diff options
Diffstat (limited to 'src/armnn/test/optimizations')
10 files changed, 61 insertions, 39 deletions
diff --git a/src/armnn/test/optimizations/AddBroadcastReshapeLayerTests.cpp b/src/armnn/test/optimizations/AddBroadcastReshapeLayerTests.cpp index 36a4507fc3..7573005518 100644 --- a/src/armnn/test/optimizations/AddBroadcastReshapeLayerTests.cpp +++ b/src/armnn/test/optimizations/AddBroadcastReshapeLayerTests.cpp @@ -290,7 +290,7 @@ TEST_CASE("ReshapeParentConstLayerTest") { Graph graph; const TensorInfo info0({ 1, 2, 3, 5 }, DataType::QAsymmU8); - const TensorInfo info1({ 5 }, DataType::QAsymmU8); + const TensorInfo info1({ 5 }, DataType::QAsymmU8, 0.0f, 0, true); const TensorInfo outputInfo({ 1, 2, 3, 5 }, DataType::QAsymmU8); auto input = graph.AddLayer<InputLayer>(0, "input"); @@ -346,7 +346,7 @@ TEST_CASE("ReshapeParentConstAddLayerMultipleConnectionsTest") // What we'll do is have two sequential add layers both using the same const tensor. Graph graph; const TensorInfo inputInfo({ 1, 512 }, DataType::Float32); - const TensorInfo constantTermInfo({ 1 }, DataType::Float32); + const TensorInfo constantTermInfo({ 1 }, DataType::Float32, 0.0f, 0, true); const TensorInfo outputInfo({ 1, 512 }, DataType::Float32); auto input = graph.AddLayer<InputLayer>(0, "input"); diff --git a/src/armnn/test/optimizations/ConvertConstantsBFloatTests.cpp b/src/armnn/test/optimizations/ConvertConstantsBFloatTests.cpp index b78a1bf207..7b326fa8bc 100644 --- a/src/armnn/test/optimizations/ConvertConstantsBFloatTests.cpp +++ b/src/armnn/test/optimizations/ConvertConstantsBFloatTests.cpp @@ -32,7 +32,7 @@ TEST_CASE("ConvertConstantsFloatToBFloatTest") -3.1055E+29f, // 0xF07ADC3C Round up -9.149516E-10f // 0xB07B7FFF Round down }; - armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float32), floatWeights); + armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float32, 0.0f, 0, true), floatWeights); // Create simple test network auto input = graph.AddLayer<armnn::InputLayer>(0, "input"); @@ -88,7 +88,7 @@ TEST_CASE("ConvertConstantsBFloatToFloatTest") std::vector<uint16_t> bfWeights(8); armnnUtils::FloatingPointConverter::ConvertFloat32ToBFloat16(convWeightsData.data(), convWeightsData.size(), bfWeights.data()); - armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::BFloat16), bfWeights); + armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::BFloat16, 0.0f, 0, true), bfWeights); //Create the simple test network auto input = graph.AddLayer<armnn::InputLayer>(0, "input"); diff --git a/src/armnn/test/optimizations/ConvertConstantsFloatToHalfTests.cpp b/src/armnn/test/optimizations/ConvertConstantsFloatToHalfTests.cpp index e6cca4f7bf..f74ab0f308 100644 --- a/src/armnn/test/optimizations/ConvertConstantsFloatToHalfTests.cpp +++ b/src/armnn/test/optimizations/ConvertConstantsFloatToHalfTests.cpp @@ -25,7 +25,7 @@ TEST_CASE("ConvertConstantsFloatToHalfTest") // Create const tensor from fp32 data unsigned int dims[] = { 4, 1, 1, 1 }; std::vector<float> floatWeights{ 1.0f, 2.0f, 3.0f, 4.0f }; - armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float32), floatWeights); + armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float32, 0.0f, 0, true), floatWeights); // Create simple test network auto input = graph.AddLayer<armnn::InputLayer>(0, "input"); diff --git a/src/armnn/test/optimizations/ConvertConstantsHalfToFloatTests.cpp