diff options
-rw-r--r-- | src/backends/reference/test/RefLayerTests.cpp | 21 | ||||
-rwxr-xr-x | src/backends/test/Conv2dTestImpl.hpp | 143 | ||||
-rwxr-xr-x | src/backends/test/LayerTests.cpp | 10 | ||||
-rw-r--r-- | src/backends/test/LayerTests.hpp | 7 |
4 files changed, 114 insertions, 67 deletions
diff --git a/src/backends/reference/test/RefLayerTests.cpp b/src/backends/reference/test/RefLayerTests.cpp index 6a92b67e57..40944bf5a6 100644 --- a/src/backends/reference/test/RefLayerTests.cpp +++ b/src/backends/reference/test/RefLayerTests.cpp @@ -55,11 +55,24 @@ ARMNN_AUTO_TEST_CASE(SimpleConvolution2dAsymmetricPaddingNhwc, ARMNN_AUTO_TEST_CASE(SimpleConvolution2dSquareNhwc, SimpleConvolution2d3x3NhwcTest, false) // Depthwise Convolution -ARMNN_AUTO_TEST_CASE(DepthwiseConvolution2d, DepthwiseConvolution2dTest, true) -ARMNN_AUTO_TEST_CASE(DepthwiseConvolution2dUint8, DepthwiseConvolution2dUint8Test, true) +ARMNN_AUTO_TEST_CASE(DepthwiseConvolution2d, DepthwiseConvolution2dTest, true, armnn::DataLayout::NCHW) +ARMNN_AUTO_TEST_CASE(DepthwiseConvolution2dUint8, DepthwiseConvolution2dUint8Test, true, armnn::DataLayout::NCHW) -ARMNN_AUTO_TEST_CASE(UnbiasedDepthwiseConvolution2d, DepthwiseConvolution2dTest, false) -ARMNN_AUTO_TEST_CASE(UnbiasedDepthwiseConvolution2dUint8, DepthwiseConvolution2dUint8Test, false) +ARMNN_AUTO_TEST_CASE(UnbiasedDepthwiseConvolution2d, DepthwiseConvolution2dTest, false, armnn::DataLayout::NCHW) +ARMNN_AUTO_TEST_CASE(UnbiasedDepthwiseConvolution2dUint8, + DepthwiseConvolution2dUint8Test, + false, + armnn::DataLayout::NCHW) + +// NHWC Depthwise Convolution +ARMNN_AUTO_TEST_CASE(DepthwiseConvolution2dNhwc, DepthwiseConvolution2dTest, true, armnn::DataLayout::NHWC) +ARMNN_AUTO_TEST_CASE(DepthwiseConvolution2dUint8Nhwc, DepthwiseConvolution2dUint8Test, true, armnn::DataLayout::NHWC) + +ARMNN_AUTO_TEST_CASE(UnbiasedDepthwiseConvolution2dNhwc, DepthwiseConvolution2dTest, false, armnn::DataLayout::NHWC) +ARMNN_AUTO_TEST_CASE(UnbiasedDepthwiseConvolution2dUint8Nhwc, + DepthwiseConvolution2dUint8Test, + false, + armnn::DataLayout::NHWC) ARMNN_AUTO_TEST_CASE(DepthwiseConvolution2dDepthMul1, DepthwiseConvolution2dDepthMul1Test, true) ARMNN_AUTO_TEST_CASE(DepthwiseConvolution2dDepthMul1Uint8, DepthwiseConvolution2dDepthMul1Uint8Test, true) diff --git a/src/backends/test/Conv2dTestImpl.hpp b/src/backends/test/Conv2dTestImpl.hpp index 993794e325..7a3f452515 100755 --- a/src/backends/test/Conv2dTestImpl.hpp +++ b/src/backends/test/Conv2dTestImpl.hpp @@ -123,7 +123,6 @@ LayerTestResult<T, 4> SimpleConvolution2dTestImpl(armnn::IWorkloadFactory& workl // Note these tensors will use two (identical) batches. - // NOTE: if layout is unknown we will get an exception at this point. armnn::TensorInfo inputTensorInfo = GetTensorInfo<T>(2*inputNum, inputChannels, inputHeight, inputWidth, layout); armnn::TensorInfo outputTensorInfo = GetTensorInfo<T>( 2*outputNum, outputChannels, outputHeight, outputWidth, layout); @@ -600,7 +599,8 @@ template<typename T, typename B> LayerTestResult<T, 4> DepthwiseConvolution2dTestImpl(armnn::IWorkloadFactory& workloadFactory, float qScale, int32_t qOffset, - bool biasEnabled) + bool