opencv/modules/dnn/src/cuda/kernel_dispatcher.hpp

95 lines
5.1 KiB
C++
Raw Normal View History

Merge pull request #14827 from YashasSamaga:cuda4dnn-csl-low CUDA backend for the DNN module * stub cuda4dnn design * minor fixes for tests and doxygen * add csl public api directory to module headers * add low-level CSL components * add high-level CSL components * integrate csl::Tensor into backbone code * switch to CPU iff unsupported; otherwise, fail on error * add fully connected layer * add softmax layer * add activation layers * support arbitary rank TensorDescriptor * pass input wrappers to `initCUDA()` * add 1d/2d/3d-convolution * add pooling layer * reorganize and refactor code * fixes for gcc, clang and doxygen; remove cxx14/17 code * add blank_layer * add LRN layer * add rounding modes for pooling layer * split tensor.hpp into tensor.hpp and tensor_ops.hpp * add concat layer * add scale layer * add batch normalization layer * split math.cu into activations.cu and math.hpp * add eltwise layer * add flatten layer * add tensor transform api * add asymmetric padding support for convolution layer * add reshape layer * fix rebase issues * add permute layer * add padding support for concat layer * refactor and reorganize code * add normalize layer * optimize bias addition in scale layer * add prior box layer * fix and optimize normalize layer * add asymmetric padding support for pooling layer * add event API * improve pooling performance for some padding scenarios * avoid over-allocation of compute resources to kernels * improve prior box performance * enable layer fusion * add const layer * add resize layer * add slice layer * add padding layer * add deconvolution layer * fix channelwise ReLU initialization * add vector traits * add vectorized versions of relu, clipped_relu, power * add vectorized concat kernels * improve concat_with_offsets performance * vectorize scale and bias kernels * add support for multi-billion element tensors * vectorize prior box kernels * fix address alignment check * improve bias addition performance of conv/deconv/fc layers * restructure code for supporting multiple targets * add DNN_TARGET_CUDA_FP64 * add DNN_TARGET_FP16 * improve vectorization * add region layer * improve tensor API, add dynamic ranks 1. use ManagedPtr instead of a Tensor in backend wrapper 2. add new methods to tensor classes - size_range: computes the combined size of for a given axis range - tensor span/view can be constructed from a raw pointer and shape 3. the tensor classes can change their rank at runtime (previously rank was fixed at compile-time) 4. remove device code from tensor classes (as they are unused) 5. enforce strict conditions on tensor class APIs to improve debugging ability * fix parametric relu activation * add squeeze/unsqueeze tensor API * add reorg layer * optimize permute and enable 2d permute * enable 1d and 2d slice * add split layer * add shuffle channel layer * allow tensors of different ranks in reshape primitive * patch SliceOp to allow Crop Layer * allow extra shape inputs in reshape layer * use `std::move_backward` instead of `std::move` for insert in resizable_static_array * improve workspace management * add spatial LRN * add nms (cpu) to region layer * add max pooling with argmax ( and a fix to limits.hpp) * add max unpooling layer * rename DNN_TARGET_CUDA_FP32 to DNN_TARGET_CUDA * update supportBackend to be more rigorous * remove stray include from preventing non-cuda build * include op_cuda.hpp outside condition #if * refactoring, fixes and many optimizations * drop DNN_TARGET_CUDA_FP64 * fix gcc errors * increase max. tensor rank limit to six * add Interp layer * drop custom layers; use BackendNode * vectorize activation kernels * fixes for gcc * remove wrong assertion * fix broken assertion in unpooling primitive * fix build errors in non-CUDA build * completely remove workspace from public API * fix permute layer * enable accuracy and perf. tests for DNN_TARGET_CUDA * add asynchronous forward * vectorize eltwise ops * vectorize fill kernel * fixes for gcc * remove CSL headers from public API * remove csl header source group from cmake * update min. cudnn version in cmake * add numerically stable FP32 log1pexp * refactor code * add FP16 specialization to cudnn based tensor addition * vectorize scale1 and bias1 + minor refactoring * fix doxygen build * fix invalid alignment assertion * clear backend wrappers before allocateLayers * ignore memory lock failures * do not allocate internal blobs * integrate NVTX * add numerically stable half precision log1pexp * fix indentation, following coding style, improve docs * remove accidental modification of IE code * Revert "add asynchronous forward" This reverts commit 1154b9da9da07e9b52f8a81bdcea48cf31c56f70. * [cmake] throw error for unsupported CC versions * fix rebase issues * add more docs, refactor code, fix bugs * minor refactoring and fixes * resolve warnings/errors from clang * remove haveCUDA() checks from supportBackend() * remove NVTX integration * changes based on review comments * avoid exception when no CUDA device is present * add color code for CUDA in Net::dump
2019-10-21 19:28:00 +08:00
// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
#ifndef OPENCV_DNN_SRC_CUDA_KERNEL_DISPATCHER_HPP
#define OPENCV_DNN_SRC_CUDA_KERNEL_DISPATCHER_HPP
#include <cstddef>
#include <type_traits>
/* The performance of many kernels are highly dependent on the tensor rank. Instead of having
* one kernel which can work with the maximally ranked tensors, we make one kernel for each supported
* tensor rank. This is to ensure that the requirements of the maximally ranked tensors do not take a
* toll on the performance of the operation for low ranked tensors. Hence, many kernels take the tensor
* rank as a template parameter.
