 [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago  [WIP] vulkan compute (#618)
* vulkan infrastructure
* vkallocator and vkmat
* layer interface for vulkan compute
* wip...
* default vulkan device, command wrapper, upload model weight in load_model to simplify layer interface
* simplify command api, vkmat holds staging buffer, relu works
* initialize specialization constant, simplify command dispatch, fix staging buffer copy with different shape, convolution works
* init extension functions
* dynamic local size and group count
* group count=1 is invalid
* regard device max workgroup size limit
* fix relu oooops
* decouple command record and staging allocation
* create result blob
* add pooling shader
* buffer is faster than image :)
* fix pooling shader
* add innerproduct shader
* readonly writeonly decoration
* simplify buffer creation
* decouple command and layer, VK_KHR_descriptor_update_template extension makes descriptor binding update easy :D
* fix vulkan building issues in visual studio (#1)
* fix building issues on visual studio
* ignore benchmark
* cancel changes
* ... ...
* decouple paramdict and vulkandevice
* fix staging buffer destroy in model loading
* remove vkdev member in option
* add padding shader
* simplify vulkan layer creation, simplify convolution and pooling shader for no padding, less debug output
* add convolutiondepthwise and softmax shader
* specialization float type, add leakyrelu
* add dropout shader
* add batchnorm shader
* split vulkan forward
* add scale shader
* push constant type can be int or float
* set_optimal_local_size_xyz
* add eltwise shader
* concat vulkan forward
* fix convolution without bias
* add dummy shader for concat and split, more fix ...
* optional VK_KHR_descriptor_update_template and VK_KHR_push_descriptor
* check VK_KHR_push_descriptor for vkCmdPushDescriptorSetWithTemplateKHR
* binaryop and unaryop shader
* hide raw command buffer
* simple vkbenchncnn benchmark
* create device with transfer queue
* rename command to vkcompute, add vktransfer and layer upload_model interface
* external VkMat, copy and map wrt buffer offset
* command copy respect offset and size
* decouple weight upload and load, simplify upload weight api, use one big staging buffer for uploading weights
* fix build on android
* binding count can not vary :(
* barrier check state, fix sub-op destruction
* declare local_size_xyz constant, fix crash on radv
* fix local_size_xyz, second try
* more barrier and state fix
* fix softmax
* reconstruct buffer memory allocator, reuse blob buffer, less verbose output
* find unified memory type index
* weight staging buffer allocator and weight buffer allocator, respect descriptor buffer offset alignment
* use VK_KHR_descriptor_update_template for faster descriptor update if available, multithread pipeline creation
* find more useful vulkan extensions and enable them
* fix msvc build
* respect VK_KHR_dedicated_allocation for weight buffer allocation
* fix android build
* fix bias name conflicts with metal
* decouple pipeline and layer, building shader sources into shader module, dedicated create_pipeline api, simplify pipeline recording
* drop dummy shader, inplace softmax, multiple shader module works
* fix unique queue family index error
* flatten support vulkan
* mnasnet run
* find shader module by name, each entry point per shader module, fix attribute/id conflict on moltenvk
* some minor changes
* add some high level api
* use dedicated transfer queue to upload weight model
* prefer mappable buffer on unified memory
* global pooling and convolution fc, reuse staging buffer
* implement ring-buffer style blob allocator, add VkBufferMemory capacity
* use blob allocator for workspace blob, it works fine :)
* vulkan option off
* Update layer.cpp
* fix build with vulkan off
* less verbose output, fix crash on vulkan_compute off
* merge benchncnn tool
* allocator clear api, use new weight buffer allocator per net
* add default locked allocator
* mapped mat ptr api, persistent mapped memory works generally :)
* travis ci linux vulkan
* travis ci vulkan wip ...
* more gpu wip ...
* more gpu wip ...
* wip...
* wip...
* wip... ...
* wip... ios vulkan build...
* find glslangValidator on ios build
* use dynamic moltenvk library
* travis ci wip ...
* ios simulator does not support metal at all
* fix cpu only extractor
* optimize workgroup size, first try
* optimize workgroup size, second try
* conv1x1s1d1 vec4
* revert build system
* fix ncnn2mem build
* fix ncnn2mem build
7 years ago |
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429 |
- // Tencent is pleased to support the open source community by making ncnn available.