biasEnabled, + const armnn::DataLayoutIndexed& layout) { unsigned int depthMultiplier = 2; @@ -617,11 +617,12 @@ LayerTestResult<T, 4> DepthwiseConvolution2dTestImpl(armnn::IWorkloadFactory& wo unsigned int outputChannels = inputChannels * depthMultiplier; unsigned int outputBatchSize = inputBatchSize; - armnn::TensorInfo inputTensorInfo({inputBatchSize, inputChannels, inputHeight, inputWidth}, - armnn::GetDataType<T>()); - armnn::TensorInfo outputTensorInfo({outputBatchSize, outputChannels, outputHeight, outputWidth}, - armnn::GetDataType<T>()); - armnn::TensorInfo kernelDesc({depthMultiplier, inputChannels, kernelHeight, kernelWidth}, armnn::GetDataType<T>()); + armnn::TensorInfo inputTensorInfo = GetTensorInfo<T>( + inputBatchSize, inputChannels, inputHeight, inputWidth, layout); + armnn::TensorInfo outputTensorInfo = GetTensorInfo<T>( + outputBatchSize, outputChannels, outputHeight, outputWidth, layout); + armnn::TensorInfo kernelDesc = GetTensorInfo<T>( + depthMultiplier, inputChannels, kernelHeight, kernelWidth, layout); armnn::TensorInfo biasDesc({outputChannels}, armnn::GetDataType<B>()); // Set quantization parameters if the requested type is a quantized type. @@ -637,59 +638,74 @@ LayerTestResult<T, 4> DepthwiseConvolution2dTestImpl(armnn::IWorkloadFactory& wo biasDesc.SetQuantizationOffset(0); } - auto input = MakeTensor<T, 4>(inputTensorInfo, std::vector<T>( - QuantizedVector<T>(inputTensorInfo.GetQuantizationScale(), inputTensorInfo.GetQuantizationOffset(), { - 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, - 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, - 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, - 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, - 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, - 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, - 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, - 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, - 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 - }))); + // NOTE: originalInputData is in NCHW format + std::vector<T> originalInputData = std::vector<T>( + QuantizedVector<T>(inputTensorInfo.GetQuantizationScale(), inputTensorInfo.GetQuantizationOffset(), { + 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, + 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, + 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, + 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, + 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, + 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, + 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, + 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, 0.5f, + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 + })); + std::vector<T> inputData = originalInputData; + // at this point if we require it permute the input data + const armnn::PermutationVector NCHWToNHWC = { 0, 3, 1, 2 }; + if (layout.GetDataLayout() == armnn::DataLayout::NHWC) + { + armnnUtils::Permute(inputTensorInfo.GetShape(), NCHWToNHWC, originalInputData.data(), inputData.data()); + } + auto input = MakeTensor<T, 