Open Source Computer Vision Library https://opencv.org/
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

259 lines
12 KiB

Merge pull request #12703 from wzw-intel:vkcom * dnn: Add a Vulkan based backend This commit adds a new backend "DNN_BACKEND_VKCOM" and a new target "DNN_TARGET_VULKAN". VKCOM means vulkan based computation library. This backend uses Vulkan API and SPIR-V shaders to do the inference computation for layers. The layer types that implemented in DNN_BACKEND_VKCOM include: Conv, Concat, ReLU, LRN, PriorBox, Softmax, MaxPooling, AvePooling, Permute This is just a beginning work for Vulkan in OpenCV DNN, more layer types will be supported and performance tuning is on the way. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/vulkan: Add FindVulkan.cmake to detect Vulkan SDK In order to build dnn with Vulkan support, need installing Vulkan SDK and setting environment variable "VULKAN_SDK" and add "-DWITH_VULKAN=ON" to cmake command. You can download Vulkan SDK from: https://vulkan.lunarg.com/sdk/home#linux For how to install, see https://vulkan.lunarg.com/doc/sdk/latest/linux/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/windows/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/mac/getting_started.html respectively for linux, windows and mac. To run the vulkan backend, also need installing mesa driver. On Ubuntu, use this command 'sudo apt-get install mesa-vulkan-drivers' To test, use command '$BUILD_DIR/bin/opencv_test_dnn --gtest_filter=*VkCom*' Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: dynamically load Vulkan runtime No compile-time dependency on Vulkan library. If Vulkan runtime is unavailable, fallback to CPU path. Use environment "OPENCL_VULKAN_RUNTIME" to specify path to your own vulkan runtime library. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: Add a python script to compile GLSL shaders to SPIR-V shaders The SPIR-V shaders are in format of text-based 32-bit hexadecimal numbers, and inserted into .cpp files as unsigned int32 array. * dnn/Vulkan: Put Vulkan headers into 3rdparty directory and some other fixes Vulkan header files are copied from https://github.com/KhronosGroup/Vulkan-Docs/tree/master/include/vulkan to 3rdparty/include Fix the Copyright declaration issue. Refine OpenCVDetectVulkan.cmake * dnn/Vulkan: Add vulkan backend tests into existing ones. Also fixed some test failures. - Don't use bool variable as uniform for shader - Fix dispathed group number beyond max issue - Bypass "group > 1" convolution. This should be support in future. * dnn/Vulkan: Fix multiple initialization in one thread.
6 years ago
#ifndef VULKAN_FUCHSIA_H_
#define VULKAN_FUCHSIA_H_ 1
/*
** Copyright 2015-2023 The Khronos Group Inc.
Merge pull request #12703 from wzw-intel:vkcom * dnn: Add a Vulkan based backend This commit adds a new backend "DNN_BACKEND_VKCOM" and a new target "DNN_TARGET_VULKAN". VKCOM means vulkan based computation library. This backend uses Vulkan API and SPIR-V shaders to do the inference computation for layers. The layer types that implemented in DNN_BACKEND_VKCOM include: Conv, Concat, ReLU, LRN, PriorBox, Softmax, MaxPooling, AvePooling, Permute This is just a beginning work for Vulkan in OpenCV DNN, more layer types will be supported and performance tuning is on the way. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/vulkan: Add FindVulkan.cmake to detect Vulkan SDK In order to build dnn with Vulkan support, need installing Vulkan SDK and setting environment variable "VULKAN_SDK" and add "-DWITH_VULKAN=ON" to cmake command. You can download Vulkan SDK from: https://vulkan.lunarg.com/sdk/home#linux For how to install, see https://vulkan.lunarg.com/doc/sdk/latest/linux/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/windows/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/mac/getting_started.html respectively for linux, windows and mac. To run the vulkan backend, also need installing mesa driver. On Ubuntu, use this command 'sudo apt-get install mesa-vulkan-drivers' To test, use command '$BUILD_DIR/bin/opencv_test_dnn --gtest_filter=*VkCom*' Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: dynamically load Vulkan runtime No compile-time dependency on Vulkan library. If Vulkan runtime is unavailable, fallback to CPU path. Use environment "OPENCL_VULKAN_RUNTIME" to specify path to your own vulkan runtime library. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: Add a python script to compile GLSL shaders to SPIR-V shaders The SPIR-V shaders are in format of text-based 32-bit hexadecimal numbers, and inserted into .cpp files as unsigned int32 array. * dnn/Vulkan: Put Vulkan headers into 3rdparty directory and some other fixes Vulkan header files are copied from https://github.com/KhronosGroup/Vulkan-Docs/tree/master/include/vulkan to 3rdparty/include Fix the Copyright declaration issue. Refine OpenCVDetectVulkan.cmake * dnn/Vulkan: Add vulkan backend tests into existing ones. Also fixed some test failures. - Don't use bool variable as uniform for shader - Fix dispathed group number beyond max issue - Bypass "group > 1" convolution. This should be support in future. * dnn/Vulkan: Fix multiple initialization in one thread.