b/src/armnn/test/optimizations/ConvertConstantsHalfToFloatTests.cpp index 2ec1279f33..c4551525c1 100644 --- a/src/armnn/test/optimizations/ConvertConstantsHalfToFloatTests.cpp +++ b/src/armnn/test/optimizations/ConvertConstantsHalfToFloatTests.cpp @@ -25,7 +25,7 @@ TEST_CASE("ConvertConstantsHalfToFloatTest") std::vector<uint16_t> halfWeights(4); armnnUtils::FloatingPointConverter::ConvertFloat32To16(convWeightsData.data(), convWeightsData.size(), halfWeights.data()); - armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float16), halfWeights); + armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float16, 0.0f, 0, true), halfWeights); //Create the simple test network auto input = graph.AddLayer<armnn::InputLayer>(0, "input"); diff --git a/src/armnn/test/optimizations/FoldPadTests.cpp b/src/armnn/test/optimizations/FoldPadTests.cpp index 11f09e80e0..a598983706 100644 --- a/src/armnn/test/optimizations/FoldPadTests.cpp +++ b/src/armnn/test/optimizations/FoldPadTests.cpp @@ -45,7 +45,7 @@ TEST_CASE("FoldPadLayerIntoConvolution2dLayer") convolution2dDescriptor.m_DataLayout = DataLayout::NHWC; std::vector<float> weightsVector(18); - ConstTensor weights(TensorInfo(4, weightsShape, DataType::Float32), weightsVector); + ConstTensor weights(TensorInfo(4, weightsShape, DataType::Float32, 0.0f, 0, true), weightsVector); Convolution2dLayer* conv2dLayer = graph.AddLayer<Convolution2dLayer>(convolution2dDescriptor, "conv2d"); conv2dLayer->m_Weight = std::make_unique<ScopedTensorHandle>(weights); @@ -122,7 +122,7 @@ TEST_CASE("FoldPadLayerIntoDepthwiseConvolution2dLayer") depthwiseConvolution2dDescriptor.m_DataLayout = DataLayout::NHWC; std::vector<float> weightsVector(18); - ConstTensor weights(TensorInfo(4, weightsShape, DataType::Float32), weightsVector); + ConstTensor weights(TensorInfo(4, weightsShape, DataType::Float32, 0.0f, 0, true), weightsVector); auto* depthwiseConv2dLayer = graph.AddLayer<DepthwiseConvolution2dLayer>(depthwiseConvolution2dDescriptor, "depthwiseConv2d"); @@ -526,7 +526,9 @@ TEST_CASE("FoldPadLayerIntoPooling2dLayer_ExecuteInferenceWithAndWithoutOptimiza NetworkId networkIdentifier; CHECK(run->LoadNetwork(networkIdentifier, std::move(optimizedNetwork)) == Status::Success); - InputTensors inputTensors{{0, ConstTensor(run->GetInputTensorInfo(networkIdentifier, 0), inputData.data())}}; + TensorInfo inputTensorInfo = run->GetInputTensorInfo(networkIdentifier, 0); + inputTensorInfo.SetConstant(true); + InputTensors inputTensors{{0, ConstTensor(inputTensorInfo, inputData.data())}}; // Set the initial values of the data to different values to the golden data just in case the inference fails. std::vector<float> optimizedData(32, -std::numeric_limits<float>::infinity()); @@ -614,10 +616,10 @@ TEST_CASE("FoldPadLayerIntoConv2dLayer_ExecuteInferenceWithAndWithoutOptimizatio 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42}; - TensorInfo weightsInfo(4, weightsShape, DataType::Float32); + TensorInfo weightsInfo(4, weightsShape, DataType::Float32, 0.0f, 0, true); ConstTensor weights(weightsInfo, weightsData); std::vector<float> biasVector = {5, 6, 7, 8}; - TensorInfo biasInfo({4}, DataType::Float32); + TensorInfo biasInfo({4}, DataType::Float32, 0.0f, 0, true); ConstTensor bias(biasInfo, biasVector); Optional<ConstTensor> optionalBias = Optional<ConstTensor>(bias); @@ -644,7 +646,9 @@ TEST_CASE("FoldPadLayerIntoConv2dLayer_ExecuteInferenceWithAndWithoutOptimizatio