*
* The kernel is a template and we have different instantiations for each rank. This causes the following pattern
* to arise frequently:
*
* if(rank == 3)
* kernel<T, 3>();
* else if(rank == 2)
* kernel<T, 2>();
* else
* kernel<T, 1>();
*
* The rank is a runtime variable. To facilitate creation of such structures, we use GENERATE_KERNEL_DISPATCHER.
* This macro creates a function which selects the correct kernel instantiation at runtime.
*
* Example:
*
* // function which setups the kernel and launches it
* template <class T, std::size_t Rank>
* void launch_some_kernel(...);
*
* // creates the dispatcher named "some_dispatcher" which invokves the correct instantiation of "launch_some_kernel"
* GENERATE_KERNEL_DISPATCHER(some_dispatcher, launch_some_kernel);
*
* // internal API function
* template <class T>
* void some(...) {
* // ...
* auto rank = input.rank();
* some_dispatcher<T, MIN_RANK, MAX_RANK>(rank, ...);
* }
*/
/*
* name name of the dispatcher function that is generated
* func template function that requires runtime selection
*
* T first template parameter to `func`
* start starting rank
* end ending rank (inclusive)
*
* Executes func<T, selector> based on runtime `selector` argument given `selector` lies
* within the range [start, end]. If outside the range, no instantiation of `func` is executed.
*/
#define GENERATE_KERNEL_DISPATCHER(name,func); \
template <class T, std::size_t start, std::size_t end, class... Args> static \
typename std::enable_if<start == end, void> \
::type name(int selector, Args&& ...args) { \
if(selector == start) \
func<T, start>(std::forward<Args>(args)...); \
} \
\
template <class T, std::size_t start, std::size_t end, class... Args> static \
typename std::enable_if<start != end, void> \
::type name(int selector, Args&& ...args) { \
if(selector == start) \
func<T, start>(std::forward<Args>(args)...); \
else \
name<T, start + 1, end, Args...>(selector, std::forward<Args>(args)...); \
}
2021-10-04 15:08:45 +08:00
// Same as GENERATE_KERNEL_DISPATCHER but takes two class template parameters T and TP1 instead of just T
#define GENERATE_KERNEL_DISPATCHER_2TP(name,func); \
template <class TP1, class TP2, std::size_t start, std::size_t end, class... Args> static \
typename std::enable_if<start == end, void> \
::type name(int selector, Args&& ...args) { \
if(selector == start) \
func<TP1, TP2, start>(std::forward<Args>(args)...); \
} \
\
template <class TP1, class TP2, std::size_t start, std::size_t end, class... Args> static \
typename std::enable_if<start != end, void> \
::type name(int selector, Args&& ...args) { \
if(selector == start) \
func<TP1, TP2, start>(std::forward<Args>(args)...); \
else \
name<TP1, TP2, start + 1, end, Args...>(selector, std::forward<Args>(args)...); \
}
Merge pull request #14827 from YashasSamaga:cuda4dnn-csl-low CUDA backend for the DNN module * stub cuda4dnn design * minor fixes for tests and doxygen * add csl public api directory to module headers * add low-level CSL components * add high-level CSL components * integrate csl::Tensor into backbone code * switch to CPU iff unsupported; otherwise, fail on error * add fully connected layer * add softmax layer * add activation layers * support arbitary rank TensorDescriptor * pass input wrappers to `initCUDA()` * add 1d/2d/3d-convolution * add pooling layer * reorganize and refactor code * fixes for gcc, clang and doxygen; remove cxx14/17 code * add blank_layer * add LRN layer * add rounding modes for pooling layer * split tensor.hpp into tensor.hpp and tensor_ops.hpp * add concat layer * add scale layer * add batch normalization layer * split math.cu into activations.cu and math.hpp * add eltwise layer * add flatten layer * add tensor transform api * add asymmetric padding support for convolution layer * add reshape layer * fix rebase issues * add permute layer * add padding support for concat layer * refactor and reorganize code * add normalize layer * optimize bias