6 years ago
**
** SPDX-License-Identifier: Apache-2.0
Merge pull request #12703 from wzw-intel:vkcom * dnn: Add a Vulkan based backend This commit adds a new backend "DNN_BACKEND_VKCOM" and a new target "DNN_TARGET_VULKAN". VKCOM means vulkan based computation library. This backend uses Vulkan API and SPIR-V shaders to do the inference computation for layers. The layer types that implemented in DNN_BACKEND_VKCOM include: Conv, Concat, ReLU, LRN, PriorBox, Softmax, MaxPooling, AvePooling, Permute This is just a beginning work for Vulkan in OpenCV DNN, more layer types will be supported and performance tuning is on the way. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/vulkan: Add FindVulkan.cmake to detect Vulkan SDK In order to build dnn with Vulkan support, need installing Vulkan SDK and setting environment variable "VULKAN_SDK" and add "-DWITH_VULKAN=ON" to cmake command. You can download Vulkan SDK from: https://vulkan.lunarg.com/sdk/home#linux For how to install, see https://vulkan.lunarg.com/doc/sdk/latest/linux/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/windows/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/mac/getting_started.html respectively for linux, windows and mac. To run the vulkan backend, also need installing mesa driver. On Ubuntu, use this command 'sudo apt-get install mesa-vulkan-drivers' To test, use command '$BUILD_DIR/bin/opencv_test_dnn --gtest_filter=*VkCom*' Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: dynamically load Vulkan runtime No compile-time dependency on Vulkan library. If Vulkan runtime is unavailable, fallback to CPU path. Use environment "OPENCL_VULKAN_RUNTIME" to specify path to your own vulkan runtime library. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: Add a python script to compile GLSL shaders to SPIR-V shaders The SPIR-V shaders are in format of text-based 32-bit hexadecimal numbers, and inserted into .cpp files as unsigned int32 array. * dnn/Vulkan: Put Vulkan headers into 3rdparty directory and some other fixes Vulkan header files are copied from https://github.com/KhronosGroup/Vulkan-Docs/tree/master/include/vulkan to 3rdparty/include Fix the Copyright declaration issue. Refine OpenCVDetectVulkan.cmake * dnn/Vulkan: Add vulkan backend tests into existing ones. Also fixed some test failures. - Don't use bool variable as uniform for shader - Fix dispathed group number beyond max issue - Bypass "group > 1" convolution. This should be support in future. * dnn/Vulkan: Fix multiple initialization in one thread.