NetworkId networkIdentifier; CHECK(run->LoadNetwork(networkIdentifier, std::move(optimizedNetwork)) == Status::Success); - InputTensors inputTensors{{0, ConstTensor(run->GetInputTensorInfo(networkIdentifier, 0), inputData.data())}}; + TensorInfo inputTensorInfo = run->GetInputTensorInfo(networkIdentifier, 0); + inputTensorInfo.SetConstant(true); + InputTensors inputTensors{{0, ConstTensor(inputTensorInfo, inputData.data())}}; // Set the initial values of the data to different values to the golden data just in case the inference fails. std::vector<float> optimizedData(100, -std::numeric_limits<float>::infinity()); @@ -732,10 +736,10 @@ TEST_CASE("FoldPadLayerIntoDepthwiseConv2dLayer_ExecuteInferenceWithAndWithoutOp 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42}; - TensorInfo weightsInfo(4, weightsShape, DataType::Float32); + TensorInfo weightsInfo(4, weightsShape, DataType::Float32, 0.0f, 0, true); ConstTensor weights(weightsInfo, weightsData); std::vector<float> biasVector = {5, 6, 7, 8, 9, 10, 11, 12, 5, 6, 7, 8}; - TensorInfo biasInfo({12}, DataType::Float32); + TensorInfo biasInfo({12}, DataType::Float32, 0.0f, 0, true); ConstTensor bias(biasInfo, biasVector); Optional<ConstTensor> optionalBias = Optional<ConstTensor>(bias); @@ -762,7 +766,9 @@ TEST_CASE("FoldPadLayerIntoDepthwiseConv2dLayer_ExecuteInferenceWithAndWithoutOp NetworkId networkIdentifier; CHECK(run->LoadNetwork(networkIdentifier, std::move(optimizedNetwork)) == Status::Success); - InputTensors inputTensors{{0, ConstTensor(run->GetInputTensorInfo(networkIdentifier, 0), inputData.data())}}; + TensorInfo inputTensorInfo = run->GetInputTensorInfo(networkIdentifier, 0); + inputTensorInfo.SetConstant(true); + InputTensors inputTensors{{0, ConstTensor(inputTensorInfo, inputData.data())}}; // Set the initial values of the data to different values to the golden data just in case the inference fails. std::vector<float> optimizedData(300, -std::numeric_limits<float>::infinity()); diff --git a/src/armnn/test/optimizations/Fp32NetworkToBf16ConverterTests.cpp b/src/armnn/test/optimizations/Fp32NetworkToBf16ConverterTests.cpp index 384b14c0cf..63cd170f02 100644 --- a/src/armnn/test/optimizations/Fp32NetworkToBf16ConverterTests.cpp +++ b/src/armnn/test/optimizations/Fp32NetworkToBf16ConverterTests.cpp @@ -59,12 +59,12 @@ TEST_CASE("Fp32NetworkToBf16OptimizationConv2DTest") -3.1055E+29f, // 0xF07ADC3C Round up -9.149516E-10f // 0xB07B7FFF Round down }; - armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float32), floatWeights); + armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float32, 0.0f, 0, true), floatWeights); // Create const bias fp32 data unsigned int biasDims[] {4}; std::vector<float> floatBias{ 1.0f, 2.0f, 3.0f, 4.0f }; - armnn::ConstTensor bias(armnn::TensorInfo(1, biasDims, armnn::DataType::Float32), floatBias); + armnn::ConstTensor bias(armnn::TensorInfo(1, biasDims, armnn::DataType::Float32, 0.0f, 0, true), floatBias); // A network with Convolution2d layer auto input = graph.AddLayer<armnn::InputLayer>(0, "input"); @@ -129,12 +129,12 @@ TEST_CASE("Fp32NetworkToBf16OptimizationFullyConnectedTest") -3.1055E+29f, // 0xF07ADC3C Round up -9.149516E-10f // 0xB07B7FFF Round down }; - armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float32), floatWeights); + armnn::ConstTensor weights(armnn::TensorInfo(4, dims, armnn::DataType::Float32, 0.0f, 0, true), floatWeights); // Create