool retainObject = false) :
+ Image(image, retainObject) { }
+
+ ImageGL& operator = (const cl_mem& rhs)
+ {
+ Image::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ ImageGL(const ImageGL& img) : Image(img) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ ImageGL& operator = (const ImageGL &img)
+ {
+ Image::operator=(img);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ ImageGL(ImageGL&& img) CL_HPP_NOEXCEPT_ : Image(std::move(img)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ ImageGL& operator = (ImageGL &&img)
+ {
+ Image::operator=(std::move(img));
+ return *this;
+ }
+};
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+/*! \brief Class interface for Pipe Memory Objects.
+*
+* See Memory for details about copy semantics, etc.
+*
+* \see Memory
+*/
+class Pipe : public Memory
+{
+public:
+
+ /*! \brief Constructs a Pipe in a specified context.
+ *
+ * Wraps clCreatePipe().
+ * @param context Context in which to create the pipe.
+ * @param flags Bitfield. Only CL_MEM_READ_WRITE and CL_MEM_HOST_NO_ACCESS are valid.
+ * @param packet_size Size in bytes of a single packet of the pipe.
+ * @param max_packets Number of packets that may be stored in the pipe.
+ *
+ */
+ Pipe(
+ const Context& context,
+ cl_uint packet_size,
+ cl_uint max_packets,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ cl_mem_flags flags = CL_MEM_READ_WRITE | CL_MEM_HOST_NO_ACCESS;
+ object_ = ::clCreatePipe(context(), flags, packet_size, max_packets, nullptr, &error);
+
+ detail::errHandler(error, __CREATE_PIPE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+ /*! \brief Constructs a Pipe in a the default context.
+ *
+ * Wraps clCreatePipe().
+ * @param flags Bitfield. Only CL_MEM_READ_WRITE and CL_MEM_HOST_NO_ACCESS are valid.
+ * @param packet_size Size in bytes of a single packet of the pipe.
+ * @param max_packets Number of packets that may be stored in the pipe.
+ *
+ */
+ Pipe(
+ cl_uint packet_size,
+ cl_uint max_packets,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ Context context = Context::getDefault(err);
+
+ cl_mem_flags flags = CL_MEM_READ_WRITE | CL_MEM_HOST_NO_ACCESS;
+ object_ = ::clCreatePipe(context(), flags, packet_size, max_packets, nullptr, &error);
+
+ detail::errHandler(error, __CREATE_PIPE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+ //! \brief Default constructor - initializes to NULL.
+ Pipe() : Memory() { }
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with earlier versions.
+ *
+ * See Memory for further details.
+ */
+ explicit Pipe(const cl_mem& pipe, bool retainObject = false) :
+ Memory(pipe, retainObject) { }
+
+ /*! \brief Assignment from cl_mem - performs shallow copy.
+ *
+ * See Memory for further details.
+ */
+ Pipe& operator = (const cl_mem& rhs)
+ {
+ Memory::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Pipe(const Pipe& pipe) : Memory(pipe) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Pipe& operator = (const Pipe &pipe)
+ {
+ Memory::operator=(pipe);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Pipe(Pipe&& pipe) CL_HPP_NOEXCEPT_ : Memory(std::move(pipe)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Pipe& operator = (Pipe &&pipe)
+ {
+ Memory::operator=(std::move(pipe));
+ return *this;
+ }
+
+ //! \brief Wrapper for clGetMemObjectInfo().
+ template <typename T>
+ cl_int getInfo(cl_pipe_info name, T* param) const
+ {
+ return detail::errHandler(
+ detail::getInfo(&::clGetPipeInfo, object_, name, param),
+ __GET_PIPE_INFO_ERR);
+ }
+
+ //! \brief Wrapper for clGetMemObjectInfo() that returns by value.
+ template <cl_int name> typename
+ detail::param_traits<detail::cl_pipe_info, name>::param_type
+ getInfo(cl_int* err = NULL) const
+ {
+ typename detail::param_traits<
+ detail::cl_pipe_info, name>::param_type param;
+ cl_int result = getInfo(name, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+}; // class Pipe
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+
+/*! \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<cl_sampler>
+{
+public:
+ //! \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;
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ cl_sampler_properties sampler_properties[] = {
+ CL_SAMPLER_NORMALIZED_COORDS, normalized_coords,
+ CL_SAMPLER_ADDRESSING_MODE, addressing_mode,
+ CL_SAMPLER_FILTER_MODE, filter_mode,
+ 0 };
+ object_ = ::clCreateSamplerWithProperties(
+ context(),
+ sampler_properties,
+ &error);
+
+ detail::errHandler(error, __CREATE_SAMPLER_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+#else
+ object_ = ::clCreateSampler(
+ context(),
+ normalized_coords,
+ addressing_mode,
+ filter_mode,
+ &error);
+
+ detail::errHandler(error, __CREATE_SAMPLER_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+#endif
+ }
+
+ /*! \brief Constructor from cl_sampler - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * This effectively transfers ownership of a refcount on the cl_sampler
+ * into the new Sampler object.
+ */
+ explicit Sampler(const cl_sampler& sampler, bool retainObject = false) :
+ detail::Wrapper<cl_type>(sampler, retainObject) { }
+
+ /*! \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<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Sampler(const Sampler& sam) : detail::Wrapper<cl_type>(sam) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Sampler& operator = (const Sampler &sam)
+ {
+ detail::Wrapper<cl_type>::operator=(sam);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Sampler(Sampler&& sam) CL_HPP_NOEXCEPT_ : detail::Wrapper<cl_type>(std::move(sam)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Sampler& operator = (Sampler &&sam)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(sam));
+ return *this;
+ }
+
+ //! \brief Wrapper for clGetSamplerInfo().
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_sampler_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+};
+
+class Program;
+class CommandQueue;
+class DeviceCommandQueue;
+class Kernel;
+
+//! \brief Class interface for specifying NDRange values.
+class NDRange
+{
+private:
+ size_type sizes_[3];
+ cl_uint dimensions_;
+
+public:
+ //! \brief Default constructor - resulting range has zero dimensions.
+ NDRange()
+ : dimensions_(0)
+ {
+ sizes_[0] = 0;
+ sizes_[1] = 0;
+ sizes_[2] = 0;
+ }
+
+ //! \brief Constructs one-dimensional range.
+ NDRange(size_type size0)
+ : dimensions_(1)
+ {
+ sizes_[0] = size0;
+ sizes_[1] = 1;
+ sizes_[2] = 1;
+ }
+
+ //! \brief Constructs two-dimensional range.
+ NDRange(size_type size0, size_type size1)
+ : dimensions_(2)
+ {
+ sizes_[0] = size0;
+ sizes_[1] = size1;
+ sizes_[2] = 1;
+ }
+
+ //! \brief Constructs three-dimensional range.
+ NDRange(size_type size0, size_type size1, size_type size2)
+ : dimensions_(3)
+ {
+ sizes_[0] = size0;
+ sizes_[1] = size1;
+ sizes_[2] = size2;
+ }
+
+ /*! \brief Conversion operator to const size_type *.
+ *
+ * \returns a pointer to the size of the first dimension.
+ */
+ operator const size_type*() const {
+ return sizes_;
+ }
+
+ //! \brief Queries the number of dimensions in the range.
+ size_type dimensions() const
+ {
+ return dimensions_;
+ }
+
+ //! \brief Returns the size of the object in bytes based on the
+ // runtime number of dimensions
+ size_type size() const
+ {
+ return dimensions_*sizeof(size_type);
+ }
+
+ size_type* get()
+ {
+ return sizes_;
+ }
+
+ const size_type* get() const
+ {
+ return sizes_;
+ }
+};
+
+//! \brief A zero-dimensional range.
+static const NDRange NullRange;
+
+//! \brief Local address wrapper for use with Kernel::setArg
+struct LocalSpaceArg
+{
+ size_type size_;
+};
+
+namespace detail {
+
+template <typename T, class Enable = void>
+struct KernelArgumentHandler;
+
+// Enable for objects that are not subclasses of memory
+// Pointers, constants etc
+template <typename T>
+struct KernelArgumentHandler<T, typename std::enable_if<!std::is_base_of<cl::Memory, T>::value>::type>
+{
+ static size_type size(const T&) { return sizeof(T); }
+ static const T* ptr(const T& value) { return &value; }
+};
+
+// Enable for subclasses of memory where we want to get a reference to the cl_mem out
+// and pass that in for safety
+template <typename T>
+struct KernelArgumentHandler<T, typename std::enable_if<std::is_base_of<cl::Memory, T>::value>::type>
+{
+ static size_type size(const T&) { return sizeof(cl_mem); }
+ static const cl_mem* ptr(const T& value) { return &(value()); }
+};
+
+// Specialization for DeviceCommandQueue defined later
+
+template <>
+struct KernelArgumentHandler<LocalSpaceArg, void>
+{
+ static size_type size(const LocalSpaceArg& value) { return value.size_; }
+ static const void* ptr(const LocalSpaceArg&) { return NULL; }
+};
+
+}
+//! \endcond
+
+/*! Local
+ * \brief Helper function for generating LocalSpaceArg objects.
+ */
+inline LocalSpaceArg
+Local(size_type size)
+{
+ LocalSpaceArg ret = { size };
+ return ret;
+}
+
+/*! \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<cl_kernel>
+{
+public:
+ inline Kernel(const Program& program, const char* name, cl_int* err = NULL);
+
+ //! \brief Default constructor - initializes to NULL.
+ Kernel() { }
+
+ /*! \brief Constructor from cl_kernel - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ * This effectively transfers ownership of a refcount on the cl_kernel
+ * into the new Kernel object.
+ */
+ explicit Kernel(const cl_kernel& kernel, bool retainObject = false) :
+ detail::Wrapper<cl_type>(kernel, retainObject) { }
+
+ /*! \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<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Kernel(const Kernel& kernel) : detail::Wrapper<cl_type>(kernel) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Kernel& operator = (const Kernel &kernel)
+ {
+ detail::Wrapper<cl_type>::operator=(kernel);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Kernel(Kernel&& kernel) CL_HPP_NOEXCEPT_ : detail::Wrapper<cl_type>(std::move(kernel)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Kernel& operator = (Kernel &&kernel)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(kernel));
+ return *this;
+ }
+
+ template <typename T>
+ cl_int getInfo(cl_kernel_info name, T* param) const
+ {
+ return detail::errHandler(
+ detail::getInfo(&::clGetKernelInfo, object_, name, param),
+ __GET_KERNEL_INFO_ERR);
+ }
+
+ template <cl_int name> typename
+ detail::param_traits<detail::cl_kernel_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_kernel_arg_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_kernel_work_group_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+#if defined(CL_HPP_USE_CL_SUB_GROUPS_KHR)
+ cl_int getSubGroupInfo(const cl::Device &dev, cl_kernel_sub_group_info name, const cl::NDRange &range, size_type* param) const
+ {
+ typedef clGetKernelSubGroupInfoKHR_fn PFN_clGetKernelSubGroupInfoKHR;
+ static PFN_clGetKernelSubGroupInfoKHR pfn_clGetKernelSubGroupInfoKHR = NULL;
+ CL_HPP_INIT_CL_EXT_FCN_PTR_(clGetKernelSubGroupInfoKHR);
+
+ return detail::errHandler(
+ pfn_clGetKernelSubGroupInfoKHR(object_, dev(), name, range.size(), range.get(), sizeof(size_type), param, nullptr),
+ __GET_KERNEL_ARG_INFO_ERR);
+ }
+
+ template <cl_int name>
+ size_type getSubGroupInfo(const cl::Device &dev, const cl::NDRange &range, cl_int* err = NULL) const
+ {
+ size_type param;
+ cl_int result = getSubGroupInfo(dev, name, range, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+#endif // #if defined(CL_HPP_USE_CL_SUB_GROUPS_KHR)
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ /*! \brief setArg overload taking a shared_ptr type
+ */
+ template<typename T, class D>
+ cl_int setArg(cl_uint index, const cl::pointer<T, D> &argPtr)
+ {
+ return detail::errHandler(
+ ::clSetKernelArgSVMPointer(object_, index, argPtr.get()),
+ __SET_KERNEL_ARGS_ERR);
+ }
+
+ /*! \brief setArg overload taking a vector type.
+ */
+ template<typename T, class Alloc>
+ cl_int setArg(cl_uint index, const cl::vector<T, Alloc> &argPtr)
+ {
+ return detail::errHandler(
+ ::clSetKernelArgSVMPointer(object_, index, argPtr.data()),
+ __SET_KERNEL_ARGS_ERR);
+ }
+
+ /*! \brief setArg overload taking a pointer type
+ */
+ template<typename T>
+ typename std::enable_if<std::is_pointer<T>::value, cl_int>::type
+ setArg(cl_uint index, const T argPtr)
+ {
+ return detail::errHandler(
+ ::clSetKernelArgSVMPointer(object_, index, argPtr),
+ __SET_KERNEL_ARGS_ERR);
+ }
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+ /*! \brief setArg overload taking a POD type
+ */
+ template <typename T>
+ typename std::enable_if<!std::is_pointer<T>::value, cl_int>::type
+ setArg(cl_uint index, const T &value)
+ {
+ return detail::errHandler(
+ ::clSetKernelArg(
+ object_,
+ index,
+ detail::KernelArgumentHandler<T>::size(value),
+ detail::KernelArgumentHandler<T>::ptr(value)),
+ __SET_KERNEL_ARGS_ERR);
+ }
+
+ cl_int setArg(cl_uint index, size_type size, const void* argPtr)
+ {
+ return detail::errHandler(
+ ::clSetKernelArg(object_, index, size, argPtr),
+ __SET_KERNEL_ARGS_ERR);
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ /*!
+ * Specify a vector of SVM pointers that the kernel may access in
+ * addition to its arguments.
+ */
+ cl_int setSVMPointers(const vector<void*> &pointerList)
+ {
+ return detail::errHandler(
+ ::clSetKernelExecInfo(
+ object_,
+ CL_KERNEL_EXEC_INFO_SVM_PTRS,
+ sizeof(void*)*pointerList.size(),
+ pointerList.data()));
+ }
+
+ /*!
+ * Specify a std::array of SVM pointers that the kernel may access in
+ * addition to its arguments.
+ */
+ template<int ArrayLength>
+ cl_int setSVMPointers(const std::array<void*, ArrayLength> &pointerList)
+ {
+ return detail::errHandler(
+ ::clSetKernelExecInfo(
+ object_,
+ CL_KERNEL_EXEC_INFO_SVM_PTRS,
+ sizeof(void*)*pointerList.size(),
+ pointerList.data()));
+ }
+
+ /*! \brief Enable fine-grained system SVM.
+ *
+ * \note It is only possible to enable fine-grained system SVM if all devices
+ * in the context associated with kernel support it.
+ *
+ * \param svmEnabled True if fine-grained system SVM is requested. False otherwise.
+ * \return CL_SUCCESS if the function was executed succesfully. CL_INVALID_OPERATION
+ * if no devices in the context support fine-grained system SVM.
+ *
+ * \see clSetKernelExecInfo
+ */
+ cl_int enableFineGrainedSystemSVM(bool svmEnabled)
+ {
+ cl_bool svmEnabled_ = svmEnabled ? CL_TRUE : CL_FALSE;
+ return detail::errHandler(
+ ::clSetKernelExecInfo(
+ object_,
+ CL_KERNEL_EXEC_INFO_SVM_FINE_GRAIN_SYSTEM,
+ sizeof(cl_bool),
+ &svmEnabled_
+ )
+ );
+ }
+
+ template<int index, int ArrayLength, class D, typename T0, typename T1, typename... Ts>
+ void setSVMPointersHelper(std::array<void*, ArrayLength> &pointerList, const pointer<T0, D> &t0, const pointer<T1, D> &t1, Ts & ... ts)
+ {
+ pointerList[index] = static_cast<void*>(t0.get());
+ setSVMPointersHelper<index + 1, ArrayLength>(pointerList, t1, ts...);
+ }
+
+ template<int index, int ArrayLength, typename T0, typename T1, typename... Ts>
+ typename std::enable_if<std::is_pointer<T0>::value, void>::type
+ setSVMPointersHelper(std::array<void*, ArrayLength> &pointerList, T0 t0, T1 t1, Ts... ts)
+ {
+ pointerList[index] = static_cast<void*>(t0);
+ setSVMPointersHelper<index + 1, ArrayLength>(pointerList, t1, ts...);
+ }
+
+ template<int index, int ArrayLength, typename T0, class D>
+ void setSVMPointersHelper(std::array<void*, ArrayLength> &pointerList, const pointer<T0, D> &t0)
+ {
+ pointerList[index] = static_cast<void*>(t0.get());
+ }
+
+
+ template<int index, int ArrayLength, typename T0>
+ typename std::enable_if<std::is_pointer<T0>::value, void>::type
+ setSVMPointersHelper(std::array<void*, ArrayLength> &pointerList, T0 t0)
+ {
+ pointerList[index] = static_cast<void*>(t0);
+ }
+
+ template<typename T0, typename... Ts>
+ cl_int setSVMPointers(const T0 &t0, Ts & ... ts)
+ {
+ std::array<void*, 1 + sizeof...(Ts)> pointerList;
+
+ setSVMPointersHelper<0, 1 + sizeof...(Ts)>(pointerList, t0, ts...);
+ return detail::errHandler(
+ ::clSetKernelExecInfo(
+ object_,
+ CL_KERNEL_EXEC_INFO_SVM_PTRS,
+ sizeof(void*)*(1 + sizeof...(Ts)),
+ pointerList.data()));
+ }
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+};
+
+/*! \class Program
+ * \brief Program interface that implements cl_program.
+ */
+class Program : public detail::Wrapper<cl_program>
+{
+public:
+#if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ typedef vector<vector<unsigned char>> Binaries;
+ typedef vector<string> Sources;
+#else // #if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ typedef vector<std::pair<const void*, size_type> > Binaries;
+ typedef vector<std::pair<const char*, size_type> > Sources;
+#endif // #if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+
+ Program(
+ const string& source,
+ bool build = false,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ const char * strings = source.c_str();
+ const size_type 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,
+#if !defined(CL_HPP_CL_1_2_DEFAULT_BUILD)
+ "-cl-std=CL2.0",
+#else
+ "",
+#endif // #if !defined(CL_HPP_CL_1_2_DEFAULT_BUILD)
+ NULL,
+ NULL);
+
+ detail::buildErrHandler(error, __BUILD_PROGRAM_ERR, getBuildInfo<CL_PROGRAM_BUILD_LOG>());
+ }
+
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+ Program(
+ const Context& context,
+ const string& source,
+ bool build = false,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ const char * strings = source.c_str();
+ const size_type 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,
+#if !defined(CL_HPP_CL_1_2_DEFAULT_BUILD)
+ "-cl-std=CL2.0",
+#else
+ "",
+#endif // #if !defined(CL_HPP_CL_1_2_DEFAULT_BUILD)
+ NULL,
+ NULL);
+
+ detail::buildErrHandler(error, __BUILD_PROGRAM_ERR, getBuildInfo<CL_PROGRAM_BUILD_LOG>());
+ }
+
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+ /**
+ * Create a program from a vector of source strings and the default context.
+ * Does not compile or link the program.
+ */
+ Program(
+ const Sources& sources,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+ Context context = Context::getDefault(err);
+
+ const size_type n = (size_type)sources.size();
+
+ vector<size_type> lengths(n);
+ vector<const char*> strings(n);
+
+ for (size_type i = 0; i < n; ++i) {
+#if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ strings[i] = sources[(int)i].data();
+ lengths[i] = sources[(int)i].length();
+#else // #if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ strings[i] = sources[(int)i].first;
+ lengths[i] = sources[(int)i].second;
+#endif // #if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ }
+
+ object_ = ::clCreateProgramWithSource(
+ context(), (cl_uint)n, strings.data(), lengths.data(), &error);
+
+ detail::errHandler(error, __CREATE_PROGRAM_WITH_SOURCE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+ /**
+ * Create a program from a vector of source strings and a provided context.
+ * Does not compile or link the program.
+ */
+ Program(
+ const Context& context,
+ const Sources& sources,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ const size_type n = (size_type)sources.size();
+
+ vector<size_type> lengths(n);
+ vector<const char*> strings(n);
+
+ for (size_type i = 0; i < n; ++i) {
+#if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ strings[i] = sources[(int)i].data();
+ lengths[i] = sources[(int)i].length();
+#else // #if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ strings[i] = sources[(int)i].first;
+ lengths[i] = sources[(int)i].second;
+#endif // #if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ }
+
+ object_ = ::clCreateProgramWithSource(
+ context(), (cl_uint)n, strings.data(), lengths.data(), &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<Device>& devices,
+ const Binaries& binaries,
+ vector<cl_int>* binaryStatus = NULL,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ const size_type 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;
+ }
+
+
+ vector<size_type> lengths(numDevices);
+ vector<const unsigned char*> images(numDevices);
+#if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ for (size_type i = 0; i < numDevices; ++i) {
+ images[i] = binaries[i].data();
+ lengths[i] = binaries[(int)i].size();
+ }
+#else // #if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+ for (size_type i = 0; i < numDevices; ++i) {
+ images[i] = (const unsigned char*)binaries[i].first;
+ lengths[i] = binaries[(int)i].second;
+ }
+#endif // #if !defined(CL_HPP_ENABLE_PROGRAM_CONSTRUCTION_FROM_ARRAY_COMPATIBILITY)
+
+ vector<cl_device_id> deviceIDs(numDevices);
+ for( size_type deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) {
+ deviceIDs[deviceIndex] = (devices[deviceIndex])();
+ }
+
+ if(binaryStatus) {
+ binaryStatus->resize(numDevices);
+ }
+
+ object_ = ::clCreateProgramWithBinary(
+ context(), (cl_uint) devices.size(),
+ deviceIDs.data(),
+ lengths.data(), images.data(), (binaryStatus != NULL && numDevices > 0)
+ ? &binaryStatus->front()
+ : NULL, &error);
+
+ detail::errHandler(error, __CREATE_PROGRAM_WITH_BINARY_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ /**
+ * Create program using builtin kernels.