6 years ago
*/
/*
** This header is generated from the Khronos Vulkan XML API Registry.
**
*/
#ifdef __cplusplus
extern "C" {
#endif
Merge pull request #12703 from wzw-intel:vkcom * dnn: Add a Vulkan based backend This commit adds a new backend "DNN_BACKEND_VKCOM" and a new target "DNN_TARGET_VULKAN". VKCOM means vulkan based computation library. This backend uses Vulkan API and SPIR-V shaders to do the inference computation for layers. The layer types that implemented in DNN_BACKEND_VKCOM include: Conv, Concat, ReLU, LRN, PriorBox, Softmax, MaxPooling, AvePooling, Permute This is just a beginning work for Vulkan in OpenCV DNN, more layer types will be supported and performance tuning is on the way. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/vulkan: Add FindVulkan.cmake to detect Vulkan SDK In order to build dnn with Vulkan support, need installing Vulkan SDK and setting environment variable "VULKAN_SDK" and add "-DWITH_VULKAN=ON" to cmake command. You can download Vulkan SDK from: https://vulkan.lunarg.com/sdk/home#linux For how to install, see https://vulkan.lunarg.com/doc/sdk/latest/linux/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/windows/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/mac/getting_started.html respectively for linux, windows and mac. To run the vulkan backend, also need installing mesa driver. On Ubuntu, use this command 'sudo apt-get install mesa-vulkan-drivers' To test, use command '$BUILD_DIR/bin/opencv_test_dnn --gtest_filter=*VkCom*' Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: dynamically load Vulkan runtime No compile-time dependency on Vulkan library. If Vulkan runtime is unavailable, fallback to CPU path. Use environment "OPENCL_VULKAN_RUNTIME" to specify path to your own vulkan runtime library. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: Add a python script to compile GLSL shaders to SPIR-V shaders The SPIR-V shaders are in format of text-based 32-bit hexadecimal numbers, and inserted into .cpp files as unsigned int32 array. * dnn/Vulkan: Put Vulkan headers into 3rdparty directory and some other fixes Vulkan header files are copied from https://github.com/KhronosGroup/Vulkan-Docs/tree/master/include/vulkan to 3rdparty/include Fix the Copyright declaration issue. Refine OpenCVDetectVulkan.cmake * dnn/Vulkan: Add vulkan backend tests into existing ones. Also fixed some test failures. - Don't use bool variable as uniform for shader - Fix dispathed group number beyond max issue - Bypass "group > 1" convolution. This should be support in future. * dnn/Vulkan: Fix multiple initialization in one thread.
6 years ago
#define VK_FUCHSIA_imagepipe_surface 1
#define VK_FUCHSIA_IMAGEPIPE_SURFACE_SPEC_VERSION 1
#define VK_FUCHSIA_IMAGEPIPE_SURFACE_EXTENSION_NAME "VK_FUCHSIA_imagepipe_surface"
typedef VkFlags VkImagePipeSurfaceCreateFlagsFUCHSIA;
typedef struct VkImagePipeSurfaceCreateInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkImagePipeSurfaceCreateFlagsFUCHSIA flags;
zx_handle_t imagePipeHandle;
} VkImagePipeSurfaceCreateInfoFUCHSIA;
typedef VkResult (VKAPI_PTR *PFN_vkCreateImagePipeSurfaceFUCHSIA)(VkInstance instance, const VkImagePipeSurfaceCreateInfoFUCHSIA* pCreateInfo, const VkAllocationCallbacks* pAllocator, VkSurfaceKHR* pSurface);
#ifndef VK_NO_PROTOTYPES
VKAPI_ATTR VkResult VKAPI_CALL vkCreateImagePipeSurfaceFUCHSIA(
VkInstance instance,
const VkImagePipeSurfaceCreateInfoFUCHSIA* pCreateInfo,
const VkAllocationCallbacks* pAllocator,
VkSurfaceKHR* pSurface);
#endif
#define VK_FUCHSIA_external_memory 1
#define VK_FUCHSIA_EXTERNAL_MEMORY_SPEC_VERSION 1