- //
- // Copyright (C) 2019 THL A29 Limited, a Tencent company. All rights reserved.
- //
- // Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
- // in compliance with the License. You may obtain a copy of the License at
- //
- // https://opensource.org/licenses/BSD-3-Clause
- //
- // Unless required by applicable law or agreed to in writing, software distributed
- // under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
- // CONDITIONS OF ANY KIND, either express or implied. See the License for the
- // specific language governing permissions and limitations under the License.
-
- #include "pipeline.h"
- #include <stdio.h>
- #include <math.h>
- #include <algorithm>
- #include <string>
- #include "mat.h"
- #include "option.h"
-
- namespace ncnn {
-
- #if NCNN_VULKAN
- Pipeline::Pipeline(const VulkanDevice* _vkdev) : vkdev(_vkdev)
- {
- local_shader_module = 0;
-
- descriptorset_layout = 0;
- pipeline_layout = 0;
- pipeline = 0;
- descriptor_update_template = 0;
-
- local_size_x = 1;
- local_size_y = 1;
- local_size_z = 1;
- }
-
- Pipeline::~Pipeline()
- {
- destroy();
- }
-
- int Pipeline::create(const uint32_t* spv_data, size_t spv_data_size, const char* entry_name, const std::vector<vk_specialization_type>& specializations, int binding_count, int push_constant_count)
- {
- local_shader_module = vkdev->compile_shader_module(spv_data, spv_data_size);
-
- // fprintf(stderr, "local_shader_module %p %s created\n", local_shader_module, entry_name);
-
- return create(local_shader_module, entry_name, specializations, binding_count, push_constant_count);
- }
-
- int Pipeline::create(VkShaderModule shader_module, const char* entry_name, const std::vector<vk_specialization_type>& specializations, int binding_count, int push_constant_count)
- {
- create_descriptorset_layout(binding_count);
-
- create_pipeline_layout(push_constant_count);
-
- create_pipeline(shader_module, entry_name, specializations);
-
- if (vkdev->info.support_VK_KHR_descriptor_update_template)
- {
- create_descriptor_update_template(binding_count);
- }
-
- return 0;
- }
-
- int Pipeline::create(const char* _name, const Option& opt, const std::vector<vk_specialization_type>& specializations, int binding_count, int push_constant_count)
- {
- std::string name = _name;
-
- if (opt.use_fp16_arithmetic)
- {
- name += "_fp16a";
- }
- else if (opt.use_fp16_storage)
- {
- name += "_fp16s";
- }
- else if (opt.use_fp16_packed)
- {
- name += "_fp16p";
- }
-
- VkShaderModule shader_module = vkdev->get_shader_module(name.c_str());
-
- return create(shader_module, name.c_str(), specializations, binding_count, push_constant_count);
- }
-
- void Pipeline::destroy()
- {
- if (vkdev->info.support_VK_KHR_descriptor_update_template)
- {
- if (descriptor_update_template)
- {
- vkdev->vkDestroyDescriptorUpdateTemplateKHR(vkdev->vkdevice(), descriptor_update_template, 0);
- descriptor_update_template = 0;
- }
- }
-
- if (pipeline)
- {
- vkDestroyPipeline(vkdev->vkdevice(), pipeline, 0);
- pipeline = 0;
- }
-
- if (pipeline_layout)
- {
- vkDestroyPipelineLayout(vkdev->vkdevice(), pipeline_layout, 0);
- pipeline_layout = 0;
- }
-
- if (descriptorset_layout)
- {
- vkDestroyDescriptorSetLayout(vkdev->vkdevice(), descriptorset_layout, 0);
- descriptorset_layout = 0;
- }
-
- if (local_shader_module)
- {
- vkDestroyShaderModule(vkdev->vkdevice(), local_shader_module, 0);
- local_shader_module = 0;
- }
- }
-
- void Pipeline::set_optimal_local_size_xyz(int w, int h, int c)
- {
- if (c > 0)
- {
- local_size_z = vkdev->info.max_workgroup_size[2];