+ * \param kernelNames Semi-colon separated list of builtin kernel names
+ */
+ Program(
+ const Context& context,
+ const vector<Device>& devices,
+ const string& kernelNames,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+
+ size_type numDevices = devices.size();
+ vector<cl_device_id> deviceIDs(numDevices);
+ for( size_type deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) {
+ deviceIDs[deviceIndex] = (devices[deviceIndex])();
+ }
+
+ object_ = ::clCreateProgramWithBuiltInKernels(
+ context(),
+ (cl_uint) devices.size(),
+ deviceIDs.data(),
+ kernelNames.c_str(),
+ &error);
+
+ detail::errHandler(error, __CREATE_PROGRAM_WITH_BUILT_IN_KERNELS_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+ Program() { }
+
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ */
+ explicit Program(const cl_program& program, bool retainObject = false) :
+ detail::Wrapper<cl_type>(program, retainObject) { }
+
+ Program& operator = (const cl_program& rhs)
+ {
+ detail::Wrapper<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Program(const Program& program) : detail::Wrapper<cl_type>(program) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ Program& operator = (const Program &program)
+ {
+ detail::Wrapper<cl_type>::operator=(program);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Program(Program&& program) CL_HPP_NOEXCEPT_ : detail::Wrapper<cl_type>(std::move(program)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ Program& operator = (Program &&program)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(program));
+ return *this;
+ }
+
+ cl_int build(
+ const vector<Device>& devices,
+ const char* options = NULL,
+ void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL,
+ void* data = NULL) const
+ {
+ size_type numDevices = devices.size();
+ vector<cl_device_id> deviceIDs(numDevices);
+
+ for( size_type deviceIndex = 0; deviceIndex < numDevices; ++deviceIndex ) {
+ deviceIDs[deviceIndex] = (devices[deviceIndex])();
+ }
+
+ cl_int buildError = ::clBuildProgram(
+ object_,
+ (cl_uint)
+ devices.size(),
+ deviceIDs.data(),
+ options,
+ notifyFptr,
+ data);
+
+ return detail::buildErrHandler(buildError, __BUILD_PROGRAM_ERR, getBuildInfo<CL_PROGRAM_BUILD_LOG>());
+ }
+
+ cl_int build(
+ const char* options = NULL,
+ void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL,
+ void* data = NULL) const
+ {
+ cl_int buildError = ::clBuildProgram(
+ object_,
+ 0,
+ NULL,
+ options,
+ notifyFptr,
+ data);
+
+
+ return detail::buildErrHandler(buildError, __BUILD_PROGRAM_ERR, getBuildInfo<CL_PROGRAM_BUILD_LOG>());
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ cl_int compile(
+ const char* options = NULL,
+ void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL,
+ void* data = NULL) const
+ {
+ cl_int error = ::clCompileProgram(
+ object_,
+ 0,
+ NULL,
+ options,
+ 0,
+ NULL,
+ NULL,
+ notifyFptr,
+ data);
+ return detail::buildErrHandler(error, __COMPILE_PROGRAM_ERR, getBuildInfo<CL_PROGRAM_BUILD_LOG>());
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+ template <typename T>
+ cl_int getInfo(cl_program_info name, T* param) const
+ {
+ return detail::errHandler(
+ detail::getInfo(&::clGetProgramInfo, object_, name, param),
+ __GET_PROGRAM_INFO_ERR);
+ }
+
+ template <cl_int name> typename
+ detail::param_traits<detail::cl_program_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_program_build_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+ /**
+ * Build info function that returns a vector of device/info pairs for the specified
+ * info type and for all devices in the program.
+ * On an error reading the info for any device, an empty vector of info will be returned.
+ */
+ template <cl_int name>
+ vector<std::pair<cl::Device, typename detail::param_traits<detail::cl_program_build_info, name>::param_type>>
+ getBuildInfo(cl_int *err = NULL) const
+ {
+ cl_int result = CL_SUCCESS;
+
+ auto devs = getInfo<CL_PROGRAM_DEVICES>(&result);
+ vector<std::pair<cl::Device, typename detail::param_traits<detail::cl_program_build_info, name>::param_type>>
+ devInfo;
+
+ // If there was an initial error from getInfo return the error
+ if (result != CL_SUCCESS) {
+ if (err != NULL) {
+ *err = result;
+ }
+ return devInfo;
+ }
+
+ for (const cl::Device &d : devs) {
+ typename detail::param_traits<
+ detail::cl_program_build_info, name>::param_type param;
+ result = getBuildInfo(d, name, &param);
+ devInfo.push_back(
+ std::pair<cl::Device, typename detail::param_traits<detail::cl_program_build_info, name>::param_type>
+ (d, param));
+ if (result != CL_SUCCESS) {
+ // On error, leave the loop and return the error code
+ break;
+ }
+ }
+ if (err != NULL) {
+ *err = result;
+ }
+ if (result != CL_SUCCESS) {
+ devInfo.clear();
+ }
+ return devInfo;
+ }
+
+ cl_int createKernels(vector<Kernel>* 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);
+ }
+
+ vector<cl_kernel> value(numKernels);
+
+ err = ::clCreateKernelsInProgram(
+ object_, numKernels, value.data(), NULL);
+ if (err != CL_SUCCESS) {
+ return detail::errHandler(err, __CREATE_KERNELS_IN_PROGRAM_ERR);
+ }
+
+ if (kernels) {
+ kernels->resize(value.size());
+
+ // Assign to param, constructing with retain behaviour
+ // to correctly capture each underlying CL object
+ for (size_type i = 0; i < value.size(); i++) {
+ // We do not need to retain because this kernel is being created
+ // by the runtime
+ (*kernels)[i] = Kernel(value[i], false);
+ }
+ }
+ return CL_SUCCESS;
+ }
+};
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+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 error_local = CL_SUCCESS;
+
+ cl_program programs[2] = { input1(), input2() };
+
+ Context ctx = input1.getInfo<CL_PROGRAM_CONTEXT>(&error_local);
+ if(error_local!=CL_SUCCESS) {
+ detail::errHandler(error_local, __LINK_PROGRAM_ERR);
+ }
+
+ cl_program prog = ::clLinkProgram(
+ ctx(),
+ 0,
+ NULL,
+ options,
+ 2,
+ programs,
+ notifyFptr,
+ data,
+ &error_local);
+
+ detail::errHandler(error_local,__COMPILE_PROGRAM_ERR);
+ if (err != NULL) {
+ *err = error_local;
+ }
+
+ return Program(prog);
+}
+
+inline Program linkProgram(
+ vector<Program> inputPrograms,
+ const char* options = NULL,
+ void (CL_CALLBACK * notifyFptr)(cl_program, void *) = NULL,
+ void* data = NULL,
+ cl_int* err = NULL)
+{
+ cl_int error_local = CL_SUCCESS;
+
+ vector<cl_program> programs(inputPrograms.size());
+
+ for (unsigned int i = 0; i < inputPrograms.size(); i++) {
+ programs[i] = inputPrograms[i]();
+ }
+
+ Context ctx;
+ if(inputPrograms.size() > 0) {
+ ctx = inputPrograms[0].getInfo<CL_PROGRAM_CONTEXT>(&error_local);
+ if(error_local!=CL_SUCCESS) {
+ detail::errHandler(error_local, __LINK_PROGRAM_ERR);
+ }
+ }
+ cl_program prog = ::clLinkProgram(
+ ctx(),
+ 0,
+ NULL,
+ options,
+ (cl_uint)inputPrograms.size(),
+ programs.data(),
+ notifyFptr,
+ data,
+ &error_local);
+
+ detail::errHandler(error_local,__COMPILE_PROGRAM_ERR);
+ if (err != NULL) {
+ *err = error_local;
+ }
+
+ return Program(prog, false);
+}
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+// Template specialization for CL_PROGRAM_BINARIES
+template <>
+inline cl_int cl::Program::getInfo(cl_program_info name, vector<vector<unsigned char>>* param) const
+{
+ if (name != CL_PROGRAM_BINARIES) {
+ return CL_INVALID_VALUE;
+ }
+ if (param) {
+ // Resize the parameter array appropriately for each allocation
+ // and pass down to the helper
+
+ vector<size_type> sizes = getInfo<CL_PROGRAM_BINARY_SIZES>();
+ size_type numBinaries = sizes.size();
+
+ // Resize the parameter array and constituent arrays
+ param->resize(numBinaries);
+ for (size_type i = 0; i < numBinaries; ++i) {
+ (*param)[i].resize(sizes[i]);
+ }
+
+ return detail::errHandler(
+ detail::getInfo(&::clGetProgramInfo, object_, name, param),
+ __GET_PROGRAM_INFO_ERR);
+ }
+
+ return CL_SUCCESS;
+}
+
+template<>
+inline vector<vector<unsigned char>> cl::Program::getInfo<CL_PROGRAM_BINARIES>(cl_int* err) const
+{
+ vector<vector<unsigned char>> binariesVectors;
+
+ cl_int result = getInfo(CL_PROGRAM_BINARIES, &binariesVectors);
+ if (err != NULL) {
+ *err = result;
+ }
+ return binariesVectors;
+}
+
+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;
+ }
+
+}
+
+enum class QueueProperties : cl_command_queue_properties
+{
+ None = 0,
+ Profiling = CL_QUEUE_PROFILING_ENABLE,
+ OutOfOrder = CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE,
+};
+
+inline QueueProperties operator|(QueueProperties lhs, QueueProperties rhs)
+{
+ return static_cast<QueueProperties>(static_cast<cl_command_queue_properties>(lhs) | static_cast<cl_command_queue_properties>(rhs));
+}
+
+/*! \class CommandQueue
+ * \brief CommandQueue interface for cl_command_queue.
+ */
+class CommandQueue : public detail::Wrapper<cl_command_queue>
+{
+private:
+ static std::once_flag default_initialized_;
+ static CommandQueue default_;
+ static cl_int default_error_;
+
+ /*! \brief Create the default command queue returned by @ref getDefault.
+ *
+ * It sets default_error_ to indicate success or failure. It does not throw
+ * @c cl::Error.
+ */
+ static void makeDefault()
+ {
+ /* We don't want to throw an error from this function, so we have to
+ * catch and set the error flag.
+ */
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ try
+#endif
+ {
+ int error;
+ Context context = Context::getDefault(&error);
+
+ if (error != CL_SUCCESS) {
+ default_error_ = error;
+ }
+ else {
+ Device device = Device::getDefault();
+ default_ = CommandQueue(context, device, 0, &default_error_);
+ }
+ }
+#if defined(CL_HPP_ENABLE_EXCEPTIONS)
+ catch (cl::Error &e) {
+ default_error_ = e.err();
+ }
+#endif
+ }
+
+ /*! \brief Create the default command queue.
+ *
+ * This sets @c default_. It does not throw
+ * @c cl::Error.
+ */
+ static void makeDefaultProvided(const CommandQueue &c) {
+ default_ = c;
+ }
+
+public:
+#ifdef CL_HPP_UNIT_TEST_ENABLE
+ /*! \brief Reset the default.
+ *
+ * This sets @c default_ to an empty value to support cleanup in
+ * the unit test framework.
+ * This function is not thread safe.
+ */
+ static void unitTestClearDefault() {
+ default_ = CommandQueue();
+ }
+#endif // #ifdef CL_HPP_UNIT_TEST_ENABLE
+
+
+ /*!
+ * \brief Constructs a CommandQueue based on passed properties.
+ * Will return an CL_INVALID_QUEUE_PROPERTIES error if CL_QUEUE_ON_DEVICE is specified.
+ */
+ CommandQueue(
+ cl_command_queue_properties properties,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ Context context = Context::getDefault(&error);
+ detail::errHandler(error, __CREATE_CONTEXT_ERR);
+
+ if (error != CL_SUCCESS) {
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+ else {
+ Device device = context.getInfo<CL_CONTEXT_DEVICES>()[0];
+ bool useWithProperties;
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200 && CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ // Run-time decision based on the actual platform
+ {
+ cl_uint version = detail::getContextPlatformVersion(context());
+ useWithProperties = (version >= 0x20000); // OpenCL 2.0 or above
+ }
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 200
+ useWithProperties = true;
+#else
+ useWithProperties = false;
+#endif
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ if (useWithProperties) {
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, properties, 0 };
+ if ((properties & CL_QUEUE_ON_DEVICE) == 0) {
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error);
+ }
+ else {
+ error = CL_INVALID_QUEUE_PROPERTIES;
+ }
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ if (!useWithProperties) {
+ object_ = ::clCreateCommandQueue(
+ context(), device(), properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ }
+ }
+
+ /*!
+ * \brief Constructs a CommandQueue based on passed properties.
+ * Will return an CL_INVALID_QUEUE_PROPERTIES error if CL_QUEUE_ON_DEVICE is specified.
+ */
+ CommandQueue(
+ QueueProperties properties,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ Context context = Context::getDefault(&error);
+ detail::errHandler(error, __CREATE_CONTEXT_ERR);
+
+ if (error != CL_SUCCESS) {
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+ else {
+ Device device = context.getInfo<CL_CONTEXT_DEVICES>()[0];
+ bool useWithProperties;
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200 && CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ // Run-time decision based on the actual platform
+ {
+ cl_uint version = detail::getContextPlatformVersion(context());
+ useWithProperties = (version >= 0x20000); // OpenCL 2.0 or above
+ }
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 200
+ useWithProperties = true;
+#else
+ useWithProperties = false;
+#endif
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ if (useWithProperties) {
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, static_cast<cl_queue_properties>(properties), 0 };
+
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ if (!useWithProperties) {
+ object_ = ::clCreateCommandQueue(
+ context(), device(), static_cast<cl_command_queue_properties>(properties), &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 200
+
+ }
+ }
+
+ /*!
+ * \brief Constructs a CommandQueue for an implementation defined device in the given context
+ * Will return an CL_INVALID_QUEUE_PROPERTIES error if CL_QUEUE_ON_DEVICE is specified.
+ */
+ explicit CommandQueue(
+ const Context& context,
+ cl_command_queue_properties properties = 0,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+ bool useWithProperties;
+ vector<cl::Device> devices;
+ error = context.getInfo(CL_CONTEXT_DEVICES, &devices);
+
+ detail::errHandler(error, __CREATE_CONTEXT_ERR);
+
+ if (error != CL_SUCCESS)
+ {
+ if (err != NULL) {
+ *err = error;
+ }
+ return;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200 && CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ // Run-time decision based on the actual platform
+ {
+ cl_uint version = detail::getContextPlatformVersion(context());
+ useWithProperties = (version >= 0x20000); // OpenCL 2.0 or above
+ }
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 200
+ useWithProperties = true;
+#else
+ useWithProperties = false;
+#endif
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ if (useWithProperties) {
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, properties, 0 };
+ if ((properties & CL_QUEUE_ON_DEVICE) == 0) {
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), devices[0](), queue_properties, &error);
+ }
+ else {
+ error = CL_INVALID_QUEUE_PROPERTIES;
+ }
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ if (!useWithProperties) {
+ object_ = ::clCreateCommandQueue(
+ context(), devices[0](), properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ }
+
+ /*!
+ * \brief Constructs a CommandQueue for an implementation defined device in the given context
+ * Will return an CL_INVALID_QUEUE_PROPERTIES error if CL_QUEUE_ON_DEVICE is specified.
+ */
+ explicit CommandQueue(
+ const Context& context,
+ QueueProperties properties,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+ bool useWithProperties;
+ vector<cl::Device> devices;
+ error = context.getInfo(CL_CONTEXT_DEVICES, &devices);
+
+ detail::errHandler(error, __CREATE_CONTEXT_ERR);
+
+ if (error != CL_SUCCESS)
+ {
+ if (err != NULL) {
+ *err = error;
+ }
+ return;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200 && CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ // Run-time decision based on the actual platform
+ {
+ cl_uint version = detail::getContextPlatformVersion(context());
+ useWithProperties = (version >= 0x20000); // OpenCL 2.0 or above
+ }
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 200
+ useWithProperties = true;
+#else
+ useWithProperties = false;
+#endif
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ if (useWithProperties) {
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, static_cast<cl_queue_properties>(properties), 0 };
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), devices[0](), queue_properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ if (!useWithProperties) {
+ object_ = ::clCreateCommandQueue(
+ context(), devices[0](), static_cast<cl_command_queue_properties>(properties), &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ }
+
+ /*!
+ * \brief Constructs a CommandQueue for a passed device and context
+ * Will return an CL_INVALID_QUEUE_PROPERTIES error if CL_QUEUE_ON_DEVICE is specified.
+ */
+ CommandQueue(
+ const Context& context,
+ const Device& device,
+ cl_command_queue_properties properties = 0,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+ bool useWithProperties;
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200 && CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ // Run-time decision based on the actual platform
+ {
+ cl_uint version = detail::getContextPlatformVersion(context());
+ useWithProperties = (version >= 0x20000); // OpenCL 2.0 or above
+ }
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 200
+ useWithProperties = true;
+#else
+ useWithProperties = false;
+#endif
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ if (useWithProperties) {
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, properties, 0 };
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ if (!useWithProperties) {
+ object_ = ::clCreateCommandQueue(
+ context(), device(), properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ }
+
+ /*!
+ * \brief Constructs a CommandQueue for a passed device and context
+ * Will return an CL_INVALID_QUEUE_PROPERTIES error if CL_QUEUE_ON_DEVICE is specified.
+ */
+ CommandQueue(
+ const Context& context,
+ const Device& device,
+ QueueProperties properties,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+ bool useWithProperties;
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200 && CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ // Run-time decision based on the actual platform
+ {
+ cl_uint version = detail::getContextPlatformVersion(context());
+ useWithProperties = (version >= 0x20000); // OpenCL 2.0 or above
+ }
+#elif CL_HPP_TARGET_OPENCL_VERSION >= 200
+ useWithProperties = true;
+#else
+ useWithProperties = false;
+#endif
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ if (useWithProperties) {
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, static_cast<cl_queue_properties>(properties), 0 };
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+#if CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ if (!useWithProperties) {
+ object_ = ::clCreateCommandQueue(
+ context(), device(), static_cast<cl_command_queue_properties>(properties), &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+#endif // CL_HPP_MINIMUM_OPENCL_VERSION < 200
+ }
+
+ static CommandQueue getDefault(cl_int * err = NULL)
+ {
+ std::call_once(default_initialized_, makeDefault);
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ detail::errHandler(default_error_, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+#else // CL_HPP_TARGET_OPENCL_VERSION >= 200
+ detail::errHandler(default_error_, __CREATE_COMMAND_QUEUE_ERR);
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 200
+ if (err != NULL) {
+ *err = default_error_;
+ }
+ return default_;
+ }
+
+ /**
+ * Modify the default command queue to be used by
+ * subsequent operations.
+ * Will only set the default if no default was previously created.
+ * @return updated default command queue.
+ * Should be compared to the passed value to ensure that it was updated.
+ */
+ static CommandQueue setDefault(const CommandQueue &default_queue)
+ {
+ std::call_once(default_initialized_, makeDefaultProvided, std::cref(default_queue));
+ detail::errHandler(default_error_);
+ return default_;
+ }
+
+ CommandQueue() { }
+
+
+ /*! \brief Constructor from cl_mem - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ */
+ explicit CommandQueue(const cl_command_queue& commandQueue, bool retainObject = false) :
+ detail::Wrapper<cl_type>(commandQueue, retainObject) { }
+
+ CommandQueue& operator = (const cl_command_queue& rhs)
+ {
+ detail::Wrapper<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ CommandQueue(const CommandQueue& queue) : detail::Wrapper<cl_type>(queue) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ CommandQueue& operator = (const CommandQueue &queue)
+ {
+ detail::Wrapper<cl_type>::operator=(queue);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ CommandQueue(CommandQueue&& queue) CL_HPP_NOEXCEPT_ : detail::Wrapper<cl_type>(std::move(queue)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ CommandQueue& operator = (CommandQueue &&queue)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(queue));
+ return *this;
+ }
+
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_command_queue_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+ cl_int enqueueReadBuffer(
+ const Buffer& buffer,
+ cl_bool blocking,
+ size_type offset,
+ size_type size,
+ void* ptr,
+ const vector<Event>* 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_type offset,
+ size_type size,
+ const void* ptr,
+ const vector<Event>* 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_type src_offset,
+ size_type dst_offset,
+ size_type size,
+ const vector<Event>* 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;
+ }
+#if CL_HPP_TARGET_OPENCL_VERSION >= 110
+ cl_int enqueueReadBufferRect(
+ const Buffer& buffer,
+ cl_bool blocking,
+ const array<size_type, 3>& buffer_offset,
+ const array<size_type, 3>& host_offset,
+ const array<size_type, 3>& region,
+ size_type buffer_row_pitch,
+ size_type buffer_slice_pitch,
+ size_type host_row_pitch,
+ size_type host_slice_pitch,
+ void *ptr,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueReadBufferRect(
+ object_,
+ buffer(),
+ blocking,
+ buffer_offset.data(),
+ host_offset.data(),
+ region.data(),
+ 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 array<size_type, 3>& buffer_offset,
+ const array<size_type, 3>& host_offset,
+ const array<size_type, 3>& region,
+ size_type buffer_row_pitch,
+ size_type buffer_slice_pitch,
+ size_type host_row_pitch,
+ size_type host_slice_pitch,
+ const void *ptr,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueWriteBufferRect(
+ object_,
+ buffer(),
+ blocking,
+ buffer_offset.data(),
+ host_offset.data(),
+ region.data(),
+ 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 array<size_type, 3>& src_origin,
+ const array<size_type, 3>& dst_origin,
+ const array<size_type, 3>& region,
+ size_type src_row_pitch,
+ size_type src_slice_pitch,
+ size_type dst_row_pitch,
+ size_type dst_slice_pitch,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueCopyBufferRect(
+ object_,
+ src(),
+ dst(),
+ src_origin.data(),
+ dst_origin.data(),
+ region.data(),
+ 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;
+ }
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 110
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ /**
+ * Enqueue a command to fill a buffer object with a pattern
+ * of a given size. The pattern is specified as a vector type.
