From 68dd25fbe6e4d3c3513fa5993863419769aa08fc Mon Sep 17 00:00:00 2001 From: Sang-Hoon Park Date: Mon, 19 Oct 2020 16:00:11 +0100 Subject: COMPMID-3637: Move utility headers from arm_compute to src Signed-off-by: Georgios Pinitas Change-Id: If9d6fa8c900b68c4b6fd373f2fc1f9abb83ea917 Signed-off-by: Michalis Spyrou Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4145 Tested-by: Arm Jenkins Reviewed-by: Sang-Hoon Park Comments-Addressed: Arm Jenkins --- src/runtime/CL/CLHelpers.cpp | 3 +- src/runtime/CL/CLMemory.cpp | 4 +- src/runtime/CL/CLRuntimeContext.cpp | 2 + src/runtime/CL/CLTensorAllocator.cpp | 4 +- src/runtime/CL/functions/CLArgMinMaxLayer.cpp | 11 ++-- src/runtime/CL/functions/CLConcatenateLayer.cpp | 1 + .../functions/CLConvertFullyConnectedWeights.cpp | 2 + src/runtime/CL/functions/CLConvolutionLayer.cpp | 2 + src/runtime/CL/functions/CLCropResize.cpp | 4 ++ src/runtime/CL/functions/CLDeconvolutionLayer.cpp | 2 + .../CL/functions/CLDirectConvolutionLayer.cpp | 2 +- .../CL/functions/CLDirectDeconvolutionLayer.cpp | 1 + src/runtime/CL/functions/CLFFT1D.cpp | 2 +- src/runtime/CL/functions/CLFFTConvolutionLayer.cpp | 5 +- src/runtime/CL/functions/CLFill.cpp | 2 + src/runtime/CL/functions/CLFullyConnectedLayer.cpp | 2 +- src/runtime/CL/functions/CLGEMM.cpp | 15 +++-- .../CL/functions/CLGEMMConvolutionLayer.cpp | 3 +- .../CL/functions/CLGEMMLowpMatrixMultiplyCore.cpp | 7 ++- .../CL/functions/CLGenerateProposalsLayer.cpp | 1 + .../CL/functions/CLInstanceNormalizationLayer.cpp | 2 + src/runtime/CL/functions/CLLSTMLayerQuantized.cpp | 1 + src/runtime/CL/functions/CLPriorBoxLayer.cpp | 2 + src/runtime/CL/functions/CLQLSTMLayer.cpp | 1 + src/runtime/CL/functions/CLReduceMean.cpp | 3 +- src/runtime/CL/functions/CLReductionOperation.cpp | 8 ++- src/runtime/CL/functions/CLRemap.cpp | 2 +- src/runtime/CL/functions/CLSelect.cpp | 2 + src/runtime/CL/functions/CLSoftmaxLayer.cpp | 7 ++- src/runtime/CL/functions/CLSplit.cpp | 1 + .../CL/functions/CLWinogradConvolutionLayer.cpp | 2 +- src/runtime/CL/gemm/CLGEMMKernelSelection.h | 65 ++++++++++++++++++++++ .../CL/gemm/CLGEMMKernelSelectionBifrost.cpp | 4 +- src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h | 55 ++++++++++++++++++ .../CL/gemm/CLGEMMKernelSelectionMidgard.cpp | 4 +- src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.h | 53 ++++++++++++++++++ .../CL/gemm/CLGEMMKernelSelectionValhall.cpp | 4 +- src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.h | 53 ++++++++++++++++++ src/runtime/CL/tuners/BifrostTuner.cpp | 2 +- src/runtime/CL/tuners/MidgardTuner.cpp | 2 +- 40 files changed, 308 insertions(+), 40 deletions(-) create mode 100644 src/runtime/CL/gemm/CLGEMMKernelSelection.h create mode 100644 src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h create mode 100644 src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.h create mode 100644 src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.h (limited to 'src/runtime/CL') diff --git a/src/runtime/CL/CLHelpers.cpp b/src/runtime/CL/CLHelpers.cpp index adfdc3c917..5f1842f76d 100644 --- a/src/runtime/CL/CLHelpers.cpp +++ b/src/runtime/CL/CLHelpers.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2019 Arm Limited. + * Copyright (c) 2019-2020 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -26,6 +26,7 @@ #include "arm_compute/core/CL/CLHelpers.h" #include "arm_compute/core/Error.h" +#include "arm_compute/core/Validate.h" #include "arm_compute/runtime/CL/CLRuntimeContext.h" namespace diff --git a/src/runtime/CL/CLMemory.cpp b/src/runtime/CL/CLMemory.cpp index efbc68f50e..a1743c56e6 100644 --- a/src/runtime/CL/CLMemory.cpp +++ b/src/runtime/CL/CLMemory.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2018-2019 Arm Limited. + * Copyright (c) 2018-2020 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -24,7 +24,7 @@ #include "arm_compute/runtime/CL/CLMemory.h" #include "arm_compute/core/Error.h" -#include "arm_compute/core/utils/misc/Cast.h" +#include "support/Cast.h" namespace arm_compute { diff --git a/src/runtime/CL/CLRuntimeContext.cpp b/src/runtime/CL/CLRuntimeContext.cpp index 2fc7f93adf..571e30931c 100644 --- a/src/runtime/CL/CLRuntimeContext.cpp +++ b/src/runtime/CL/CLRuntimeContext.cpp @@ -26,6 +26,8 @@ #include "arm_compute/runtime/CL/CLHelpers.