diff options
| author | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2021-04-16 15:08:59 +0100 |
|---|---|---|
| committer | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2021-07-02 15:56:45 +0000 |
| commit | 8155c0253c00aa9e26651361460c66feb39829a6 (patch) | |
| tree | 41dacc432d4d1f1daa32d20d15e5120c11b9fa56 /tests/validation/CL/DepthwiseConvolutionLayer.cpp | |
| parent | 2eb5d16b839cbc28c6cb7f0de7a0bf15290b425a (diff) | |
| download | ComputeLibrary-8155c0253c00aa9e26651361460c66feb39829a6.tar.gz | |
Rework OpenCL Depthwise Convolution
- Remove dedicated kernels for NCHW. Now we only use NHWC with permute
- Remove specialized kernels for 3x3 NHWC
- Simplify CLDepthwiseConvolutionLayer.cpp to call just the native
implementation for both floating-point and quantized data types
- Develop two parametric opencl kernels for depthwise convolution layer NHWC
(floating-point and quantized)
- Add support to export the weights to cl_image
- Extend test for depthwise convolution on opencl
Resolves COMPMID-4417
Change-Id: Ibe533f79c2860f9cac8e921895d5a8f947753a5c
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5893
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'tests/validation/CL/DepthwiseConvolutionLayer.cpp')
| -rw-r--r-- | tests/validation/CL/DepthwiseConvolutionLayer.cpp | 7 |
1 files changed, 3 insertions, 4 deletions
diff --git a/tests/validation/CL/DepthwiseConvolutionLayer.cpp b/tests/validation/CL/DepthwiseConvolutionLayer.cpp index 22922f41a2..79a2678b44 100644 --- a/tests/validation/CL/DepthwiseConvolutionLayer.cpp +++ b/tests/validation/CL/DepthwiseConvolutionLayer.cpp @@ -48,14 +48,13 @@ constexpr RelativeTolerance<float> tolerance_f32(0.01f); /**< constexpr AbsoluteTolerance<uint8_t> tolerance_qasymm8(0); /**< Tolerance value for comparing reference's output against implementation's output for DataType::QASYMM8 */ constexpr float tolerance_num = 0.05f; /**< Tolerance number */ -const auto depth_multipliers = framework::dataset::make("DepthMultiplier", { 1, 2, 5 }); +const auto depth_multipliers = framework::dataset::make("DepthMultiplier", { 1, 5 }); const auto large_depth_multipliers = framework::dataset::make("DepthMultiplier", { 1, 2, 5, 8 }); //Activation Functions const auto ActivationFunctionsDataset = framework::dataset::make("ActivationInfo", { ActivationLayerInfo(), - ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU), ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::BOUNDED_RELU, 6.f, 0.f) }); } // namespace @@ -481,8 +480,8 @@ FIXTURE_DATA_TEST_CASE_NEW(RunSmall, CLDepthwiseConvolutionLayerQuantizedFixture combine(combine(combine(combine(combine(combine(datasets::SmallDepthwiseConvolutionLayerDataset(), depth_multipliers), framework::dataset::make("DataType", DataType::QASYMM8)), - framework::dataset::make("SrcQuantizationInfo", { QuantizationInfo(0.3f, 10), QuantizationInfo(2.2f, 10) })), - framework::dataset::make("DstQuantizationInfo", { QuantizationInfo(0.5f, 4) })), + framework::dataset::make("SrcQuantizationInfo", { QuantizationInfo(0.5f, 128), QuantizationInfo(2.2f, 10) })), + framework::dataset::make("DstQuantizationInfo", { QuantizationInfo(1.f, 128) })), framework::dataset::make("DataLayout", { DataLayout::NHWC })), // NCHW is tested with int8 ActivationFunctionsDataset)) { |