+ * \tparam PatternType The datatype of the pattern field.
+ * The pattern type must be an accepted OpenCL data type.
+ * \tparam offset Is the offset in bytes into the buffer at
+ * which to start filling. This must be a multiple of
+ * the pattern size.
+ * \tparam size Is the size in bytes of the region to fill.
+ * This must be a multiple of the pattern size.
+ */
+ template<typename PatternType>
+ cl_int enqueueFillBuffer(
+ const Buffer& buffer,
+ PatternType pattern,
+ size_type offset,
+ size_type size,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueFillBuffer(
+ object_,
+ buffer(),
+ static_cast<void*>(&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 // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+ cl_int enqueueReadImage(
+ const Image& image,
+ cl_bool blocking,
+ const array<size_type, 3>& origin,
+ const array<size_type, 3>& region,
+ size_type row_pitch,
+ size_type slice_pitch,
+ void* ptr,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueReadImage(
+ object_,
+ image(),
+ blocking,
+ origin.data(),
+ region.data(),
+ 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 array<size_type, 3>& origin,
+ const array<size_type, 3>& region,
+ size_type row_pitch,
+ size_type slice_pitch,
+ const void* ptr,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueWriteImage(
+ object_,
+ image(),
+ blocking,
+ origin.data(),
+ region.data(),
+ 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 array<size_type, 3>& src_origin,
+ const array<size_type, 3>& dst_origin,
+ const array<size_type, 3>& region,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueCopyImage(
+ object_,
+ src(),
+ dst(),
+ src_origin.data(),
+ dst_origin.data(),
+ region.data(),
+ (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 CL_HPP_TARGET_OPENCL_VERSION >= 120
+ /**
+ * 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 array<size_type, 3>& origin,
+ const array<size_type, 3>& region,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueFillImage(
+ object_,
+ image(),
+ static_cast<void*>(&fillColor),
+ origin.data(),
+ region.data(),
+ (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 array<size_type, 3>& origin,
+ const array<size_type, 3>& region,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueFillImage(
+ object_,
+ image(),
+ static_cast<void*>(&fillColor),
+ origin.data(),
+ region.data(),
+ (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 array<size_type, 3>& origin,
+ const array<size_type, 3>& region,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueFillImage(
+ object_,
+ image(),
+ static_cast<void*>(&fillColor),
+ origin.data(),
+ region.data(),
+ (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 // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+ cl_int enqueueCopyImageToBuffer(
+ const Image& src,
+ const Buffer& dst,
+ const array<size_type, 3>& src_origin,
+ const array<size_type, 3>& region,
+ size_type dst_offset,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueCopyImageToBuffer(
+ object_,
+ src(),
+ dst(),
+ src_origin.data(),
+ region.data(),
+ 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_type src_offset,
+ const array<size_type, 3>& dst_origin,
+ const array<size_type, 3>& region,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueCopyBufferToImage(
+ object_,
+ src(),
+ dst(),
+ src_offset,
+ dst_origin.data(),
+ region.data(),
+ (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_type offset,
+ size_type size,
+ const vector<Event>* events = NULL,
+ Event* event = NULL,
+ cl_int* err = NULL) const
+ {
+ cl_event tmp;
+ 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,
+ (event != NULL) ? &tmp : NULL,
+ &error);
+
+ detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ if (event != NULL && error == CL_SUCCESS)
+ *event = tmp;
+
+ return result;
+ }
+
+ void* enqueueMapImage(
+ const Image& buffer,
+ cl_bool blocking,
+ cl_map_flags flags,
+ const array<size_type, 3>& origin,
+ const array<size_type, 3>& region,
+ size_type * row_pitch,
+ size_type * slice_pitch,
+ const vector<Event>* events = NULL,
+ Event* event = NULL,
+ cl_int* err = NULL) const
+ {
+ cl_event tmp;
+ cl_int error;
+ void * result = ::clEnqueueMapImage(
+ object_, buffer(), blocking, flags,
+ origin.data(),
+ region.data(),
+ row_pitch, slice_pitch,
+ (events != NULL) ? (cl_uint) events->size() : 0,
+ (events != NULL && events->size() > 0) ? (cl_event*) &events->front() : NULL,
+ (event != NULL) ? &tmp : NULL,
+ &error);
+
+ detail::errHandler(error, __ENQUEUE_MAP_IMAGE_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ if (event != NULL && error == CL_SUCCESS)
+ *event = tmp;
+ return result;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ /**
+ * Enqueues a command that will allow the host to update a region of a coarse-grained SVM buffer.
+ * This variant takes a raw SVM pointer.
+ */
+ template<typename T>
+ cl_int enqueueMapSVM(
+ T* ptr,
+ cl_bool blocking,
+ cl_map_flags flags,
+ size_type size,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(::clEnqueueSVMMap(
+ object_, blocking, flags, static_cast<void*>(ptr), size,
+ (events != NULL) ? (cl_uint)events->size() : 0,
+ (events != NULL && events->size() > 0) ? (cl_event*)&events->front() : NULL,
+ (event != NULL) ? &tmp : NULL),
+ __ENQUEUE_MAP_BUFFER_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
+ }
+
+
+ /**
+ * Enqueues a command that will allow the host to update a region of a coarse-grained SVM buffer.
+ * This variant takes a cl::pointer instance.
+ */
+ template<typename T, class D>
+ cl_int enqueueMapSVM(
+ cl::pointer<T, D> &ptr,
+ cl_bool blocking,
+ cl_map_flags flags,
+ size_type size,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(::clEnqueueSVMMap(
+ object_, blocking, flags, static_cast<void*>(ptr.get()), size,
+ (events != NULL) ? (cl_uint)events->size() : 0,
+ (events != NULL && events->size() > 0) ? (cl_event*)&events->front() : NULL,
+ (event != NULL) ? &tmp : NULL),
+ __ENQUEUE_MAP_BUFFER_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
+ }
+
+ /**
+ * Enqueues a command that will allow the host to update a region of a coarse-grained SVM buffer.
+ * This variant takes a cl::vector instance.
+ */
+ template<typename T, class Alloc>
+ cl_int enqueueMapSVM(
+ cl::vector<T, Alloc> &container,
+ cl_bool blocking,
+ cl_map_flags flags,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(::clEnqueueSVMMap(
+ object_, blocking, flags, static_cast<void*>(container.data()), container.size(),
+ (events != NULL) ? (cl_uint)events->size() : 0,
+ (events != NULL && events->size() > 0) ? (cl_event*)&events->front() : NULL,
+ (event != NULL) ? &tmp : NULL),
+ __ENQUEUE_MAP_BUFFER_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
+ }
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+ cl_int enqueueUnmapMemObject(
+ const Memory& memory,
+ void* mapped_ptr,
+ const vector<Event>* 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 CL_HPP_TARGET_OPENCL_VERSION >= 200
+ /**
+ * Enqueues a command that will release a coarse-grained SVM buffer back to the OpenCL runtime.
+ * This variant takes a raw SVM pointer.
+ */
+ template<typename T>
+ cl_int enqueueUnmapSVM(
+ T* ptr,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueSVMUnmap(
+ object_, static_cast<void*>(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;
+ }
+
+ /**
+ * Enqueues a command that will release a coarse-grained SVM buffer back to the OpenCL runtime.
+ * This variant takes a cl::pointer instance.
+ */
+ template<typename T, class D>
+ cl_int enqueueUnmapSVM(
+ cl::pointer<T, D> &ptr,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueSVMUnmap(
+ object_, static_cast<void*>(ptr.get()),
+ (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;
+ }
+
+ /**
+ * Enqueues a command that will release a coarse-grained SVM buffer back to the OpenCL runtime.
+ * This variant takes a cl::vector instance.
+ */
+ template<typename T, class Alloc>
+ cl_int enqueueUnmapSVM(
+ cl::vector<T, Alloc> &container,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueSVMUnmap(
+ object_, static_cast<void*>(container.data()),
+ (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 CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ /**
+ * 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<Event> *events = 0,
+ Event *event = 0) const
+ {
+ 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<Event> *events = 0,
+ Event *event = 0) const
+ {
+ 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<Memory> &memObjects,
+ cl_mem_migration_flags flags,
+ const vector<Event>* events = NULL,
+ Event* event = NULL
+ ) const
+ {
+ cl_event tmp;
+
+ vector<cl_mem> localMemObjects(memObjects.size());
+
+ for( int i = 0; i < (int)memObjects.size(); ++i ) {
+ localMemObjects[i] = memObjects[i]();
+ }
+
+
+ cl_int err = detail::errHandler(
+ ::clEnqueueMigrateMemObjects(
+ object_,
+ (cl_uint)memObjects.size(),
+ localMemObjects.data(),
+ 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 // CL_HPP_TARGET_OPENCL_VERSION >= 120
+
+ cl_int enqueueNDRangeKernel(
+ const Kernel& kernel,
+ const NDRange& offset,
+ const NDRange& global,
+ const NDRange& local = NullRange,
+ const vector<Event>* 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_type*) offset : NULL,
+ (const size_type*) global,
+ local.dimensions() != 0 ? (const size_type*) 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;
+ }
+
+#if defined(CL_USE_DEPRECATED_OPENCL_1_2_APIS)
+ CL_EXT_PREFIX__VERSION_1_2_DEPRECATED cl_int enqueueTask(
+ const Kernel& kernel,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED
+ {
+ 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;
+ }
+#endif // #if defined(CL_USE_DEPRECATED_OPENCL_1_2_APIS)
+
+ cl_int enqueueNativeKernel(
+ void (CL_CALLBACK *userFptr)(void *),
+ std::pair<void*, size_type> args,
+ const vector<Memory>* mem_objects = NULL,
+ const vector<const void*>* mem_locs = NULL,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ size_type elements = 0;
+ if (mem_objects != NULL) {
+ elements = mem_objects->size();
+ }
+ vector<cl_mem> mems(elements);
+ for (unsigned int i = 0; i < elements; 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.data(),
+ (mem_locs != NULL && mem_locs->size() > 0) ? (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)
+ CL_EXT_PREFIX__VERSION_1_1_DEPRECATED
+ cl_int enqueueMarker(Event* event = NULL) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED
+ {
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ ::clEnqueueMarker(
+ object_,
+ (event != NULL) ? &tmp : NULL),
+ __ENQUEUE_MARKER_ERR);
+
+ if (event != NULL && err == CL_SUCCESS)
+ *event = tmp;
+
+ return err;
+ }
+
+ CL_EXT_PREFIX__VERSION_1_1_DEPRECATED
+ cl_int enqueueWaitForEvents(const vector<Event>& events) const CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED
+ {
+ return detail::errHandler(
+ ::clEnqueueWaitForEvents(
+ object_,
+ (cl_uint) events.size(),
+ events.size() > 0 ? (const cl_event*) &events.front() : NULL),
+ __ENQUEUE_WAIT_FOR_EVENTS_ERR);
+ }
+#endif // defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+
+ cl_int enqueueAcquireGLObjects(
+ const vector<Memory>* mem_objects = NULL,
+ const vector<Event>* 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 && mem_objects->size() > 0) ? (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<Memory>* mem_objects = NULL,
+ const vector<Event>* 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 && mem_objects->size() > 0) ? (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 (CL_HPP_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<Memory>* mem_objects = NULL,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ static PFN_clEnqueueAcquireD3D10ObjectsKHR pfn_clEnqueueAcquireD3D10ObjectsKHR = NULL;
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ cl_context context = getInfo<CL_QUEUE_CONTEXT>();
+ cl::Device device(getInfo<CL_QUEUE_DEVICE>());
+ cl_platform_id platform = device.getInfo<CL_DEVICE_PLATFORM>();
+ CL_HPP_INIT_CL_EXT_FCN_PTR_PLATFORM_(platform, clEnqueueAcquireD3D10ObjectsKHR);
+#endif
+#if CL_HPP_TARGET_OPENCL_VERSION >= 110
+ CL_HPP_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 && mem_objects->size() > 0) ? (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<Memory>* mem_objects = NULL,
+ const vector<Event>* events = NULL,
+ Event* event = NULL) const
+ {
+ static PFN_clEnqueueReleaseD3D10ObjectsKHR pfn_clEnqueueReleaseD3D10ObjectsKHR = NULL;
+#if CL_HPP_TARGET_OPENCL_VERSION >= 120
+ cl_context context = getInfo<CL_QUEUE_CONTEXT>();
+ cl::Device device(getInfo<CL_QUEUE_DEVICE>());
+ cl_platform_id platform = device.getInfo<CL_DEVICE_PLATFORM>();
+ CL_HPP_INIT_CL_EXT_FCN_PTR_PLATFORM_(platform, clEnqueueReleaseD3D10ObjectsKHR);
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 120
+#if CL_HPP_TARGET_OPENCL_VERSION >= 110
+ CL_HPP_INIT_CL_EXT_FCN_PTR_(clEnqueueReleaseD3D10ObjectsKHR);
+#endif // CL_HPP_TARGET_OPENCL_VERSION >= 110
+
+ cl_event tmp;
+ cl_int err = detail::errHandler(
+ pfn_clEnqueueReleaseD3D10ObjectsKHR(
+ object_,
+ (mem_objects != NULL) ? (cl_uint) mem_objects->size() : 0,
+ (mem_objects != NULL && mem_objects->size() > 0) ? (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;
+ }
+#endif
+
+/**
+ * Deprecated APIs for 1.2
+ */
+#if defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+ 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 // CL_USE_DEPRECATED_OPENCL_1_1_APIS
+
+ cl_int flush() const
+ {
+ return detail::errHandler(::clFlush(object_), __FLUSH_ERR);
+ }
+
+ cl_int finish() const
+ {
+ return detail::errHandler(::clFinish(object_), __FINISH_ERR);
+ }
+}; // CommandQueue
+
+CL_HPP_DEFINE_STATIC_MEMBER_ std::once_flag CommandQueue::default_initialized_;
+CL_HPP_DEFINE_STATIC_MEMBER_ CommandQueue CommandQueue::default_;
+CL_HPP_DEFINE_STATIC_MEMBER_ cl_int CommandQueue::default_error_ = CL_SUCCESS;
+
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+enum class DeviceQueueProperties : cl_command_queue_properties
+{
+ None = 0,
+ Profiling = CL_QUEUE_PROFILING_ENABLE,
+};
+
+inline DeviceQueueProperties operator|(DeviceQueueProperties lhs, DeviceQueueProperties rhs)
+{
+ return static_cast<DeviceQueueProperties>(static_cast<cl_command_queue_properties>(lhs) | static_cast<cl_command_queue_properties>(rhs));
+}
+
+/*! \class DeviceCommandQueue
+ * \brief DeviceCommandQueue interface for device cl_command_queues.
+ */
+class DeviceCommandQueue : public detail::Wrapper<cl_command_queue>
+{
+public:
+
+ /*!
+ * Trivial empty constructor to create a null queue.
+ */
+ DeviceCommandQueue() { }
+
+ /*!
+ * Default construct device command queue on default context and device
+ */
+ DeviceCommandQueue(DeviceQueueProperties properties, cl_int* err = NULL)
+ {
+ cl_int error;
+ cl::Context context = cl::Context::getDefault();
+ cl::Device device = cl::Device::getDefault();
+
+ cl_command_queue_properties mergedProperties =
+ CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE | CL_QUEUE_ON_DEVICE | static_cast<cl_command_queue_properties>(properties);
+
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, mergedProperties, 0 };
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+ /*!
+ * Create a device command queue for a specified device in the passed context.
+ */
+ DeviceCommandQueue(
+ const Context& context,
+ const Device& device,
+ DeviceQueueProperties properties = DeviceQueueProperties::None,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ cl_command_queue_properties mergedProperties =
+ CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE | CL_QUEUE_ON_DEVICE | static_cast<cl_command_queue_properties>(properties);
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, mergedProperties, 0 };
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+ /*!
+ * Create a device command queue for a specified device in the passed context.
+ */
+ DeviceCommandQueue(
+ const Context& context,
+ const Device& device,
+ cl_uint queueSize,
+ DeviceQueueProperties properties = DeviceQueueProperties::None,
+ cl_int* err = NULL)
+ {
+ cl_int error;
+
+ cl_command_queue_properties mergedProperties =
+ CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE | CL_QUEUE_ON_DEVICE | static_cast<cl_command_queue_properties>(properties);
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, mergedProperties,
+ CL_QUEUE_SIZE, queueSize,
+ 0 };
+ object_ = ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error);
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+ }
+
+ /*! \brief Constructor from cl_command_queue - takes ownership.
+ *
+ * \param retainObject will cause the constructor to retain its cl object.
+ * Defaults to false to maintain compatibility with
+ * earlier versions.
+ */
+ explicit DeviceCommandQueue(const cl_command_queue& commandQueue, bool retainObject = false) :
+ detail::Wrapper<cl_type>(commandQueue, retainObject) { }
+
+ DeviceCommandQueue& operator = (const cl_command_queue& rhs)
+ {
+ detail::Wrapper<cl_type>::operator=(rhs);
+ return *this;
+ }
+
+ /*! \brief Copy constructor to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ DeviceCommandQueue(const DeviceCommandQueue& queue) : detail::Wrapper<cl_type>(queue) {}
+
+ /*! \brief Copy assignment to forward copy to the superclass correctly.
+ * Required for MSVC.
+ */
+ DeviceCommandQueue& operator = (const DeviceCommandQueue &queue)
+ {
+ detail::Wrapper<cl_type>::operator=(queue);
+ return *this;
+ }
+
+ /*! \brief Move constructor to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ DeviceCommandQueue(DeviceCommandQueue&& queue) CL_HPP_NOEXCEPT_ : detail::Wrapper<cl_type>(std::move(queue)) {}
+
+ /*! \brief Move assignment to forward move to the superclass correctly.
+ * Required for MSVC.
+ */
+ DeviceCommandQueue& operator = (DeviceCommandQueue &&queue)
+ {
+ detail::Wrapper<cl_type>::operator=(std::move(queue));
+ return *this;
+ }
+
+ template <typename T>
+ 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 <cl_int name> typename
+ detail::param_traits<detail::cl_command_queue_info, name>::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, &param);
+ if (err != NULL) {
+ *err = result;
+ }
+ return param;
+ }
+
+ /*!
+ * Create a new default device command queue for the default device,
+ * in the default context and of the default size.
+ * If there is already a default queue for the specified device this
+ * function will return the pre-existing queue.
+ */
+ static DeviceCommandQueue makeDefault(
+ cl_int *err = nullptr)
+ {
+ cl_int error;
+ cl::Context context = cl::Context::getDefault();
+ cl::Device device = cl::Device::getDefault();
+
+ cl_command_queue_properties properties =
+ CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE | CL_QUEUE_ON_DEVICE | CL_QUEUE_ON_DEVICE_DEFAULT;
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, properties,
+ 0 };
+ DeviceCommandQueue deviceQueue(
+ ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error));
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+
+ return deviceQueue;
+ }
+
+ /*!
+ * Create a new default device command queue for the specified device
+ * and of the default size.
+ * If there is already a default queue for the specified device this
+ * function will return the pre-existing queue.
+ */
+ static DeviceCommandQueue makeDefault(
+ const Context &context, const Device &device, cl_int *err = nullptr)
+ {
+ cl_int error;
+
+ cl_command_queue_properties properties =
+ CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE | CL_QUEUE_ON_DEVICE | CL_QUEUE_ON_DEVICE_DEFAULT;
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, properties,
+ 0 };
+ DeviceCommandQueue deviceQueue(
+ ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error));
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+
+ return deviceQueue;
+ }
+
+ /*!
+ * Create a new default device command queue for the specified device
+ * and of the requested size in bytes.
+ * If there is already a default queue for the specified device this
+ * function will return the pre-existing queue.
+ */
+ static DeviceCommandQueue makeDefault(
+ const Context &context, const Device &device, cl_uint queueSize, cl_int *err = nullptr)
+ {
+ cl_int error;
+
+ cl_command_queue_properties properties =
+ CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE | CL_QUEUE_ON_DEVICE | CL_QUEUE_ON_DEVICE_DEFAULT;
+ cl_queue_properties queue_properties[] = {
+ CL_QUEUE_PROPERTIES, properties,
+ CL_QUEUE_SIZE, queueSize,
+ 0 };
+ DeviceCommandQueue deviceQueue(
+ ::clCreateCommandQueueWithProperties(
+ context(), device(), queue_properties, &error));
+
+ detail::errHandler(error, __CREATE_COMMAND_QUEUE_WITH_PROPERTIES_ERR);
+ if (err != NULL) {
+ *err = error;
+ }
+
+ return deviceQueue;
+ }
+}; // DeviceCommandQueue
+
+namespace detail
+{
+ // Specialization for device command queue
+ template <>
+ struct KernelArgumentHandler<cl::DeviceCommandQueue, void>
+ {
+ static size_type size(const cl::DeviceCommandQueue&) { return sizeof(cl_command_queue); }
+ static const cl_command_queue* ptr(const cl::DeviceCommandQueue& value) { return &(value()); }
+ };
+} // namespace detail
+
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+
+template< typename IteratorType >
+Buffer::Buffer(
+ const Context &context,
+ IteratorType startIterator,
+ IteratorType endIterator,
+ bool readOnly,
+ bool useHostPtr,
+ cl_int* err)
+{
+ typedef typename std::iterator_traits<IteratorType>::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_type size = sizeof(DataType)*(endIterator - startIterator);
+
+ if( useHostPtr ) {
+ object_ = ::clCreateBuffer(context(), flags, size, static_cast<DataType*>(&*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;
+ }
+ }
+}
+
+template< typename IteratorType >
+Buffer::Buffer(
+ const CommandQueue &queue,
+ IteratorType startIterator,
+ IteratorType endIterator,
+ bool readOnly,
+ bool useHostPtr,
+ cl_int* err)
+{
+ typedef typename std::iterator_traits<IteratorType>::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_type size = sizeof(DataType)*(endIterator - startIterator);
+
+ Context context = queue.getInfo<CL_QUEUE_CONTEXT>();
+
+ if (useHostPtr) {
+ object_ = ::clCreateBuffer(context(), flags, size, static_cast<DataType*>(&*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(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_type offset,
+ size_type size,
+ void* ptr,
+ const vector<Event>* 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_type offset,
+ size_type size,
+ const void* ptr,
+ const vector<Event>* 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_type offset,
+ size_type size,
+ const vector<Event>* 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;
+}
+
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+/**
+ * Enqueues to the default queue a command that will allow the host to
+ * update a region of a coarse-grained SVM buffer.
