diff options
Diffstat (limited to 'filelist.json')
| -rw-r--r-- | filelist.json | 2670 |
1 files changed, 2389 insertions, 281 deletions
diff --git a/filelist.json b/filelist.json index c375b1daef..2c3621cd8b 100644 --- a/filelist.json +++ b/filelist.json @@ -1,292 +1,2400 @@ { - "common" : [ - "src/common/cpuinfo/target/CpuInfoSveUtils.cpp", - "src/common/cpuinfo/CpuInfo.cpp", - "src/common/cpuinfo/CpuModel.cpp", - "src/common/cpuinfo/CpuIsaInfo.cpp", - "src/common/utils/LegacySupport.cpp", - "src/common/AllocatorWrapper.cpp", - "src/common/ITensorV2.cpp", - "src/common/TensorPack.cpp" + "common": [ + "src/common/cpuinfo/CpuInfo.cpp", + "src/common/cpuinfo/CpuModel.cpp", + "src/common/cpuinfo/CpuIsaInfo.cpp", + "src/common/utils/LegacySupport.cpp", + "src/common/AllocatorWrapper.cpp", + "src/common/ITensorV2.cpp", + "src/common/TensorPack.cpp", + "src/common/IOperator.cpp", + "src/core/AccessWindowAutoPadding.cpp", + "src/core/AccessWindowStatic.cpp", + "src/core/AccessWindowTranspose.cpp", + "src/core/Error.cpp", + "src/core/GPUTarget.cpp", + "src/core/Helpers.cpp", + "src/core/helpers/LUTManager.cpp", + "src/core/IAccessWindow.cpp", + "src/core/IKernel.cpp", + "src/core/ITensor.cpp", + "src/core/ITensorPack.cpp", + "src/core/Rounding.cpp", + "src/core/Size2D.cpp", + "src/core/SubTensorInfo.cpp", + "src/core/TensorInfo.cpp", + "src/core/Utils.cpp", + "src/core/Validate.cpp", + "src/core/Version.cpp", + "src/core/helpers/SoftmaxHelpers.cpp", + "src/core/helpers/Utils.cpp", + "src/core/helpers/WindowHelpers.cpp", + "src/core/utils/ActivationFunctionUtils.cpp", + "src/core/utils/AssemblyUtils.cpp", + "src/core/utils/DataTypeUtils.cpp", + "src/core/utils/DataLayoutUtils.cpp", + "src/core/utils/InterpolationPolicyUtils.cpp", + "src/core/utils/FormatUtils.cpp", + "src/core/utils/Math.cpp", + "src/core/utils/ScaleUtils.cpp", + "src/core/utils/StringUtils.cpp", + "src/core/utils/helpers/fft.cpp", + "src/core/utils/helpers/tensor_transform.cpp", + "src/core/utils/io/FileHandler.cpp", + "src/core/utils/misc/MMappedFile.cpp", + "src/core/utils/quantization/AsymmHelpers.cpp", + "src/core/CPP/CPPTypes.cpp", + "src/core/CPP/kernels/CPPBoxWithNonMaximaSuppressionLimitKernel.cpp", + "src/core/CPP/kernels/CPPNonMaximumSuppressionKernel.cpp", + "src/core/CPP/kernels/CPPPermuteKernel.cpp", + "src/core/CPP/kernels/CPPTopKVKernel.cpp", + "src/core/CPP/kernels/CPPUpsampleKernel.cpp", + "src/runtime/Allocator.cpp", + "src/runtime/BlobLifetimeManager.cpp", + "src/runtime/BlobMemoryPool.cpp", + "src/runtime/ISimpleLifetimeManager.cpp", + "src/runtime/ITensorAllocator.cpp", + "src/runtime/IWeightsManager.cpp", + "src/runtime/IScheduler.cpp", + "src/runtime/Memory.cpp", + "src/runtime/MemoryManagerOnDemand.cpp", + "src/runtime/OffsetLifetimeManager.cpp", + "src/runtime/OffsetMemoryPool.cpp", + "src/runtime/OperatorTensor.cpp", + "src/runtime/PoolManager.cpp", + "src/runtime/RuntimeContext.cpp", + "src/runtime/Scheduler.cpp", + "src/runtime/SchedulerFactory.cpp", + "src/runtime/SchedulerUtils.cpp", + "src/runtime/SubTensor.cpp", + "src/runtime/Tensor.cpp", + "src/runtime/TensorAllocator.cpp", + "src/runtime/Utils.cpp", + "src/runtime/CPP/ICPPSimpleFunction.cpp", + "src/runtime/CPP/functions/CPPBoxWithNonMaximaSuppressionLimit.cpp", + "src/runtime/CPP/functions/CPPDetectionOutputLayer.cpp", + "src/runtime/CPP/functions/CPPDetectionPostProcessLayer.cpp", + "src/runtime/CPP/functions/CPPNonMaximumSuppression.cpp", + "src/runtime/CPP/functions/CPPPermute.cpp", + "src/runtime/CPP/functions/CPPTopKV.cpp", + "src/runtime/CPP/functions/CPPUpsample.cpp" + ], + "logging": [ + "src/core/utils/logging/FilePrinter.cpp", + "src/core/utils/logging/Helpers.cpp", + "src/core/utils/logging/Logger.cpp", + "src/core/utils/logging/LoggerRegistry.cpp" + ], + "scheduler": { + "single": [ "src/runtime/CPP/SingleThreadScheduler.cpp" ], + "threads": [ "src/runtime/CPP/CPPScheduler.cpp" ], + "omp": [ "src/runtime/OMP/OMPScheduler.cpp"] + }, + "c_api": { + "common": [ + "src/c/AclContext.cpp", + "src/c/AclOperator.cpp", + "src/c/AclQueue.cpp", + "src/c/AclTensor.cpp", + "src/c/AclTensorPack.cpp", + "src/c/AclVersion.cpp" ], - "c_api" : - { - "cpu": [ - "src/c/AclContext.cpp", - "src/c/AclQueue.cpp", - "src/c/AclTensor.cpp", - "src/c/AclTensorPack.cpp", - "src/c/AclVersion.cpp" - ], - "gpu": [ - "src/c/cl/AclOpenClExt.cpp" + "cpu": [ + ], + "gpu": [ + "src/c/cl/AclOpenClExt.cpp" + ], + "operators": + [ + "src/c/operators/AclActivation.cpp" + ] + }, + "high_priority": [ + "Activation", + "DepthwiseConv2d", + "Conv2d", + "Permute", + "Pool2d", + "Reshape", + "MatMul" + ], + "gpu": { + "common": [ + "src/core/CL/CLCommandBuffer.cpp", + "src/core/CL/CLCompatCommandBuffer.cpp", + "src/core/CL/CLCompileContext.cpp", + "src/core/CL/CLMutableCommandBuffer.cpp", + "src/core/CL/DefaultLWSHeuristics.cpp", + "src/core/CL/CLHelpers.cpp", + "src/core/CL/CLKernelLibrary.cpp", + "src/core/CL/CLUtils.cpp", + "src/core/CL/ICLKernel.cpp", + "src/core/CL/ICLSimple2DKernel.cpp", + "src/core/CL/ICLSimple3DKernel.cpp", + "src/core/CL/ICLSimpleKernel.cpp", + "src/core/CL/ICLTensor.cpp", + "src/core/CL/OpenCL.cpp", + "src/gpu/cl/ClContext.cpp", + "src/gpu/cl/ClKernelLibrary.cpp", + "src/gpu/cl/ClQueue.cpp", + "src/gpu/cl/ClTensor.cpp", + "src/core/CL/kernels/CLFillBorderKernel.cpp", + "src/runtime/CL/CLBufferAllocator.cpp", + "src/runtime/CL/CLGEMMHeuristicsHandle.cpp", + "src/runtime/CL/CLHelpers.cpp", + "src/runtime/CL/CLMemory.cpp", + "src/runtime/CL/CLMemoryRegion.cpp", + "src/runtime/CL/CLOperator.cpp", + "src/runtime/CL/CLRuntimeContext.cpp", + "src/runtime/CL/CLScheduler.cpp", + "src/runtime/CL/CLSubTensor.cpp", + "src/runtime/CL/CLTensor.cpp", + "src/runtime/CL/CLTensorAllocator.cpp", + "src/runtime/CL/CLTuner.cpp", + "src/runtime/CL/ICLSimpleFunction.cpp", + "src/runtime/CL/Utils.cpp", + "src/runtime/CL/mlgo/HeuristicTree.cpp", + "src/runtime/CL/mlgo/MLGOHeuristics.cpp", + "src/runtime/CL/mlgo/MLGOParser.cpp", + "src/runtime/CL/mlgo/Utils.cpp", + "src/runtime/CL/tuners/CLTuningParametersList.cpp" + ], + "operators": { + "Activation":{ + "files": { + "common": [ + "src/gpu/cl/kernels/ClActivationKernel.cpp", + "src/gpu/cl/operators/ClActivation.cpp", + "src/runtime/CL/functions/CLActivationLayer.cpp" + ] + } + }, + "ArgMinMax": { + "deps": [ "Reshape" ], + "files": { + "common": [ + "src/core/CL/kernels/CLArgMinMaxLayerKernel.cpp", + "src/runtime/CL/functions/CLArgMinMaxLayer.cpp" + ] + } + }, + "Add": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClElementwiseKernel.cpp", + "src/gpu/cl/operators/ClAdd.cpp" + ] + } + }, + "BatchNormalization": { + "files": { + "common": [ + "src/core/CL/kernels/CLBatchNormalizationLayerKernel.cpp", + "src/core/CL/kernels/CLFuseBatchNormalizationKernel.cpp", + "src/runtime/CL/functions/CLBatchNormalizationLayer.cpp", + "src/runtime/CL/functions/CLFuseBatchNormalization.cpp" + ] + } + }, + "BatchToSpace": { + "files": { + "common": [ + "src/core/CL/kernels/CLBatchToSpaceLayerKernel.cpp", + "src/runtime/CL/functions/CLBatchToSpaceLayer.cpp" + ] + } + }, + "Bitwise": { + "files": { + "common": [ "src/core/CL/kernels/CLBitwiseKernel.cpp" ] + } + }, + "BitwiseAnd": { + "deps": [ "Bitwise" ], + "files": { + "common": [ "src/runtime/CL/functions/CLBitwiseAnd.cpp" ] + } + }, + "BitwiseNot": { + "deps": [ "Bitwise" ], + "files": { + "common": [ "src/runtime/CL/functions/CLBitwiseNot.cpp" ] + } + }, + "BitwiseOr": { + "deps": [ "Bitwise" ], + "files": { + "common": [ "src/runtime/CL/functions/CLBitwiseOr.cpp" ] + } + }, + "BitwiseXor": { + "deps": [ "Bitwise" ], + "files": { + "common": [ "src/runtime/CL/functions/CLBitwiseXor.cpp" ] + } + }, + "BoundingBoxTransform": { + "files": { + "common": [ + "src/core/CL/kernels/CLBoundingBoxTransformKernel.cpp", + "src/runtime/CL/functions/CLBoundingBoxTransform.cpp" + ] + } + }, + "Cast": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClCastKernel.cpp", + "src/gpu/cl/operators/ClCast.cpp", + "src/runtime/CL/functions/CLCast.cpp" + ] + } + }, + "ChannelShuffle": { + "files": { + "common": [ + "src/core/CL/kernels/CLChannelShuffleLayerKernel.cpp", + "src/runtime/CL/functions/CLChannelShuffleLayer.cpp" + ] + } + }, + "Comparison": { + "files": { + "common": [ + "src/core/CL/kernels/CLComparisonKernel.cpp", + "src/runtime/CL/functions/CLComparison.cpp" + ] + } + }, + "Concatenate": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClBatchConcatenateKernel.cpp", + "src/gpu/cl/kernels/ClDepthConcatenateKernel.cpp", + "src/gpu/cl/kernels/ClHeightConcatenateKernel.cpp", + "src/gpu/cl/kernels/ClWidthConcatenate2TensorsKernel.cpp", + "src/gpu/cl/kernels/ClWidthConcatenate4TensorsKernel.cpp", + "src/gpu/cl/kernels/ClWidthConcatenateKernel.cpp", + "src/gpu/cl/operators/ClConcatenate.cpp", + "src/runtime/CL/functions/CLConcatenateLayer.cpp" + ] + } + }, + "Conv2d": { + "deps": [ + "Activation", + "ElementwiseBinary", + "FFT2D", + "Gemm", + "Mul", + "Pad", + "Permute", + "Reduction", + "Reshape", + "Reverse", + "Slice" + ], + "files": { + "common": [ + "src/gpu/cl/kernels/ClDirectConv2dKernel.cpp", + "src/gpu/cl/kernels/ClWinogradFilterTransformKernel.cpp", + "src/gpu/cl/kernels/ClWinogradInputTransformKernel.cpp", + "src/gpu/cl/kernels/ClWinogradOutputTransformKernel.cpp", + "src/gpu/cl/kernels/ClIm2ColKernel.cpp", + "src/gpu/cl/kernels/ClIndirectConv2dKernel.cpp", + "src/gpu/cl/kernels/ClIndirectConv2dAddressPrecalculationKernel.cpp", + "src/gpu/cl/kernels/ClCol2ImKernel.cpp", + "src/gpu/cl/operators/ClConv2d.cpp", + "src/gpu/cl/operators/ClDirectConv2d.cpp", + "src/gpu/cl/operators/ClGemmConv2d.cpp", + "src/gpu/cl/operators/ClIndirectConv2d.cpp", + "src/gpu/cl/operators/ClWinogradConv2d.cpp", + "src/gpu/cl/kernels/ClWeightsReshapeKernel.cpp", + "src/runtime/CL/functions/CLConvolutionLayer.cpp", + "src/runtime/CL/functions/CLDirectConvolutionLayer.cpp", + "src/runtime/CL/functions/CLFFTConvolutionLayer.cpp", + "src/runtime/CL/functions/CLGEMMConvolutionLayer.cpp", + "src/runtime/CL/functions/CLIndirectConvolutionLayer.cpp", + "src/runtime/CL/functions/CLWinogradConvolutionLayer.cpp" + ] + } + }, + "Conv3d": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClDirectConv3dKernel.cpp", + "src/gpu/cl/operators/ClDirectConv3d.cpp", + "src/runtime/CL/functions/CLConv3D.cpp" + ] + } + }, + "Copy": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClCopyKernel.cpp", + "src/gpu/cl/operators/ClCopy.cpp", + "src/runtime/CL/functions/CLCopy.cpp" + ] + } + }, + "CropResize": { + "deps": [ "Copy", "Fill", "Scale" ], + "files": { + "common": [ + "src/gpu/cl/kernels/ClCropKernel.cpp", + "src/gpu/cl/operators/ClCrop.cpp", + "src/runtime/CL/functions/CLCrop.cpp", + "src/runtime/CL/functions/CLCropResize.cpp" + ] + } + }, + "Deconv2d": { + "deps": [ "Conv2d", "Reverse", "Transpose"], + "files": { + "common": [ + "src/core/CL/kernels/CLDeconvolutionLayerUpsampleKernel.cpp", + "src/core/CL/kernels/CLDeconvolutionReshapeOutputKernel.cpp", + "src/gpu/cl/kernels/ClTransposedConvolutionKernel.cpp", + "src/gpu/cl/operators/ClTransposedConvolution.cpp", + "src/runtime/CL/functions/CLDeconvolutionLayer.cpp", + "src/runtime/CL/functions/CLDeconvolutionLayerUpsample.cpp", + "src/runtime/CL/functions/CLDirectDeconvolutionLayer.cpp", + "src/runtime/CL/functions/CLGEMMDeconvolutionLayer.cpp" + ] + } + }, + "DepthConvert": { + "deps": [ "Cast"], + "files": { + "common": [ "src/runtime/CL/functions/CLDepthConvertLayer.cpp" ] + } + }, + "DepthToSpace": { + "files": { + "common": [ + "src/core/CL/kernels/CLDepthToSpaceLayerKernel.cpp", + "src/runtime/CL/functions/CLDepthToSpaceLayer.cpp" + ] + } + }, + "DepthwiseConv2d": { + "deps": [ "Permute" ], + "files": { + "common": [ + "src/core/CL/kernels/CLDepthwiseConvolutionLayerNativeKernel.cpp", + "src/runtime/CL/functions/CLDepthwiseConvolutionLayer.cpp" + ] + } + }, + "Dequantize": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClDequantizeKernel.cpp", + "src/gpu/cl/operators/ClDequantize.cpp", + "src/runtime/CL/functions/CLDequantizationLayer.cpp" + ] + } + }, + "ElementwiseBinary": { + "deps": ["Add", "Sub"], + "files": { + "common": [ + "src/gpu/cl/kernels/ClElementwiseKernel.cpp", + "src/gpu/cl/operators/ClElementwiseOperations.cpp", + "src/runtime/CL/functions/CLElementwiseOperations.cpp" + ] + } + }, + "ElementwiseUnary":{ + "files": { + "common": [ + "src/gpu/cl/kernels/ClElementwiseUnaryKernel.cpp", + "src/gpu/cl/operators/ClElementwiseUnary.cpp", + "src/runtime/CL/functions/CLElementwiseUnaryLayer.cpp" + ] + } + }, + "FFT1D": { + "files": { + "common": [ + "src/core/CL/kernels/CLFFTDigitReverseKernel.cpp", + "src/core/CL/kernels/CLFFTRadixStageKernel.cpp", + "src/core/CL/kernels/CLFFTScaleKernel.cpp", + "src/runtime/CL/functions/CLFFT1D.cpp" + ] + } + }, + "FFT2D": { + "deps": [ "FFT1D" ], + "files": { + "common": [ "src/runtime/CL/functions/CLFFT2D.cpp" ] + } + }, + "Fill": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClFillKernel.cpp", + "src/gpu/cl/operators/ClFill.cpp", + "src/runtime/CL/functions/CLFill.cpp" + ] + } + }, + "Flatten": { + "files": { + "common": [ + "src/gpu/cl/operators/ClFlatten.cpp", + "src/runtime/CL/functions/CLFlattenLayer.cpp" + ] + } + }, + "Floor": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClFloorKernel.cpp", + "src/gpu/cl/operators/ClFloor.cpp", + "src/runtime/CL/functions/CLFloor.cpp" + ] + } + }, + "FullyConnected": { + "deps": [ "Flatten", "Gemm", "Transpose"], + "files": { + "common": [ + "src/gpu/cl/kernels/ClConvertFullyConnectedWeightsKernel.cpp", + "src/gpu/cl/operators/ClConvertFullyConnectedWeights.cpp", + "src/gpu/cl/operators/ClFullyConnected.cpp", + "src/runtime/CL/functions/CLConvertFullyConnectedWeights.cpp", + "src/runtime/CL/functions/CLFullyConnectedLayer.cpp" + ] + } + }, + "Gather": { + "files": { + "common": [ + "src/core/CL/kernels/CLGatherKernel.cpp", + "src/runtime/CL/functions/CLGather.cpp"] + } + }, + "Gemm": { + "deps": [ "Cast" ], + "files": { + "common": [ + "src/gpu/cl/kernels/gemm/ClGemmHelpers.cpp", + "src/gpu/cl/kernels/gemm/native/ClGemmDefaultConfigNativeBifrost.cpp", + "src/gpu/cl/kernels/gemm/native/ClGemmDefaultConfigNativeMidgard.cpp", + "src/gpu/cl/kernels/gemm/native/ClGemmDefaultConfigNativeValhall.cpp", + "src/gpu/cl/kernels/gemm/reshaped/ClGemmDefaultConfigReshapedBifrost.cpp", + "src/gpu/cl/kernels/gemm/reshaped/ClGemmDefaultConfigReshapedValhall.cpp", + "src/gpu/cl/kernels/gemm/reshaped_only_rhs/ClGemmDefaultConfigReshapedRhsOnlyBifrost.cpp", + "src/gpu/cl/kernels/gemm/reshaped_only_rhs/ClGemmDefaultConfigReshapedRhsOnlyValhall.cpp", + "src/gpu/cl/kernels/ClGemmLowpMatrixMultiplyNativeKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpMatrixMultiplyReshapedKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpMatrixMultiplyReshapedOnlyRhsKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpMatrixMultiplyReshapedOnlyRhsMMULKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpOffsetContributionKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpOffsetContributionOutputStageKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpQuantizeDownInt32ScaleByFixedPointKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpQuantizeDownInt32ScaleByFloatKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpQuantizeDownInt32ScaleKernel.cpp", + "src/gpu/cl/kernels/ClGemmMatrixMultiplyNativeKernel.cpp", + "src/gpu/cl/kernels/ClGemmMatrixMultiplyReshapedOnlyRhsMMULKernel.cpp", + "src/gpu/cl/kernels/ClGemmMatrixMultiplyReshapedKernel.cpp", + "src/gpu/cl/kernels/ClGemmMatrixMultiplyReshapedOnlyRhsKernel.cpp", + "src/gpu/cl/kernels/ClGemmReshapeLhsMatrixKernel.cpp", + "src/gpu/cl/kernels/ClGemmReshapeRhsMatrixKernel.cpp", + "src/gpu/cl/kernels/ClGemmLowpReductionKernel.cpp", + "src/gpu/cl/operators/ClGemm.cpp", + "src/gpu/cl/operators/ClGemmLowpMatrixMultiplyCore.cpp", + "src/gpu/cl/operators/ClGemmLowpOutputStage.cpp", + "src/runtime/CL/gemm/CLGEMMDefaultTypeBifrost.cpp", + "src/runtime/CL/gemm/CLGEMMDefaultTypeMidgard.cpp", + "src/runtime/CL/gemm/CLGEMMDefaultTypeValhall.cpp", + "src/runtime/CL/gemm_auto_heuristics/CLGEMMAutoHeuristics.cpp", + "src/runtime/CL/functions/CLGEMM.cpp", + "src/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