const bias fp32 data unsigned int biasDims[] {4}; std::vector<float> floatBias{ 1.0f, 2.0f, 3.0f, 4.0f }; - armnn::ConstTensor bias(armnn::TensorInfo(1, biasDims, armnn::DataType::Float32), floatBias); + armnn::ConstTensor bias(armnn::TensorInfo(1, biasDims, armnn::DataType::Float32, 0.0f, 0, true), floatBias); // A network with FullyConnected layer auto input = graph.AddLayer<armnn::InputLayer>(0, "input"); diff --git a/src/armnn/test/optimizations/FuseActivationTests.cpp b/src/armnn/test/optimizations/FuseActivationTests.cpp index 2352a3c498..54a9d9a189 100644 --- a/src/armnn/test/optimizations/FuseActivationTests.cpp +++ b/src/armnn/test/optimizations/FuseActivationTests.cpp @@ -66,7 +66,7 @@ struct Convolution2dTest 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42}; std::vector<T> weightsVector = armnnUtils::QuantizedVector<T>(weightsData, scale, offset); - TensorInfo weightsInfo(GetWeightsShape(), ArmnnType, scale, offset); + TensorInfo weightsInfo(GetWeightsShape(), ArmnnType, scale, offset, true); ConstTensor weights(weightsInfo, weightsVector); Optional<ConstTensor> optionalBias; @@ -115,7 +115,7 @@ public: 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42}; std::vector<T> weightsVector = armnnUtils::QuantizedVector<T>(weightsData, scale, offset); - TensorInfo weightsInfo(GetWeightsShape(), ArmnnType, scale, offset); + TensorInfo weightsInfo(GetWeightsShape(), ArmnnType, scale, offset, true); ConstTensor weights(weightsInfo, weightsVector); Optional<ConstTensor> optionalBias; @@ -212,10 +212,10 @@ public: std::vector<T> varianceVector = GetVector<T>(GetOutputShape()[3], 1.0f, 0.1f); const unsigned int outputChannelSize[] = { GetOutputShape()[3] }; - ConstTensor beta(TensorInfo(1, outputChannelSize, ArmnnType), betaVector); - ConstTensor gamma(TensorInfo(1, outputChannelSize, ArmnnType), gammaVector); - ConstTensor mean(TensorInfo(1, outputChannelSize, ArmnnType), meanVector); - ConstTensor variance(TensorInfo(1, outputChannelSize, ArmnnType), varianceVector); + ConstTensor beta(TensorInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true), betaVector); + ConstTensor gamma(TensorInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true), gammaVector); + ConstTensor mean(TensorInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true), meanVector); + ConstTensor variance(TensorInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true), varianceVector); return network->AddBatchNormalizationLayer(descriptor, mean, variance, beta, gamma, name); } @@ -491,8 +491,11 @@ void FuseActivationIntoPreviousLayerTest(ActivationDescriptor activationDescript std::vector<T> inputDataFused = armnnUtils::QuantizedVector<T>(data, scale, offset); std::vector<T> outputDataFused(LayerTest::outputSize); + armnn::TensorInfo inputTensorInfo = run->GetInputTensorInfo(networkIdentifier, 0); + inputTensorInfo.SetConstant(true); + InputTensors inputTensorsFused{ - {0, ConstTensor(run->GetInputTensorInfo(networkIdentifier, 0), inputDataFused.data())}}; + {0, ConstTensor(inputTensorInfo, inputDataFused.data())}}; OutputTensors outputTensorsFused{ {0, Tensor(run->GetOutputTensorInfo(networkIdentifier, 0), outputDataFused.data())}}; @@ -545,8 +548,11 @@ void FuseActivationIntoPreviousLayerTest(ActivationDescriptor activationDescript std::vector<T> outputDataNotFused(LayerTest::outputSize); std::vector<T> outputData2NotFused(LayerTest::outputSize); + TensorInfo