- while ((uint32_t)c < local_size_z)
- {
- local_size_z /= 2;
- }
- }
- else
- {
- local_size_z = std::min((uint32_t)128, vkdev->info.max_workgroup_size[2]);
- }
-
- uint32_t max_local_size_xy = vkdev->info.max_workgroup_invocations / local_size_z;
-
- if (h == w || (h < 0 && w < 0))
- {
- uint32_t local_size_xy = sqrt(max_local_size_xy);
- uint32_t local_size_xy_prefer = 128;
- while (local_size_xy < local_size_xy_prefer)
- {
- local_size_xy_prefer /= 2;
- }
- local_size_x = local_size_xy_prefer;
- local_size_y = local_size_xy_prefer;
- }
- if (h > 0 && w > 0)
- {
- if (h > w)
- {
- float ps = h / (float)w;
- float local_size_xy = sqrt(max_local_size_xy / ps);
- local_size_y = local_size_xy * ps;
- local_size_x = std::max((uint32_t)local_size_xy, (uint32_t)1);
- }
- else
- {
- float ps = w / (float)h;
- float local_size_xy = sqrt(max_local_size_xy / ps);
- local_size_y = std::max((uint32_t)local_size_xy, (uint32_t)1);
- local_size_x = local_size_xy * ps;
- }
-
- uint32_t local_size_y_prefer = std::min((uint32_t)128, vkdev->info.max_workgroup_size[1]);
- while (local_size_y < local_size_y_prefer)
- {
- local_size_y_prefer /= 2;
- }
-
- uint32_t local_size_x_prefer = std::min((uint32_t)128, vkdev->info.max_workgroup_size[0]);
- while (local_size_x < local_size_x_prefer)
- {
- local_size_x_prefer /= 2;
- }
-
- local_size_y = local_size_y_prefer;
- local_size_x = local_size_x_prefer;
- }
- else if (h > 0)
- {
- local_size_y = std::min(max_local_size_xy, vkdev->info.max_workgroup_size[1]);
- while ((uint32_t)h < local_size_y)
- {
- local_size_y /= 2;
- }
-
- uint32_t max_local_size_x = max_local_size_xy / local_size_y;
- local_size_x = std::min(max_local_size_x, vkdev->info.max_workgroup_size[0]);
- }
- else if (w > 0)
- {
- local_size_x = std::min(max_local_size_xy, vkdev->info.max_workgroup_size[0]);
- while ((uint32_t)w < local_size_x)
- {
- local_size_x /= 2;
- }
-
- uint32_t max_local_size_y = max_local_size_xy / local_size_x;
- local_size_y = std::min(max_local_size_y, vkdev->info.max_workgroup_size[1]);
- }
-
- // fprintf(stderr, "local size = %d %d %d\n", local_size_x, local_size_y, local_size_z);
- }
-
- void Pipeline::set_local_size_xyz(int w, int h, int c)
- {
- local_size_x = w;
- local_size_y = h;
- local_size_z = c;
- }
-
- int Pipeline::create_descriptorset_layout(int binding_count)
- {
- if (binding_count == 0)
- {
- descriptorset_layout = 0;
- return 0;
- }
-
- std::vector<VkDescriptorSetLayoutBinding> descriptorSetLayoutBindings(binding_count);
- for (int i=0; i<binding_count; i++)
- {
- descriptorSetLayoutBindings[i].binding = i;
- descriptorSetLayoutBindings[i].descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER;
- descriptorSetLayoutBindings[i].descriptorCount = 1;
- descriptorSetLayoutBindings[i].stageFlags = VK_SHADER_STAGE_COMPUTE_BIT;
- descriptorSetLayoutBindings[i].pImmutableSamplers = 0;
- }
-
- VkDescriptorSetLayoutCreateInfo descriptorSetLayoutCreateInfo;
- descriptorSetLayoutCreateInfo.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO;
- descriptorSetLayoutCreateInfo.pNext = 0;
- descriptorSetLayoutCreateInfo.flags = 0;
- descriptorSetLayoutCreateInfo.bindingCount = binding_count;
- descriptorSetLayoutCreateInfo.pBindings = descriptorSetLayoutBindings.data();
-
- if (vkdev->info.support_VK_KHR_push_descriptor)
- {
- descriptorSetLayoutCreateInfo.flags |= VK_DESCRIPTOR_SET_LAYOUT_CREATE_PUSH_DESCRIPTOR_BIT_KHR;
- }
-
- VkResult ret = vkCreateDescriptorSetLayout(vkdev->vkdevice(), &descriptorSetLayoutCreateInfo, 0, &descriptorset_layout);