addition in scale layer * add prior box layer * fix and optimize normalize layer * add asymmetric padding support for pooling layer * add event API * improve pooling performance for some padding scenarios * avoid over-allocation of compute resources to kernels * improve prior box performance * enable layer fusion * add const layer * add resize layer * add slice layer * add padding layer * add deconvolution layer * fix channelwise ReLU initialization * add vector traits * add vectorized versions of relu, clipped_relu, power * add vectorized concat kernels * improve concat_with_offsets performance * vectorize scale and bias kernels * add support for multi-billion element tensors * vectorize prior box kernels * fix address alignment check * improve bias addition performance of conv/deconv/fc layers * restructure code for supporting multiple targets * add DNN_TARGET_CUDA_FP64 * add DNN_TARGET_FP16 * improve vectorization * add region layer * improve tensor API, add dynamic ranks 1. use ManagedPtr instead of a Tensor in backend wrapper 2. add new methods to tensor classes - size_range: computes the combined size of for a given axis range - tensor span/view can be constructed from a raw pointer and shape 3. the tensor classes can change their rank at runtime (previously rank was fixed at compile-time) 4. remove device code from tensor classes (as they are unused) 5. enforce strict conditions on tensor class APIs to improve debugging ability * fix parametric relu activation * add squeeze/unsqueeze tensor API * add reorg layer * optimize permute and enable 2d permute * enable 1d and 2d slice * add split layer * add shuffle channel layer * allow tensors of different ranks in reshape primitive * patch SliceOp to allow Crop Layer * allow extra shape inputs in reshape layer * use `std::move_backward` instead of `std::move` for insert in resizable_static_array * improve workspace management * add spatial LRN * add nms (cpu) to region layer * add max pooling with argmax ( and a fix to limits.hpp) * add max unpooling layer * rename DNN_TARGET_CUDA_FP32 to DNN_TARGET_CUDA * update supportBackend to be more rigorous * remove stray include from preventing non-cuda build * include op_cuda.hpp outside condition #if * refactoring, fixes and many optimizations * drop DNN_TARGET_CUDA_FP64 * fix gcc errors * increase max. tensor rank limit to six * add Interp layer * drop custom layers; use BackendNode * vectorize activation kernels * fixes for gcc * remove wrong assertion * fix broken assertion in unpooling primitive * fix build errors in non-CUDA build * completely remove workspace from public API * fix permute layer * enable accuracy and perf. tests for DNN_TARGET_CUDA * add asynchronous forward * vectorize eltwise ops * vectorize fill kernel * fixes for gcc * remove CSL headers from public API * remove csl header source group from cmake * update min. cudnn version in cmake * add numerically stable FP32 log1pexp * refactor code * add FP16 specialization to cudnn based tensor addition * vectorize scale1 and bias1 + minor refactoring * fix doxygen build * fix invalid alignment assertion * clear backend wrappers before allocateLayers * ignore memory lock failures * do not allocate internal blobs * integrate NVTX * add numerically stable half precision log1pexp * fix indentation, following coding style, improve docs * remove accidental modification of IE code * Revert "add asynchronous forward" This reverts commit 1154b9da9da07e9b52f8a81bdcea48cf31c56f70. * [cmake] throw error for unsupported CC versions * fix rebase issues * add more docs, refactor code, fix bugs * minor refactoring and fixes * resolve warnings/errors from clang * remove haveCUDA() checks from supportBackend() * remove NVTX integration * changes based on review comments * avoid exception when no CUDA device is present * add color code for CUDA in Net::dump
2019-10-21 19:28:00 +08:00
#endif /* OPENCV_DNN_SRC_CUDA_KERNEL_DISPATCHER_HPP */