diff options
| author | Michele Di Giorgio <michele.digiorgio@arm.com> | 2018-11-20 16:03:01 +0000 |
|---|---|---|
| committer | Michalis Spyrou <michalis.spyrou@arm.com> | 2018-11-21 14:08:19 +0000 |
| commit | 1c948d47f55ff8a39aa527f63ea7df93a13dd38e (patch) | |
| tree | 4579798cb95ecaf62b5d7fe61b5d3753301e2804 /tests/validation/NEON/YOLOLayer.cpp | |
| parent | 8b2814ab7b9dc00278132d74d2f738b843b6c0c7 (diff) | |
| download | ComputeLibrary-1c948d47f55ff8a39aa527f63ea7df93a13dd38e.tar.gz | |
COMPMID-1800: (Nightly) Mismatches in SC9863 board for NEON FP16
Fixes for:
- ReduceMean, reduction on the X axis for FP16 with 8 elements was
performed only up to a certain point. The fix now takes into account the
number of elements of the vector and does as many reductions as
necessary.
- YOLOLayer, activation for FP16 has to be performed on 32 bits until
the FP16 approximations is fixed.
Change-Id: I75373f4edd37de476e6fe1a56de3ef386b65c619
Diffstat (limited to 'tests/validation/NEON/YOLOLayer.cpp')
| -rw-r--r-- | tests/validation/NEON/YOLOLayer.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/tests/validation/NEON/YOLOLayer.cpp b/tests/validation/NEON/YOLOLayer.cpp index 926a2dad86..6225dc170e 100644 --- a/tests/validation/NEON/YOLOLayer.cpp +++ b/tests/validation/NEON/YOLOLayer.cpp @@ -46,7 +46,7 @@ namespace /** Tolerance */ constexpr AbsoluteTolerance<float> tolerance_f32(1e-6f); #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC -constexpr RelativeTolerance<float> tolerance_f16(0.001f); +constexpr RelativeTolerance<float> tolerance_f16(0.01f); #endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */ /** Floating point data sets. */ |