diff options
| author | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-06-21 18:43:12 +0100 |
|---|---|---|
| committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-07-05 15:30:24 +0000 |
| commit | 37d080f2f11cfd734104b76512e1fb191486216e (patch) | |
| tree | d5df067c826aacc0676e7e9557a54b61a9a3b7eb /src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp | |
| parent | 11de30da8a9f79943255ddba7bb70a66b076673b (diff) | |
| download | ComputeLibrary-37d080f2f11cfd734104b76512e1fb191486216e.tar.gz | |
COMPMID-2378: Sanitize GEMM configuration for NEON
Change-Id: I7859b82b2059e14685f8792424648ac5eacd67f1
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/1418
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp')
| -rw-r--r-- | src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp | 9 |
1 files changed, 4 insertions, 5 deletions
diff --git a/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp b/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp index d8773e37ab..f10f114287 100644 --- a/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp +++ b/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp @@ -87,7 +87,7 @@ void NEGEMMLowpMatrixMultiplyCore::configure(const ITensor *a, const ITensor *b, case DataType::U8: case DataType::S8: { - _asm_glue.configure(a, b, _fuse_output_stage ? &_mm_result_s32 : output, 1.f, 0.f, _reshape_b_only_on_first_run); + _asm_glue.configure(a, b, _fuse_output_stage ? &_mm_result_s32 : output, 1.f, 0.f, gemm_info); _dot_product_path = _asm_glue.is_configured(); break; } @@ -224,9 +224,8 @@ Status NEGEMMLowpMatrixMultiplyCore::validate(const ITensorInfo *a, const ITenso TensorInfo tmp_b_info{}; TensorInfo mm_result_s32_info{}; - int32_t a_offset = a->quantization_info().uniform().offset; - int32_t b_offset = b->quantization_info().uniform().offset; - const bool reshape_b_only_on_first_run = gemm_info.reshape_b_only_on_first_run(); + int32_t a_offset = a->quantization_info().uniform().offset; + int32_t b_offset = b->quantization_info().uniform().offset; bool fuse_output_stage = gemm_info.gemmlowp_output_stage().type != GEMMLowpOutputStageType::NONE; if(fuse_output_stage) @@ -235,7 +234,7 @@ Status NEGEMMLowpMatrixMultiplyCore::validate(const ITensorInfo *a, const ITenso } // Check if we need to run the optimized assembly kernel - const bool run_optimised = bool(NEGEMMAssemblyDispatch::validate(a, b, fuse_output_stage ? &mm_result_s32_info : output, 1.f, 0.f, reshape_b_only_on_first_run)); + const bool run_optimised = bool(NEGEMMAssemblyDispatch::validate(a, b, fuse_output_stage ? &mm_result_s32_info : output, 1.f, 0.f, gemm_info)); if(run_optimised) { |