inputTensorInfoNotFused = runNotFused->GetInputTensorInfo(networkIdentifierNotFused, 0); + inputTensorInfoNotFused.SetConstant(true); + InputTensors inputTensorsNotFused{ - {0, ConstTensor(runNotFused->GetInputTensorInfo(networkIdentifierNotFused, 0), inputDataNotFused.data())}}; + {0, ConstTensor(inputTensorInfoNotFused, inputDataNotFused.data())}}; OutputTensors outputTensorsNotFused{ {0, Tensor(runNotFused->GetOutputTensorInfo(networkIdentifierNotFused, 0), outputDataNotFused.data())}, {1, Tensor(runNotFused->GetOutputTensorInfo(networkIdentifierNotFused, 1), outputData2NotFused.data())}}; @@ -591,8 +597,11 @@ bool FuseActivationSimpleTest(ActivationDescriptor activationDescriptor, Compute std::vector<T> inputDataFused = armnnUtils::QuantizedVector<T>(data, scale, offset); std::vector<T> outputDataFused(LayerTest::outputSize); + TensorInfo inputTensorInfo = run->GetInputTensorInfo(networkIdentifier, 0); + inputTensorInfo.SetConstant(true); + InputTensors inputTensorsFused{ - {0, ConstTensor(run->GetInputTensorInfo(networkIdentifier, 0), inputDataFused.data())}}; + {0, ConstTensor(inputTensorInfo, inputDataFused.data())}}; OutputTensors outputTensorsFused{ {0, Tensor(run->GetOutputTensorInfo(networkIdentifier, 0), outputDataFused.data())}}; diff --git a/src/armnn/test/optimizations/FuseBatchNormTests.cpp b/src/armnn/test/optimizations/FuseBatchNormTests.cpp index 20d2940b81..0e969c1a5c 100644 --- a/src/armnn/test/optimizations/FuseBatchNormTests.cpp +++ b/src/armnn/test/optimizations/FuseBatchNormTests.cpp @@ -107,11 +107,11 @@ INetworkPtr CreatNetwork(bool depthwise, bool preventFusing) 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42}; std::vector<T> weightsVector(begin(weightsIntVector), end(weightsIntVector)); - TensorInfo weightsInfo(4, weightsDimensionSizes, ArmnnType); + TensorInfo weightsInfo(4, weightsDimensionSizes, ArmnnType, 0.0f, 0, true); ConstTensor weights(weightsInfo, weightsVector); std::vector<T> biasVector = GetVector<T>(outputDimensionSizes[3], 3.3f, 0.1f); - TensorInfo biasInfo(1, outputChannelSize, ArmnnType); + TensorInfo biasInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true); ConstTensor bias(biasInfo, biasVector); Optional<ConstTensor> optionalBias = Optional<ConstTensor>(bias); @@ -120,10 +120,10 @@ INetworkPtr CreatNetwork(bool depthwise, bool preventFusing) std::vector<T> meanVector = GetVector<T>(outputDimensionSizes[3], 0.1f, 0.1f); std::vector<T> varianceVector = GetVector<T>(outputDimensionSizes[3], 1.0f, 0.1f); - ConstTensor beta (TensorInfo(1, outputChannelSize, ArmnnType), betaVector); - ConstTensor gamma (TensorInfo(1, outputChannelSize, ArmnnType), gammaVector); - ConstTensor mean (TensorInfo(1, outputChannelSize, ArmnnType), meanVector); - ConstTensor variance(TensorInfo(1, outputChannelSize, ArmnnType), varianceVector); + ConstTensor beta (TensorInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true), betaVector); + ConstTensor gamma (TensorInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true), gammaVector); + ConstTensor mean (TensorInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true), meanVector); + ConstTensor variance(TensorInfo(1, outputChannelSize, ArmnnType, 0.0f, 0, true), varianceVector); // Create a network INetworkPtr network = INetwork::Create(); @@ -215,8 +215,10 @@ void FuseBatchNormIntoConvTest(bool depthwise, float tolerance, armnn::Compute b