#define VK_FUCHSIA_EXTERNAL_MEMORY_EXTENSION_NAME "VK_FUCHSIA_external_memory"
typedef struct VkImportMemoryZirconHandleInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkExternalMemoryHandleTypeFlagBits handleType;
zx_handle_t handle;
} VkImportMemoryZirconHandleInfoFUCHSIA;
typedef struct VkMemoryZirconHandlePropertiesFUCHSIA {
VkStructureType sType;
void* pNext;
uint32_t memoryTypeBits;
} VkMemoryZirconHandlePropertiesFUCHSIA;
typedef struct VkMemoryGetZirconHandleInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkDeviceMemory memory;
VkExternalMemoryHandleTypeFlagBits handleType;
} VkMemoryGetZirconHandleInfoFUCHSIA;
typedef VkResult (VKAPI_PTR *PFN_vkGetMemoryZirconHandleFUCHSIA)(VkDevice device, const VkMemoryGetZirconHandleInfoFUCHSIA* pGetZirconHandleInfo, zx_handle_t* pZirconHandle);
typedef VkResult (VKAPI_PTR *PFN_vkGetMemoryZirconHandlePropertiesFUCHSIA)(VkDevice device, VkExternalMemoryHandleTypeFlagBits handleType, zx_handle_t zirconHandle, VkMemoryZirconHandlePropertiesFUCHSIA* pMemoryZirconHandleProperties);
#ifndef VK_NO_PROTOTYPES
VKAPI_ATTR VkResult VKAPI_CALL vkGetMemoryZirconHandleFUCHSIA(
VkDevice device,
const VkMemoryGetZirconHandleInfoFUCHSIA* pGetZirconHandleInfo,
zx_handle_t* pZirconHandle);
VKAPI_ATTR VkResult VKAPI_CALL vkGetMemoryZirconHandlePropertiesFUCHSIA(
VkDevice device,
VkExternalMemoryHandleTypeFlagBits handleType,
zx_handle_t zirconHandle,
VkMemoryZirconHandlePropertiesFUCHSIA* pMemoryZirconHandleProperties);
#endif
#define VK_FUCHSIA_external_semaphore 1
#define VK_FUCHSIA_EXTERNAL_SEMAPHORE_SPEC_VERSION 1
#define VK_FUCHSIA_EXTERNAL_SEMAPHORE_EXTENSION_NAME "VK_FUCHSIA_external_semaphore"
typedef struct VkImportSemaphoreZirconHandleInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkSemaphore semaphore;
VkSemaphoreImportFlags flags;
VkExternalSemaphoreHandleTypeFlagBits handleType;
zx_handle_t zirconHandle;
} VkImportSemaphoreZirconHandleInfoFUCHSIA;
typedef struct VkSemaphoreGetZirconHandleInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkSemaphore semaphore;
VkExternalSemaphoreHandleTypeFlagBits handleType;
} VkSemaphoreGetZirconHandleInfoFUCHSIA;
typedef VkResult (VKAPI_PTR *PFN_vkImportSemaphoreZirconHandleFUCHSIA)(VkDevice device, const VkImportSemaphoreZirconHandleInfoFUCHSIA* pImportSemaphoreZirconHandleInfo);
typedef VkResult (VKAPI_PTR *PFN_vkGetSemaphoreZirconHandleFUCHSIA)(VkDevice device, const VkSemaphoreGetZirconHandleInfoFUCHSIA* pGetZirconHandleInfo, zx_handle_t* pZirconHandle);
#ifndef VK_NO_PROTOTYPES
VKAPI_ATTR VkResult VKAPI_CALL vkImportSemaphoreZirconHandleFUCHSIA(
VkDevice device,
const VkImportSemaphoreZirconHandleInfoFUCHSIA* pImportSemaphoreZirconHandleInfo);
VKAPI_ATTR VkResult VKAPI_CALL vkGetSemaphoreZirconHandleFUCHSIA(
VkDevice device,
const VkSemaphoreGetZirconHandleInfoFUCHSIA* pGetZirconHandleInfo,
zx_handle_t* pZirconHandle);
#endif
#define VK_FUCHSIA_buffer_collection 1
VK_DEFINE_NON_DISPATCHABLE_HANDLE(VkBufferCollectionFUCHSIA)
#define VK_FUCHSIA_BUFFER_COLLECTION_SPEC_VERSION 2
#define VK_FUCHSIA_BUFFER_COLLECTION_EXTENSION_NAME "VK_FUCHSIA_buffer_collection"
typedef VkFlags VkImageFormatConstraintsFlagsFUCHSIA;
typedef enum VkImageConstraintsInfoFlagBitsFUCHSIA {
VK_IMAGE_CONSTRAINTS_INFO_CPU_READ_RARELY_FUCHSIA = 0x00000001,
VK_IMAGE_CONSTRAINTS_INFO_CPU_READ_OFTEN_FUCHSIA = 0x00000002,
VK_IMAGE_CONSTRAINTS_INFO_CPU_WRITE_RARELY_FUCHSIA = 0x00000004,
VK_IMAGE_CONSTRAINTS_INFO_CPU_WRITE_OFTEN_FUCHSIA = 0x00000008,
VK_IMAGE_CONSTRAINTS_INFO_PROTECTED_OPTIONAL_FUCHSIA = 0x00000010,
VK_IMAGE_CONSTRAINTS_INFO_FLAG_BITS_MAX_ENUM_FUCHSIA = 0x7FFFFFFF
} VkImageConstraintsInfoFlagBitsFUCHSIA;