- if (ret != VK_SUCCESS)
- {
- fprintf(stderr, "vkCreateDescriptorSetLayout failed %d\n", ret);
- return -1;
- }
-
- return 0;
- }
-
- int Pipeline::create_pipeline_layout(int push_constant_count)
- {
- VkPushConstantRange pushConstantRange;
- pushConstantRange.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT;
- pushConstantRange.offset = 0;
- pushConstantRange.size = sizeof(int) * push_constant_count;
-
- VkPipelineLayoutCreateInfo pipelineLayoutCreateInfo;
- pipelineLayoutCreateInfo.sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO;
- pipelineLayoutCreateInfo.pNext = 0;
- pipelineLayoutCreateInfo.flags = 0;
-
- if (descriptorset_layout)
- {
- pipelineLayoutCreateInfo.setLayoutCount = 1;
- pipelineLayoutCreateInfo.pSetLayouts = &descriptorset_layout;
- }
- else
- {
- pipelineLayoutCreateInfo.setLayoutCount = 0;
- pipelineLayoutCreateInfo.pSetLayouts = 0;
- }
-
- if (push_constant_count > 0)
- {
- pipelineLayoutCreateInfo.pushConstantRangeCount = 1;
- pipelineLayoutCreateInfo.pPushConstantRanges = &pushConstantRange;
- }
- else
- {
- pipelineLayoutCreateInfo.pushConstantRangeCount = 0;
- pipelineLayoutCreateInfo.pPushConstantRanges = 0;
- }
-
- VkResult ret = vkCreatePipelineLayout(vkdev->vkdevice(), &pipelineLayoutCreateInfo, 0, &pipeline_layout);
- if (ret != VK_SUCCESS)
- {
- fprintf(stderr, "vkCreatePipelineLayout failed %d\n", ret);
- return -1;
- }
-
- return 0;
- }
-
- int Pipeline::create_pipeline(VkShaderModule shader_module, const char* entry_name, const std::vector<vk_specialization_type>& specializations)
- {
- const int specialization_count = specializations.size();
-
- // +3 for local_size_xyz
- std::vector<VkSpecializationMapEntry> specializationMapEntries;
- specializationMapEntries.resize(specialization_count + 3);
-
- for (int i=0; i<specialization_count; i++)
- {
- specializationMapEntries[i].constantID = i;
- specializationMapEntries[i].offset = i * sizeof(vk_specialization_type);
- specializationMapEntries[i].size = sizeof(vk_specialization_type);
- }
-
- std::vector<vk_specialization_type> specialization_data = specializations;
-
- // append local_size_xyz specialization
- {
- VkSpecializationMapEntry* local_size_xyz_entries = specializationMapEntries.data() + specialization_count;
-
- local_size_xyz_entries[0].constantID = 233;
- local_size_xyz_entries[0].offset = (specialization_count+0) * sizeof(vk_specialization_type);
- local_size_xyz_entries[0].size = sizeof(vk_specialization_type);
-
- local_size_xyz_entries[1].constantID = 234;
- local_size_xyz_entries[1].offset = (specialization_count+1) * sizeof(vk_specialization_type);
- local_size_xyz_entries[1].size = sizeof(vk_specialization_type);
-
- local_size_xyz_entries[2].constantID = 235;
- local_size_xyz_entries[2].offset = (specialization_count+2) * sizeof(vk_specialization_type);
- local_size_xyz_entries[2].size = sizeof(vk_specialization_type);
-
- specialization_data.resize(specialization_count + 3);
- specialization_data[ specialization_count+0 ].u32 = local_size_x;
- specialization_data[ specialization_count+1 ].u32 = local_size_y;
- specialization_data[ specialization_count+2 ].u32 = local_size_z;
- }
-
- VkSpecializationInfo specializationInfo;
- specializationInfo.mapEntryCount = specializationMapEntries.size();
- specializationInfo.pMapEntries = specializationMapEntries.data();
- specializationInfo.dataSize = specialization_data.size() * sizeof(vk_specialization_type);
- specializationInfo.pData = specialization_data.data();
-
- VkPipelineShaderStageCreateInfo pipelineShaderStageCreateInfo;
