diff options
| author | Matthew Bentham <matthew.bentham@arm.com> | 2020-03-05 23:37:48 +0000 |
|---|---|---|
| committer | Michele Di Giorgio <michele.digiorgio@arm.com> | 2020-03-10 18:41:04 +0000 |
| commit | 758b5ba3e6d22509d4deab3d8b0b9c2f03418130 (patch) | |
| tree | c16734e12b105819a919e8015a4d281dffd8817c /src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp | |
| parent | 6b3865ad038d60a126fe1f90df815a480527a29f (diff) | |
| download | ComputeLibrary-758b5ba3e6d22509d4deab3d8b0b9c2f03418130.tar.gz | |
COMPMID-3069: Improve build time by splitting up ToolchainSupport.h
Split out the parts of ToolchainSupport coming from <memory> and
the parts coming from <string> into their own new header files.
This accounts for 99% of uses of ToolchainSupport, which means that
expensive header files such as arm_neon.h don't need to be included
everywhere.
Knocks about 10% of compilation time off kernel files.
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Change-Id: I2ae718fe766b5ff28608812b0f686f30eeac1b21
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2852
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Diffstat (limited to 'src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp')
| -rw-r--r-- | src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp | 12 |
1 files changed, 6 insertions, 6 deletions
diff --git a/src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp b/src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp index 21971903eb..1890cf7e04 100644 --- a/src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp +++ b/src/core/GLES_COMPUTE/kernels/GCIm2ColKernel.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017-2018 ARM Limited. + * Copyright (c) 2017-2020 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -35,7 +35,7 @@ #include "arm_compute/core/TensorInfo.h" #include "arm_compute/core/Types.h" #include "arm_compute/core/Validate.h" -#include "support/ToolchainSupport.h" +#include "support/StringSupport.h" #include <cmath> #include <tuple> @@ -91,7 +91,7 @@ void GCIm2ColKernel::configure(const IGCTensor *input, IGCTensor *output, const int stride_y = 0; std::tie(stride_x, stride_y) = conv_info.stride(); - _kernel_dims = std::make_pair(kernel_dims.width, kernel_dims.height); + _kernel_dims = std::make_pair(kernel_dims.width, kernel_dims.height); const bool run_img2col_reduced = (output->info()->dimension(0) == (input->info()->dimension(0) * input->info()->dimension(1) * input->info()->dimension(2))) && (TensorShape::num_max_dimensions >= 4) && (std::equal(input->info()->tensor_shape().cbegin() + 3, @@ -109,9 +109,9 @@ void GCIm2ColKernel::configure(const IGCTensor *input, IGCTensor *output, const } build_opts.emplace("#define IM2COL_GENERIC"); - _convolved_dims = scaled_dimensions(input->info()->dimension(0), input->info()->dimension(1), - kernel_dims.width, kernel_dims.height, - conv_info, dilation); + _convolved_dims = scaled_dimensions(input->info()->dimension(0), input->info()->dimension(1), + kernel_dims.width, kernel_dims.height, + conv_info, dilation); _num_elems_processed_per_iteration = (input->info()->data_type() == DataType::F32) ? 1 : 2; build_opts.emplace("#define KERNEL_WIDTH " + support::cpp11::to_string(kernel_dims.width)); |