4>(inputTensorInfo, inputData); std::vector<B> biasV(QuantizedVector<B>(biasDesc.GetQuantizationScale(), biasDesc.GetQuantizationOffset(), {0, 2, 1, -1})); auto bias = MakeTensor<B, 1>(biasDesc, biasV); - auto kernel = MakeTensor<T, 4>(kernelDesc, std::vector<T>( - QuantizedVector<T>(kernelDesc.GetQuantizationScale(), kernelDesc.GetQuantizationOffset(), { - 1, 1, 1, - 1, -1, 1, - 1, 1, 1, - 1, 1, 1, - 1, 1, 1, - - 2, 2, 2, - 2, 2, 2, - 2, 2, 2, - 2, 2, 2, - 2, 2, 2, - - 0, 0, 0, - 0, -1, 0, - 0, 0, 0, - 0, 0, 0, - 0, 0, 0, - - 0, 0, 0, - 0, 0, 0, - 0, 1, 0, - 0, 0, 0, - 0, 0, 0 - }))); + std::vector<T> originalKernelData = std::vector<T>( + QuantizedVector<T>(kernelDesc.GetQuantizationScale(), kernelDesc.GetQuantizationOffset(), { + 1, 1, 1, + 1, -1, 1, + 1, 1, 1, + 1, 1, 1, + 1, 1, 1, + + 2, 2, 2, + 2, 2, 2, + 2, 2, 2, + 2, 2, 2, + 2, 2, 2, + + 0, 0, 0, + 0, -1, 0, + 0, 0, 0, + 0, 0, 0, + 0, 0, 0, + + 0, 0, 0, + 0, 0, 0, + 0, 1, 0, + 0, 0, 0, + 0, 0, 0 + })); + std::vector<T> kernelData = originalKernelData; + if (layout.GetDataLayout() == armnn::DataLayout::NHWC) + { + armnnUtils::Permute(kernelDesc.GetShape(), NCHWToNHWC, originalKernelData.data(), kernelData.data()); + } + auto kernel = MakeTensor<T, 4>(kernelDesc, kernelData); // Manually calculated. - std::vector<T> outputImage = std::vector<T>( + std::vector<T> originalOutputImage = std::vector<T>( QuantizedVector<T>(outputTensorInfo.GetQuantizationScale(), outputTensorInfo.GetQuantizationOffset(), { 3.5f, 3.5f, 3.5f, 3.5f, 3.5f, 3.5f, 3.5f, 6.0f, 6.0f, 6.0f, 6.0f, 6.0f, 6.0f, 6.0f, @@ -723,12 +739,23 @@ LayerTestResult<T, 4> DepthwiseConvolution2dTestImpl(armnn::IWorkloadFactory& wo // Optionally apply bias to output image. if(biasEnabled) { - ApplyBias(outputImage, outputTensorInfo.GetQuantizationScale(), outputTensorInfo.GetQuantizationOffset(), - biasV, biasDesc.GetQuantizationScale(), biasDesc.GetQuantizationOffset(), - outputWidth, outputHeight); + ApplyBias(originalOutputImage, + outputTensorInfo.GetQuantizationScale(), + outputTensorInfo.GetQuantizationOffset(), + biasV, + biasDesc.GetQuantizationScale(), + biasDesc.GetQuantizationOffset(), + outputWidth, + outputHeight); } LayerTestResult<T, 4> ret(outputTensorInfo); + std::vector<T> outputImage = originalOutputImage; + if (layout.GetDataLayout() == armnn::DataLayout::NHWC) + { + armnnUtils::Permute(outputTensorInfo.GetShape(), NCHWToNHWC, originalOutputImage.data(), outputImage.data()); + } + ret.outputExpected = MakeTensor<T, 4>(outputTensorInfo, outputImage); std::unique_ptr<armnn::ITensorHandle> inputHandle = workloadFactory.CreateTensorHandle(inputTensorInfo); @@ -754,6 +781,7 @@ LayerTestResult<T, 4> DepthwiseConvolution2dTestImpl(armnn::IWorkloadFactory& wo data.m_Parameters.m_PadTop = 1; data.m_Parameters.m_PadBottom = 1; data.m_Parameters.m_BiasEnabled = biasEnabled; + data.m_Parameters.m_DataLayout = layout.GetDataLayout(); std::unique_ptr<armnn::IWorkload> workload = workloadFactory.CreateDepthwiseConvolution2d(data, info); inputHandle->Allocate(); @@ -1132,7 +1160,8 @@ LayerTestResult<T, 4> CompareDepthwiseConvolution2dTestImpl(armnn::IWorkloadFact auto input = MakeRandomTensor<T, 4>(inputTensorInfo, 124908, 0.0f, 255.0f); auto