- pipelineShaderStageCreateInfo.sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO;
- pipelineShaderStageCreateInfo.pNext = 0;
- pipelineShaderStageCreateInfo.flags = 0;
- pipelineShaderStageCreateInfo.stage = VK_SHADER_STAGE_COMPUTE_BIT;
- pipelineShaderStageCreateInfo.module = shader_module;
- pipelineShaderStageCreateInfo.pName = entry_name;
- pipelineShaderStageCreateInfo.pSpecializationInfo = &specializationInfo;
-
- VkComputePipelineCreateInfo computePipelineCreateInfo;
- computePipelineCreateInfo.sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO;
- computePipelineCreateInfo.pNext = 0;
- computePipelineCreateInfo.flags = 0;
- computePipelineCreateInfo.stage = pipelineShaderStageCreateInfo;
- computePipelineCreateInfo.layout = pipeline_layout;
- computePipelineCreateInfo.basePipelineHandle = 0;
- computePipelineCreateInfo.basePipelineIndex = 0;
-
- VkResult ret = vkCreateComputePipelines(vkdev->vkdevice(), 0, 1, &computePipelineCreateInfo, 0, &pipeline);
- if (ret != VK_SUCCESS)
- {
- fprintf(stderr, "vkCreateComputePipelines failed %d\n", ret);
- return -1;
- }
-
- return 0;
- }
-
- int Pipeline::create_descriptor_update_template(int binding_count)
- {
- if (binding_count == 0)
- {
- descriptor_update_template = 0;
- return 0;
- }
-
- std::vector<VkDescriptorUpdateTemplateEntryKHR> descriptorUpdateTemplateEntries(binding_count);
- for (int i=0; i<binding_count; i++)// TODO do not update weights
- {
- descriptorUpdateTemplateEntries[i].dstBinding = i;
- descriptorUpdateTemplateEntries[i].dstArrayElement = 0;
- descriptorUpdateTemplateEntries[i].descriptorCount = 1;
- descriptorUpdateTemplateEntries[i].descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER;
- descriptorUpdateTemplateEntries[i].offset = i * sizeof(VkDescriptorBufferInfo);
- descriptorUpdateTemplateEntries[i].stride = sizeof(VkDescriptorBufferInfo);
- }
-
- VkDescriptorUpdateTemplateCreateInfoKHR descriptorUpdateTemplateCreateInfo;
- descriptorUpdateTemplateCreateInfo.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_UPDATE_TEMPLATE_CREATE_INFO_KHR;
- descriptorUpdateTemplateCreateInfo.pNext = 0;
- descriptorUpdateTemplateCreateInfo.flags = 0;
- descriptorUpdateTemplateCreateInfo.descriptorUpdateEntryCount = binding_count;// TODO do not update weights
- descriptorUpdateTemplateCreateInfo.pDescriptorUpdateEntries = descriptorUpdateTemplateEntries.data();
- if (vkdev->info.support_VK_KHR_push_descriptor)
- {
- descriptorUpdateTemplateCreateInfo.templateType = VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_PUSH_DESCRIPTORS_KHR;
- }
- else
- {
- descriptorUpdateTemplateCreateInfo.templateType = VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_DESCRIPTOR_SET_KHR;
- }
- // descriptorSetLayout should be ignored if VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_PUSH_DESCRIPTORS_KHR
- // FIXME HACK WARNING TODO NOTE but crash on radv if set NULL :(
- descriptorUpdateTemplateCreateInfo.descriptorSetLayout = descriptorset_layout;
- descriptorUpdateTemplateCreateInfo.pipelineBindPoint = VK_PIPELINE_BIND_POINT_COMPUTE;
- descriptorUpdateTemplateCreateInfo.pipelineLayout = pipeline_layout;
- descriptorUpdateTemplateCreateInfo.set = 0;
-
- VkResult ret = vkdev->vkCreateDescriptorUpdateTemplateKHR(vkdev->vkdevice(), &descriptorUpdateTemplateCreateInfo, 0, &descriptor_update_template);
- if (ret != VK_SUCCESS)
- {
- fprintf(stderr, "vkCreateDescriptorUpdateTemplateKHR failed %d\n", ret);
- return -1;
- }
-
- return 0;
- }
- #endif // NCNN_VULKAN
-
- } // namespace ncnn
|