+ * This variant takes a raw SVM pointer.
+ */
+template<typename T>
+inline cl_int enqueueMapSVM(
+ T* ptr,
+ cl_bool blocking,
+ cl_map_flags flags,
+ size_type size,
+ const vector<Event>* events,
+ Event* event)
+{
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ if (error != CL_SUCCESS) {
+ return detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR);
+ }
+
+ return queue.enqueueMapSVM(
+ ptr, blocking, flags, size, events, event);
+}
+
+/**
+ * Enqueues to the default queue a command that will allow the host to
+ * update a region of a coarse-grained SVM buffer.
+ * This variant takes a cl::pointer instance.
+ */
+template<typename T, class D>
+inline cl_int enqueueMapSVM(
+ cl::pointer<T, D> ptr,
+ cl_bool blocking,
+ cl_map_flags flags,
+ size_type size,
+ const vector<Event>* events = NULL,
+ Event* event = NULL)
+{
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ if (error != CL_SUCCESS) {
+ return detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR);
+ }
+
+ return queue.enqueueMapSVM(
+ ptr, blocking, flags, size, events, event);
+}
+
+/**
+ * Enqueues to the default queue a command that will allow the host to
+ * update a region of a coarse-grained SVM buffer.
+ * This variant takes a cl::vector instance.
+ */
+template<typename T, class Alloc>
+inline cl_int enqueueMapSVM(
+ cl::vector<T, Alloc> container,
+ cl_bool blocking,
+ cl_map_flags flags,
+ const vector<Event>* events = NULL,
+ Event* event = NULL)
+{
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ if (error != CL_SUCCESS) {
+ return detail::errHandler(error, __ENQUEUE_MAP_BUFFER_ERR);
+ }
+
+ return queue.enqueueMapSVM(
+ container, blocking, flags, events, event);
+}
+
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+inline cl_int enqueueUnmapMemObject(
+ const Memory& memory,
+ void* mapped_ptr,
+ const vector<Event>* 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;
+}
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+/**
+ * Enqueues to the default queue a command that will release a coarse-grained
+ * SVM buffer back to the OpenCL runtime.
+ * This variant takes a raw SVM pointer.
+ */
+template<typename T>
+inline cl_int enqueueUnmapSVM(
+ T* ptr,
+ const vector<Event>* events = NULL,
+ Event* event = NULL)
+{
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ if (error != CL_SUCCESS) {
+ return detail::errHandler(error, __ENQUEUE_UNMAP_MEM_OBJECT_ERR);
+ }
+
+ return detail::errHandler(queue.enqueueUnmapSVM(ptr, events, event),
+ __ENQUEUE_UNMAP_MEM_OBJECT_ERR);
+
+}
+
+/**
+ * Enqueues to the default queue a command that will release a coarse-grained
+ * SVM buffer back to the OpenCL runtime.
+ * This variant takes a cl::pointer instance.
+ */
+template<typename T, class D>
+inline cl_int enqueueUnmapSVM(
+ cl::pointer<T, D> &ptr,
+ const vector<Event>* events = NULL,
+ Event* event = NULL)
+{
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ if (error != CL_SUCCESS) {
+ return detail::errHandler(error, __ENQUEUE_UNMAP_MEM_OBJECT_ERR);
+ }
+
+ return detail::errHandler(queue.enqueueUnmapSVM(ptr, events, event),
+ __ENQUEUE_UNMAP_MEM_OBJECT_ERR);
+}
+
+/**
+ * Enqueues to the default queue a command that will release a coarse-grained
+ * SVM buffer back to the OpenCL runtime.
+ * This variant takes a cl::vector instance.
+ */
+template<typename T, class Alloc>
+inline cl_int enqueueUnmapSVM(
+ cl::vector<T, Alloc> &container,
+ const vector<Event>* events = NULL,
+ Event* event = NULL)
+{
+ cl_int error;
+ CommandQueue queue = CommandQueue::getDefault(&error);
+ if (error != CL_SUCCESS) {
+ return detail::errHandler(error, __ENQUEUE_UNMAP_MEM_OBJECT_ERR);
+ }
+
+ return detail::errHandler(queue.enqueueUnmapSVM(container, events, event),
+ __ENQUEUE_UNMAP_MEM_OBJECT_ERR);
+}
+
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+inline cl_int enqueueCopyBuffer(
+ const Buffer& src,
+ const Buffer& dst,
+ size_type src_offset,
+ size_type dst_offset,
+ size_type size,
+ const vector<Event>* 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<IteratorType>::value_type DataType;
+ cl_int error;
+
+ size_type length = endIterator-startIterator;
+ size_type byteLength = length*sizeof(DataType);
+
+ DataType *pointer =
+ static_cast<DataType*>(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<DataType*>(
+ 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<IteratorType>::value_type DataType;
+ cl_int error;
+
+ size_type length = endIterator-startIterator;
+ size_type byteLength = length*sizeof(DataType);
+
+ DataType *pointer =
+ static_cast<DataType*>(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 CL_HPP_TARGET_OPENCL_VERSION >= 200
+/**
+ * Blocking SVM map operation - performs a blocking map underneath.
+ */
+template<typename T, class Alloc>
+inline cl_int mapSVM(cl::vector<T, Alloc> &container)
+{
+ return enqueueMapSVM(container, CL_TRUE, CL_MAP_READ | CL_MAP_WRITE);
+}
+
+/**
+* Blocking SVM map operation - performs a blocking map underneath.
+*/
+template<typename T, class Alloc>
+inline cl_int unmapSVM(cl::vector<T, Alloc> &container)
+{
+ return enqueueUnmapSVM(container);
+}
+
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 110
+inline cl_int enqueueReadBufferRect(
+ const Buffer& buffer,
+ cl_bool blocking,
+ const array<size_type, 3>& buffer_offset,
+ const array<size_type, 3>& host_offset,
+ const array<size_type, 3>& region,
+ size_type buffer_row_pitch,
+ size_type buffer_slice_pitch,
+ size_type host_row_pitch,
+ size_type host_slice_pitch,
+ void *ptr,
+ const vector<Event>* 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 array<size_type, 3>& buffer_offset,
+ const array<size_type, 3>& host_offset,
+ const array<size_type, 3>& region,
+ size_type buffer_row_pitch,
+ size_type buffer_slice_pitch,
+ size_type host_row_pitch,
+ size_type host_slice_pitch,
+ const void *ptr,
+ const vector<Event>* 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 array<size_type, 3>& src_origin,
+ const array<size_type, 3>& dst_origin,
+ const array<size_type, 3>& region,
+ size_type src_row_pitch,
+ size_type src_slice_pitch,
+ size_type dst_row_pitch,
+ size_type dst_slice_pitch,
+ const vector<Event>* 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 // CL_HPP_TARGET_OPENCL_VERSION >= 110
+
+inline cl_int enqueueReadImage(
+ const Image& image,
+ cl_bool blocking,
+ const array<size_type, 3>& origin,
+ const array<size_type, 3>& region,
+ size_type row_pitch,
+ size_type slice_pitch,
+ void* ptr,
+ const vector<Event>* 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 array<size_type, 3>& origin,
+ const array<size_type, 3>& region,
+ size_type row_pitch,
+ size_type slice_pitch,
+ const void* ptr,
+ const vector<Event>* 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 array<size_type, 3>& src_origin,
+ const array<size_type, 3>& dst_origin,
+ const array<size_type, 3>& region,
+ const vector<Event>* 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 array<size_type, 3>& src_origin,
+ const array<size_type, 3>& region,
+ size_type dst_offset,
+ const vector<Event>* 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_type src_offset,
+ const array<size_type, 3>& dst_origin,
+ const array<size_type, 3>& region,
+ const vector<Event>* 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();
+}
+
+class EnqueueArgs
+{
+private:
+ CommandQueue queue_;
+ const NDRange offset_;
+ const NDRange global_;
+ const NDRange local_;
+ vector<Event> events_;
+
+ template<typename... Ts>
+ friend class KernelFunctor;
+
+public:
+ 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<Event> &events, NDRange global) :
+ queue_(CommandQueue::getDefault()),
+ offset_(NullRange),
+ global_(global),
+ local_(NullRange),
+ events_(events)
+ {
+
+ }
+
+ EnqueueArgs(const vector<Event> &events, NDRange global, NDRange local) :
+ queue_(CommandQueue::getDefault()),
+ offset_(NullRange),
+ global_(global),
+ local_(local),
+ events_(events)
+ {
+
+ }
+
+ EnqueueArgs(const vector<Event> &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<Event> &events, NDRange global) :
+ queue_(queue),
+ offset_(NullRange),
+ global_(global),
+ local_(NullRange),
+ events_(events)
+ {
+
+ }
+
+ EnqueueArgs(CommandQueue &queue, const vector<Event> &events, NDRange global, NDRange local) :
+ queue_(queue),
+ offset_(NullRange),
+ global_(global),
+ local_(local),
+ events_(events)
+ {
+
+ }
+
+ EnqueueArgs(CommandQueue &queue, const vector<Event> &events, NDRange offset, NDRange global, NDRange local) :
+ queue_(queue),
+ offset_(offset),
+ global_(global),
+ local_(local),
+ events_(events)
+ {
+
+ }
+};
+
+
+//----------------------------------------------------------------------------------------------
+
+
+/**
+ * Type safe kernel functor.
+ *
+ */
+template<typename... Ts>
+class KernelFunctor
+{
+private:
+ Kernel kernel_;
+
+ template<int index, typename T0, typename... T1s>
+ void setArgs(T0&& t0, T1s&&... t1s)
+ {
+ kernel_.setArg(index, t0);
+ setArgs<index + 1, T1s...>(std::forward<T1s>(t1s)...);
+ }
+
+ template<int index, typename T0>
+ void setArgs(T0&& t0)
+ {
+ kernel_.setArg(index, t0);
+ }
+
+ template<int index>
+ void setArgs()
+ {
+ }
+
+
+public:
+ KernelFunctor(Kernel kernel) : kernel_(kernel)
+ {}
+
+ KernelFunctor(
+ const Program& program,
+ const string name,
+ cl_int * err = NULL) :
+ kernel_(program, name.c_str(), err)
+ {}
+
+ //! \brief Return type of the functor
+ typedef Event result_type;
+
+ /**
+ * Enqueue kernel.
+ * @param args Launch parameters of the kernel.
+ * @param t0... List of kernel arguments based on the template type of the functor.
+ */
+ Event operator() (
+ const EnqueueArgs& args,
+ Ts... ts)
+ {
+ Event event;
+ setArgs<0>(std::forward<Ts>(ts)...);
+
+ args.queue_.enqueueNDRangeKernel(
+ kernel_,
+ args.offset_,
+ args.global_,
+ args.local_,
+ &args.events_,
+ &event);
+
+ return event;
+ }
+
+ /**
+ * Enqueue kernel with support for error code.
+ * @param args Launch parameters of the kernel.
+ * @param t0... List of kernel arguments based on the template type of the functor.
+ * @param error Out parameter returning the error code from the execution.
+ */
+ Event operator() (
+ const EnqueueArgs& args,
+ Ts... ts,
+ cl_int &error)
+ {
+ Event event;
+ setArgs<0>(std::forward<Ts>(ts)...);
+
+ error = args.queue_.enqueueNDRangeKernel(
+ kernel_,
+ args.offset_,
+ args.global_,
+ args.local_,
+ &args.events_,
+ &event);
+
+ return event;
+ }
+
+#if CL_HPP_TARGET_OPENCL_VERSION >= 200
+ cl_int setSVMPointers(const vector<void*> &pointerList)
+ {
+ return kernel_.setSVMPointers(pointerList);
+ }
+
+ template<typename T0, typename... T1s>
+ cl_int setSVMPointers(const T0 &t0, T1s &... ts)
+ {
+ return kernel_.setSVMPointers(t0, ts...);
+ }
+#endif // #if CL_HPP_TARGET_OPENCL_VERSION >= 200
+
+ Kernel getKernel()
+ {
+ return kernel_;
+ }
+};
+
+namespace compatibility {
+ /**
+ * Backward compatibility class to ensure that cl.hpp code works with cl2.hpp.
+ * Please use KernelFunctor directly.
+ */
+ template<typename... Ts>
+ struct make_kernel
+ {
+ typedef KernelFunctor<Ts...> FunctorType;
+
+ FunctorType functor_;
+
+ make_kernel(
+ const Program& program,
+ const string name,
+ cl_int * err = NULL) :
+ functor_(FunctorType(program, name, err))
+ {}
+
+ make_kernel(
+ const Kernel kernel) :
+ functor_(FunctorType(kernel))
+ {}
+
+ //! \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&,
+ Ts...);
+
+ Event operator()(
+ const EnqueueArgs& enqueueArgs,
+ Ts... args)
+ {
+ return functor_(
+ enqueueArgs, args...);
+ }
+ };
+} // namespace compatibility
+
+
+//----------------------------------------------------------------------------------------------------------------------
+
+#undef CL_HPP_ERR_STR_
+#if !defined(CL_HPP_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 __UNLOAD_COMPILER_ERR
+#undef __CREATE_SUB_DEVICES_ERR
+
+#undef __CREATE_PIPE_ERR
+#undef __GET_PIPE_INFO_ERR
+
+#endif //CL_HPP_USER_OVERRIDE_ERROR_STRINGS
+
+// Extensions
+#undef CL_HPP_INIT_CL_EXT_FCN_PTR_
+#undef CL_HPP_INIT_CL_EXT_FCN_PTR_PLATFORM_
+
+#if defined(CL_HPP_USE_CL_DEVICE_FISSION)
+#undef CL_HPP_PARAM_NAME_DEVICE_FISSION_
+#endif // CL_HPP_USE_CL_DEVICE_FISSION
+
+#undef CL_HPP_NOEXCEPT_
+#undef CL_HPP_DEFINE_STATIC_MEMBER_
+
+} // namespace cl
+
+#endif // CL_HPP_
diff --git a/include/CL/cl_d3d10.h b/include/CL/cl_d3d10.h
index b6c90b328d9..d5960a43f72 100644
--- a/include/CL/cl_d3d10.h
+++ b/include/CL/cl_d3d10.h
@@ -1,5 +1,5 @@
/**********************************************************************************
- * Copyright (c) 2008-2012 The Khronos Group Inc.
+ * Copyright (c) 2008-2015 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
diff --git a/include/CL/cl_d3d11.h b/include/CL/cl_d3d11.h
index 2e0a63fd3e0..39f9072398a 100644
--- a/include/CL/cl_d3d11.h
+++ b/include/CL/cl_d3d11.h
@@ -1,5 +1,5 @@
/**********************************************************************************
- * Copyright (c) 2008-2012 The Khronos Group Inc.
+ * Copyright (c) 2008-2015 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
diff --git a/include/CL/cl_dx9_media_sharing.h b/include/CL/cl_dx9_media_sharing.h
index 23f16317869..2729e8b9e89 100644
--- a/include/CL/cl_dx9_media_sharing.h
+++ b/include/CL/cl_dx9_media_sharing.h
@@ -1,5 +1,5 @@
/**********************************************************************************
- * Copyright (c) 2008-2012 The Khronos Group Inc.
+ * Copyright (c) 2008-2015 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -33,7 +38,7 @@
extern "C" {
#endif
-/******************************************************************************
+/******************************************************************************/
/* cl_khr_dx9_media_sharing */
#define cl_khr_dx9_media_sharing 1
diff --git a/include/CL/cl_dx9_media_sharing_intel.h b/include/CL/cl_dx9_media_sharing_intel.h
new file mode 100644
index 00000000000..737e6856490
--- /dev/null
+++ b/include/CL/cl_dx9_media_sharing_intel.h
@@ -0,0 +1,182 @@
+/**********************************************************************************
+ * Copyright (c) 2008-2019 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.
+ *
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
+ * 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.
+ **********************************************************************************/
+/*****************************************************************************\
+
+Copyright (c) 2013-2019 Intel Corporation All Rights Reserved.
+
+THESE MATERIALS ARE PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL OR ITS
+CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY OR TORT (INCLUDING
+NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THESE
+MATERIALS, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+File Name: cl_dx9_media_sharing_intel.h
+
+Abstract:
+
+Notes:
+
+\*****************************************************************************/
+
+#ifndef __OPENCL_CL_DX9_MEDIA_SHARING_INTEL_H
+#define __OPENCL_CL_DX9_MEDIA_SHARING_INTEL_H
+
+#include <CL/cl.h>
+#include <CL/cl_platform.h>
+#include <d3d9.h>
+#include <dxvahd.h>
+#include <wtypes.h>
+#include <d3d9types.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/***************************************
+* cl_intel_dx9_media_sharing extension *
+****************************************/
+
+#define cl_intel_dx9_media_sharing 1
+
+typedef cl_uint cl_dx9_device_source_intel;
+typedef cl_uint cl_dx9_device_set_intel;
+
+/* error codes */
+#define CL_INVALID_DX9_DEVICE_INTEL -1010
+#define CL_INVALID_DX9_RESOURCE_INTEL -1011
+#define CL_DX9_RESOURCE_ALREADY_ACQUIRED_INTEL -1012
+#define CL_DX9_RESOURCE_NOT_ACQUIRED_INTEL -1013
+
+/* cl_dx9_device_source_intel */
+#define CL_D3D9_DEVICE_INTEL 0x4022
+#define CL_D3D9EX_DEVICE_INTEL 0x4070
+#define CL_DXVA_DEVICE_INTEL 0x4071
+
+/* cl_dx9_device_set_intel */
+#define CL_PREFERRED_DEVICES_FOR_DX9_INTEL 0x4024
+#define CL_ALL_DEVICES_FOR_DX9_INTEL 0x4025
+
+/* cl_context_info */
+#define CL_CONTEXT_D3D9_DEVICE_INTEL 0x4026
+#define CL_CONTEXT_D3D9EX_DEVICE_INTEL 0x4072
+#define CL_CONTEXT_DXVA_DEVICE_INTEL 0x4073
+
+/* cl_mem_info */
+#define CL_MEM_DX9_RESOURCE_INTEL 0x4027
+#define CL_MEM_DX9_SHARED_HANDLE_INTEL 0x4074
+
+/* cl_image_info */
+#define CL_IMAGE_DX9_PLANE_INTEL 0x4075
+
+/* cl_command_type */
+#define CL_COMMAND_ACQUIRE_DX9_OBJECTS_INTEL 0x402A
+#define CL_COMMAND_RELEASE_DX9_OBJECTS_INTEL 0x402B
+/******************************************************************************/
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clGetDeviceIDsFromDX9INTEL(
+ cl_platform_id platform,
+ cl_dx9_device_source_intel dx9_device_source,
+ void* dx9_object,
+ cl_dx9_device_set_intel dx9_device_set,
+ cl_uint num_entries,
+ cl_device_id* devices,
+ cl_uint* num_devices) CL_EXT_SUFFIX__VERSION_1_1;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL* clGetDeviceIDsFromDX9INTEL_fn)(
+ cl_platform_id platform,
+ cl_dx9_device_source_intel dx9_device_source,
+ void* dx9_object,
+ cl_dx9_device_set_intel dx9_device_set,
+ cl_uint num_entries,
+ cl_device_id* devices,
+ cl_uint* num_devices) CL_EXT_SUFFIX__VERSION_1_1;
+
+extern CL_API_ENTRY cl_mem CL_API_CALL
+clCreateFromDX9MediaSurfaceINTEL(
+ cl_context context,
+ cl_mem_flags flags,
+ IDirect3DSurface9* resource,
+ HANDLE sharedHandle,
+ UINT plane,
+ cl_int* errcode_ret) CL_EXT_SUFFIX__VERSION_1_1;
+
+typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromDX9MediaSurfaceINTEL_fn)(
+ cl_context context,
+ cl_mem_flags flags,
+ IDirect3DSurface9* resource,
+ HANDLE sharedHandle,
+ UINT plane,
+ cl_int* errcode_ret) CL_EXT_SUFFIX__VERSION_1_1;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueAcquireDX9ObjectsINTEL(
+ 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_EXT_SUFFIX__VERSION_1_1;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireDX9ObjectsINTEL_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_EXT_SUFFIX__VERSION_1_1;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueReleaseDX9ObjectsINTEL(
+ cl_command_queue command_queue,
+ cl_uint num_objects,
+ cl_mem* mem_objects,
+ cl_uint num_events_in_wait_list,
+ const cl_event* event_wait_list,
+ cl_event* event) CL_EXT_SUFFIX__VERSION_1_1;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseDX9ObjectsINTEL_fn)(
+ cl_command_queue command_queue,
+ cl_uint num_objects,
+ cl_mem* mem_objects,
+ cl_uint num_events_in_wait_list,
+ const cl_event* event_wait_list,
+ cl_event* event) CL_EXT_SUFFIX__VERSION_1_1;
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* __OPENCL_CL_DX9_MEDIA_SHARING_INTEL_H */
+
diff --git a/include/CL/cl_egl.h b/include/CL/cl_egl.h
index 93e6c9cfba5..bc4d998eb3e 100644
--- a/include/CL/cl_egl.h
+++ b/include/CL/cl_egl.h
@@ -1,5 +1,5 @@
/*******************************************************************************
- * Copyright (c) 2008-2010 The Khronos Group Inc.