outputDataFused.resize(108); } + TensorInfo inputTensorInfo = run->GetInputTensorInfo(networkIdentifier, 0); + inputTensorInfo.SetConstant(true); InputTensors inputTensorsFused { - {0, ConstTensor(run->GetInputTensorInfo (networkIdentifier, 0), inputDataFused.data())}}; + {0, ConstTensor(inputTensorInfo, inputDataFused.data())}}; OutputTensors outputTensorsFused{ {0, Tensor(run->GetOutputTensorInfo(networkIdentifier, 0), outputDataFused.data())}}; @@ -259,8 +261,11 @@ void FuseBatchNormIntoConvTest(bool depthwise, float tolerance, armnn::Compute b outputDataNotFused.resize(108); outputData2NotFused.resize(108); } + + TensorInfo inputTensorInfo2 = runNotFused->GetInputTensorInfo(networkIdentifierNotFused, 0); + inputTensorInfo2.SetConstant(true); InputTensors inputTensorsNotFused{ - {0, ConstTensor(runNotFused->GetInputTensorInfo(networkIdentifierNotFused, 0), inputDataNotFused.data())}}; + {0, ConstTensor(inputTensorInfo2, inputDataNotFused.data())}}; OutputTensors outputTensorsNotFused{ {0, Tensor(runNotFused->GetOutputTensorInfo(networkIdentifierNotFused, 0), outputDataNotFused.data())}, {1, Tensor(runNotFused->GetOutputTensorInfo(networkIdentifierNotFused, 1), outputData2NotFused.data())}}; diff --git a/src/armnn/test/optimizations/PermuteAndBatchToSpaceAsDepthToSpaceTests.cpp b/src/armnn/test/optimizations/PermuteAndBatchToSpaceAsDepthToSpaceTests.cpp index e91e16f132..f862315220 100644 --- a/src/armnn/test/optimizations/PermuteAndBatchToSpaceAsDepthToSpaceTests.cpp +++ b/src/armnn/test/optimizations/PermuteAndBatchToSpaceAsDepthToSpaceTests.cpp @@ -232,7 +232,7 @@ TEST_CASE("PermuteAndBatchToSpaceAsDepthToSpaceCorrectnessTest") -1.0f, -2.0f, -3.0f, -4.0f, -10.0f, -20.0f, -30.0f, -40.0f, -100.0f, -200.0f, -300.0f, -400.0f, // clang-format on }; - ConstTensor input(TensorInfo({ 1, 2, 3, 4 }, DataType::Float32), inputData); + ConstTensor input(TensorInfo({ 1, 2, 3, 4 }, DataType::Float32, 0.0f, 0, true), inputData); InputTensors inputs = { { 0, input } }; std::vector<float> outputData(4 * 6); Tensor output(TensorInfo({ 1, 4, 6, 1 }, DataType::Float32), outputData.data()); @@ -279,7 +279,7 @@ TEST_CASE("TransposeAndBatchToSpaceAsDepthToSpaceCorrectnessTest") -1.0f, -2.0f, -3.0f, -4.0f, -10.0f, -20.0f, -30.0f, -40.0f, -100.0f, -200.0f, -300.0f, -400.0f, // clang-format on }; - ConstTensor input(TensorInfo({ 1, 2, 3, 4 }, DataType::Float32), inputData); + ConstTensor input(TensorInfo({ 1, 2, 3, 4 }, DataType::Float32, 0.0f, 0, true), inputData); InputTensors inputs = { { 0, input } }; std::vector<float> outputData(4 * 6); Tensor output(TensorInfo({ 1, 4, 6, 1 }, DataType::Float32), outputData.data()); diff --git a/src/armnn/test/optimizations/ReduceMultipleAxesTests.cpp b/src/armnn/test/optimizations/ReduceMultipleAxesTests.cpp index 0be8857224..692f371356 100644 --- a/src/armnn/test/optimizations/ReduceMultipleAxesTests.cpp +++ b/src/armnn/test/optimizations/ReduceMultipleAxesTests.cpp @@ -94,9 +94,11 @@ void ReduceWithMultipleAxesTest(INetworkPtr& network, // Create input and output tensors std::vector<float> outputData(expectedOutput.size()); + armnn::TensorInfo inputTensorInfo = run->GetInputTensorInfo(networkIdentifier, 0); + inputTensorInfo.SetConstant(true); InputTensors inputTensors { - {0, armnn::ConstTensor(run->GetInputTensorInfo(networkIdentifier, 0), inputData.data())} + {0, armnn::ConstTensor(inputTensorInfo, inputData.data())} }; OutputTensors outputTensors { |