diff options
author | Michalis Spyrou <michalis.spyrou@arm.com> | 2020-05-26 09:59:18 +0100 |
---|---|---|
committer | Michalis Spyrou <michalis.spyrou@arm.com> | 2020-05-26 14:44:56 +0000 |
commit | c8bdeaacc2f02625b900be7b6ccb26a4b663eead (patch) | |
tree | ee2be4043d49618c5ecfde615f999c5245d90626 /src | |
parent | 075253afb56a8a29ec54131e8a867d1ee0a0935c (diff) | |
download | ComputeLibrary-c8bdeaacc2f02625b900be7b6ccb26a4b663eead.tar.gz |
COMPMID-3069: Fix reduction in GEMMLowp
We shouldn't perform reduction on matrix b when we
have a configured fused assembly kernel
Change-Id: I1f26c2afb387ee6ebbd54263e7255dab276ea08f
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3261
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src')
-rw-r--r-- | src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp b/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp index f0fa61657b..993907c29a 100644 --- a/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp +++ b/src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp @@ -574,7 +574,7 @@ void NEGEMMLowpMatrixMultiplyCore::prepare() } // Run matrix B reduction kernel only if _a_offset is not equal to 0 - if(_a_offset != 0 && _reshape_b_only_on_first_run) + if(!_fused_assembly_path && _a_offset != 0 && _reshape_b_only_on_first_run) { _vector_sum_col.allocator()->allocate(); NEScheduler::get().schedule(&_mtx_b_reduction_kernel, Window::DimX); |