+ * Copyright (c) 2008-2019 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -24,13 +29,7 @@
#ifndef __OPENCL_CL_EGL_H
#define __OPENCL_CL_EGL_H
-#ifdef __APPLE__
-
-#else
#include <CL/cl.h>
-#include <EGL/egl.h>
-#include <EGL/eglext.h>
-#endif
#ifdef __cplusplus
extern "C" {
@@ -62,69 +61,69 @@ typedef intptr_t cl_egl_image_properties_khr;
#define cl_khr_egl_image 1
extern CL_API_ENTRY cl_mem CL_API_CALL
-clCreateFromEGLImageKHR(cl_context /* context */,
- CLeglDisplayKHR /* egldisplay */,
- CLeglImageKHR /* eglimage */,
- cl_mem_flags /* flags */,
- const cl_egl_image_properties_khr * /* properties */,
- cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+clCreateFromEGLImageKHR(cl_context context,
+ CLeglDisplayKHR egldisplay,
+ CLeglImageKHR eglimage,
+ cl_mem_flags flags,
+ const cl_egl_image_properties_khr * properties,
+ cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0;
typedef CL_API_ENTRY cl_mem (CL_API_CALL *clCreateFromEGLImageKHR_fn)(
- cl_context context,
- CLeglDisplayKHR egldisplay,
- CLeglImageKHR eglimage,
- cl_mem_flags flags,
- const cl_egl_image_properties_khr * properties,
- cl_int * errcode_ret);
+ cl_context context,
+ CLeglDisplayKHR egldisplay,
+ CLeglImageKHR eglimage,
+ cl_mem_flags flags,
+ const cl_egl_image_properties_khr * properties,
+ cl_int * errcode_ret);
extern CL_API_ENTRY cl_int CL_API_CALL
-clEnqueueAcquireEGLObjectsKHR(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;
+clEnqueueAcquireEGLObjectsKHR(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 *clEnqueueAcquireEGLObjectsKHR_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_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);
extern CL_API_ENTRY cl_int CL_API_CALL
-clEnqueueReleaseEGLObjectsKHR(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;
+clEnqueueReleaseEGLObjectsKHR(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 *clEnqueueReleaseEGLObjectsKHR_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_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);
#define cl_khr_egl_event 1
extern CL_API_ENTRY cl_event CL_API_CALL
-clCreateEventFromEGLSyncKHR(cl_context /* context */,
- CLeglSyncKHR /* sync */,
- CLeglDisplayKHR /* display */,
- cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+clCreateEventFromEGLSyncKHR(cl_context context,
+ CLeglSyncKHR sync,
+ CLeglDisplayKHR display,
+ cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0;
typedef CL_API_ENTRY cl_event (CL_API_CALL *clCreateEventFromEGLSyncKHR_fn)(
- cl_context context,
- CLeglSyncKHR sync,
- CLeglDisplayKHR display,
- cl_int * errcode_ret);
+ cl_context context,
+ CLeglSyncKHR sync,
+ CLeglDisplayKHR display,
+ cl_int * errcode_ret);
#ifdef __cplusplus
}
diff --git a/include/CL/cl_ext.h b/include/CL/cl_ext.h
index 710bea8837d..c1bf42b62b5 100644
--- a/include/CL/cl_ext.h
+++ b/include/CL/cl_ext.h
@@ -1,5 +1,5 @@
/*******************************************************************************
- * Copyright (c) 2008-2013 The Khronos Group Inc.
+ * Copyright (c) 2008-2019 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -21,8 +26,6 @@
* 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. */
@@ -33,11 +36,13 @@
extern "C" {
#endif
-#ifdef __APPLE__
- #include <OpenCL/cl.h>
- #include <AvailabilityMacros.h>
-#else
- #include <CL/cl.h>
+#include <CL/cl.h>
+
+/* cl_khr_fp64 extension - no extension #define since it has no functions */
+/* CL_DEVICE_DOUBLE_FP_CONFIG is defined in CL.h for OpenCL >= 120 */
+
+#if CL_TARGET_OPENCL_VERSION <= 110
+#define CL_DEVICE_DOUBLE_FP_CONFIG 0x1032
#endif
/* cl_khr_fp16 extension - no extension #define since it has no functions */
@@ -47,12 +52,12 @@ extern "C" {
*
* 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
+ * 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.
@@ -61,9 +66,9 @@ extern "C" {
* 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;
+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
@@ -72,29 +77,29 @@ cl_int CL_API_ENTRY clSetMemObjectDestructorAPPLE( cl_mem /* memobj */,
* 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
+ * clLogMessagesToSystemLog forwards 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;
+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;
+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;
+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 *
+/************************
+* cl_khr_icd extension *
************************/
#define cl_khr_icd 1
@@ -105,15 +110,42 @@ extern void CL_API_ENTRY clLogMessagesToStderrAPPLE( const char * /* errstr */
#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 */);
+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);
+
+
+/*******************************
+ * cl_khr_il_program extension *
+ *******************************/
+#define cl_khr_il_program 1
+
+/* New property to clGetDeviceInfo for retrieving supported intermediate
+ * languages
+ */
+#define CL_DEVICE_IL_VERSION_KHR 0x105B
-typedef CL_API_ENTRY cl_int (CL_API_CALL *clIcdGetPlatformIDsKHR_fn)(
- cl_uint /* num_entries */,
- cl_platform_id * /* platforms */,
- cl_uint * /* num_platforms */);
+/* New property to clGetProgramInfo for retrieving for retrieving the IL of a
+ * program
+ */
+#define CL_PROGRAM_IL_KHR 0x1169
+
+extern CL_API_ENTRY cl_program CL_API_CALL
+clCreateProgramWithILKHR(cl_context context,
+ const void * il,
+ size_t length,
+ cl_int * errcode_ret);
+typedef CL_API_ENTRY cl_program
+(CL_API_CALL *clCreateProgramWithILKHR_fn)(cl_context context,
+ const void * il,
+ size_t length,
+ cl_int * errcode_ret) CL_EXT_SUFFIX__VERSION_1_2;
/* Extension: cl_khr_image2D_buffer
*
@@ -129,31 +161,33 @@ typedef CL_API_ENTRY cl_int (CL_API_CALL *clIcdGetPlatformIDsKHR_fn)(
* 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_initialize_memory extension *
+ **************************************/
+
+#define CL_CONTEXT_MEMORY_INITIALIZE_KHR 0x2030
+
+
/**************************************
* cl_khr_terminate_context extension *
**************************************/
-
-#define CL_DEVICE_TERMINATE_CAPABILITY_KHR 0x200F
-#define CL_CONTEXT_TERMINATE_KHR 0x2010
+
+#define CL_DEVICE_TERMINATE_CAPABILITY_KHR 0x2031
+#define CL_CONTEXT_TERMINATE_KHR 0x2032
#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;
+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;
+
-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
+ * This extension adds support to create an OpenCL program object from a
* Standard Portable Intermediate Representation (SPIR) instance
*/
@@ -161,9 +195,30 @@ typedef CL_API_ENTRY cl_int (CL_API_CALL *clTerminateContextKHR_fn)(cl_context /
#define CL_PROGRAM_BINARY_TYPE_INTERMEDIATE 0x40E1
+/*****************************************
+ * cl_khr_create_command_queue extension *
+ *****************************************/
+#define cl_khr_create_command_queue 1
+
+typedef cl_bitfield cl_queue_properties_khr;
+
+extern CL_API_ENTRY cl_command_queue CL_API_CALL
+clCreateCommandQueueWithPropertiesKHR(cl_context context,
+ cl_device_id device,
+ const cl_queue_properties_khr* properties,
+ cl_int* errcode_ret) CL_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_command_queue
+(CL_API_CALL *clCreateCommandQueueWithPropertiesKHR_fn)(cl_context context,
+ cl_device_id device,
+ const cl_queue_properties_khr* properties,
+ cl_int* errcode_ret) CL_EXT_SUFFIX__VERSION_1_2;
+
+
/******************************************
* 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
@@ -173,88 +228,124 @@ typedef CL_API_ENTRY cl_int (CL_API_CALL *clTerminateContextKHR_fn)(cl_context /
#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
+
/*********************************
* cl_arm_printf extension
*********************************/
+
#define CL_PRINTF_CALLBACK_ARM 0x40B0
#define CL_PRINTF_BUFFERSIZE_ARM 0x40B1
-#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_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_ext_migrate_memobject extension definitions
+ ***********************************/
+#define cl_ext_migrate_memobject 1
+
+typedef cl_bitfield cl_mem_migration_flags_ext;
+
+#define CL_MIGRATE_MEM_OBJECT_HOST_EXT 0x1
+
+#define CL_COMMAND_MIGRATE_MEM_OBJECT_EXT 0x4040
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueMigrateMemObjectEXT(cl_command_queue command_queue,
+ cl_uint num_mem_objects,
+ const cl_mem * mem_objects,
+ cl_mem_migration_flags_ext flags,
+ 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 *clEnqueueMigrateMemObjectEXT_fn)(cl_command_queue command_queue,
+ cl_uint num_mem_objects,
+ const cl_mem * mem_objects,
+ cl_mem_migration_flags_ext flags,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event);
+
/*********************************
* cl_qcom_ext_host_ptr extension
*********************************/
+#define cl_qcom_ext_host_ptr 1
#define CL_MEM_EXT_HOST_PTR_QCOM (1 << 29)
-#define CL_DEVICE_EXT_MEM_PADDING_IN_BYTES_QCOM 0x40A0
+#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
@@ -280,12 +371,21 @@ 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. */
+
+ /* Host cache policy for this external memory allocation. */
cl_uint host_cache_policy;
} cl_mem_ext_host_ptr;
+
+/*******************************************
+* cl_qcom_ext_host_ptr_iocoherent extension
+********************************************/
+
+/* Cache policy specifying io-coherence */
+#define CL_MEM_HOST_IOCOHERENT_QCOM 0x40A9
+
+
/*********************************
* cl_qcom_ion_host_ptr extension
*********************************/
@@ -300,13 +400,339 @@ typedef struct _cl_mem_ion_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 */
+
+/*********************************
+* cl_qcom_android_native_buffer_host_ptr extension
+*********************************/
+
+#define CL_MEM_ANDROID_NATIVE_BUFFER_HOST_PTR_QCOM 0x40C6
+
+typedef struct _cl_mem_android_native_buffer_host_ptr
+{
+ /* Type of external memory allocation. */
+ /* Must be CL_MEM_ANDROID_NATIVE_BUFFER_HOST_PTR_QCOM for Android native buffers. */
+ cl_mem_ext_host_ptr ext_host_ptr;
+
+ /* Virtual pointer to the android native buffer */
+ void* anb_ptr;
+
+} cl_mem_android_native_buffer_host_ptr;
+
+
+/******************************************
+ * cl_img_yuv_image extension *
+ ******************************************/
+
+/* Image formats used in clCreateImage */
+#define CL_NV21_IMG 0x40D0
+#define CL_YV12_IMG 0x40D1
+
+
+/******************************************
+ * cl_img_cached_allocations extension *
+ ******************************************/
+
+/* Flag values used by clCreateBuffer */
+#define CL_MEM_USE_UNCACHED_CPU_MEMORY_IMG (1 << 26)
+#define CL_MEM_USE_CACHED_CPU_MEMORY_IMG (1 << 27)
+
+
+/******************************************
+ * cl_img_use_gralloc_ptr extension *
+ ******************************************/
+#define cl_img_use_gralloc_ptr 1
+
+/* Flag values used by clCreateBuffer */
+#define CL_MEM_USE_GRALLOC_PTR_IMG (1 << 28)
+
+/* To be used by clGetEventInfo: */
+#define CL_COMMAND_ACQUIRE_GRALLOC_OBJECTS_IMG 0x40D2
+#define CL_COMMAND_RELEASE_GRALLOC_OBJECTS_IMG 0x40D3
+
+/* Error code from clEnqueueReleaseGrallocObjectsIMG */
+#define CL_GRALLOC_RESOURCE_NOT_ACQUIRED_IMG 0x40D4
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueAcquireGrallocObjectsIMG(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_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueReleaseGrallocObjectsIMG(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_EXT_SUFFIX__VERSION_1_2;
+
+
+/*********************************
+* cl_khr_subgroups extension
+*********************************/
+#define cl_khr_subgroups 1
+
+#if !defined(CL_VERSION_2_1)
+/* For OpenCL 2.1 and newer, cl_kernel_sub_group_info is declared in CL.h.
+ In hindsight, there should have been a khr suffix on this type for
+ the extension, but keeping it un-suffixed to maintain backwards
+ compatibility. */
+typedef cl_uint cl_kernel_sub_group_info;
+#endif
+
+/* cl_kernel_sub_group_info */
+#define CL_KERNEL_MAX_SUB_GROUP_SIZE_FOR_NDRANGE_KHR 0x2033
+#define CL_KERNEL_SUB_GROUP_COUNT_FOR_NDRANGE_KHR 0x2034
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clGetKernelSubGroupInfoKHR(cl_kernel in_kernel,
+ cl_device_id in_device,
+ cl_kernel_sub_group_info param_name,
+ size_t input_value_size,
+ const void * input_value,
+ size_t param_value_size,
+ void * param_value,
+ size_t * param_value_size_ret) CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED;
+
+typedef CL_API_ENTRY cl_int
+(CL_API_CALL * clGetKernelSubGroupInfoKHR_fn)(cl_kernel in_kernel,
+ cl_device_id in_device,
+ cl_kernel_sub_group_info param_name,
+ size_t input_value_size,
+ const void * input_value,
+ size_t param_value_size,
+ void * param_value,
+ size_t * param_value_size_ret) CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED;
+
+
+/*********************************
+* cl_khr_mipmap_image extension
+*********************************/
+
+/* cl_sampler_properties */
+#define CL_SAMPLER_MIP_FILTER_MODE_KHR 0x1155
+#define CL_SAMPLER_LOD_MIN_KHR 0x1156
+#define CL_SAMPLER_LOD_MAX_KHR 0x1157
+
+
+/*********************************
+* cl_khr_priority_hints extension
+*********************************/
+/* This extension define is for backwards compatibility.
+ It shouldn't be required since this extension has no new functions. */
+#define cl_khr_priority_hints 1
+
+typedef cl_uint cl_queue_priority_khr;
+
+/* cl_command_queue_properties */
+#define CL_QUEUE_PRIORITY_KHR 0x1096
+
+/* cl_queue_priority_khr */
+#define CL_QUEUE_PRIORITY_HIGH_KHR (1<<0)
+#define CL_QUEUE_PRIORITY_MED_KHR (1<<1)
+#define CL_QUEUE_PRIORITY_LOW_KHR (1<<2)
+
+
+/*********************************
+* cl_khr_throttle_hints extension
+*********************************/
+/* This extension define is for backwards compatibility.
+ It shouldn't be required since this extension has no new functions. */
+#define cl_khr_throttle_hints 1
+
+typedef cl_uint cl_queue_throttle_khr;
+
+/* cl_command_queue_properties */
+#define CL_QUEUE_THROTTLE_KHR 0x1097
+
+/* cl_queue_throttle_khr */
+#define CL_QUEUE_THROTTLE_HIGH_KHR (1<<0)
+#define CL_QUEUE_THROTTLE_MED_KHR (1<<1)
+#define CL_QUEUE_THROTTLE_LOW_KHR (1<<2)
+
+
+/*********************************
+* cl_khr_subgroup_named_barrier
+*********************************/
+/* This extension define is for backwards compatibility.
+ It shouldn't be required since this extension has no new functions. */
+#define cl_khr_subgroup_named_barrier 1
+
+/* cl_device_info */
+#define CL_DEVICE_MAX_NAMED_BARRIER_COUNT_KHR 0x2035
+
+
+/**********************************
+ * cl_arm_import_memory extension *
+ **********************************/
+#define cl_arm_import_memory 1
+
+typedef intptr_t cl_import_properties_arm;
+
+/* Default and valid proporties name for cl_arm_import_memory */
+#define CL_IMPORT_TYPE_ARM 0x40B2
+
+/* Host process memory type default value for CL_IMPORT_TYPE_ARM property */
+#define CL_IMPORT_TYPE_HOST_ARM 0x40B3
+
+/* DMA BUF memory type value for CL_IMPORT_TYPE_ARM property */
+#define CL_IMPORT_TYPE_DMA_BUF_ARM 0x40B4
+
+/* Protected DMA BUF memory type value for CL_IMPORT_TYPE_ARM property */
+#define CL_IMPORT_TYPE_PROTECTED_ARM 0x40B5
+
+/* This extension adds a new function that allows for direct memory import into
+ * OpenCL via the clImportMemoryARM function.
+ *
+ * Memory imported through this interface will be mapped into the device's page
+ * tables directly, providing zero copy access. It will never fall back to copy
+ * operations and aliased buffers.
+ *
+ * Types of memory supported for import are specified as additional extension
+ * strings.
+ *
+ * This extension produces cl_mem allocations which are compatible with all other
+ * users of cl_mem in the standard API.
+ *
+ * This extension maps pages with the same properties as the normal buffer creation
+ * function clCreateBuffer.
+ */
+extern CL_API_ENTRY cl_mem CL_API_CALL
+clImportMemoryARM( cl_context context,
+ cl_mem_flags flags,
+ const cl_import_properties_arm *properties,
+ void *memory,
+ size_t size,
+ cl_int *errcode_ret) CL_EXT_SUFFIX__VERSION_1_0;
+
+
+/******************************************
+ * cl_arm_shared_virtual_memory extension *
+ ******************************************/
+#define cl_arm_shared_virtual_memory 1
+
+/* Used by clGetDeviceInfo */
+#define CL_DEVICE_SVM_CAPABILITIES_ARM 0x40B6
+
+/* Used by clGetMemObjectInfo */
+#define CL_MEM_USES_SVM_POINTER_ARM 0x40B7
+
+/* Used by clSetKernelExecInfoARM: */
+#define CL_KERNEL_EXEC_INFO_SVM_PTRS_ARM 0x40B8
+#define CL_KERNEL_EXEC_INFO_SVM_FINE_GRAIN_SYSTEM_ARM 0x40B9
+
+/* To be used by clGetEventInfo: */
+#define CL_COMMAND_SVM_FREE_ARM 0x40BA
+#define CL_COMMAND_SVM_MEMCPY_ARM 0x40BB
+#define CL_COMMAND_SVM_MEMFILL_ARM 0x40BC
+#define CL_COMMAND_SVM_MAP_ARM 0x40BD
+#define CL_COMMAND_SVM_UNMAP_ARM 0x40BE
+
+/* Flag values returned by clGetDeviceInfo with CL_DEVICE_SVM_CAPABILITIES_ARM as the param_name. */
+#define CL_DEVICE_SVM_COARSE_GRAIN_BUFFER_ARM (1 << 0)
+#define CL_DEVICE_SVM_FINE_GRAIN_BUFFER_ARM (1 << 1)
+#define CL_DEVICE_SVM_FINE_GRAIN_SYSTEM_ARM (1 << 2)
+#define CL_DEVICE_SVM_ATOMICS_ARM (1 << 3)
+
+/* Flag values used by clSVMAllocARM: */
+#define CL_MEM_SVM_FINE_GRAIN_BUFFER_ARM (1 << 10)
+#define CL_MEM_SVM_ATOMICS_ARM (1 << 11)
+
+typedef cl_bitfield cl_svm_mem_flags_arm;
+typedef cl_uint cl_kernel_exec_info_arm;
+typedef cl_bitfield cl_device_svm_capabilities_arm;
+
+extern CL_API_ENTRY void * CL_API_CALL
+clSVMAllocARM(cl_context context,
+ cl_svm_mem_flags_arm flags,
+ size_t size,
+ cl_uint alignment) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY void CL_API_CALL
+clSVMFreeARM(cl_context context,
+ void * svm_pointer) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMFreeARM(cl_command_queue command_queue,
+ cl_uint num_svm_pointers,
+ void * svm_pointers[],
+ void (CL_CALLBACK * pfn_free_func)(cl_command_queue queue,
+ cl_uint num_svm_pointers,
+ void * svm_pointers[],
+ void * user_data),
+ void * user_data,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMMemcpyARM(cl_command_queue command_queue,
+ cl_bool blocking_copy,
+ void * dst_ptr,
+ const void * src_ptr,
+ size_t size,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMMemFillARM(cl_command_queue command_queue,
+ void * svm_ptr,
+ const void * pattern,
+ size_t pattern_size,
+ size_t size,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMMapARM(cl_command_queue command_queue,
+ cl_bool blocking_map,
+ cl_map_flags flags,
+ void * svm_ptr,
+ size_t size,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueSVMUnmapARM(cl_command_queue command_queue,
+ void * svm_ptr,
+ cl_uint num_events_in_wait_list,
+ const cl_event * event_wait_list,
+ cl_event * event) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clSetKernelArgSVMPointerARM(cl_kernel kernel,
+ cl_uint arg_index,
+ const void * arg_value) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clSetKernelExecInfoARM(cl_kernel kernel,
+ cl_kernel_exec_info_arm param_name,
+ size_t param_value_size,
+ const void * param_value) CL_EXT_SUFFIX__VERSION_1_2;
+
+/********************************
+ * cl_arm_get_core_id extension *
+ ********************************/
+
+#ifdef CL_VERSION_1_2
+
+#define cl_arm_get_core_id 1
+
+/* Device info property for bitfield of cores present */
+#define CL_DEVICE_COMPUTE_UNITS_BITFIELD_ARM 0x40BF
+
+#endif /* CL_VERSION_1_2 */
#ifdef __cplusplus
}
diff --git a/include/CL/cl_ext_intel.h b/include/CL/cl_ext_intel.h
new file mode 100644
index 00000000000..9d1e4b58773
--- /dev/null
+++ b/include/CL/cl_ext_intel.h
@@ -0,0 +1,423 @@
+/*******************************************************************************
+ * Copyright (c) 2008-2019 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.