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "support/MemorySupport.h" + namespace arm_compute { CLRuntimeContext::CLRuntimeContext() diff --git a/src/runtime/CL/CLTensorAllocator.cpp b/src/runtime/CL/CLTensorAllocator.cpp index 90d77883f6..f37fc779fe 100644 --- a/src/runtime/CL/CLTensorAllocator.cpp +++ b/src/runtime/CL/CLTensorAllocator.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2016-2019 Arm Limited. + * Copyright (c) 2016-2020 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -28,6 +28,8 @@ #include "arm_compute/runtime/CL/CLRuntimeContext.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "support/MemorySupport.h" + namespace arm_compute { const cl::Buffer CLTensorAllocator::_empty_buffer = cl::Buffer(); diff --git a/src/runtime/CL/functions/CLArgMinMaxLayer.cpp b/src/runtime/CL/functions/CLArgMinMaxLayer.cpp index ad6e7ba97b..57c4f685f6 100644 --- a/src/runtime/CL/functions/CLArgMinMaxLayer.cpp +++ b/src/runtime/CL/functions/CLArgMinMaxLayer.cpp @@ -24,13 +24,14 @@ #include "arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h" -#include "arm_compute/core/CL/CLValidate.h" #include "arm_compute/core/Error.h" #include "arm_compute/core/TensorInfo.h" #include "arm_compute/core/Types.h" #include "arm_compute/core/Validate.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" -#include "arm_compute/runtime/Utils.h" +#include "src/core/CL/CLValidate.h" +#include "src/core/helpers/AutoConfiguration.h" +#include "src/runtime/Utils.h" namespace arm_compute { @@ -47,7 +48,7 @@ Status CLArgMinMaxLayer::validate(const ITensorInfo *input, int axis, const ITen ARM_COMPUTE_RETURN_ERROR_ON_MSG(op != ReductionOperation::ARG_IDX_MAX && op != ReductionOperation::ARG_IDX_MIN, "Invalid reduction operation"); ARM_COMPUTE_RETURN_ERROR_ON_MSG(axis >= static_cast(TensorShape::num_max_dimensions), "Reduction axis greater than max number of dimensions"); ARM_COMPUTE_RETURN_ERROR_ON_MSG(axis > 3, "Unsupported reduction axis"); - const unsigned int num_of_stages = calculate_number_of_stages_only_x_axis(input->dimension(0), axis); + const unsigned int num_of_stages = utils::calculate_number_of_stages_only_x_axis(input->dimension(0), axis); DataType output_data_type = DataType::S32; TensorInfo not_reshaped_output; @@ -115,7 +116,7 @@ void CLArgMinMaxLayer::configure(const ICLTensor *input, int axis, ICLTensor *ou void CLArgMinMaxLayer::configure(const CLCompileContext &compile_context, const ICLTensor *input, int axis, ICLTensor *output, const ReductionOperation &op) { ARM_COMPUTE_ERROR_ON_NULLPTR(input, output); - _num_of_stages = calculate_number_of_stages_only_x_axis(input->info()->dimension(0), axis); + _num_of_stages = utils::calculate_number_of_stages_only_x_axis(input->info()->dimension(0), axis); _reduction_axis = axis; const TensorShape output_shape = arm_compute::misc::shape_calculator::compute_reduced_shape(input->info()->tensor_shape(), axis, false); @@ -172,4 +173,4 @@ void CLArgMinMaxLayer::run() } _reshape.run(); } -} // namespace arm_compute \ No newline at end of file +} // namespace arm_compute diff --git a/src/runtime/CL/functions/CLConcatenateLayer.cpp b/src/runtime/CL/functions/CLConcatenateLayer.cpp index 4214813446..2eb310b893 100644 --- a/src/runtime/CL/functions/CLConcatenateLayer.cpp +++ b/src/runtime/CL/functions/CLConcatenateLayer.cpp @@ -36,6 +36,7 @@ #include "arm_compute/core/Error.h" #include "arm_compute/core/TensorInfo.h" #include "arm_compute/core/Types.h" +#include "src/core/helpers/AutoConfiguration.h" #include "support/MemorySupport.h" namespace arm_compute diff --git a/src/runtime/CL/functions/CLConvertFullyConnectedWeights.cpp b/src/runtime/CL/functions/CLConvertFullyConnectedWeights.cpp index 4c787673b5..b291ae5b88 100644 --- a/src/runtime/CL/functions/CLConvertFullyConnectedWeights.cpp +++ b/src/runtime/CL/functions/CLConvertFullyConnectedWeights.cpp @@ -23,6 +23,8 @@ */ #include "arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h" +#include "support/MemorySupport.h" + namespace arm_compute { void CLConvertFullyConnectedWeights::configure(const ICLTensor *input, ICLTensor *output, const TensorShape &original_input_shape, diff --git a/src/runtime/CL/functions/CLConvolutionLayer.cpp b/src/runtime/CL/functions/CLConvolutionLayer.cpp index 630352e4e6..85355f0f17 100644 --- a/src/runtime/CL/functions/CLConvolutionLayer.cpp +++ b/src/runtime/CL/functions/CLConvolutionLayer.cpp @@ -30,6 +30,8 @@ #include "arm_compute/core/utils/quantization/AsymmHelpers.