diff options
| author | Gunes Bayir <gunes.bayir@arm.com> | 2024-04-26 16:51:54 +0100 |
|---|---|---|
| committer | Gunes Bayir <gunes.bayir@arm.com> | 2024-04-29 09:31:44 +0000 |
| commit | e5ef8c159a14872dda5e36e320f07b0963858d8c (patch) | |
| tree | 2d359eeff83b54b3e7e65484abc37e531aedbdf2 /tests/validation/NEON | |
| parent | 499b5bca1a897461d4105ba52e4c766ddb5f564a (diff) | |
| download | ComputeLibrary-e5ef8c159a14872dda5e36e320f07b0963858d8c.tar.gz | |
Disable SME2 Gemmlowp s8f32 kernel selection in case results needs to be accumulated
Similar to https://review.mlplatform.org/c/ml/ComputeLibrary/+/11500, s8f32 kernels do not support accumulate mode. This patch modifies the kernel selection and also adds more tests to stress these test cases better.
Partially Resolves: COMPMID-6995
Change-Id: I40e19446c012eb7334e4511e254cce0d635aa234
Signed-off-by: Gunes Bayir <gunes.bayir@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/11503
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Radu Salavat <radu.salavat@arm.com>
Reviewed-by: Jakub Sujak <jakub.sujak@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'tests/validation/NEON')
| -rw-r--r-- | tests/validation/NEON/GEMMLowp.cpp | 12 |
1 files changed, 10 insertions, 2 deletions
diff --git a/tests/validation/NEON/GEMMLowp.cpp b/tests/validation/NEON/GEMMLowp.cpp index 9b1da61ed7..d25f43a330 100644 --- a/tests/validation/NEON/GEMMLowp.cpp +++ b/tests/validation/NEON/GEMMLowp.cpp @@ -360,13 +360,21 @@ TEST_SUITE_END() // DynamicQuantization // Deqaunt tests involve returning F32 from the MatrixMultiplyCore kernels and is only implemented in aarch64 TEST_SUITE(Dequant) constexpr AbsoluteTolerance<float> tolerance_dequantized(0.01f); -FIXTURE_DATA_TEST_CASE(RunSmall, NEGEMMLowpDequantizedMatrixMultiplyValidationFixture, framework::DatasetMode::ALL, datasets::SmallGEMMLowpDataset()) +FIXTURE_DATA_TEST_CASE(RunSmall, NEGEMMLowpDequantizedMatrixMultiplyValidationFixture, framework::DatasetMode::ALL, + combine( + datasets::SmallGEMMLowpDataset(), + make("accumulate", {true, false}) + )) { // Validate output validate(Accessor(_target), _reference, tolerance_dequantized); } -FIXTURE_DATA_TEST_CASE(RunLarge, NEGEMMLowpDequantizedMatrixMultiplyValidationFixture, framework::DatasetMode::NIGHTLY, datasets::LargeGEMMLowpDataset()) +FIXTURE_DATA_TEST_CASE(RunLarge, NEGEMMLowpDequantizedMatrixMultiplyValidationFixture, framework::DatasetMode::NIGHTLY, + combine( + datasets::LargeGEMMLowpDataset(), + make("accumulate", {false}) + )) { // Validate output validate(Accessor(_target), _reference, tolerance_dequantized); |