+ *
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
+ * 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.
+ ******************************************************************************/
+/*****************************************************************************\
+
+Copyright (c) 2013-2019 Intel Corporation All Rights Reserved.
+
+THESE MATERIALS ARE PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL OR ITS
+CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY OR TORT (INCLUDING
+NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THESE
+MATERIALS, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+File Name: cl_ext_intel.h
+
+Abstract:
+
+Notes:
+
+\*****************************************************************************/
+
+#ifndef __CL_EXT_INTEL_H
+#define __CL_EXT_INTEL_H
+
+#include <CL/cl.h>
+#include <CL/cl_platform.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/***************************************
+* cl_intel_thread_local_exec extension *
+****************************************/
+
+#define cl_intel_thread_local_exec 1
+
+#define CL_QUEUE_THREAD_LOCAL_EXEC_ENABLE_INTEL (((cl_bitfield)1) << 31)
+
+/***********************************************
+* cl_intel_device_partition_by_names extension *
+************************************************/
+
+#define cl_intel_device_partition_by_names 1
+
+#define CL_DEVICE_PARTITION_BY_NAMES_INTEL 0x4052
+#define CL_PARTITION_BY_NAMES_LIST_END_INTEL -1
+
+/************************************************
+* cl_intel_accelerator extension *
+* cl_intel_motion_estimation extension *
+* cl_intel_advanced_motion_estimation extension *
+*************************************************/
+
+#define cl_intel_accelerator 1
+#define cl_intel_motion_estimation 1
+#define cl_intel_advanced_motion_estimation 1
+
+typedef struct _cl_accelerator_intel* cl_accelerator_intel;
+typedef cl_uint cl_accelerator_type_intel;
+typedef cl_uint cl_accelerator_info_intel;
+
+typedef struct _cl_motion_estimation_desc_intel {
+ cl_uint mb_block_type;
+ cl_uint subpixel_mode;
+ cl_uint sad_adjust_mode;
+ cl_uint search_path_type;
+} cl_motion_estimation_desc_intel;
+
+/* error codes */
+#define CL_INVALID_ACCELERATOR_INTEL -1094
+#define CL_INVALID_ACCELERATOR_TYPE_INTEL -1095
+#define CL_INVALID_ACCELERATOR_DESCRIPTOR_INTEL -1096
+#define CL_ACCELERATOR_TYPE_NOT_SUPPORTED_INTEL -1097
+
+/* cl_accelerator_type_intel */
+#define CL_ACCELERATOR_TYPE_MOTION_ESTIMATION_INTEL 0x0
+
+/* cl_accelerator_info_intel */
+#define CL_ACCELERATOR_DESCRIPTOR_INTEL 0x4090
+#define CL_ACCELERATOR_REFERENCE_COUNT_INTEL 0x4091
+#define CL_ACCELERATOR_CONTEXT_INTEL 0x4092
+#define CL_ACCELERATOR_TYPE_INTEL 0x4093
+
+/* cl_motion_detect_desc_intel flags */
+#define CL_ME_MB_TYPE_16x16_INTEL 0x0
+#define CL_ME_MB_TYPE_8x8_INTEL 0x1
+#define CL_ME_MB_TYPE_4x4_INTEL 0x2
+
+#define CL_ME_SUBPIXEL_MODE_INTEGER_INTEL 0x0
+#define CL_ME_SUBPIXEL_MODE_HPEL_INTEL 0x1
+#define CL_ME_SUBPIXEL_MODE_QPEL_INTEL 0x2
+
+#define CL_ME_SAD_ADJUST_MODE_NONE_INTEL 0x0
+#define CL_ME_SAD_ADJUST_MODE_HAAR_INTEL 0x1
+
+#define CL_ME_SEARCH_PATH_RADIUS_2_2_INTEL 0x0
+#define CL_ME_SEARCH_PATH_RADIUS_4_4_INTEL 0x1
+#define CL_ME_SEARCH_PATH_RADIUS_16_12_INTEL 0x5
+
+#define CL_ME_SKIP_BLOCK_TYPE_16x16_INTEL 0x0
+#define CL_ME_CHROMA_INTRA_PREDICT_ENABLED_INTEL 0x1
+#define CL_ME_LUMA_INTRA_PREDICT_ENABLED_INTEL 0x2
+#define CL_ME_SKIP_BLOCK_TYPE_8x8_INTEL 0x4
+
+#define CL_ME_FORWARD_INPUT_MODE_INTEL 0x1
+#define CL_ME_BACKWARD_INPUT_MODE_INTEL 0x2
+#define CL_ME_BIDIRECTION_INPUT_MODE_INTEL 0x3
+
+#define CL_ME_BIDIR_WEIGHT_QUARTER_INTEL 16
+#define CL_ME_BIDIR_WEIGHT_THIRD_INTEL 21
+#define CL_ME_BIDIR_WEIGHT_HALF_INTEL 32
+#define CL_ME_BIDIR_WEIGHT_TWO_THIRD_INTEL 43
+#define CL_ME_BIDIR_WEIGHT_THREE_QUARTER_INTEL 48
+
+#define CL_ME_COST_PENALTY_NONE_INTEL 0x0
+#define CL_ME_COST_PENALTY_LOW_INTEL 0x1
+#define CL_ME_COST_PENALTY_NORMAL_INTEL 0x2
+#define CL_ME_COST_PENALTY_HIGH_INTEL 0x3
+
+#define CL_ME_COST_PRECISION_QPEL_INTEL 0x0
+#define CL_ME_COST_PRECISION_HPEL_INTEL 0x1
+#define CL_ME_COST_PRECISION_PEL_INTEL 0x2
+#define CL_ME_COST_PRECISION_DPEL_INTEL 0x3
+
+#define CL_ME_LUMA_PREDICTOR_MODE_VERTICAL_INTEL 0x0
+#define CL_ME_LUMA_PREDICTOR_MODE_HORIZONTAL_INTEL 0x1
+#define CL_ME_LUMA_PREDICTOR_MODE_DC_INTEL 0x2
+#define CL_ME_LUMA_PREDICTOR_MODE_DIAGONAL_DOWN_LEFT_INTEL 0x3
+
+#define CL_ME_LUMA_PREDICTOR_MODE_DIAGONAL_DOWN_RIGHT_INTEL 0x4
+#define CL_ME_LUMA_PREDICTOR_MODE_PLANE_INTEL 0x4
+#define CL_ME_LUMA_PREDICTOR_MODE_VERTICAL_RIGHT_INTEL 0x5
+#define CL_ME_LUMA_PREDICTOR_MODE_HORIZONTAL_DOWN_INTEL 0x6
+#define CL_ME_LUMA_PREDICTOR_MODE_VERTICAL_LEFT_INTEL 0x7
+#define CL_ME_LUMA_PREDICTOR_MODE_HORIZONTAL_UP_INTEL 0x8
+
+#define CL_ME_CHROMA_PREDICTOR_MODE_DC_INTEL 0x0
+#define CL_ME_CHROMA_PREDICTOR_MODE_HORIZONTAL_INTEL 0x1
+#define CL_ME_CHROMA_PREDICTOR_MODE_VERTICAL_INTEL 0x2
+#define CL_ME_CHROMA_PREDICTOR_MODE_PLANE_INTEL 0x3
+
+/* cl_device_info */
+#define CL_DEVICE_ME_VERSION_INTEL 0x407E
+
+#define CL_ME_VERSION_LEGACY_INTEL 0x0
+#define CL_ME_VERSION_ADVANCED_VER_1_INTEL 0x1
+#define CL_ME_VERSION_ADVANCED_VER_2_INTEL 0x2
+
+extern CL_API_ENTRY cl_accelerator_intel CL_API_CALL
+clCreateAcceleratorINTEL(
+ cl_context context,
+ cl_accelerator_type_intel accelerator_type,
+ size_t descriptor_size,
+ const void* descriptor,
+ cl_int* errcode_ret) CL_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_accelerator_intel (CL_API_CALL *clCreateAcceleratorINTEL_fn)(
+ cl_context context,
+ cl_accelerator_type_intel accelerator_type,
+ size_t descriptor_size,
+ const void* descriptor,
+ cl_int* errcode_ret) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clGetAcceleratorInfoINTEL(
+ cl_accelerator_intel accelerator,
+ cl_accelerator_info_intel param_name,
+ size_t param_value_size,
+ void* param_value,
+ size_t* param_value_size_ret) CL_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL *clGetAcceleratorInfoINTEL_fn)(
+ cl_accelerator_intel accelerator,
+ cl_accelerator_info_intel param_name,
+ size_t param_value_size,
+ void* param_value,
+ size_t* param_value_size_ret) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clRetainAcceleratorINTEL(
+ cl_accelerator_intel accelerator) CL_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL *clRetainAcceleratorINTEL_fn)(
+ cl_accelerator_intel accelerator) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clReleaseAcceleratorINTEL(
+ cl_accelerator_intel accelerator) CL_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL *clReleaseAcceleratorINTEL_fn)(
+ cl_accelerator_intel accelerator) CL_EXT_SUFFIX__VERSION_1_2;
+
+/******************************************
+* cl_intel_simultaneous_sharing extension *
+*******************************************/
+
+#define cl_intel_simultaneous_sharing 1
+
+#define CL_DEVICE_SIMULTANEOUS_INTEROPS_INTEL 0x4104
+#define CL_DEVICE_NUM_SIMULTANEOUS_INTEROPS_INTEL 0x4105
+
+/***********************************
+* cl_intel_egl_image_yuv extension *
+************************************/
+
+#define cl_intel_egl_image_yuv 1
+
+#define CL_EGL_YUV_PLANE_INTEL 0x4107
+
+/********************************
+* cl_intel_packed_yuv extension *
+*********************************/
+
+#define cl_intel_packed_yuv 1
+
+#define CL_YUYV_INTEL 0x4076
+#define CL_UYVY_INTEL 0x4077
+#define CL_YVYU_INTEL 0x4078
+#define CL_VYUY_INTEL 0x4079
+
+/********************************************
+* cl_intel_required_subgroup_size extension *
+*********************************************/
+
+#define cl_intel_required_subgroup_size 1
+
+#define CL_DEVICE_SUB_GROUP_SIZES_INTEL 0x4108
+#define CL_KERNEL_SPILL_MEM_SIZE_INTEL 0x4109
+#define CL_KERNEL_COMPILE_SUB_GROUP_SIZE_INTEL 0x410A
+
+/****************************************
+* cl_intel_driver_diagnostics extension *
+*****************************************/
+
+#define cl_intel_driver_diagnostics 1
+
+typedef cl_uint cl_diagnostics_verbose_level;
+
+#define CL_CONTEXT_SHOW_DIAGNOSTICS_INTEL 0x4106
+
+#define CL_CONTEXT_DIAGNOSTICS_LEVEL_ALL_INTEL ( 0xff )
+#define CL_CONTEXT_DIAGNOSTICS_LEVEL_GOOD_INTEL ( 1 )
+#define CL_CONTEXT_DIAGNOSTICS_LEVEL_BAD_INTEL ( 1 << 1 )
+#define CL_CONTEXT_DIAGNOSTICS_LEVEL_NEUTRAL_INTEL ( 1 << 2 )
+
+/********************************
+* cl_intel_planar_yuv extension *
+*********************************/
+
+#define CL_NV12_INTEL 0x410E
+
+#define CL_MEM_NO_ACCESS_INTEL ( 1 << 24 )
+#define CL_MEM_ACCESS_FLAGS_UNRESTRICTED_INTEL ( 1 << 25 )
+
+#define CL_DEVICE_PLANAR_YUV_MAX_WIDTH_INTEL 0x417E
+#define CL_DEVICE_PLANAR_YUV_MAX_HEIGHT_INTEL 0x417F
+
+/*******************************************************
+* cl_intel_device_side_avc_motion_estimation extension *
+********************************************************/
+
+#define CL_DEVICE_AVC_ME_VERSION_INTEL 0x410B
+#define CL_DEVICE_AVC_ME_SUPPORTS_TEXTURE_SAMPLER_USE_INTEL 0x410C
+#define CL_DEVICE_AVC_ME_SUPPORTS_PREEMPTION_INTEL 0x410D
+
+#define CL_AVC_ME_VERSION_0_INTEL 0x0; // No support.
+#define CL_AVC_ME_VERSION_1_INTEL 0x1; // First supported version.
+
+#define CL_AVC_ME_MAJOR_16x16_INTEL 0x0
+#define CL_AVC_ME_MAJOR_16x8_INTEL 0x1
+#define CL_AVC_ME_MAJOR_8x16_INTEL 0x2
+#define CL_AVC_ME_MAJOR_8x8_INTEL 0x3
+
+#define CL_AVC_ME_MINOR_8x8_INTEL 0x0
+#define CL_AVC_ME_MINOR_8x4_INTEL 0x1
+#define CL_AVC_ME_MINOR_4x8_INTEL 0x2
+#define CL_AVC_ME_MINOR_4x4_INTEL 0x3
+
+#define CL_AVC_ME_MAJOR_FORWARD_INTEL 0x0
+#define CL_AVC_ME_MAJOR_BACKWARD_INTEL 0x1
+#define CL_AVC_ME_MAJOR_BIDIRECTIONAL_INTEL 0x2
+
+#define CL_AVC_ME_PARTITION_MASK_ALL_INTEL 0x0
+#define CL_AVC_ME_PARTITION_MASK_16x16_INTEL 0x7E
+#define CL_AVC_ME_PARTITION_MASK_16x8_INTEL 0x7D
+#define CL_AVC_ME_PARTITION_MASK_8x16_INTEL 0x7B
+#define CL_AVC_ME_PARTITION_MASK_8x8_INTEL 0x77
+#define CL_AVC_ME_PARTITION_MASK_8x4_INTEL 0x6F
+#define CL_AVC_ME_PARTITION_MASK_4x8_INTEL 0x5F
+#define CL_AVC_ME_PARTITION_MASK_4x4_INTEL 0x3F
+
+#define CL_AVC_ME_SEARCH_WINDOW_EXHAUSTIVE_INTEL 0x0
+#define CL_AVC_ME_SEARCH_WINDOW_SMALL_INTEL 0x1
+#define CL_AVC_ME_SEARCH_WINDOW_TINY_INTEL 0x2
+#define CL_AVC_ME_SEARCH_WINDOW_EXTRA_TINY_INTEL 0x3
+#define CL_AVC_ME_SEARCH_WINDOW_DIAMOND_INTEL 0x4
+#define CL_AVC_ME_SEARCH_WINDOW_LARGE_DIAMOND_INTEL 0x5
+#define CL_AVC_ME_SEARCH_WINDOW_RESERVED0_INTEL 0x6
+#define CL_AVC_ME_SEARCH_WINDOW_RESERVED1_INTEL 0x7
+#define CL_AVC_ME_SEARCH_WINDOW_CUSTOM_INTEL 0x8
+#define CL_AVC_ME_SEARCH_WINDOW_16x12_RADIUS_INTEL 0x9
+#define CL_AVC_ME_SEARCH_WINDOW_4x4_RADIUS_INTEL 0x2
+#define CL_AVC_ME_SEARCH_WINDOW_2x2_RADIUS_INTEL 0xa
+
+#define CL_AVC_ME_SAD_ADJUST_MODE_NONE_INTEL 0x0
+#define CL_AVC_ME_SAD_ADJUST_MODE_HAAR_INTEL 0x2
+
+#define CL_AVC_ME_SUBPIXEL_MODE_INTEGER_INTEL 0x0
+#define CL_AVC_ME_SUBPIXEL_MODE_HPEL_INTEL 0x1
+#define CL_AVC_ME_SUBPIXEL_MODE_QPEL_INTEL 0x3
+
+#define CL_AVC_ME_COST_PRECISION_QPEL_INTEL 0x0
+#define CL_AVC_ME_COST_PRECISION_HPEL_INTEL 0x1
+#define CL_AVC_ME_COST_PRECISION_PEL_INTEL 0x2
+#define CL_AVC_ME_COST_PRECISION_DPEL_INTEL 0x3
+
+#define CL_AVC_ME_BIDIR_WEIGHT_QUARTER_INTEL 0x10
+#define CL_AVC_ME_BIDIR_WEIGHT_THIRD_INTEL 0x15
+#define CL_AVC_ME_BIDIR_WEIGHT_HALF_INTEL 0x20
+#define CL_AVC_ME_BIDIR_WEIGHT_TWO_THIRD_INTEL 0x2B
+#define CL_AVC_ME_BIDIR_WEIGHT_THREE_QUARTER_INTEL 0x30
+
+#define CL_AVC_ME_BORDER_REACHED_LEFT_INTEL 0x0
+#define CL_AVC_ME_BORDER_REACHED_RIGHT_INTEL 0x2
+#define CL_AVC_ME_BORDER_REACHED_TOP_INTEL 0x4
+#define CL_AVC_ME_BORDER_REACHED_BOTTOM_INTEL 0x8
+
+#define CL_AVC_ME_SKIP_BLOCK_PARTITION_16x16_INTEL 0x0
+#define CL_AVC_ME_SKIP_BLOCK_PARTITION_8x8_INTEL 0x4000
+
+#define CL_AVC_ME_SKIP_BLOCK_16x16_FORWARD_ENABLE_INTEL ( 0x1 << 24 )
+#define CL_AVC_ME_SKIP_BLOCK_16x16_BACKWARD_ENABLE_INTEL ( 0x2 << 24 )
+#define CL_AVC_ME_SKIP_BLOCK_16x16_DUAL_ENABLE_INTEL ( 0x3 << 24 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_FORWARD_ENABLE_INTEL ( 0x55 << 24 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_BACKWARD_ENABLE_INTEL ( 0xAA << 24 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_DUAL_ENABLE_INTEL ( 0xFF << 24 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_0_FORWARD_ENABLE_INTEL ( 0x1 << 24 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_0_BACKWARD_ENABLE_INTEL ( 0x2 << 24 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_1_FORWARD_ENABLE_INTEL ( 0x1 << 26 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_1_BACKWARD_ENABLE_INTEL ( 0x2 << 26 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_2_FORWARD_ENABLE_INTEL ( 0x1 << 28 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_2_BACKWARD_ENABLE_INTEL ( 0x2 << 28 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_3_FORWARD_ENABLE_INTEL ( 0x1 << 30 )
+#define CL_AVC_ME_SKIP_BLOCK_8x8_3_BACKWARD_ENABLE_INTEL ( 0x2 << 30 )
+
+#define CL_AVC_ME_BLOCK_BASED_SKIP_4x4_INTEL 0x00
+#define CL_AVC_ME_BLOCK_BASED_SKIP_8x8_INTEL 0x80
+
+#define CL_AVC_ME_INTRA_16x16_INTEL 0x0
+#define CL_AVC_ME_INTRA_8x8_INTEL 0x1
+#define CL_AVC_ME_INTRA_4x4_INTEL 0x2
+
+#define CL_AVC_ME_INTRA_LUMA_PARTITION_MASK_16x16_INTEL 0x6
+#define CL_AVC_ME_INTRA_LUMA_PARTITION_MASK_8x8_INTEL 0x5
+#define CL_AVC_ME_INTRA_LUMA_PARTITION_MASK_4x4_INTEL 0x3
+
+#define CL_AVC_ME_INTRA_NEIGHBOR_LEFT_MASK_ENABLE_INTEL 0x60
+#define CL_AVC_ME_INTRA_NEIGHBOR_UPPER_MASK_ENABLE_INTEL 0x10
+#define CL_AVC_ME_INTRA_NEIGHBOR_UPPER_RIGHT_MASK_ENABLE_INTEL 0x8
+#define CL_AVC_ME_INTRA_NEIGHBOR_UPPER_LEFT_MASK_ENABLE_INTEL 0x4
+
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_VERTICAL_INTEL 0x0
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_HORIZONTAL_INTEL 0x1
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_DC_INTEL 0x2
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_DIAGONAL_DOWN_LEFT_INTEL 0x3
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_DIAGONAL_DOWN_RIGHT_INTEL 0x4
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_PLANE_INTEL 0x4
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_VERTICAL_RIGHT_INTEL 0x5
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_HORIZONTAL_DOWN_INTEL 0x6
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_VERTICAL_LEFT_INTEL 0x7
+#define CL_AVC_ME_LUMA_PREDICTOR_MODE_HORIZONTAL_UP_INTEL 0x8
+#define CL_AVC_ME_CHROMA_PREDICTOR_MODE_DC_INTEL 0x0
+#define CL_AVC_ME_CHROMA_PREDICTOR_MODE_HORIZONTAL_INTEL 0x1
+#define CL_AVC_ME_CHROMA_PREDICTOR_MODE_VERTICAL_INTEL 0x2
+#define CL_AVC_ME_CHROMA_PREDICTOR_MODE_PLANE_INTEL 0x3
+
+#define CL_AVC_ME_FRAME_FORWARD_INTEL 0x1
+#define CL_AVC_ME_FRAME_BACKWARD_INTEL 0x2
+#define CL_AVC_ME_FRAME_DUAL_INTEL 0x3
+
+#define CL_AVC_ME_SLICE_TYPE_PRED_INTEL 0x0
+#define CL_AVC_ME_SLICE_TYPE_BPRED_INTEL 0x1
+#define CL_AVC_ME_SLICE_TYPE_INTRA_INTEL 0x2
+
+#define CL_AVC_ME_INTERLACED_SCAN_TOP_FIELD_INTEL 0x0
+#define CL_AVC_ME_INTERLACED_SCAN_BOTTOM_FIELD_INTEL 0x1
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* __CL_EXT_INTEL_H */
diff --git a/include/CL/cl_gl.h b/include/CL/cl_gl.h
index e52c1b693d4..fbdaf629778 100644
--- a/include/CL/cl_gl.h
+++ b/include/CL/cl_gl.h
@@ -1,5 +1,5 @@
/**********************************************************************************
- * Copyright (c) 2008 - 2012 The Khronos Group Inc.