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "support/MemorySupport.h" + #include #include #include diff --git a/src/runtime/CL/functions/CLCropResize.cpp b/src/runtime/CL/functions/CLCropResize.cpp index 529f7bfb3e..6167e9de0a 100644 --- a/src/runtime/CL/functions/CLCropResize.cpp +++ b/src/runtime/CL/functions/CLCropResize.cpp @@ -25,6 +25,10 @@ #include "arm_compute/core/CL/CLHelpers.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "src/core/helpers/AutoConfiguration.h" +#include "src/core/helpers/WindowHelpers.h" + +#include "support/MemorySupport.h" #include diff --git a/src/runtime/CL/functions/CLDeconvolutionLayer.cpp b/src/runtime/CL/functions/CLDeconvolutionLayer.cpp index cd55336d9a..e6717b6d01 100644 --- a/src/runtime/CL/functions/CLDeconvolutionLayer.cpp +++ b/src/runtime/CL/functions/CLDeconvolutionLayer.cpp @@ -29,6 +29,8 @@ #include "arm_compute/core/utils/quantization/AsymmHelpers.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "support/MemorySupport.h" + #include #include #include diff --git a/src/runtime/CL/functions/CLDirectConvolutionLayer.cpp b/src/runtime/CL/functions/CLDirectConvolutionLayer.cpp index c1055dda36..07e7a18941 100644 --- a/src/runtime/CL/functions/CLDirectConvolutionLayer.cpp +++ b/src/runtime/CL/functions/CLDirectConvolutionLayer.cpp @@ -43,7 +43,7 @@ void CLDirectConvolutionLayer::configure(ICLTensor *input, const ICLTensor *weig } void CLDirectConvolutionLayer::configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, - const PadStrideInfo &conv_info, + const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info) { // Set GPU target diff --git a/src/runtime/CL/functions/CLDirectDeconvolutionLayer.cpp b/src/runtime/CL/functions/CLDirectDeconvolutionLayer.cpp index 3515c25d82..0ffafa0221 100644 --- a/src/runtime/CL/functions/CLDirectDeconvolutionLayer.cpp +++ b/src/runtime/CL/functions/CLDirectDeconvolutionLayer.cpp @@ -28,6 +28,7 @@ #include "arm_compute/core/Validate.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "src/core/helpers/AutoConfiguration.h" #include #include diff --git a/src/runtime/CL/functions/CLFFT1D.cpp b/src/runtime/CL/functions/CLFFT1D.cpp index 7d15d33ab5..1269cba90d 100644 --- a/src/runtime/CL/functions/CLFFT1D.cpp +++ b/src/runtime/CL/functions/CLFFT1D.cpp @@ -25,8 +25,8 @@ #include "arm_compute/core/CL/ICLTensor.h" #include "arm_compute/core/Validate.h" -#include "arm_compute/core/utils/helpers/fft.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "src/core/utils/helpers/fft.h" namespace arm_compute { diff --git a/src/runtime/CL/functions/CLFFTConvolutionLayer.cpp b/src/runtime/CL/functions/CLFFTConvolutionLayer.cpp index 1def674bb6..4d0eab81ee 100644 --- a/src/runtime/CL/functions/CLFFTConvolutionLayer.cpp +++ b/src/runtime/CL/functions/CLFFTConvolutionLayer.cpp @@ -26,10 +26,13 @@ #include "arm_compute/core/CL/ICLTensor.h" #include "arm_compute/core/Utils.h" #include "arm_compute/core/Validate.h" -#include "arm_compute/core/utils/helpers/fft.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/runtime/CL/CLScheduler.h" #include "arm_compute/runtime/CPP/CPPScheduler.h" +#include "src/core/helpers/AutoConfiguration.h" +#include "src/core/utils/helpers/fft.h" + +#include "support/MemorySupport.h" namespace arm_compute { diff --git a/src/runtime/CL/functions/CLFill.cpp b/src/runtime/CL/functions/CLFill.cpp index 6c0f1786f0..a89383ec31 100644 --- a/src/runtime/CL/functions/CLFill.cpp +++ b/src/runtime/CL/functions/CLFill.cpp @@ -26,6 +26,8 @@ #include "arm_compute/core/CL/kernels/CLMemsetKernel.h" #include "arm_compute/core/Types.h" +#include "support/MemorySupport.h" + #include namespace arm_compute diff --git a/src/runtime/CL/functions/CLFullyConnectedLayer.cpp b/src/runtime/CL/functions/CLFullyConnectedLayer.cpp index 4f365b6a61..75e87c382b 100644 --- a/src/runtime/CL/functions/CLFullyConnectedLayer.cpp +++ b/src/runtime/CL/functions/CLFullyConnectedLayer.cpp @@ -25,10 +25,10 @@ #include "arm_compute/core/Size2D.h" #include "arm_compute/core/Validate.h" -#include "arm_compute/core/utils/misc/Cast.