diff options
| author | Sang-Hoon Park <sang-hoon.park@arm.com> | 2020-08-20 13:57:49 +0100 |
|---|---|---|
| committer | Sang-Hoon Park <sang-hoon.park@arm.com> | 2020-08-25 08:30:05 +0000 |
| commit | 90251bc1e162925c4b85e8a2923af153af23da93 (patch) | |
| tree | bd46caaa2dee8e1c9a3cb50cfa934781fa1dfdee /tests | |
| parent | 747a2c6204f369bfad90b4373da5ebe7e58bdd97 (diff) | |
| download | ComputeLibrary-90251bc1e162925c4b85e8a2923af153af23da93.tar.gz | |
COMPMID-3749: Adjust FP32 tolerance for NEScale validation
To prevent unexpected failures caused by some cases,
the bigger tolerance value is used, which is matched
to CL's relative tolerance value.
Change-Id: If6e3bc2f30651c54769dcd8dd647a3233a88c488
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3826
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'tests')
| -rw-r--r-- | tests/validation/NEON/Scale.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/tests/validation/NEON/Scale.cpp b/tests/validation/NEON/Scale.cpp index 9d9a2821dc..1b59faaf07 100644 --- a/tests/validation/NEON/Scale.cpp +++ b/tests/validation/NEON/Scale.cpp @@ -79,7 +79,7 @@ const auto QuantizationInfoSet = framework::dataset::make("QuantizationInfo", /** Tolerance */ constexpr AbsoluteTolerance<uint8_t> tolerance_u8(1); constexpr AbsoluteTolerance<int16_t> tolerance_s16(1); -RelativeTolerance<float> tolerance_f32(0.01); +RelativeTolerance<float> tolerance_f32(0.05); #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC RelativeTolerance<half> tolerance_f16(half(0.1)); #endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */ |