diff options
author | Radu Salavat <radu.salavat@arm.com> | 2024-02-27 18:32:26 +0000 |
---|---|---|
committer | Radu Salavat <radu.salavat@arm.com> | 2024-04-11 08:47:50 +0000 |
commit | f1f1f87132690a8061801ef1a4638d637c780df7 (patch) | |
tree | 8ad4c3739217b3bc6281f4e0b9a7a63fe6c3f9bb /tests/validation/CL/GEMMLowp.cpp | |
parent | 1322065a3fbd15b00dbfb0969d6b438b5ba15530 (diff) | |
download | ComputeLibrary-f1f1f87132690a8061801ef1a4638d637c780df7.tar.gz |
Add in place summation to CPU GEMM kernels
Instead of dispatching the sum postop for GEMM kernels to a
separate kernel + add, that requires an extra destination sized
allocation, plus 3 extra load/stores per element,
just do it in the GEMM kernel.
Resolves: ONCPUML-1442
Signed-off-by: Radu Salavat <radu.salavat@arm.com>
Co-authored-by: Milos Puzovic <milos.puzovic@arm.com>
Change-Id: I7a1f2da3300875fa1ac88b705a34390969518077
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/11298
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'tests/validation/CL/GEMMLowp.cpp')
-rw-r--r-- | tests/validation/CL/GEMMLowp.cpp | 13 |
1 files changed, 7 insertions, 6 deletions
diff --git a/tests/validation/CL/GEMMLowp.cpp b/tests/validation/CL/GEMMLowp.cpp index 1ae9e96626..78d794a9bb 100644 --- a/tests/validation/CL/GEMMLowp.cpp +++ b/tests/validation/CL/GEMMLowp.cpp @@ -71,7 +71,7 @@ FIXTURE_DATA_TEST_CASE(RunLarge, CLGEMMLowpMatrixMultiplyCoreFixture, framework: } using CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputFixtureBatchedUnsigned = - GEMMLowpMatrixMultiplyCoreFusedOffsetOutputGenericValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, false, false, uint8_t, uint8_t, true>; + GEMMLowpBatchedMatrixMultiplyCoreFusedOffsetOutputFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, false, false, uint8_t, uint8_t, true>; TEST_SUITE(BatchedMatMul) TEST_SUITE(QASYMM8) FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputFixtureBatchedUnsigned, framework::DatasetMode::ALL, @@ -84,7 +84,7 @@ FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputFi TEST_SUITE_END() // QASYMM8 using CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputFixtureBatchedSigned = - GEMMLowpMatrixMultiplyCoreFusedOffsetOutputGenericValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, false, false, int8_t, int8_t, true>; + GEMMLowpBatchedMatrixMultiplyCoreFusedOffsetOutputFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, false, false, int8_t, int8_t, true>; TEST_SUITE(QASYMM8_SIGNED) FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputFixtureBatchedSigned, framework::DatasetMode::ALL, combine(datasets::SmallGEMMLowpFusedBatchedMatMulDataset(), @@ -98,7 +98,7 @@ TEST_SUITE_END() // BatchedMatMul TEST_SUITE(FusedOffsetOutput) TEST_SUITE(QASYMM8) -using CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputUint8Fixture = GEMMLowpMatrixMultiplyCoreFusedOffsetOutputGenericValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore>; +using CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputUint8Fixture = GEMMLowpMatrixMultiplyCoreFusedOffsetOutputValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore>; FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputUint8Fixture, framework::DatasetMode::ALL, combine(datasets::SmallGEMMLowpFusedOffsetOutputUint8Dataset(), make("DataType", { DataType::QASYMM8 }), @@ -110,7 +110,7 @@ FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputUi TEST_SUITE(Output3D) using CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputOutput3DUint8Fixture = - GEMMLowpMatrixMultiplyCoreFusedOffsetOutputGenericValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, false, true>; + GEMMLowpMatrixMultiplyCoreFusedOffsetOutputValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, false, true>; FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputOutput3DUint8Fixture, framework::DatasetMode::ALL, combine(datasets::SmallGEMMLowpFusedOffsetOutputOutput3DUint8Dataset(), make("DataType", { DataType::QASYMM8 }), @@ -123,7 +123,7 @@ TEST_SUITE_END() // Output3D TEST_SUITE(InputOutput3D) using CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputInputOutput3DUint8Fixture = - GEMMLowpMatrixMultiplyCoreFusedOffsetOutputGenericValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, true, true>; + GEMMLowpMatrixMultiplyCoreFusedOffsetOutputValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, true, true>; FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputInputOutput3DUint8Fixture, framework::DatasetMode::ALL, combine(datasets::SmallGEMMLowpFusedOffsetOutputInputOutput3DUint8Dataset(), make("DataType", { DataType::QASYMM8 }), @@ -148,7 +148,8 @@ using CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputInt8Fixture = GEMMLowpMatrixMultiplyCoreFusedOffsetOutputValidationFixture<CLTensor, CLAccessor, CLGEMMLowpMatrixMultiplyCore, false, false, int8_t, int8_t>; FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMLowpMatrixMultiplyCoreFusedOffsetOutputInt8Fixture, framework::DatasetMode::ALL, combine(datasets::SmallGEMMLowpFusedOffsetOutputInt8Dataset(), - make("DataType", { DataType::QASYMM8_SIGNED }))) + make("DataType", { DataType::QASYMM8_SIGNED }), + make("reshape_b_only_on_first_run", { false }))) { // Validate output validate(CLAccessor(_target), _reference, tolerance_quant); |