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/core/utils/quantization/AsymmHelpers.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "support/Cast.h" #include "support/MemorySupport.h" #include diff --git a/src/runtime/CL/functions/CLGEMM.cpp b/src/runtime/CL/functions/CLGEMM.cpp index d56b341abf..ccae6713a6 100644 --- a/src/runtime/CL/functions/CLGEMM.cpp +++ b/src/runtime/CL/functions/CLGEMM.cpp @@ -23,10 +23,7 @@ */ #include "arm_compute/runtime/CL/functions/CLGEMM.h" -#include "arm_compute/core/CL/ICLGEMMKernelConfiguration.h" #include "arm_compute/core/CL/ICLTensor.h" -#include "arm_compute/core/CL/gemm/reshaped/CLGEMMReshapedKernelConfiguration.h" -#include "arm_compute/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfiguration.h" #include "arm_compute/core/Error.h" #include "arm_compute/core/GPUTarget.h" #include "arm_compute/core/Helpers.h" @@ -35,12 +32,18 @@ #include "arm_compute/core/Types.h" #include "arm_compute/core/Utils.h" #include "arm_compute/core/Validate.h" -#include "arm_compute/core/utils/helpers/float_ops.h" -#include "arm_compute/core/utils/misc/Cast.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/runtime/CL/CLScheduler.h" -#include "arm_compute/runtime/CL/gemm/CLGEMMKernelSelection.h" #include "arm_compute/runtime/ITensorAllocator.h" +#include "src/core/CL/ICLGEMMKernelConfiguration.h" +#include "src/core/CL/gemm/reshaped/CLGEMMReshapedKernelConfiguration.h" +#include "src/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfiguration.h" +#include "src/core/helpers/AutoConfiguration.h" +#include "src/core/utils/helpers/float_ops.h" +#include "src/runtime/CL/gemm/CLGEMMKernelSelection.h" +#include "support/Cast.h" + +#include "support/MemorySupport.h" namespace arm_compute { diff --git a/src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp b/src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp index ee90b39c2b..e871b39805 100644 --- a/src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp +++ b/src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp @@ -27,10 +27,11 @@ #include "arm_compute/core/Size2D.h" #include "arm_compute/core/Utils.h" #include "arm_compute/core/Validate.h" -#include "arm_compute/core/utils/misc/Cast.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/core/utils/quantization/AsymmHelpers.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "src/core/helpers/AutoConfiguration.h" +#include "support/Cast.h" #include #include diff --git a/src/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.cpp b/src/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.cpp index 30dce5b8fe..7a8de6c1f5 100644 --- a/src/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.cpp +++ b/src/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.cpp @@ -24,8 +24,6 @@ #include "arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h" #include "arm_compute/core/CL/ICLTensor.h" -#include "arm_compute/core/CL/gemm/native/CLGEMMNativeKernelConfiguration.h" -#include "arm_compute/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfiguration.h" #include "arm_compute/core/Error.h" #include "arm_compute/core/Helpers.h" #include "arm_compute/core/KernelDescriptors.h" @@ -35,7 +33,10 @@ #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/core/utils/quantization/AsymmHelpers.h" #include "arm_compute/runtime/CL/CLScheduler.h" -#include "arm_compute/runtime/CL/gemm/CLGEMMKernelSelection.h" +#include "src/core/CL/gemm/native/CLGEMMNativeKernelConfiguration.h" +#include "src/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfiguration.h" +#include "src/core/helpers/AutoConfiguration.h" +#include "src/runtime/CL/gemm/CLGEMMKernelSelection.h" namespace arm_compute { diff --git a/src/runtime/CL/functions/CLGenerateProposalsLayer.cpp b/src/runtime/CL/functions/CLGenerateProposalsLayer.cpp index 45dc402449..5291de074a 100644 --- a/src/runtime/CL/functions/CLGenerateProposalsLayer.cpp +++ b/src/runtime/CL/functions/CLGenerateProposalsLayer.cpp @@ -25,6 +25,7 @@ #include "arm_compute/core/CL/ICLTensor.h" #include "arm_compute/core/Types.h" +#include "src/core/helpers/AutoConfiguration.h" namespace arm_compute { diff --git a/src/runtime/CL/functions/CLInstanceNormalizationLayer.cpp b/src/runtime/CL/functions/CLInstanceNormalizationLayer.cpp index