typedef VkFlags VkImageConstraintsInfoFlagsFUCHSIA;
typedef struct VkBufferCollectionCreateInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
zx_handle_t collectionToken;
} VkBufferCollectionCreateInfoFUCHSIA;
typedef struct VkImportMemoryBufferCollectionFUCHSIA {
VkStructureType sType;
const void* pNext;
VkBufferCollectionFUCHSIA collection;
uint32_t index;
} VkImportMemoryBufferCollectionFUCHSIA;
typedef struct VkBufferCollectionImageCreateInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkBufferCollectionFUCHSIA collection;
uint32_t index;
} VkBufferCollectionImageCreateInfoFUCHSIA;
typedef struct VkBufferCollectionConstraintsInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
uint32_t minBufferCount;
uint32_t maxBufferCount;
uint32_t minBufferCountForCamping;
uint32_t minBufferCountForDedicatedSlack;
uint32_t minBufferCountForSharedSlack;
} VkBufferCollectionConstraintsInfoFUCHSIA;
typedef struct VkBufferConstraintsInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkBufferCreateInfo createInfo;
VkFormatFeatureFlags requiredFormatFeatures;
VkBufferCollectionConstraintsInfoFUCHSIA bufferCollectionConstraints;
} VkBufferConstraintsInfoFUCHSIA;
typedef struct VkBufferCollectionBufferCreateInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkBufferCollectionFUCHSIA collection;
uint32_t index;
} VkBufferCollectionBufferCreateInfoFUCHSIA;
typedef struct VkSysmemColorSpaceFUCHSIA {
VkStructureType sType;
const void* pNext;
uint32_t colorSpace;
} VkSysmemColorSpaceFUCHSIA;
typedef struct VkBufferCollectionPropertiesFUCHSIA {
VkStructureType sType;
void* pNext;
uint32_t memoryTypeBits;
uint32_t bufferCount;
uint32_t createInfoIndex;
uint64_t sysmemPixelFormat;
VkFormatFeatureFlags formatFeatures;
VkSysmemColorSpaceFUCHSIA sysmemColorSpaceIndex;
VkComponentMapping samplerYcbcrConversionComponents;
VkSamplerYcbcrModelConversion suggestedYcbcrModel;
VkSamplerYcbcrRange suggestedYcbcrRange;
VkChromaLocation suggestedXChromaOffset;
VkChromaLocation suggestedYChromaOffset;
} VkBufferCollectionPropertiesFUCHSIA;
typedef struct VkImageFormatConstraintsInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
VkImageCreateInfo imageCreateInfo;
VkFormatFeatureFlags requiredFormatFeatures;
VkImageFormatConstraintsFlagsFUCHSIA flags;
uint64_t sysmemPixelFormat;
uint32_t colorSpaceCount;
const VkSysmemColorSpaceFUCHSIA* pColorSpaces;
} VkImageFormatConstraintsInfoFUCHSIA;
typedef struct VkImageConstraintsInfoFUCHSIA {
VkStructureType sType;
const void* pNext;
uint32_t formatConstraintsCount;
const VkImageFormatConstraintsInfoFUCHSIA* pFormatConstraints;
VkBufferCollectionConstraintsInfoFUCHSIA bufferCollectionConstraints;
VkImageConstraintsInfoFlagsFUCHSIA flags;
} VkImageConstraintsInfoFUCHSIA;
typedef VkResult (VKAPI_PTR *PFN_vkCreateBufferCollectionFUCHSIA)(VkDevice device, const VkBufferCollectionCreateInfoFUCHSIA* pCreateInfo, const VkAllocationCallbacks* pAllocator, VkBufferCollectionFUCHSIA* pCollection);
typedef VkResult (VKAPI_PTR *PFN_vkSetBufferCollectionImageConstraintsFUCHSIA)(VkDevice device, VkBufferCollectionFUCHSIA collection, const VkImageConstraintsInfoFUCHSIA* pImageConstraintsInfo);
typedef VkResult (VKAPI_PTR *PFN_vkSetBufferCollectionBufferConstraintsFUCHSIA)(VkDevice device, VkBufferCollectionFUCHSIA collection, const VkBufferConstraintsInfoFUCHSIA* pBufferConstraintsInfo);
typedef void (VKAPI_PTR *PFN_vkDestroyBufferCollectionFUCHSIA)(VkDevice device, VkBufferCollectionFUCHSIA collection, const VkAllocationCallbacks* pAllocator);
