diff options
| author | Manuel Bottini <manuel.bottini@arm.com> | 2020-07-02 11:27:27 +0100 |
|---|---|---|
| committer | Manuel Bottini <manuel.bottini@arm.com> | 2020-07-09 15:03:17 +0000 |
| commit | c8e6e2c48e558da0c0698428fe496491e18c022a (patch) | |
| tree | 3b55d0c9a24d587d9a2970e720170ec2959fb404 /src | |
| parent | d9eaf614975afad18fd13b9e8c7a7dd21ff6a1dd (diff) | |
| download | ComputeLibrary-c8e6e2c48e558da0c0698428fe496491e18c022a.tar.gz | |
COMPMID-3325: Add support in gemm_tuner for cl_image
Change-Id: I78f815005516ca0e83366bab017884530c1d2e86
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3518
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src')
| -rw-r--r-- | src/core/CL/gemm/CLGEMMHelpers.cpp | 17 | ||||
| -rw-r--r-- | src/core/CL/kernels/CLGEMMReshapeRHSMatrixKernel.cpp | 11 |
2 files changed, 18 insertions, 10 deletions
diff --git a/src/core/CL/gemm/CLGEMMHelpers.cpp b/src/core/CL/gemm/CLGEMMHelpers.cpp index 8f54d3f670..1165e70273 100644 --- a/src/core/CL/gemm/CLGEMMHelpers.cpp +++ b/src/core/CL/gemm/CLGEMMHelpers.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2019 Arm Limited. + * Copyright (c) 2019-2020 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -23,6 +23,8 @@ */ #include "arm_compute/core/CL/gemm/CLGEMMHelpers.h" +#include "arm_compute/core/CL/CLKernelLibrary.h" + #include <utility> namespace arm_compute @@ -51,5 +53,18 @@ std::pair<GEMMLHSMatrixInfo, GEMMRHSMatrixInfo> configure_lhs_rhs_info(unsigned return std::make_pair(lhs_info, rhs_info); } + +void update_padding_for_cl_image(ITensorInfo *tensor) +{ + constexpr unsigned int num_floats_per_pixel = 4; + + const unsigned int stride_y_in_elements = tensor->strides_in_bytes()[1] / tensor->element_size(); + const unsigned int pixel_aligment = get_cl_image_pitch_alignment(CLKernelLibrary::get().get_device()); + const unsigned int row_pitch_alignment = pixel_aligment * num_floats_per_pixel; + const unsigned int round_up_width = ((stride_y_in_elements + row_pitch_alignment - 1) / row_pitch_alignment) * row_pitch_alignment; + const unsigned int padding = round_up_width - stride_y_in_elements; + + tensor->extend_padding(PaddingSize(0, padding, 0, 0)); +} } // namespace cl_gemm } // namespace arm_compute
\ No newline at end of file diff --git a/src/core/CL/kernels/CLGEMMReshapeRHSMatrixKernel.cpp b/src/core/CL/kernels/CLGEMMReshapeRHSMatrixKernel.cpp index 5e09899ec8..c57066ae03 100644 --- a/src/core/CL/kernels/CLGEMMReshapeRHSMatrixKernel.cpp +++ b/src/core/CL/kernels/CLGEMMReshapeRHSMatrixKernel.cpp @@ -29,6 +29,7 @@ #include "arm_compute/core/CL/CLValidate.h" #include "arm_compute/core/CL/ICLTensor.h" #include "arm_compute/core/CL/OpenCL.h" +#include "arm_compute/core/CL/gemm/CLGEMMHelpers.h" #include "arm_compute/core/Error.h" #include "arm_compute/core/Helpers.h" #include "arm_compute/core/TensorInfo.h" @@ -107,15 +108,7 @@ std::pair<Status, Window> validate_and_configure_window(ITensorInfo *input, ITen if(rhs_info.export_to_cl_image) { - constexpr unsigned int num_floats_per_pixel = 4; - - const unsigned int stride_y_in_elements = output->strides_in_bytes()[1] / output->element_size(); - const unsigned int pixel_aligment = get_cl_image_pitch_alignment(CLKernelLibrary::get().get_device()); - const unsigned int row_pitch_alignment = pixel_aligment * num_floats_per_pixel; - const unsigned int round_up_width = ((stride_y_in_elements + row_pitch_alignment - 1) / row_pitch_alignment) * row_pitch_alignment; - const unsigned int padding = round_up_width - stride_y_in_elements; - - output->extend_padding(PaddingSize(0, padding, 0, 0)); + arm_compute::cl_gemm::update_padding_for_cl_image(output); } // Collapse along the Z direction |