+ * Copyright (c) 2008-2019 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -24,11 +29,7 @@
#ifndef __OPENCL_CL_GL_H
#define __OPENCL_CL_GL_H
-#ifdef __APPLE__
-#include <OpenCL/cl.h>
-#else
#include <CL/cl.h>
-#endif
#ifdef __cplusplus
extern "C" {
@@ -44,110 +45,118 @@ typedef struct __GLsync *cl_GLsync;
#define CL_GL_OBJECT_TEXTURE2D 0x2001
#define CL_GL_OBJECT_TEXTURE3D 0x2002
#define CL_GL_OBJECT_RENDERBUFFER 0x2003
+#ifdef CL_VERSION_1_2
#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
+#endif
/* cl_gl_texture_info */
#define CL_GL_TEXTURE_TARGET 0x2004
#define CL_GL_MIPMAP_LEVEL 0x2005
+#ifdef CL_VERSION_1_2
#define CL_GL_NUM_SAMPLES 0x2012
+#endif
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;
+clCreateFromGLBuffer(cl_context context,
+ cl_mem_flags flags,
+ cl_GLuint bufobj,
+ cl_int * errcode_ret) CL_API_SUFFIX__VERSION_1_0;
+
+#ifdef CL_VERSION_1_2
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;
-
+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;
+
+#endif
+
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;
+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;
-
+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;
+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;
+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;
+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;
-
+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;
-
+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;
-
+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,
diff --git a/include/CL/cl_gl_ext.h b/include/CL/cl_gl_ext.h
index 77d53536f62..c26d31abed0 100644
--- a/include/CL/cl_gl_ext.h
+++ b/include/CL/cl_gl_ext.h
@@ -1,5 +1,5 @@
/**********************************************************************************
- * Copyright (c) 2008-2012 The Khronos Group Inc.
+ * Copyright (c) 2008-2019 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -21,11 +26,6 @@
* 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
@@ -33,34 +33,17 @@
extern "C" {
#endif
-#ifdef __APPLE__
- #include <OpenCL/cl_gl.h>
-#else
- #include <CL/cl_gl.h>
-#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.
- */
+#include <CL/cl_gl.h>
/*
- * cl_khr_gl_event extension
- * See section 9.9 in the OpenCL 1.1 spec for more information
+ * cl_khr_gl_event extension
*/
#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;
+clCreateEventFromGLsyncKHR(cl_context context,
+ cl_GLsync cl_GLsync,
+ cl_int * errcode_ret) CL_EXT_SUFFIX__VERSION_1_1;
#ifdef __cplusplus
}
diff --git a/include/CL/cl_platform.h b/include/CL/cl_platform.h
index 7f6f5e8a743..184db2516dc 100644
--- a/include/CL/cl_platform.h
+++ b/include/CL/cl_platform.h
@@ -1,5 +1,5 @@
/**********************************************************************************
- * Copyright (c) 2008-2012 The Khronos Group Inc.
+ * Copyright (c) 2008-2018 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -21,15 +26,10 @@
* 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 <AvailabilityMacros.h>
-#endif
+#include <CL/cl_version.h>
#ifdef __cplusplus
extern "C" {
@@ -45,76 +45,123 @@ extern "C" {
#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
+/*
+ * Deprecation flags refer to the last version of the header in which the
+ * feature was not deprecated.
+ *
+ * E.g. VERSION_1_1_DEPRECATED means the feature is present in 1.1 without
+ * deprecation but is deprecated in versions later than 1.1.
+ */
+
+#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
+#define CL_API_SUFFIX__VERSION_2_0
+#define CL_EXT_SUFFIX__VERSION_2_0
+#define CL_API_SUFFIX__VERSION_2_1
+#define CL_EXT_SUFFIX__VERSION_2_1
+#define CL_API_SUFFIX__VERSION_2_2
+#define CL_EXT_SUFFIX__VERSION_2_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
- #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
+ #define CL_EXT_SUFFIX__VERSION_1_1_DEPRECATED __attribute__((deprecated))
+ #define CL_EXT_PREFIX__VERSION_1_1_DEPRECATED
#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
+
+ #ifdef CL_USE_DEPRECATED_OPENCL_1_2_APIS
+ #define CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_1_2_DEPRECATED
#else
+ #define CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED __attribute__((deprecated))
+ #define CL_EXT_PREFIX__VERSION_1_2_DEPRECATED
+ #endif
+
+ #ifdef CL_USE_DEPRECATED_OPENCL_2_0_APIS
+ #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED
+ #else
+ #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED __attribute__((deprecated))
+ #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED
+ #endif
+
+ #ifdef CL_USE_DEPRECATED_OPENCL_2_1_APIS
+ #define CL_EXT_SUFFIX__VERSION_2_1_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_2_1_DEPRECATED
+ #else
+ #define CL_EXT_SUFFIX__VERSION_2_1_DEPRECATED __attribute__((deprecated))
+ #define CL_EXT_PREFIX__VERSION_2_1_DEPRECATED
+ #endif
+#elif defined(_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
+
+ #ifdef CL_USE_DEPRECATED_OPENCL_1_2_APIS
+ #define CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_1_2_DEPRECATED
+ #else
+ #define CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_1_2_DEPRECATED __declspec(deprecated)
+ #endif
+
+ #ifdef CL_USE_DEPRECATED_OPENCL_2_0_APIS
+ #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED
+ #else
+ #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED __declspec(deprecated)
+ #endif
+
+ #ifdef CL_USE_DEPRECATED_OPENCL_2_1_APIS
+ #define CL_EXT_SUFFIX__VERSION_2_1_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_2_1_DEPRECATED
+ #else
+ #define CL_EXT_SUFFIX__VERSION_2_1_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_2_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
+
+ #define CL_EXT_SUFFIX__VERSION_1_2_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_1_2_DEPRECATED
+
+ #define CL_EXT_SUFFIX__VERSION_2_0_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_2_0_DEPRECATED
+
+ #define CL_EXT_SUFFIX__VERSION_2_1_DEPRECATED
+ #define CL_EXT_PREFIX__VERSION_2_1_DEPRECATED
#endif
#if (defined (_WIN32) && defined(_MSC_VER))
@@ -159,7 +206,18 @@ typedef double cl_double;
#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_FLT_EPSILON 1.1920928955078125e-7f
+
+#define CL_HALF_DIG 3
+#define CL_HALF_MANT_DIG 11
+#define CL_HALF_MAX_10_EXP +4
+#define CL_HALF_MAX_EXP +16
+#define CL_HALF_MIN_10_EXP -4
+#define CL_HALF_MIN_EXP -13
+#define CL_HALF_RADIX 2
+#define CL_HALF_MAX 65504.0f
+#define CL_HALF_MIN 6.103515625e-05f
+#define CL_HALF_EPSILON 9.765625e-04f
#define CL_DBL_DIG 15
#define CL_DBL_MANT_DIG 53
@@ -168,37 +226,37 @@ typedef double cl_double;
#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_MAX 1.7976931348623158e+308
#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_M_E 2.7182818284590452354
+#define CL_M_LOG2E 1.4426950408889634074
+#define CL_M_LOG10E 0.43429448190325182765
+#define CL_M_LN2 0.69314718055994530942
+#define CL_M_LN10 2.30258509299404568402
+#define CL_M_PI 3.14159265358979323846
+#define CL_M_PI_2 1.57079632679489661923
+#define CL_M_PI_4 0.78539816339744830962
+#define CL_M_1_PI 0.31830988618379067154
+#define CL_M_2_PI 0.63661977236758134308
+#define CL_M_2_SQRTPI 1.12837916709551257390
+#define CL_M_SQRT2 1.41421356237309504880
+#define CL_M_SQRT1_2 0.70710678118654752440
+
+#define CL_M_E_F 2.718281828f
+#define CL_M_LOG2E_F 1.442695041f
+#define CL_M_LOG10E_F 0.434294482f
+#define CL_M_LN2_F 0.693147181f
+#define CL_M_LN10_F 2.302585093f
+#define CL_M_PI_F 3.141592654f
+#define CL_M_PI_2_F 1.570796327f
+#define CL_M_PI_4_F 0.785398163f
+#define CL_M_1_PI_F 0.318309886f
+#define CL_M_2_PI_F 0.636619772f
+#define CL_M_2_SQRTPI_F 1.128379167f
+#define CL_M_SQRT2_F 1.414213562f
+#define CL_M_SQRT1_2_F 0.707106781f
#define CL_NAN (CL_INFINITY - CL_INFINITY)
#define CL_HUGE_VALF ((cl_float) 1e50)
@@ -248,9 +306,20 @@ typedef double cl_double __attribute__((aligned(8)));
#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_FLT_MAX 340282346638528859811704183484516925440.0f
+#define CL_FLT_MIN 1.175494350822287507969e-38f
+#define CL_FLT_EPSILON 1.1920928955078125e-7f
+
+#define CL_HALF_DIG 3
+#define CL_HALF_MANT_DIG 11
+#define CL_HALF_MAX_10_EXP +4
+#define CL_HALF_MAX_EXP +16
+#define CL_HALF_MIN_10_EXP -4
+#define CL_HALF_MIN_EXP -13
+#define CL_HALF_RADIX 2
+#define CL_HALF_MAX 65504.0f
+#define CL_HALF_MIN 6.103515625e-05f
+#define CL_HALF_EPSILON 9.765625e-04f
#define CL_DBL_DIG 15
#define CL_DBL_MANT_DIG 53
@@ -259,37 +328,37 @@ typedef double cl_double __attribute__((aligned(8)));
#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
+#define CL_DBL_MAX 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.0
+#define CL_DBL_MIN 2.225073858507201383090e-308
+#define CL_DBL_EPSILON 2.220446049250313080847e-16
+
+#define CL_M_E 2.7182818284590452354
+#define CL_M_LOG2E 1.4426950408889634074
+#define CL_M_LOG10E 0.43429448190325182765
+#define CL_M_LN2 0.69314718055994530942
+#define CL_M_LN10 2.30258509299404568402
+#define CL_M_PI 3.14159265358979323846
+#define CL_M_PI_2 1.57079632679489661923
+#define CL_M_PI_4 0.78539816339744830962
+#define CL_M_1_PI 0.31830988618379067154
+#define CL_M_2_PI 0.63661977236758134308
+#define CL_M_2_SQRTPI 1.12837916709551257390
+#define CL_M_SQRT2 1.41421356237309504880
+#define CL_M_SQRT1_2 0.70710678118654752440
+
+#define CL_M_E_F 2.718281828f
+#define CL_M_LOG2E_F 1.442695041f
+#define CL_M_LOG10E_F 0.434294482f
+#define CL_M_LN2_F 0.693147181f
+#define CL_M_LN10_F 2.302585093f
+#define CL_M_PI_F 3.141592654f
+#define CL_M_PI_2_F 1.570796327f
+#define CL_M_PI_4_F 0.785398163f
+#define CL_M_1_PI_F 0.318309886f
+#define CL_M_2_PI_F 0.636619772f
+#define CL_M_2_SQRTPI_F 1.128379167f
+#define CL_M_SQRT2_F 1.414213562f
+#define CL_M_SQRT1_2_F 0.707106781f
#if defined( __GNUC__ )
#define CL_HUGE_VALF __builtin_huge_valf()
@@ -299,7 +368,7 @@ typedef double cl_double __attribute__((aligned(8)));
#define CL_HUGE_VALF ((cl_float) 1e50)
#define CL_HUGE_VAL ((cl_double) 1e500)
float nanf( const char * );
- #define CL_NAN nanf( "" )
+ #define CL_NAN nanf( "" )
#endif
#define CL_MAXFLOAT CL_FLT_MAX
#define CL_INFINITY CL_HUGE_VALF
@@ -314,17 +383,17 @@ typedef int cl_GLint;
typedef unsigned int cl_GLenum;
/*
- * Vector types
+ * Vector types
*
- * Note: OpenCL requires that all types be naturally aligned.
+ * 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
+ * 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.
+ * the start of the struct must itself be 16-byte aligned.
*
* Maintaining proper alignment is the user's responsibility.
*/
@@ -332,13 +401,13 @@ typedef unsigned int cl_GLenum;
/* Define basic vector types */
#if defined( __VEC__ )
#include <altivec.h> /* 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;
+ 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
@@ -438,7 +507,7 @@ typedef unsigned int cl_GLenum;
#if defined( __MINGW64__ )
#include <intrin.h>
#else
- #include <immintrin.h>
+ #include <immintrin.h>
#endif
#if defined( __GNUC__ )
typedef cl_float __cl_float8 __attribute__((vector_size(32)));
@@ -452,18 +521,23 @@ typedef unsigned int cl_GLenum;
#endif
/* Define capabilities for anonymous struct members. */
-#if defined( __GNUC__) && ! defined( __STRICT_ANSI__ )
+#if !defined(__cplusplus) && defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
+#define __CL_HAS_ANON_STRUCT__ 1
+#define __CL_ANON_STRUCT__
+#elif defined( __GNUC__) && ! defined( __STRICT_ANSI__ )
#define __CL_HAS_ANON_STRUCT__ 1
#define __CL_ANON_STRUCT__ __extension__
-#elif defined( _WIN32) && (_MSC_VER >= 1500)
+#elif defined( _WIN32) && defined(_MSC_VER)
+ #if _MSC_VER >= 1500
/* Microsoft Developer Studio 2008 supports anonymous structs, but
* complains by default. */
-#define __CL_HAS_ANON_STRUCT__ 1
-#define __CL_ANON_STRUCT__
+ #define __CL_HAS_ANON_STRUCT__ 1
+ #define __CL_ANON_STRUCT__
/* Disable warning C4201: nonstandard extension used : nameless
* struct/union */
-#pragma warning( push )
-#pragma warning( disable : 4201 )
+ #pragma warning( push )
+ #pragma warning( disable : 4201 )
+ #endif
#else
#define __CL_HAS_ANON_STRUCT__ 0
#define __CL_ANON_STRUCT__
@@ -502,7 +576,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_char s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_char lo, hi; };
#endif
-#if defined( __CL_CHAR2__)
+#if defined( __CL_CHAR2__)
__cl_char2 v2;
#endif
}cl_char2;
@@ -515,10 +589,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_char s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_char2 lo, hi; };
#endif
-#if defined( __CL_CHAR2__)
+#if defined( __CL_CHAR2__)
__cl_char2 v2[2];
#endif
-#if defined( __CL_CHAR4__)
+#if defined( __CL_CHAR4__)
__cl_char4 v4;
#endif
}cl_char4;
@@ -534,10 +608,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_char s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_char4 lo, hi; };
#endif
-#if defined( __CL_CHAR2__)
+#if defined( __CL_CHAR2__)
__cl_char2 v2[4];
#endif
-#if defined( __CL_CHAR4__)
+#if defined( __CL_CHAR4__)
__cl_char4 v4[2];
#endif
#if defined( __CL_CHAR8__ )
@@ -553,10 +627,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_char s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_char8 lo, hi; };
#endif
-#if defined( __CL_CHAR2__)
+#if defined( __CL_CHAR2__)
__cl_char2 v2[8];
#endif
-#if defined( __CL_CHAR4__)
+#if defined( __CL_CHAR4__)
__cl_char4 v4[4];
#endif
#if defined( __CL_CHAR8__ )
@@ -577,7 +651,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_uchar s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_uchar lo, hi; };
#endif
-#if defined( __cl_uchar2__)
+#if defined( __cl_uchar2__)
__cl_uchar2 v2;
#endif
}cl_uchar2;
@@ -590,10 +664,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_uchar s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_uchar2 lo, hi; };
#endif
-#if defined( __CL_UCHAR2__)
+#if defined( __CL_UCHAR2__)
__cl_uchar2 v2[2];
#endif
-#if defined( __CL_UCHAR4__)
+#if defined( __CL_UCHAR4__)
__cl_uchar4 v4;
#endif
}cl_uchar4;
@@ -609,10 +683,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_uchar s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_uchar4 lo, hi; };
#endif
-#if defined( __CL_UCHAR2__)
+#if defined( __CL_UCHAR2__)
__cl_uchar2 v2[4];
#endif
-#if defined( __CL_UCHAR4__)
+#if defined( __CL_UCHAR4__)
__cl_uchar4 v4[2];
#endif
#if defined( __CL_UCHAR8__ )
@@ -628,10 +702,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_uchar s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_uchar8 lo, hi; };
#endif
-#if defined( __CL_UCHAR2__)
+#if defined( __CL_UCHAR2__)
__cl_uchar2 v2[8];
#endif
-#if defined( __CL_UCHAR4__)
+#if defined( __CL_UCHAR4__)
__cl_uchar4 v4[4];
#endif
#if defined( __CL_UCHAR8__ )
@@ -652,7 +726,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_short s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_short lo, hi; };
#endif
-#if defined( __CL_SHORT2__)
+#if defined( __CL_SHORT2__)
__cl_short2 v2;
#endif
}cl_short2;
@@ -665,10 +739,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_short s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_short2 lo, hi; };
#endif
-#if defined( __CL_SHORT2__)
+#if defined( __CL_SHORT2__)
__cl_short2 v2[2];
#endif
-#if defined( __CL_SHORT4__)
+#if defined( __CL_SHORT4__)
__cl_short4 v4;
#endif
}cl_short4;
@@ -684,10 +758,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_short s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_short4 lo, hi; };
#endif
-#if defined( __CL_SHORT2__)
+#if defined( __CL_SHORT2__)
__cl_short2 v2[4];
#endif
-#if defined( __CL_SHORT4__)
+#if defined( __CL_SHORT4__)
__cl_short4 v4[2];
#endif
#if defined( __CL_SHORT8__ )
@@ -703,10 +777,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_short s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_short8 lo, hi; };
#endif
-#if defined( __CL_SHORT2__)
+#if defined( __CL_SHORT2__)
__cl_short2 v2[8];
#endif
-#if defined( __CL_SHORT4__)
+#if defined( __CL_SHORT4__)
__cl_short4 v4[4];
#endif
#if defined( __CL_SHORT8__ )
@@ -727,7 +801,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_ushort s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_ushort lo, hi; };
#endif
-#if defined( __CL_USHORT2__)
+#if defined( __CL_USHORT2__)
__cl_ushort2 v2;
#endif
}cl_ushort2;
@@ -740,10 +814,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_ushort s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_ushort2 lo, hi; };
#endif
-#if defined( __CL_USHORT2__)
+#if defined( __CL_USHORT2__)
__cl_ushort2 v2[2];
#endif
-#if defined( __CL_USHORT4__)
+#if defined( __CL_USHORT4__)
__cl_ushort4 v4;
#endif
}cl_ushort4;
@@ -759,10 +833,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_ushort s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_ushort4 lo, hi; };
#endif
-#if defined( __CL_USHORT2__)
+#if defined( __CL_USHORT2__)
__cl_ushort2 v2[4];
#endif
-#if defined( __CL_USHORT4__)
+#if defined( __CL_USHORT4__)
__cl_ushort4 v4[2];
#endif
#if defined( __CL_USHORT8__ )
@@ -778,10 +852,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_ushort s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_ushort8 lo, hi; };
#endif
-#if defined( __CL_USHORT2__)
+#if defined( __CL_USHORT2__)
__cl_ushort2 v2[8];
#endif
-#if defined( __CL_USHORT4__)
+#if defined( __CL_USHORT4__)
__cl_ushort4 v4[4];
#endif
#if defined( __CL_USHORT8__ )
@@ -792,6 +866,81 @@ typedef union
#endif
}cl_ushort16;
+
+/* ---- cl_halfn ---- */
+typedef union
+{
+ cl_half CL_ALIGNED(4) s[2];
+#if __CL_HAS_ANON_STRUCT__
+ __CL_ANON_STRUCT__ struct{ cl_half x, y; };
+ __CL_ANON_STRUCT__ struct{ cl_half s0, s1; };
+ __CL_ANON_STRUCT__ struct{ cl_half lo, hi; };
+#endif
+#if defined( __CL_HALF2__)
+ __cl_half2 v2;
+#endif
+}cl_half2;
+
+typedef union
+{
+ cl_half CL_ALIGNED(8) s[4];
+#if __CL_HAS_ANON_STRUCT__
+ __CL_ANON_STRUCT__ struct{ cl_half x, y, z, w; };
+ __CL_ANON_STRUCT__ struct{ cl_half s0, s1, s2, s3; };
+ __CL_ANON_STRUCT__ struct{ cl_half2 lo, hi; };
+#endif
+#if defined( __CL_HALF2__)
+ __cl_half2 v2[2];
+#endif
+#if defined( __CL_HALF4__)
+ __cl_half4 v4;
+#endif
+}cl_half4;
+
+/* cl_half3 is identical in size, alignment and behavior to cl_half4. See section 6.1.5. */
+typedef cl_half4 cl_half3;
+
+typedef union
+{
+ cl_half CL_ALIGNED(16) s[8];
+#if __CL_HAS_ANON_STRUCT__
+ __CL_ANON_STRUCT__ struct{ cl_half x, y, z, w; };
+ __CL_ANON_STRUCT__ struct{ cl_half s0, s1, s2, s3, s4, s5, s6, s7; };
+ __CL_ANON_STRUCT__ struct{ cl_half4 lo, hi; };
+#endif
+#if defined( __CL_HALF2__)
+ __cl_half2 v2[4];
+#endif
+#if defined( __CL_HALF4__)
+ __cl_half4 v4[2];
+#endif
+#if defined( __CL_HALF8__ )
+ __cl_half8 v8;
+#endif
+}cl_half8;
+
+typedef union
+{
+ cl_half CL_ALIGNED(32) s[16];
+#if __CL_HAS_ANON_STRUCT__