fce1fe43a2..4a60ee9d08 100644 --- a/src/runtime/CL/functions/CLInstanceNormalizationLayer.cpp +++ b/src/runtime/CL/functions/CLInstanceNormalizationLayer.cpp @@ -26,6 +26,8 @@ #include "arm_compute/core/CL/kernels/CLInstanceNormalizationLayerKernel.h" #include "arm_compute/core/Types.h" +#include "support/MemorySupport.h" + namespace arm_compute { CLInstanceNormalizationLayer::CLInstanceNormalizationLayer() diff --git a/src/runtime/CL/functions/CLLSTMLayerQuantized.cpp b/src/runtime/CL/functions/CLLSTMLayerQuantized.cpp index e30b1dbb86..76a531b1c9 100644 --- a/src/runtime/CL/functions/CLLSTMLayerQuantized.cpp +++ b/src/runtime/CL/functions/CLLSTMLayerQuantized.cpp @@ -27,6 +27,7 @@ #include "arm_compute/core/Utils.h" #include "arm_compute/core/Validate.h" #include "arm_compute/core/utils/quantization/AsymmHelpers.h" +#include "src/core/helpers/AutoConfiguration.h" #include diff --git a/src/runtime/CL/functions/CLPriorBoxLayer.cpp b/src/runtime/CL/functions/CLPriorBoxLayer.cpp index 1907c7cc08..fefbff639d 100644 --- a/src/runtime/CL/functions/CLPriorBoxLayer.cpp +++ b/src/runtime/CL/functions/CLPriorBoxLayer.cpp @@ -31,6 +31,8 @@ #include "arm_compute/core/Validate.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "support/MemorySupport.h" + using namespace arm_compute; CLPriorBoxLayer::CLPriorBoxLayer() diff --git a/src/runtime/CL/functions/CLQLSTMLayer.cpp b/src/runtime/CL/functions/CLQLSTMLayer.cpp index 15a54c7928..c493471667 100644 --- a/src/runtime/CL/functions/CLQLSTMLayer.cpp +++ b/src/runtime/CL/functions/CLQLSTMLayer.cpp @@ -30,6 +30,7 @@ #include "arm_compute/core/utils/misc/InfoHelpers.h" #include "arm_compute/core/utils/quantization/AsymmHelpers.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "src/core/helpers/WindowHelpers.h" namespace arm_compute { diff --git a/src/runtime/CL/functions/CLReduceMean.cpp b/src/runtime/CL/functions/CLReduceMean.cpp index 0e2ede7167..4ea7f7642f 100644 --- a/src/runtime/CL/functions/CLReduceMean.cpp +++ b/src/runtime/CL/functions/CLReduceMean.cpp @@ -23,12 +23,13 @@ */ #include "arm_compute/runtime/CL/functions/CLReduceMean.h" -#include "arm_compute/core/CL/CLValidate.h" #include "arm_compute/core/CL/ICLTensor.h" #include "arm_compute/core/CL/kernels/CLReductionOperationKernel.h" #include "arm_compute/core/Error.h" #include "arm_compute/core/Types.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" +#include "src/core/CL/CLValidate.h" +#include "src/core/helpers/AutoConfiguration.h" namespace arm_compute { diff --git a/src/runtime/CL/functions/CLReductionOperation.cpp b/src/runtime/CL/functions/CLReductionOperation.cpp index 54e91fb8d8..208371c45d 100644 --- a/src/runtime/CL/functions/CLReductionOperation.cpp +++ b/src/runtime/CL/functions/CLReductionOperation.cpp @@ -30,7 +30,9 @@ #include "arm_compute/core/Validate.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/runtime/CL/CLScheduler.h" -#include "arm_compute/runtime/Utils.h" +#include "src/core/helpers/AutoConfiguration.h" +#include "src/runtime/Utils.h" + #include "support/MemorySupport.h" namespace arm_compute @@ -47,7 +49,7 @@ Status CLReductionOperation::validate(const ITensorInfo *input, const ITensorInf ARM_COMPUTE_RETURN_ERROR_ON_MSG(axis >= TensorShape::num_max_dimensions, "Reduction axis greater than max number of dimensions"); ARM_COMPUTE_RETURN_ERROR_ON_MSG(axis > 3, "Unsupported reduction axis"); - const unsigned int num_of_stages = calculate_number_of_stages_only_x_axis(input->dimension(0), axis); + const unsigned int num_of_stages = utils::calculate_number_of_stages_only_x_axis(input->dimension(0), axis); const bool is_serial = needs_serialized_reduction(op, input->data_type(), axis); const bool is_reshape_required = !keep_dims; @@ -194,7 +196,7 @@ void CLReductionOperation::configure(ICLTensor *input, ICLTensor *output, unsign void CLReductionOperation::configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, unsigned int axis, ReductionOperation op, bool keep_dims) { ARM_COMPUTE_ERROR_ON_NULLPTR(input, output); - _num_of_stages = calculate_number_of_stages_only_x_axis(input->info()->dimension(0), axis); + _num_of_stages = utils::calculate_number_of_stages_only_x_axis(input->info()->dimension(0), axis); _reduction_axis = axis; _is_serial = needs_serialized_reduction(op, input->info()->data_type(), axis); _is_reshape_required = !keep_dims; diff --git a/src/runtime/CL/functions/CLRemap.cpp b/src/runtime/CL/functions/CLRemap.cpp index 60b72c5f87..1e3d614402 100644 --- a/src/runtime/CL/functions/CLRemap.cpp +++ b/src/runtime/CL/functions/CLRemap.cpp @@ -42,7 +42,7 @@ void CLRemap::configure(ICLTensor *input, const ICLTensor *map_x, const ICLTenso void CLRemap::configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *map_x, const ICLTensor *map_y, ICLTensor *output, InterpolationPolicy policy, BorderMode border_mode, - uint8_t constant_border_value) + uint8_t constant_border_value) { ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::U8); ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::U8); diff --git a/src/runtime/CL/functions/CLSelect.cpp b/src/runtime/CL/functions/CLSelect.cpp index c7d7df75d2..ef8010847b 100644 --- a/src/runtime/CL/functions/CLSelect.cpp +++ b/src/runtime/CL/functions/CLSelect.cpp @@ -27,6 +27,8 @@ #include "arm_compute/core/Types.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "support/MemorySupport.h" + using namespace arm_compute; namespace arm_compute diff --git a/src/runtime/CL/functions/CLSoftmaxLayer.cpp b/src/runtime/CL/functions/CLSoftmaxLayer.cpp index 720f9111a5..759c8706a1 100644 --- a/src/runtime/CL/functions/CLSoftmaxLayer.cpp +++ b/src/runtime/CL/functions/CLSoftmaxLayer.cpp @@ -31,6 +31,7 @@ #include "arm_compute/core/Utils.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "src/core/helpers/SoftmaxHelpers.h" namespace arm_compute { @@ -63,7 +64,7 @@ void CLSoftmaxLayerGeneric::configure(const CLCompileContext &compile_co { _memory_group.manage(&_input_permuted); _memory_group.manage(&_output_permuted); - _permute_input.configure(compile_context, input, &_input_permuted, get_permutation_vector_from_softmax_axis(actual_axis)); + _permute_input.configure(compile_context, input, &_input_permuted, softmax_helpers::get_permutation_vector_from_softmax_axis(actual_axis)); tmp_output = &_output_permuted; } @@ -99,7 +100,7 @@ void CLSoftmaxLayerGeneric::configure(const CLCompileContext &compile_co _sum.allocator()->allocate(); if(_needs_permute) { - _permute_output.configure(compile_context, &_output_permuted, output, get_permutation_vector_from_softmax_axis(actual_axis)); + _permute_output.configure(compile_context, &_output_permuted, output, softmax_helpers::get_permutation_vector_from_softmax_axis(actual_axis)); _input_permuted.allocator()->allocate(); _output_permuted.allocator()->allocate(); } @@ -117,7 +118,7 @@ Status CLSoftmaxLayerGeneric::validate(const ITensorInfo *input, const I const bool needs_permute = actual_axis != 0; if(needs_permute) { - const PermutationVector permutation_vector = get_permutation_vector_from_softmax_axis(actual_axis); + const PermutationVector permutation_vector = softmax_helpers::get_permutation_vector_from_softmax_axis(actual_axis); const TensorShape permuted_shape = misc::shape_calculator::compute_permutation_output_shape(*input, permutation_vector); TensorInfo input_permuted(input->clone()->set_tensor_shape(permuted_shape)); ARM_COMPUTE_RETURN_ON_ERROR(CLPermute::validate(input, &input_permuted, permutation_vector)); diff --git a/src/runtime/CL/functions/CLSplit.cpp b/src/runtime/CL/functions/CLSplit.cpp index db0b14b9a2..0b27371e3f 100644 --- a/src/runtime/CL/functions/CLSplit.cpp +++ b/src/runtime/CL/functions/CLSplit.cpp @@ -30,6 +30,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/core/utils/misc/ShapeCalculator.h" #include "arm_compute/runtime/CL/CLScheduler.h" +#include "src/core/helpers/AutoConfiguration.h" namespace arm_compute { diff --git a/src/runtime/CL/functions/CLWinogradConvolutionLayer.cpp b/src/runtime/CL/functions/CLWinogradConvolutionLayer.cpp index 09a35a6f27..7ad017f918 100644 --- a/src/runtime/CL/functions/CLWinogradConvolutionLayer.cpp +++ b/src/runtime/CL/functions/CLWinogradConvolutionLayer.cpp @@ -102,7 +102,7 @@ void CLWinogradConvolutionLayer::configure(ICLTensor *input, const ICLTensor *we } void CLWinogradConvolutionLayer::configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, - const PadStrideInfo &conv_info, + const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info, bool enable_fast_math) { // Get indices for the width