diff options
| author | George Wort <george.wort@arm.com> | 2019-02-22 16:37:41 +0000 |
|---|---|---|
| committer | Giuseppe Rossini <giuseppe.rossini@arm.com> | 2019-03-15 13:34:00 +0000 |
| commit | 2d7e683e79c8ad328d4930c1f82a46827313faf4 (patch) | |
| tree | eb81f928ecd2543ef80af87f65d1bdef5a78ea2a /src/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.cpp | |
| parent | 3814b30623d6a9e570d850fe5ae275fe2117f3f5 (diff) | |
| download | ComputeLibrary-2d7e683e79c8ad328d4930c1f82a46827313faf4.tar.gz | |
COMPMID-1694: Fuse offset contribution with the output stage when we use NEGEMMLowpMatrixMultiplyCore
Change-Id: Ic1a681e4cc03e1eba3bf8485d9cdb17b3e926047
Signed-off-by: giuros01 <giuseppe.rossini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/561
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.cpp')
| -rw-r--r-- | src/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.cpp | 11 |
1 files changed, 4 insertions, 7 deletions
diff --git a/src/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.cpp b/src/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.cpp index 33a5b4ace3..22939266e5 100644 --- a/src/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.cpp +++ b/src/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017-2018 ARM Limited. + * Copyright (c) 2017-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -106,20 +106,17 @@ std::pair<Status, Window> validate_and_configure_window(ITensorInfo *mm_result, Window win = calculate_max_window(*mm_result, Steps(num_elems_processed_per_iteration)); AccessWindowHorizontal mm_result_access(mm_result, 0, num_elems_processed_per_iteration); - window_changed = window_changed || update_window_and_padding(win, - mm_result_access); + window_changed = window_changed || update_window_and_padding(win, mm_result_access); if(a_offset != 0) { AccessWindowHorizontal vector_sum_col_access(vector_sum_col, 0, num_elems_processed_per_iteration); - window_changed = window_changed || update_window_and_padding(win, - vector_sum_col_access); + window_changed = window_changed || update_window_and_padding(win, vector_sum_col_access); } if(b_offset != 0) { AccessWindowStatic vector_sum_row_access(vector_sum_row, 0, 0, vector_sum_row->dimension(0), 0); // NOLINT - window_changed = window_changed || update_window_and_padding(win, - vector_sum_row_access); + window_changed = window_changed || update_window_and_padding(win, vector_sum_row_access); } Status err = (window_changed) ? ARM_COMPUTE_CREATE_ERROR(ErrorCode::RUNTIME_ERROR, "Insufficient Padding!") : Status{}; |