diff options
| author | Michalis Spyrou <michalis.spyrou@arm.com> | 2020-01-22 10:27:06 +0000 |
|---|---|---|
| committer | Michele Di Giorgio <michele.digiorgio@arm.com> | 2020-01-23 09:27:17 +0000 |
| commit | 19bd412fd044197726dbd8c756dbd74a9e33fd2b (patch) | |
| tree | 611072c0419d394ef191705ac82521c8a28a1873 /src/runtime/NEON/functions/NEReductionOperation.cpp | |
| parent | b80c435d65eea7c7befdafaef883da9f66574036 (diff) | |
| download | ComputeLibrary-19bd412fd044197726dbd8c756dbd74a9e33fd2b.tar.gz | |
COMPMID-3022 Fix NEON QASYMM8 Reduction Operation
Use proper offsets in our tests and subtract the
accumulated offsets when we do a SUM operation.
Also change the reference MEAN_SUM impementation to
use floats. As a result the tolerance is increased
to 2.
Change-Id: Icac8b84680bd880fd30dc1dd82d19cce43921eb5
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/2616
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Diffstat (limited to 'src/runtime/NEON/functions/NEReductionOperation.cpp')
| -rw-r--r-- | src/runtime/NEON/functions/NEReductionOperation.cpp | 8 |
1 files changed, 6 insertions, 2 deletions
diff --git a/src/runtime/NEON/functions/NEReductionOperation.cpp b/src/runtime/NEON/functions/NEReductionOperation.cpp index 4e37659d30..43205ef762 100644 --- a/src/runtime/NEON/functions/NEReductionOperation.cpp +++ b/src/runtime/NEON/functions/NEReductionOperation.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017-2019 ARM Limited. + * Copyright (c) 2017-2020 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -194,11 +194,15 @@ void NEReductionOperation::configure(ITensor *input, ITensor *output, unsigned i } case ReductionOperation::ARG_IDX_MAX: case ReductionOperation::ARG_IDX_MIN: + { + pixelValue = PixelValue(0, input->info()->data_type(), input->info()->quantization_info()); + break; + } case ReductionOperation::MEAN_SUM: case ReductionOperation::SUM_SQUARE: case ReductionOperation::SUM: { - pixelValue = PixelValue(0, input->info()->data_type(), input->info()->quantization_info()); + pixelValue = PixelValue(static_cast<uint32_t>(0)); break; } default: |