diff options
author | Gunes Bayir <gunes.bayir@arm.com> | 2022-11-09 09:26:27 +0000 |
---|---|---|
committer | Gunes Bayir <gunes.bayir@arm.com> | 2022-11-14 15:01:27 +0000 |
commit | ec0113dd7749991959ae351934eea0c0d8077dcb (patch) | |
tree | 792f38b3a05fa85436339d54abddab096e20cc77 /src/gpu/cl/operators/ClTransposedConvolution.cpp | |
parent | 46ede332c7064b85dc538891cad165b3bb427320 (diff) | |
download | ComputeLibrary-ec0113dd7749991959ae351934eea0c0d8077dcb.tar.gz |
Optimize Transposed Convolution for CL backend (FP32/16)
This patch optimizes transposed convolution for CL backend by rewriting it in a single kernel instead of three (flip_kernel + upsample + conv). The new kernel skips the upsampling step which reduces the input space of convolution by stride_x * stride_y, resulting in significant performance improvement. It also skips the kernel flipping by traversing the weights accordingly, thus reduces the memory footprint.
Resolves: COMPMID-5676
Signed-off-by: Gunes Bayir <gunes.bayir@arm.com>
Change-Id: I8a333212dc7c5f7f0597aa58b0d56d44814baa14
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8588
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/gpu/cl/operators/ClTransposedConvolution.cpp')
-rw-r--r-- | src/gpu/cl/operators/ClTransposedConvolution.cpp | 58 |
1 files changed, 58 insertions, 0 deletions
diff --git a/src/gpu/cl/operators/ClTransposedConvolution.cpp b/src/gpu/cl/operators/ClTransposedConvolution.cpp new file mode 100644 index 0000000000..90dbe7f291 --- /dev/null +++ b/src/gpu/cl/operators/ClTransposedConvolution.cpp @@ -0,0 +1,58 @@ +/* + * Copyright (c) 2022 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#include "src/gpu/cl/operators/ClTransposedConvolution.h" + +#include "arm_compute/core/Error.h" +#include "arm_compute/runtime/CL/CLScheduler.h" +#include "src/common/utils/Log.h" +#include "src/gpu/cl/kernels/ClTransposedConvolutionKernel.h" + +namespace arm_compute +{ +namespace opencl +{ +void ClTransposedConvolution::configure(const CLCompileContext &compile_context, const ITensorInfo *input, const ITensorInfo *weights, + const ITensorInfo *biases, ITensorInfo *output, const PadStrideInfo &deconv_info) +{ + ARM_COMPUTE_ERROR_ON_NULLPTR(input); + ARM_COMPUTE_LOG_PARAMS(input, weights, biases, output, deconv_info); + auto kernel_object = std::make_unique<kernels::ClTransposedConvolutionKernel>(); + kernel_object->set_target(CLScheduler::get().target()); + kernel_object->configure(compile_context, input, weights, biases, output, deconv_info); + _transposed_conv_kernel = std::move(kernel_object); +} + +Status ClTransposedConvolution::validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, + const ITensorInfo *output, const PadStrideInfo &deconv_info) +{ + ARM_COMPUTE_RETURN_ON_ERROR(kernels::ClTransposedConvolutionKernel::validate(input, weights, biases, output, deconv_info)); + return Status{}; +} + +void ClTransposedConvolution::run(ITensorPack &tensors) +{ + CLScheduler::get().enqueue_op(*_transposed_conv_kernel.get(), tensors, false); +} +} // namespace opencl +} // namespace arm_compute |