and height diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelection.h b/src/runtime/CL/gemm/CLGEMMKernelSelection.h new file mode 100644 index 0000000000..f6fad7e4ff --- /dev/null +++ b/src/runtime/CL/gemm/CLGEMMKernelSelection.h @@ -0,0 +1,65 @@ +/* + * Copyright (c) 2020 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef SRC_CLGEMMKERNELSELECTION_H +#define SRC_CLGEMMKERNELSELECTION_H + +#include "arm_compute/runtime/CL/ICLGEMMKernelSelection.h" +#include "src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h" +#include "src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.h" +#include "src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.h" + +#include "support/MemorySupport.h" + +namespace arm_compute +{ +namespace cl_gemm +{ +/** CLGEMMKernelSelection factory class */ +class CLGEMMKernelSelectionFactory final +{ +public: + /** Static method to select the GEMM kernel accordingly with the GPU target and GEMM's dimensionality + * + * @param[in] gpu GPU target + * + * @return CLGEMMKernelSelection class + */ + static std::unique_ptr create(GPUTarget gpu) + { + switch(get_arch_from_target(gpu)) + { + case GPUTarget::MIDGARD: + return support::cpp14::make_unique(gpu); + case GPUTarget::BIFROST: + return support::cpp14::make_unique(gpu); + case GPUTarget::VALHALL: + return support::cpp14::make_unique(gpu); + default: + ARM_COMPUTE_ERROR("Not supported GPU target"); + } + } +}; +} // namespace cl_gemm +} // namespace arm_compute +#endif /* SRC_CLGEMMKERNELSELECTION_H */ diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.cpp b/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.cpp index b1dd690ca5..73b90568f5 100644 --- a/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.cpp +++ b/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.cpp @@ -21,11 +21,11 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#include "arm_compute/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h" +#include "src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h" #include "arm_compute/core/CL/CLHelpers.h" #include "arm_compute/core/CL/CLKernelLibrary.h" -#include "arm_compute/core/CL/gemm/CLGEMMHelpers.h" +#include "src/core/CL/gemm/CLGEMMHelpers.h" #include #include diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h b/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h new file mode 100644 index 0000000000..a495b48301 --- /dev/null +++ b/src/runtime/CL/gemm/CLGEMMKernelSelectionBifrost.h @@ -0,0 +1,55 @@ +/* + * Copyright (c) 2020 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef SRC_CLGEMMKERNELSELECTIONBIFROST_H +#define SRC_CLGEMMKERNELSELECTIONBIFROST_H + +#include "arm_compute/runtime/CL/ICLGEMMKernelSelection.h" + +namespace arm_compute +{ +namespace cl_gemm +{ +/** Bifrost based OpenCL GEMMKernel selection */ +class CLGEMMKernelSelectionBifrost final : public ICLGEMMKernelSelection +{ +public: + /** Constructor + * + * @param[in] gpu GPU target + */ + CLGEMMKernelSelectionBifrost(GPUTarget gpu); + + // Inherited overridden method + CLGEMMKernelType select_kernel(const CLGEMMKernelSelectionParams ¶ms) override; + +private: + CLGEMMKernelType g76_f32(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); + CLGEMMKernelType g71_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); + CLGEMMKernelType default_f32(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); + CLGEMMKernelType default_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); + CLGEMMKernelType default_q8(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); +}; +} // namespace cl_gemm +} // namespace arm_compute +#endif /* SRC_CLGEMMKERNELSELECTIONBIFROST_H */ diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.cpp b/src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.cpp index 324c2f7dca..d172a827b5 100644 --- a/src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.cpp +++ b/src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.cpp @@ -21,12 +21,12 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#include "arm_compute/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.h" +#include "src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.h" #include "arm_compute/core/CL/CLHelpers.h" #include "arm_compute/core/CL/CLKernelLibrary.h" -#include "arm_compute/core/CL/gemm/CLGEMMHelpers.h" #include "arm_compute/core/GPUTarget.h" +#include "src/core/CL/gemm/CLGEMMHelpers.h" #include #include diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.h b/src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.h new file mode 100644 index 0000000000..3f6003f7dc --- /dev/null +++ b/src/runtime/CL/gemm/CLGEMMKernelSelectionMidgard.h @@ -0,0 +1,53 @@ +/* + * Copyright (c) 2020 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef SRC_CLGEMMKERNELSELECTIONMIDGARD_H +#define SRC_CLGEMMKERNELSELECTIONMIDGARD_H + +#include "arm_compute/runtime/CL/ICLGEMMKernelSelection.h" + +namespace arm_compute +{ +namespace cl_gemm +{ +/** Midgard based OpenCL GEMMKernel selection */ +class CLGEMMKernelSelectionMidgard final : public ICLGEMMKernelSelection +{ +public: + /** Constructor + * + * @param[in] gpu GPU target + */ + CLGEMMKernelSelectionMidgard(GPUTarget gpu); + + // Inherited overridden method + CLGEMMKernelType select_kernel(const CLGEMMKernelSelectionParams ¶ms) override; + +private: + CLGEMMKernelType default_f32(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); + CLGEMMKernelType default_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); + CLGEMMKernelType default_q8(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); +}; +} // namespace cl_gemm +} // namespace arm_compute +#endif /* SRC_CLGEMMKERNELSELECTIONMIDGARD_H */ diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.cpp b/src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.cpp index c50c7ae76b..acae0e7565 100644 --- a/src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.cpp +++ b/src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.cpp @@ -21,11 +21,11 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#include "arm_compute/runtime/CL/gemm/CLGEMMKernelSelectionValhall.h" +#include "src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.h" #include "arm_compute/core/CL/CLHelpers.h" #include "arm_compute/core/CL/CLKernelLibrary.h" -#include "arm_compute/core/CL/gemm/CLGEMMHelpers.h" +#include "src/core/CL/gemm/CLGEMMHelpers.h" #include #include diff --git a/src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.h b/src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.h new file mode 100644 index 0000000000..cbea9ea548 --- /dev/null +++ b/src/runtime/CL/gemm/CLGEMMKernelSelectionValhall.h @@ -0,0 +1,53 @@ +/* + * Copyright (c) 2020 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef SRC_CLGEMMKERNELSELECTIONVALHALL_H +#define SRC_CLGEMMKERNELSELECTIONVALHALL_H + +#include "arm_compute/runtime/CL/ICLGEMMKernelSelection.h" + +namespace arm_compute +{ +namespace cl_gemm +{ +/** Valhall based OpenCL GEMMKernel selection */ +class CLGEMMKernelSelectionValhall final : public ICLGEMMKernelSelection +{ +public: + /** Constructor + * + * @param[in] gpu GPU target + */ + CLGEMMKernelSelectionValhall(GPUTarget gpu); + + // Inherited overridden method + CLGEMMKernelType select_kernel(const CLGEMMKernelSelectionParams ¶ms) override; + +private: + CLGEMMKernelType default_f32(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); + CLGEMMKernelType default_f16(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); + CLGEMMKernelType default_q8(unsigned int m, unsigned int n, unsigned int k, unsigned int b, bool is_rhs_constant); +}; +} // namespace cl_gemm +} // namespace arm_compute +#endif /* SRC_CLGEMMKERNELSELECTIONVALHALL_H */ diff --git a/src/runtime/CL/tuners/BifrostTuner.cpp b/src/runtime/CL/tuners/BifrostTuner.cpp index 52644bf192..a6474c9835 100644 --- a/src/runtime/CL/tuners/BifrostTuner.cpp +++ b/src/runtime/CL/tuners/BifrostTuner.cpp @@ -25,7 +25,7 @@ #include "arm_compute/core/CL/CLHelpers.h" #include "arm_compute/core/CL/CLKernels.h" -#include "arm_compute/core/utils/misc/Cast.h" +#include "support/Cast.h" namespace arm_compute { diff --git a/src/runtime/CL/tuners/MidgardTuner.cpp b/src/runtime/CL/tuners/MidgardTuner.cpp index e49e15508b..58b0d579d2 100644 --- a/src/runtime/CL/tuners/MidgardTuner.cpp +++ b/src/runtime/CL/tuners/MidgardTuner.cpp @@ -25,7 +25,7 @@ #include "arm_compute/core/CL/CLHelpers.h" #include "arm_compute/core/CL/CLKernels.h" -#include "arm_compute/core/utils/misc/Cast.h" +#include "support/Cast.h" namespace arm_compute { -- cgit v1.2.1