kernel = MakeRandomTensor<T, 4>(kernelDesc, 891234, 0.0f, 255.0f); - auto bias = MakeRandomTensor<typename FullyConnectedBiasTypeForInputType<T>::Type, 1>(biasDesc, 1028, 0.0f, 255.0f); + auto bias = MakeRandomTensor<typename FullyConnectedBiasTypeForInputType<T>::Type, 1>( + biasDesc, 1028, 0.0f, 255.0f); std::unique_ptr<armnn::ITensorHandle> inputHandle = workloadFactory.CreateTensorHandle(inputTensorInfo); std::unique_ptr<armnn::ITensorHandle> outputHandle = workloadFactory.CreateTensorHandle(outputTensorInfo); diff --git a/src/backends/test/LayerTests.cpp b/src/backends/test/LayerTests.cpp index 43a42f305c..ae6d16c755 100755 --- a/src/backends/test/LayerTests.cpp +++ b/src/backends/test/LayerTests.cpp @@ -627,9 +627,10 @@ LayerTestResult<float, 4> Convolution2dAsymmetricPaddingTest(armnn::IWorkloadFac } LayerTestResult<float, 4> DepthwiseConvolution2dTest(armnn::IWorkloadFactory& workloadFactory, - bool biasEnabled) + bool biasEnabled, + const armnn::DataLayoutIndexed& layout) { - return DepthwiseConvolution2dTestImpl<float, float>(workloadFactory, 0.0f, 0, biasEnabled); + return DepthwiseConvolution2dTestImpl<float, float>(workloadFactory, 0.0f, 0, biasEnabled, layout); } LayerTestResult<float, 4> DepthwiseConvolution2dDepthNhwcTest(armnn::IWorkloadFactory& workloadFactory, @@ -652,9 +653,10 @@ LayerTestResult<float, 4> DepthwiseConvolution2dAsymmetricTest(armnn::IWorkloadF } LayerTestResult<uint8_t, 4> DepthwiseConvolution2dUint8Test(armnn::IWorkloadFactory& workloadFactory, - bool biasEnabled) + bool biasEnabled, + const armnn::DataLayoutIndexed& layout) { - return DepthwiseConvolution2dTestImpl<uint8_t, int32_t>(workloadFactory, 0.5f, 50, biasEnabled); + return DepthwiseConvolution2dTestImpl<uint8_t, int32_t>(workloadFactory, 0.5f, 50, biasEnabled, layout); } LayerTestResult<uint8_t, 4> DepthwiseConvolution2dDepthMul1Uint8Test(armnn::IWorkloadFactory& workloadFactory, diff --git a/src/backends/test/LayerTests.hpp b/src/backends/test/LayerTests.hpp index fec73d08f4..0e45024e22 100644 --- a/src/backends/test/LayerTests.hpp +++ b/src/backends/test/LayerTests.hpp @@ -72,7 +72,9 @@ LayerTestResult<float, 4> Convolution1dTest(armnn::IWorkloadFactory& workloadF LayerTestResult<uint8_t, 4> Convolution1dUint8Test(armnn::IWorkloadFactory& workloadFactory, bool biasEnabled); -LayerTestResult<float, 4> DepthwiseConvolution2dTest(armnn::IWorkloadFactory& workloadFactory, bool biasEnabled); +LayerTestResult<float, 4> DepthwiseConvolution2dTest(armnn::IWorkloadFactory& workloadFactory, + bool biasEnabled, + const armnn::DataLayoutIndexed& layout); LayerTestResult<float, 4> DepthwiseConvolution2dDepthNhwcTest(armnn::IWorkloadFactory& workloadFactory, bool biasEnabled); @@ -327,7 +329,8 @@ LayerTestResult<uint8_t, 4> SimpleConvolution2d3x3Uint8Test(armnn::IWorkloadFact const armnn::DataLayoutIndexed& layout); LayerTestResult<uint8_t, 4> DepthwiseConvolution2dUint8Test(armnn::IWorkloadFactory& workloadFactory, - bool biasEnabled); + bool biasEnabled, + const armnn::DataLayoutIndexed& layout); LayerTestResult<uint8_t, 4> DepthwiseConvolution2dDepthMul1Uint8Test(armnn::IWorkloadFactory& workloadFactory, bool biasEnabled); |