diff options
| author | Manuel Bottini <manuel.bottini@arm.com> | 2019-06-10 17:06:39 +0100 |
|---|---|---|
| committer | Manuel Bottini <manuel.bottini@arm.com> | 2019-06-11 14:37:57 +0000 |
| commit | 62404a58d4b6f5ddbf0e551764b314d1b2da2478 (patch) | |
| tree | 84c0ba9c02a90049b24d503a0f03b4c5193b72c6 /src/core/CL/kernels/CLGEMMMatrixMultiplyReshapedOnlyRHSKernel.cpp | |
| parent | 5b48ad7d43c3d1c2fdbae64beac3f37bc6697338 (diff) | |
| download | ComputeLibrary-62404a58d4b6f5ddbf0e551764b314d1b2da2478.tar.gz | |
COMPMID-2376: (Nightly) NEON ArgMinMax QASYMM8
The maximum index value returned from the reference is computed as a uint8_t
Therefore the maximum value it returns is 255. In order to fix this bug for
tensor shapes with dimension along the reduced axis above 255, I separated the
computation of the arg min max reference so that it cleanly computes the
result as a uint32_t
Change-Id: I96a710177609d97c53ed12f20651d9737b3eb703
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/1318
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Diffstat (limited to 'src/core/CL/kernels/CLGEMMMatrixMultiplyReshapedOnlyRHSKernel.cpp')
0 files changed, 0 insertions, 0 deletions