typedef VkResult (VKAPI_PTR *PFN_vkGetBufferCollectionPropertiesFUCHSIA)(VkDevice device, VkBufferCollectionFUCHSIA collection, VkBufferCollectionPropertiesFUCHSIA* pProperties);
#ifndef VK_NO_PROTOTYPES
VKAPI_ATTR VkResult VKAPI_CALL vkCreateBufferCollectionFUCHSIA(
VkDevice device,
const VkBufferCollectionCreateInfoFUCHSIA* pCreateInfo,
const VkAllocationCallbacks* pAllocator,
VkBufferCollectionFUCHSIA* pCollection);
VKAPI_ATTR VkResult VKAPI_CALL vkSetBufferCollectionImageConstraintsFUCHSIA(
VkDevice device,
VkBufferCollectionFUCHSIA collection,
const VkImageConstraintsInfoFUCHSIA* pImageConstraintsInfo);
VKAPI_ATTR VkResult VKAPI_CALL vkSetBufferCollectionBufferConstraintsFUCHSIA(
VkDevice device,
VkBufferCollectionFUCHSIA collection,
const VkBufferConstraintsInfoFUCHSIA* pBufferConstraintsInfo);
VKAPI_ATTR void VKAPI_CALL vkDestroyBufferCollectionFUCHSIA(
VkDevice device,
VkBufferCollectionFUCHSIA collection,
const VkAllocationCallbacks* pAllocator);
VKAPI_ATTR VkResult VKAPI_CALL vkGetBufferCollectionPropertiesFUCHSIA(
VkDevice device,
VkBufferCollectionFUCHSIA collection,
VkBufferCollectionPropertiesFUCHSIA* pProperties);
#endif
Merge pull request #12703 from wzw-intel:vkcom * dnn: Add a Vulkan based backend This commit adds a new backend "DNN_BACKEND_VKCOM" and a new target "DNN_TARGET_VULKAN". VKCOM means vulkan based computation library. This backend uses Vulkan API and SPIR-V shaders to do the inference computation for layers. The layer types that implemented in DNN_BACKEND_VKCOM include: Conv, Concat, ReLU, LRN, PriorBox, Softmax, MaxPooling, AvePooling, Permute This is just a beginning work for Vulkan in OpenCV DNN, more layer types will be supported and performance tuning is on the way. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/vulkan: Add FindVulkan.cmake to detect Vulkan SDK In order to build dnn with Vulkan support, need installing Vulkan SDK and setting environment variable "VULKAN_SDK" and add "-DWITH_VULKAN=ON" to cmake command. You can download Vulkan SDK from: https://vulkan.lunarg.com/sdk/home#linux For how to install, see https://vulkan.lunarg.com/doc/sdk/latest/linux/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/windows/getting_started.html https://vulkan.lunarg.com/doc/sdk/latest/mac/getting_started.html respectively for linux, windows and mac. To run the vulkan backend, also need installing mesa driver. On Ubuntu, use this command 'sudo apt-get install mesa-vulkan-drivers' To test, use command '$BUILD_DIR/bin/opencv_test_dnn --gtest_filter=*VkCom*' Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: dynamically load Vulkan runtime No compile-time dependency on Vulkan library. If Vulkan runtime is unavailable, fallback to CPU path. Use environment "OPENCL_VULKAN_RUNTIME" to specify path to your own vulkan runtime library. Signed-off-by: Wu Zhiwen <zhiwen.wu@intel.com> * dnn/Vulkan: Add a python script to compile GLSL shaders to SPIR-V shaders The SPIR-V shaders are in format of text-based 32-bit hexadecimal numbers, and inserted into .cpp files as unsigned int32 array. * dnn/Vulkan: Put Vulkan headers into 3rdparty directory and some other fixes Vulkan header files are copied from https://github.com/KhronosGroup/Vulkan-Docs/tree/master/include/vulkan to 3rdparty/include Fix the Copyright declaration issue. Refine OpenCVDetectVulkan.cmake * dnn/Vulkan: Add vulkan backend tests into existing ones. Also fixed some test failures. - Don't use bool variable as uniform for shader - Fix dispathed group number beyond max issue - Bypass "group > 1" convolution. This should be support in future. * dnn/Vulkan: Fix multiple initialization in one thread.
6 years ago
#ifdef __cplusplus
}
#endif
#endif