+ __CL_ANON_STRUCT__ struct{ cl_half x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
+ __CL_ANON_STRUCT__ struct{ cl_half s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
+ __CL_ANON_STRUCT__ struct{ cl_half8 lo, hi; };
+#endif
+#if defined( __CL_HALF2__)
+ __cl_half2 v2[8];
+#endif
+#if defined( __CL_HALF4__)
+ __cl_half4 v4[4];
+#endif
+#if defined( __CL_HALF8__ )
+ __cl_half8 v8[2];
+#endif
+#if defined( __CL_HALF16__ )
+ __cl_half16 v16;
+#endif
+}cl_half16;
+
/* ---- cl_intn ---- */
typedef union
{
@@ -801,7 +950,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_int s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_int lo, hi; };
#endif
-#if defined( __CL_INT2__)
+#if defined( __CL_INT2__)
__cl_int2 v2;
#endif
}cl_int2;
@@ -814,10 +963,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_int s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_int2 lo, hi; };
#endif
-#if defined( __CL_INT2__)
+#if defined( __CL_INT2__)
__cl_int2 v2[2];
#endif
-#if defined( __CL_INT4__)
+#if defined( __CL_INT4__)
__cl_int4 v4;
#endif
}cl_int4;
@@ -833,10 +982,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_int s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_int4 lo, hi; };
#endif
-#if defined( __CL_INT2__)
+#if defined( __CL_INT2__)
__cl_int2 v2[4];
#endif
-#if defined( __CL_INT4__)
+#if defined( __CL_INT4__)
__cl_int4 v4[2];
#endif
#if defined( __CL_INT8__ )
@@ -852,10 +1001,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_int s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_int8 lo, hi; };
#endif
-#if defined( __CL_INT2__)
+#if defined( __CL_INT2__)
__cl_int2 v2[8];
#endif
-#if defined( __CL_INT4__)
+#if defined( __CL_INT4__)
__cl_int4 v4[4];
#endif
#if defined( __CL_INT8__ )
@@ -876,7 +1025,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_uint s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_uint lo, hi; };
#endif
-#if defined( __CL_UINT2__)
+#if defined( __CL_UINT2__)
__cl_uint2 v2;
#endif
}cl_uint2;
@@ -889,10 +1038,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_uint s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_uint2 lo, hi; };
#endif
-#if defined( __CL_UINT2__)
+#if defined( __CL_UINT2__)
__cl_uint2 v2[2];
#endif
-#if defined( __CL_UINT4__)
+#if defined( __CL_UINT4__)
__cl_uint4 v4;
#endif
}cl_uint4;
@@ -908,10 +1057,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_uint s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_uint4 lo, hi; };
#endif
-#if defined( __CL_UINT2__)
+#if defined( __CL_UINT2__)
__cl_uint2 v2[4];
#endif
-#if defined( __CL_UINT4__)
+#if defined( __CL_UINT4__)
__cl_uint4 v4[2];
#endif
#if defined( __CL_UINT8__ )
@@ -927,10 +1076,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_uint s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_uint8 lo, hi; };
#endif
-#if defined( __CL_UINT2__)
+#if defined( __CL_UINT2__)
__cl_uint2 v2[8];
#endif
-#if defined( __CL_UINT4__)
+#if defined( __CL_UINT4__)
__cl_uint4 v4[4];
#endif
#if defined( __CL_UINT8__ )
@@ -950,7 +1099,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_long s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_long lo, hi; };
#endif
-#if defined( __CL_LONG2__)
+#if defined( __CL_LONG2__)
__cl_long2 v2;
#endif
}cl_long2;
@@ -963,10 +1112,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_long s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_long2 lo, hi; };
#endif
-#if defined( __CL_LONG2__)
+#if defined( __CL_LONG2__)
__cl_long2 v2[2];
#endif
-#if defined( __CL_LONG4__)
+#if defined( __CL_LONG4__)
__cl_long4 v4;
#endif
}cl_long4;
@@ -982,10 +1131,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_long s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_long4 lo, hi; };
#endif
-#if defined( __CL_LONG2__)
+#if defined( __CL_LONG2__)
__cl_long2 v2[4];
#endif
-#if defined( __CL_LONG4__)
+#if defined( __CL_LONG4__)
__cl_long4 v4[2];
#endif
#if defined( __CL_LONG8__ )
@@ -1001,10 +1150,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_long s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_long8 lo, hi; };
#endif
-#if defined( __CL_LONG2__)
+#if defined( __CL_LONG2__)
__cl_long2 v2[8];
#endif
-#if defined( __CL_LONG4__)
+#if defined( __CL_LONG4__)
__cl_long4 v4[4];
#endif
#if defined( __CL_LONG8__ )
@@ -1025,7 +1174,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_ulong s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_ulong lo, hi; };
#endif
-#if defined( __CL_ULONG2__)
+#if defined( __CL_ULONG2__)
__cl_ulong2 v2;
#endif
}cl_ulong2;
@@ -1038,10 +1187,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_ulong s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_ulong2 lo, hi; };
#endif
-#if defined( __CL_ULONG2__)
+#if defined( __CL_ULONG2__)
__cl_ulong2 v2[2];
#endif
-#if defined( __CL_ULONG4__)
+#if defined( __CL_ULONG4__)
__cl_ulong4 v4;
#endif
}cl_ulong4;
@@ -1057,10 +1206,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_ulong s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_ulong4 lo, hi; };
#endif
-#if defined( __CL_ULONG2__)
+#if defined( __CL_ULONG2__)
__cl_ulong2 v2[4];
#endif
-#if defined( __CL_ULONG4__)
+#if defined( __CL_ULONG4__)
__cl_ulong4 v4[2];
#endif
#if defined( __CL_ULONG8__ )
@@ -1076,10 +1225,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_ulong s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_ulong8 lo, hi; };
#endif
-#if defined( __CL_ULONG2__)
+#if defined( __CL_ULONG2__)
__cl_ulong2 v2[8];
#endif
-#if defined( __CL_ULONG4__)
+#if defined( __CL_ULONG4__)
__cl_ulong4 v4[4];
#endif
#if defined( __CL_ULONG8__ )
@@ -1101,7 +1250,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_float s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_float lo, hi; };
#endif
-#if defined( __CL_FLOAT2__)
+#if defined( __CL_FLOAT2__)
__cl_float2 v2;
#endif
}cl_float2;
@@ -1114,10 +1263,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_float s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_float2 lo, hi; };
#endif
-#if defined( __CL_FLOAT2__)
+#if defined( __CL_FLOAT2__)
__cl_float2 v2[2];
#endif
-#if defined( __CL_FLOAT4__)
+#if defined( __CL_FLOAT4__)
__cl_float4 v4;
#endif
}cl_float4;
@@ -1133,10 +1282,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_float s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_float4 lo, hi; };
#endif
-#if defined( __CL_FLOAT2__)
+#if defined( __CL_FLOAT2__)
__cl_float2 v2[4];
#endif
-#if defined( __CL_FLOAT4__)
+#if defined( __CL_FLOAT4__)
__cl_float4 v4[2];
#endif
#if defined( __CL_FLOAT8__ )
@@ -1152,10 +1301,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_float s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_float8 lo, hi; };
#endif
-#if defined( __CL_FLOAT2__)
+#if defined( __CL_FLOAT2__)
__cl_float2 v2[8];
#endif
-#if defined( __CL_FLOAT4__)
+#if defined( __CL_FLOAT4__)
__cl_float4 v4[4];
#endif
#if defined( __CL_FLOAT8__ )
@@ -1176,7 +1325,7 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_double s0, s1; };
__CL_ANON_STRUCT__ struct{ cl_double lo, hi; };
#endif
-#if defined( __CL_DOUBLE2__)
+#if defined( __CL_DOUBLE2__)
__cl_double2 v2;
#endif
}cl_double2;
@@ -1189,10 +1338,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_double s0, s1, s2, s3; };
__CL_ANON_STRUCT__ struct{ cl_double2 lo, hi; };
#endif
-#if defined( __CL_DOUBLE2__)
+#if defined( __CL_DOUBLE2__)
__cl_double2 v2[2];
#endif
-#if defined( __CL_DOUBLE4__)
+#if defined( __CL_DOUBLE4__)
__cl_double4 v4;
#endif
}cl_double4;
@@ -1208,10 +1357,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_double s0, s1, s2, s3, s4, s5, s6, s7; };
__CL_ANON_STRUCT__ struct{ cl_double4 lo, hi; };
#endif
-#if defined( __CL_DOUBLE2__)
+#if defined( __CL_DOUBLE2__)
__cl_double2 v2[4];
#endif
-#if defined( __CL_DOUBLE4__)
+#if defined( __CL_DOUBLE4__)
__cl_double4 v4[2];
#endif
#if defined( __CL_DOUBLE8__ )
@@ -1227,10 +1376,10 @@ typedef union
__CL_ANON_STRUCT__ struct{ cl_double s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
__CL_ANON_STRUCT__ struct{ cl_double8 lo, hi; };
#endif
-#if defined( __CL_DOUBLE2__)
+#if defined( __CL_DOUBLE2__)
__cl_double2 v2[8];
#endif
-#if defined( __CL_DOUBLE4__)
+#if defined( __CL_DOUBLE4__)
__cl_double4 v4[4];
#endif
#if defined( __CL_DOUBLE8__ )
@@ -1241,9 +1390,9 @@ typedef union
#endif
}cl_double16;
-/* Macro to facilitate debugging
+/* Macro to facilitate debugging
* Usage:
- * Place CL_PROGRAM_STRING_DEBUG_INFO on the line before the first line of your source.
+ * 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 ";
@@ -1258,21 +1407,23 @@ typedef union
* } \n\
* ";
*
- * This should correctly set up the line, (column) and file information for your source
+ * 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"
-
+#define CL_PROGRAM_STRING_DEBUG_INFO "#line " _CL_STRINGIFY(__LINE__) " \"" __FILE__ "\" \n\n"
+
#ifdef __cplusplus
}
#endif
#undef __CL_HAS_ANON_STRUCT__
#undef __CL_ANON_STRUCT__
-#if defined( _WIN32) && (_MSC_VER >= 1500)
-#pragma warning( pop )
+#if defined( _WIN32) && defined(_MSC_VER)
+ #if _MSC_VER >=1500
+ #pragma warning( pop )
+ #endif
#endif
#endif /* __CL_PLATFORM_H */
diff --git a/include/CL/cl_va_api_media_sharing_intel.h b/include/CL/cl_va_api_media_sharing_intel.h
new file mode 100644
index 00000000000..934f3f52a07
--- /dev/null
+++ b/include/CL/cl_va_api_media_sharing_intel.h
@@ -0,0 +1,172 @@
+/**********************************************************************************
+ * Copyright (c) 2008-2019 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.
+ *
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
+ * 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.
+ **********************************************************************************/
+/*****************************************************************************\
+
+Copyright (c) 2013-2019 Intel Corporation All Rights Reserved.
+
+THESE MATERIALS ARE PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL OR ITS
+CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY OR TORT (INCLUDING
+NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THESE
+MATERIALS, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+File Name: cl_va_api_media_sharing_intel.h
+
+Abstract:
+
+Notes:
+
+\*****************************************************************************/
+
+
+#ifndef __OPENCL_CL_VA_API_MEDIA_SHARING_INTEL_H
+#define __OPENCL_CL_VA_API_MEDIA_SHARING_INTEL_H
+
+#include <CL/cl.h>
+#include <CL/cl_platform.h>
+#include <va/va.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/******************************************
+* cl_intel_va_api_media_sharing extension *
+*******************************************/
+
+#define cl_intel_va_api_media_sharing 1
+
+/* error codes */
+#define CL_INVALID_VA_API_MEDIA_ADAPTER_INTEL -1098
+#define CL_INVALID_VA_API_MEDIA_SURFACE_INTEL -1099
+#define CL_VA_API_MEDIA_SURFACE_ALREADY_ACQUIRED_INTEL -1100
+#define CL_VA_API_MEDIA_SURFACE_NOT_ACQUIRED_INTEL -1101
+
+/* cl_va_api_device_source_intel */
+#define CL_VA_API_DISPLAY_INTEL 0x4094
+
+/* cl_va_api_device_set_intel */
+#define CL_PREFERRED_DEVICES_FOR_VA_API_INTEL 0x4095
+#define CL_ALL_DEVICES_FOR_VA_API_INTEL 0x4096
+
+/* cl_context_info */
+#define CL_CONTEXT_VA_API_DISPLAY_INTEL 0x4097
+
+/* cl_mem_info */
+#define CL_MEM_VA_API_MEDIA_SURFACE_INTEL 0x4098
+
+/* cl_image_info */
+#define CL_IMAGE_VA_API_PLANE_INTEL 0x4099
+
+/* cl_command_type */
+#define CL_COMMAND_ACQUIRE_VA_API_MEDIA_SURFACES_INTEL 0x409A
+#define CL_COMMAND_RELEASE_VA_API_MEDIA_SURFACES_INTEL 0x409B
+
+typedef cl_uint cl_va_api_device_source_intel;
+typedef cl_uint cl_va_api_device_set_intel;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clGetDeviceIDsFromVA_APIMediaAdapterINTEL(
+ cl_platform_id platform,
+ cl_va_api_device_source_intel media_adapter_type,
+ void* media_adapter,
+ cl_va_api_device_set_intel media_adapter_set,
+ cl_uint num_entries,
+ cl_device_id* devices,
+ cl_uint* num_devices) CL_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL * clGetDeviceIDsFromVA_APIMediaAdapterINTEL_fn)(
+ cl_platform_id platform,
+ cl_va_api_device_source_intel media_adapter_type,
+ void* media_adapter,
+ cl_va_api_device_set_intel media_adapter_set,
+ cl_uint num_entries,
+ cl_device_id* devices,
+ cl_uint* num_devices) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_mem CL_API_CALL
+clCreateFromVA_APIMediaSurfaceINTEL(
+ cl_context context,
+ cl_mem_flags flags,
+ VASurfaceID* surface,
+ cl_uint plane,
+ cl_int* errcode_ret) CL_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_mem (CL_API_CALL * clCreateFromVA_APIMediaSurfaceINTEL_fn)(
+ cl_context context,
+ cl_mem_flags flags,
+ VASurfaceID* surface,
+ cl_uint plane,
+ cl_int* errcode_ret) CL_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueAcquireVA_APIMediaSurfacesINTEL(
+ 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_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueAcquireVA_APIMediaSurfacesINTEL_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_EXT_SUFFIX__VERSION_1_2;
+
+extern CL_API_ENTRY cl_int CL_API_CALL
+clEnqueueReleaseVA_APIMediaSurfacesINTEL(
+ 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_EXT_SUFFIX__VERSION_1_2;
+
+typedef CL_API_ENTRY cl_int (CL_API_CALL *clEnqueueReleaseVA_APIMediaSurfacesINTEL_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_EXT_SUFFIX__VERSION_1_2;
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* __OPENCL_CL_VA_API_MEDIA_SHARING_INTEL_H */
+
diff --git a/include/CL/cl_version.h b/include/CL/cl_version.h
new file mode 100644
index 00000000000..bb766cb9bbd
--- /dev/null
+++ b/include/CL/cl_version.h
@@ -0,0 +1,86 @@
+/*******************************************************************************
+ * Copyright (c) 2018 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.
+ *
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
+ * 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 __CL_VERSION_H
+#define __CL_VERSION_H
+
+/* Detect which version to target */
+#if !defined(CL_TARGET_OPENCL_VERSION)
+#pragma message("cl_version.h: CL_TARGET_OPENCL_VERSION is not defined. Defaulting to 220 (OpenCL 2.2)")
+#define CL_TARGET_OPENCL_VERSION 220
+#endif
+#if CL_TARGET_OPENCL_VERSION != 100 && \
+ CL_TARGET_OPENCL_VERSION != 110 && \
+ CL_TARGET_OPENCL_VERSION != 120 && \
+ CL_TARGET_OPENCL_VERSION != 200 && \
+ CL_TARGET_OPENCL_VERSION != 210 && \
+ CL_TARGET_OPENCL_VERSION != 220
+#pragma message("cl_version: CL_TARGET_OPENCL_VERSION is not a valid value (100, 110, 120, 200, 210, 220). Defaulting to 220 (OpenCL 2.2)")
+#undef CL_TARGET_OPENCL_VERSION
+#define CL_TARGET_OPENCL_VERSION 220
+#endif
+
+
+/* OpenCL Version */
+#if CL_TARGET_OPENCL_VERSION >= 220 && !defined(CL_VERSION_2_2)
+#define CL_VERSION_2_2 1
+#endif
+#if CL_TARGET_OPENCL_VERSION >= 210 && !defined(CL_VERSION_2_1)
+#define CL_VERSION_2_1 1
+#endif
+#if CL_TARGET_OPENCL_VERSION >= 200 && !defined(CL_VERSION_2_0)
+#define CL_VERSION_2_0 1
+#endif
+#if CL_TARGET_OPENCL_VERSION >= 120 && !defined(CL_VERSION_1_2)
+#define CL_VERSION_1_2 1
+#endif
+#if CL_TARGET_OPENCL_VERSION >= 110 && !defined(CL_VERSION_1_1)
+#define CL_VERSION_1_1 1
+#endif
+#if CL_TARGET_OPENCL_VERSION >= 100 && !defined(CL_VERSION_1_0)
+#define CL_VERSION_1_0 1
+#endif
+
+/* Allow deprecated APIs for older OpenCL versions. */
+#if CL_TARGET_OPENCL_VERSION <= 210 && !defined(CL_USE_DEPRECATED_OPENCL_2_1_APIS)
+#define CL_USE_DEPRECATED_OPENCL_2_1_APIS
+#endif
+#if CL_TARGET_OPENCL_VERSION <= 200 && !defined(CL_USE_DEPRECATED_OPENCL_2_0_APIS)
+#define CL_USE_DEPRECATED_OPENCL_2_0_APIS
+#endif
+#if CL_TARGET_OPENCL_VERSION <= 120 && !defined(CL_USE_DEPRECATED_OPENCL_1_2_APIS)
+#define CL_USE_DEPRECATED_OPENCL_1_2_APIS
+#endif
+#if CL_TARGET_OPENCL_VERSION <= 110 && !defined(CL_USE_DEPRECATED_OPENCL_1_1_APIS)
+#define CL_USE_DEPRECATED_OPENCL_1_1_APIS
+#endif
+#if CL_TARGET_OPENCL_VERSION <= 100 && !defined(CL_USE_DEPRECATED_OPENCL_1_0_APIS)
+#define CL_USE_DEPRECATED_OPENCL_1_0_APIS
+#endif
+
+#endif /* __CL_VERSION_H */
diff --git a/include/CL/opencl.h b/include/CL/opencl.h
index 3f00524719b..143d1d2dc67 100644
--- a/include/CL/opencl.h
+++ b/include/CL/opencl.h
@@ -1,5 +1,5 @@
/*******************************************************************************
- * Copyright (c) 2008-2012 The Khronos Group Inc.
+ * Copyright (c) 2008-2015 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
@@ -12,6 +12,11 @@
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Materials.
*
+ * MODIFICATIONS TO THIS FILE MAY MEAN IT NO LONGER ACCURATELY REFLECTS
+ * KHRONOS STANDARDS. THE UNMODIFIED, NORMATIVE VERSIONS OF KHRONOS
+ * SPECIFICATIONS AND HEADER INFORMATION ARE LOCATED AT
+ * https://www.khronos.org/registry/
+ *
* 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.
@@ -30,25 +35,13 @@
extern "C" {
#endif
-#ifdef __APPLE__
-
-#include <OpenCL/cl.h>
-#include <OpenCL/cl_gl.h>
-#include <OpenCL/cl_gl_ext.h>
-#include <OpenCL/cl_ext.h>
-
-#else
-
#include <CL/cl.h>
#include <CL/cl_gl.h>
#include <CL/cl_gl_ext.h>
#include <CL/cl_ext.h>
-#endif
-
#ifdef __cplusplus
}
#endif
#endif /* __OPENCL_H */
-
diff --git a/include/meson.build b/include/meson.build
index a9fadc09721..dbae29fe60a 100644
--- a/include/meson.build
+++ b/include/meson.build
@@ -93,14 +93,19 @@ if with_gallium_opencl and not with_opencl_icd
install_headers(
'CL/cl.h',
'CL/cl.hpp',
+ 'CL/cl2.hpp',
'CL/cl_d3d10.h',
'CL/cl_d3d11.h',
'CL/cl_dx9_media_sharing.h',
+ 'CL/cl_dx9_media_sharing_intel.h',
'CL/cl_egl.h',
'CL/cl_ext.h',
+ 'CL/cl_ext_intel.h',
'CL/cl_gl.h',
'CL/cl_gl_ext.h',
'CL/cl_platform.h',
+ 'CL/cl_va_api_media_sharing_intel.h',
+ 'CL/cl_version.h',
'CL/opencl.h',
subdir: 'CL'
)
diff --git a/src/gallium/state_trackers/clover/Makefile.am b/src/gallium/state_trackers/clover/Makefile.am
index a7befb46059..c92b2cdd274 100644
--- a/src/gallium/state_trackers/clover/Makefile.am
+++ b/src/gallium/state_trackers/clover/Makefile.am
@@ -17,14 +17,19 @@ cldir = $(includedir)/CL
cl_HEADERS = \
$(top_srcdir)/include/CL/cl.h \
$(top_srcdir)/include/CL/cl.hpp \
+ $(top_srcdir)/include/CL/cl2.hpp \
$(top_srcdir)/include/CL/cl_d3d10.h \
$(top_srcdir)/include/CL/cl_d3d11.h \
$(top_srcdir)/include/CL/cl_dx9_media_sharing.h \
+ $(top_srcdir)/include/CL/cl_dx9_media_sharing_intel.h \
$(top_srcdir)/include/CL/cl_egl.h \
$(top_srcdir)/include/CL/cl_ext.h \
+ $(top_srcdir)/include/CL/cl_ext_intel.h \
$(top_srcdir)/include/CL/cl_gl.h \
$(top_srcdir)/include/CL/cl_gl_ext.h \
$(top_srcdir)/include/CL/cl_platform.h \
+ $(top_srcdir)/include/CL/cl_va_api_media_sharing_intel.h \
+ $(top_srcdir)/include/CL/cl_version.h \
$(top_srcdir)/include/CL/opencl.h
endif
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-02 08:58:27 +0000