.cpp", + "src/runtime/CL/functions/CLGEMMLowpOutputStage.cpp", + "src/runtime/heuristics/direct_conv/ClDirectConvDefaultConfigBifrost.cpp", + "src/runtime/heuristics/direct_conv/ClDirectConvDefaultConfigValhall.cpp", + "src/runtime/heuristics/dwc_native/ClDWCNativeDefaultConfigBifrost.cpp", + "src/runtime/heuristics/dwc_native/ClDWCNativeDefaultConfigValhall.cpp", + "src/runtime/heuristics/dwc_native/ClDWCNativeHeuristicsHelpers.cpp", + "src/runtime/heuristics/indirect_conv/ClIndirectConvDefaultConfigValhall.cpp" + ] + } + }, + "MatMul": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClMatMulLowpNativeKernel.cpp", + "src/gpu/cl/kernels/ClMatMulLowpNativeMMULKernel.cpp", + "src/gpu/cl/kernels/ClMatMulNativeKernel.cpp", + "src/gpu/cl/kernels/ClMatMulNativeMMULKernel.cpp", + "src/gpu/cl/kernels/helpers/MatMulKernelHelpers.cpp", + "src/gpu/cl/operators/ClMatMul.cpp", + "src/runtime/CL/functions/CLMatMul.cpp", + "src/runtime/heuristics/matmul_native/ClMatMulNativeDefaultConfigValhall.cpp", + "src/runtime/heuristics/matmul_native/ClMatMulNativeHelpers.cpp", + "src/runtime/heuristics/matmul_native/ClMatMulNativeDefaultVariantValhall.cpp" + ] + } + }, + "GenerateProposals": { + "deps": [ "BoundingBoxTransform", "Dequantize", "Pad", "Permute", "Quantize", "Reshape" ], + "files": { + "common": [ + "src/core/CL/kernels/CLGenerateProposalsLayerKernel.cpp", + "src/runtime/CL/functions/CLGenerateProposalsLayer.cpp" + ] + } + }, + "InstanceNormalize": { + "files": { + "common": [ + "src/core/CL/kernels/CLInstanceNormalizationLayerKernel.cpp", + "src/runtime/CL/functions/CLInstanceNormalizationLayer.cpp" + ] + } + }, + "L2Normalize": { + "deps": [ "Reduction" ], + "files": { + "common": [ + "src/core/CL/kernels/CLL2NormalizeLayerKernel.cpp", + "src/runtime/CL/functions/CLL2NormalizeLayer.cpp" + ] + } + }, + "Logical": { + "files": { + "common": [ + "src/gpu/cl/operators/ClLogicalNot.cpp", + "src/runtime/CL/functions/CLLogicalAnd.cpp", + "src/runtime/CL/functions/CLLogicalNot.cpp", + "src/runtime/CL/functions/CLLogicalOr.cpp" + ] + } + }, + "LSTM": { + "deps": [ + "Activation", + "Concatenate", + "Copy", + "Dequantize", + "ElementwiseBinary", + "Fill", + "FullyConnected", + "Gemm", + "MeanStdDevNormalize", + "Mul", + "Quantize", + "Slice", + "Transpose" + ], + "files": { + "common": [ + "src/core/CL/kernels/CLQLSTMLayerNormalizationKernel.cpp", + "src/runtime/CL/functions/CLQLSTMLayer.cpp", + "src/runtime/CL/functions/CLLSTMLayer.cpp", + "src/runtime/CL/functions/CLLSTMLayerQuantized.cpp" + ] + } + }, + "MaxUnpool2d": { + "deps": [ "Fill" ], + "files": { + "common": [ + "src/core/CL/kernels/CLMaxUnpoolingLayerKernel.cpp", + "src/runtime/CL/functions/CLMaxUnpoolingLayer.cpp" + ] + } + }, + "MeanStdDevNormalize": { + "deps": [ "Reduction" ], + "files": { + "common": [ + "src/core/CL/kernels/CLMeanStdDevNormalizationKernel.cpp", + "src/runtime/CL/functions/CLMeanStdDevNormalizationLayer.cpp", + "src/runtime/CL/functions/CLReduceMean.cpp" + ] + } + }, + "Mul": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClMulKernel.cpp", + "src/gpu/cl/operators/ClMul.cpp", + "src/runtime/CL/functions/CLPixelWiseMultiplication.cpp" + ] + } + }, + "Normalize": { + "files": { + "common": [ + "src/core/CL/kernels/CLNormalizationLayerKernel.cpp", + "src/runtime/CL/functions/CLNormalizationLayer.cpp" + ] + } + }, + "Pad": { + "deps": [ "Copy" ], + "files": { + "common": [ + "src/core/CL/kernels/CLPadLayerKernel.cpp", + "src/runtime/CL/functions/CLPadLayer.cpp" + ] + } + }, + "Permute": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClPermuteKernel.cpp", + "src/gpu/cl/operators/ClPermute.cpp", + "src/runtime/CL/functions/CLPermute.cpp" + ] + } + }, + "Pool2d": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClPool2dKernel.cpp", + "src/gpu/cl/operators/ClPool2d.cpp", + "src/runtime/CL/functions/CLPoolingLayer.cpp" + ] + } + }, + "Pool3d": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClPool3dKernel.cpp", + "src/gpu/cl/operators/ClPool3d.cpp", + "src/runtime/CL/functions/CLPooling3dLayer.cpp" + ] + } + }, + "PRelu": { + "deps": [ "ElementwiseBinary" ], + "files": { + "common": [ + "src/gpu/cl/operators/ClPRelu.cpp", + "src/runtime/CL/functions/CLPReluLayer.cpp" + ] + } + }, + "PriorBox": { + "files": { + "common": [ + "src/core/CL/kernels/CLPriorBoxLayerKernel.cpp", + "src/runtime/CL/functions/CLPriorBoxLayer.cpp" + ] + } + }, + "Quantize": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClQuantizeKernel.cpp", + "src/gpu/cl/operators/ClQuantize.cpp", + "src/runtime/CL/functions/CLQuantizationLayer.cpp" + ] + } + }, + "Range": { + "files": { + "common": [ + "src/core/CL/kernels/CLRangeKernel.cpp", + "src/runtime/CL/functions/CLRange.cpp" + ] + } + }, + "Reduction": { + "deps": [ "Reshape" ], + "files": { + "common": [ + "src/core/CL/kernels/CLReductionOperationKernel.cpp", + "src/runtime/CL/functions/CLReductionOperation.cpp" + ] + } + }, + "Reorg": { + "files": { + "common": [ + "src/core/CL/kernels/CLReorgLayerKernel.cpp", + "src/runtime/CL/functions/CLReorgLayer.cpp" + ] + } + }, + "Reshape": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClReshapeKernel.cpp", + "src/gpu/cl/operators/ClReshape.cpp", + "src/runtime/CL/functions/CLReshapeLayer.cpp" + ] + } + }, + "Reverse": { + "files": { + "common": [ + "src/core/CL/kernels/CLReverseKernel.cpp", + "src/runtime/CL/functions/CLReverse.cpp" + ] + } + }, + "RNN": { + "deps": [ "Activation", "Cast", "ElementwiseBinary", "FullyConnected", "Gemm"], + "files": { + "common": [ "src/runtime/CL/functions/CLRNNLayer.cpp" ] + } + }, + "ROIAlign": { + "files": { + "common": [ + "src/core/CL/kernels/CLROIAlignLayerKernel.cpp", + "src/runtime/CL/functions/CLROIAlignLayer.cpp" + ] + } + }, + "ROIPool2d": { + "files": { + "common": [ + "src/core/CL/kernels/CLROIPoolingLayerKernel.cpp", + "src/runtime/CL/functions/CLROIPoolingLayer.cpp" + ] + } + }, + "Scale": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClScaleKernel.cpp", + "src/gpu/cl/operators/ClScale.cpp", + "src/runtime/CL/functions/CLScale.cpp" + ] + } + }, + "Scatter": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClScatterKernel.cpp", + "src/gpu/cl/operators/ClScatter.cpp", + "src/runtime/CL/functions/CLScatter.cpp" + ] + } + }, + "Select": { + "files": { + "common": [ + "src/core/CL/kernels/CLSelectKernel.cpp", + "src/runtime/CL/functions/CLSelect.cpp" ] + } }, + "Slice": { + "deps": [ "StridedSlice" ], + "files": { + "common": [ "src/runtime/CL/functions/CLSlice.cpp" ] + } + }, + "Softmax": { + "deps": [ "Permute" ], + "files": { + "common": [ + "src/gpu/cl/kernels/ClSoftmaxKernel.cpp", + "src/gpu/cl/operators/ClSoftmax.cpp", + "src/runtime/CL/functions/CLSoftmaxLayer.cpp" + ] + } + }, + "SpaceToBatch": { + "files": { + "common": [ + "src/core/CL/kernels/CLSpaceToBatchLayerKernel.cpp", + "src/runtime/CL/functions/CLSpaceToBatchLayer.cpp" + ] + } + }, + "SpaceToDepth": { + "files": { + "common": [ + "src/core/CL/kernels/CLSpaceToDepthLayerKernel.cpp", + "src/runtime/CL/functions/CLSpaceToDepthLayer.cpp" + ] + } + }, + "Split": { + "deps": [ "StridedSlice" ], + "files": { + "common": [ "src/runtime/CL/functions/CLSplit.cpp" ] + } + }, + "Stack": { + "files": { + "common": [ + "src/core/CL/kernels/CLStackLayerKernel.cpp", + "src/runtime/CL/functions/CLStackLayer.cpp" + ] + } + }, + "StridedSlice": { + "files": { + "common": [ + "src/core/CL/kernels/CLStridedSliceKernel.cpp", + "src/runtime/CL/functions/CLStridedSlice.cpp" + ] + } + }, + "Sub": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClElementwiseKernel.cpp", + "src/gpu/cl/operators/ClSub.cpp" + ] + } + }, + "Tile": { + "files": { + "common": [ + "src/core/CL/kernels/CLTileKernel.cpp", + "src/runtime/CL/functions/CLTile.cpp" + ] + } + }, + "Transpose": { + "files": { + "common": [ + "src/gpu/cl/kernels/ClTransposeKernel.cpp", + "src/gpu/cl/operators/ClTranspose.cpp", + "src/runtime/CL/functions/CLTranspose.cpp" + ] + } + }, + "Unstack": { + "deps": [ "StridedSlice" ], + "files": { + "common": [ "src/runtime/CL/functions/CLUnstack.cpp" ] + } + }, + "YUVNormalize": { + "files": { + "common": [ + "src/core/CL/kernels/CLNormalizePlanarYUVLayerKernel.cpp", + "src/runtime/CL/functions/CLNormalizePlanarYUVLayer.cpp" + ] + } + } + } +}, + "cpu": { + "common": [ + "src/cpu/CpuContext.cpp", + "src/cpu/CpuQueue.cpp", + "src/cpu/CpuTensor.cpp", + "src/core/NEON/kernels/NEFillBorderKernel.cpp", + "src/runtime/NEON/INEOperator.cpp", + "src/runtime/NEON/INESimpleFunction.cpp", + "src/runtime/NEON/INESimpleFunctionNoBorder.cpp" + ], + "operators": { + "Activation": { + "deps": [ "LUT" ], + "files": { + "common": [ + "src/cpu/operators/CpuActivation.cpp", + "src/cpu/kernels/CpuActivationKernel.cpp", + "src/runtime/NEON/functions/NEActivationLayer.cpp" + ], + "neon": { + "fp16": [ "src/cpu/kernels/activation/generic/neon/fp16.cpp" ], + "fp32": [ "src/cpu/kernels/activation/generic/neon/fp32.cpp" ], + "qasymm8": [ + "src/cpu/kernels/activation/generic/neon/qasymm8.cpp", + "src/cpu/kernels/activation/generic/neon/lut.cpp" + ], + "qasymm8_signed": [ + "src/cpu/kernels/activation/generic/neon/qasymm8_signed.cpp", + "src/cpu/kernels/activation/generic/neon/lut.cpp" + ], + "qsymm16": [ "src/cpu/kernels/activation/generic/neon/qsymm16.cpp" ] + }, + "sve": { + "fp16": [ "src/cpu/kernels/activation/generic/sve/fp16.cpp" ], + "fp32": [ "src/cpu/kernels/activation/generic/sve/fp32.cpp" ] + }, + "sve2":{ + "qasymm8": [ + "src/cpu/kernels/activation/generic/sve2/qasymm8.cpp", + "src/cpu/kernels/activation/generic/sve2/lut.cpp" + ], + "qasymm8_signed": [ "src/cpu/kernels/activation/generic/sve2/qasymm8_signed.cpp" ], + "qsymm16": [ "src/cpu/kernels/activation/generic/sve2/qsymm16.cpp" ] + } + } + }, + "ArgMinMax": { + "deps": [ "Reduction" ], + "files": { + "common": [ "src/runtime/NEON/functions/NEArgMinMaxLayer.cpp" ] + } + }, + "Add": { + "files": { + "common": [ + "src/cpu/operators/CpuAdd.cpp", + "src/cpu/kernels/CpuAddKernel.cpp", + "src/runtime/NEON/functions/NEArithmeticAddition.cpp" + ], + "neon": { + "common": ["src/cpu/kernels/add/generic/neon/impl.cpp"], + "fp32":["src/cpu/kernels/add/generic/neon/fp32.cpp"], + "fp16":["src/cpu/kernels/add/generic/neon/fp16.cpp"], + "integer":["src/cpu/kernels/add/generic/neon/integer.cpp"], + "qasymm8": ["src/cpu/kernels/add/generic/neon/qasymm8.cpp"], + "qasymm8_signed": ["src/cpu/kernels/add/generic/neon/qasymm8_signed.cpp"], + "qsymm16": ["src/cpu/kernels/add/generic/neon/qsymm16.cpp"] + }, + "sve": { + "common": [ "src/cpu/kernels/add/generic/sve/impl.cpp" ], + "integer":["src/cpu/kernels/add/generic/sve/integer.cpp"], + "fp32":["src/cpu/kernels/add/generic/sve/fp32.cpp"], + "fp16":["src/cpu/kernels/add/generic/sve/fp16.cpp"] + }, + "sve2": { + "qasymm8": [ "src/cpu/kernels/add/generic/sve2/qasymm8.cpp" ], + "qasymm8_signed": [ "src/cpu/kernels/add/generic/sve2/qasymm8_signed.cpp" ], + "qsymm16": [ "src/cpu/kernels/add/generic/sve2/qsymm16.cpp" ] + } + } + }, + "AddMulAdd": { + "files": { + "common": [ + "src/cpu/operators/CpuAddMulAdd.cpp", + "src/cpu/kernels/CpuAddMulAddKernel.cpp", + "src/runtime/NEON/functions/NEAddMulAdd.cpp" + ], + "neon": { + "fp32":["src/cpu/kernels/addmuladd/generic/neon/fp32.cpp"], + "fp16":["src/cpu/kernels/addmuladd/generic/neon/fp16.cpp"], + "qasymm8": ["src/cpu/kernels/addmuladd/generic/neon/qasymm8.cpp"], + "qasymm8_signed": ["src/cpu/kernels/addmuladd/generic/neon/qasymm8_signed.cpp"] + } + } + }, + "BatchNormalize": { + "files": { + "common": [ + "src/core/NEON/kernels/NEBatchNormalizationLayerKernel.cpp", + "src/core/NEON/kernels/NEFuseBatchNormalizationKernel.cpp", + "src/runtime/NEON/functions/NEBatchNormalizationLayer.cpp", + "src/runtime/NEON/functions/NEFuseBatchNormalization.cpp" + ], + "neon": { + "common": [ + "src/cpu/kernels/fuse_batch_normalization/nchw/all.cpp" + ], + "fp16": [ + "src/cpu/kernels/fuse_batch_normalization/generic/fp16.cpp", + "src/core/NEON/kernels/batchnormalization/impl/NEON/fp16.cpp", + "src/cpu/kernels/fuse_batch_normalization/nhwc/neon/fp16.cpp", + "src/cpu/kernels/fuse_batch_normalization/nchw/neon/fp16.cpp" - "gpu" : - { - "core" : - { - "kernels" : - { - "high_priority" : [ - "src/core/gpu/cl/kernels/ClActivationKernel.cpp", - "src/core/gpu/cl/kernels/ClDirectConv2dKernel.cpp", - "src/core/gpu/cl/kernels/ClPermuteKernel.cpp", - "src/core/gpu/cl/kernels/ClPool2dKernel.cpp", - "src/core/gpu/cl/kernels/ClReshapeKernel.cpp" - ], - "all" : [ - "src/core/gpu/cl/kernels/ClBatchConcatenateKernel.cpp", - "src/core/gpu/cl/kernels/ClCastKernel.cpp", - "src/core/gpu/cl/kernels/ClConvertFullyConnectedWeightsKernel.cpp", - "src/core/gpu/cl/kernels/ClCopyKernel.cpp", - "src/core/gpu/cl/kernels/ClCropKernel.cpp", - "src/core/gpu/cl/kernels/ClDepthConcatenateKernel.cpp", - "src/core/gpu/cl/kernels/ClDequantizeKernel.cpp", - "src/core/gpu/cl/kernels/ClElementwiseKernel.cpp", - "src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.cpp", - "src/core/gpu/cl/kernels/ClFillKernel.cpp", - "src/core/gpu/cl/kernels/ClFloorKernel.cpp", - "src/core/gpu/cl/kernels/ClGemmMatrixMultiplyNativeKernel.cpp", - "src/core/gpu/cl/kernels/ClGemmMatrixMultiplyReshapedKernel.cpp", - "src/core/gpu/cl/kernels/ClGemmMatrixMultiplyReshapedOnlyRhsKernel.cpp", - "src/core/gpu/cl/kernels/ClGemmMatrixMultiplyKernel.cpp", - "src/core/gpu/cl/kernels/ClGemmReshapeRhsMatrixKernel.cpp", - "src/core/gpu/cl/kernels/ClGemmReshapeLhsMatrixKernel.cpp", - "src/core/gpu/cl/kernels/ClHeightConcatenateKernel.cpp", - "src/core/gpu/cl/kernels/ClMulKernel.cpp", - "src/core/gpu/cl/kernels/ClQuantizeKernel.cpp", - "src/core/gpu/cl/kernels/ClScaleKernel.cpp", - "src/core/gpu/cl/kernels/ClSoftmaxKernel.cpp", - "src/core/gpu/cl/kernels/ClTransposeKernel.cpp", - "src/core/gpu/cl/kernels/ClWidthConcatenate2TensorsKernel.cpp", - "src/core/gpu/cl/kernels/ClWidthConcatenate4TensorsKernel.cpp", - "src/core/gpu/cl/kernels/ClWidthConcatenateKernel.cpp", - "src/core/gpu/cl/kernels/ClWinogradFilterTransformKernel.cpp", - "src/core/gpu/cl/kernels/ClWinogradInputTransformKernel.cpp", - "src/core/gpu/cl/kernels/ClWinogradOutputTransformKernel.cpp", - "src/core/CL/kernels/CLGenerateProposalsLayerKernel.cpp", - "src/core/CL/kernels/CLArgMinMaxLayerKernel.cpp", - "src/core/CL/kernels/CLBatchNormalizationLayerKernel.cpp", - "src/core/CL/kernels/CLBatchToSpaceLayerKernel.cpp", - "src/core/CL/kernels/CLBitwiseKernel.cpp", - "src/core/CL/kernels/CLBoundingBoxTransformKernel.cpp", - "src/core/CL/kernels/CLChannelShuffleLayerKernel.cpp", - "src/core/CL/kernels/CLCol2ImKernel.cpp", - "src/core/CL/kernels/CLComparisonKernel.cpp", - "src/core/CL/kernels/CLDeconvolutionLayerUpsampleKernel.cpp", - "src/core/CL/kernels/CLDeconvolutionReshapeOutputKernel.cpp", - "src/core/CL/kernels/CLDepthToSpaceLayerKernel.cpp", - "src/core/CL/kernels/CLDepthwiseConvolutionLayer3x3NCHWKernel.cpp", - "src/core/CL/kernels/CLDepthwiseConvolutionLayer3x3NHWCKernel.cpp", - "src/core/CL/kernels/CLDepthwiseConvolutionLayerNativeKernel.cpp", - "src/core/CL/kernels/CLFFTDigitReverseKernel.cpp", - "src/core/CL/kernels/CLFFTRadixStageKernel.cpp", - "src/core/CL/kernels/CLFFTScaleKernel.cpp", - "src/core/CL/kernels/CLFillBorderKernel.cpp", - "src/core/CL/kernels/CLFuseBatchNormalizationKernel.cpp", - "src/core/CL/kernels/CLGatherKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpMatrixMultiplyNativeKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpOffsetContributionKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpOffsetContributionOutputStageKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ScaleByFixedPointKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ScaleByFloatKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ScaleKernel.cpp", - "src/core/CL/kernels/CLGEMMLowpReductionKernel.cpp", - "src/core/CL/kernels/CLIm2ColKernel.cpp", - "src/core/CL/kernels/CLInstanceNormalizationLayerKernel.cpp", - "src/core/CL/kernels/CLL2NormalizeLayerKernel.cpp", - "src/core/CL/kernels/CLMaxUnpoolingLayerKernel.cpp", - "src/core/CL/kernels/CLMeanStdDevNormalizationKernel.cpp", - "src/core/CL/kernels/CLMinMaxLayerKernel.cpp", - "src/core/CL/kernels/CLNormalizationLayerKernel.cpp", - "src/core/CL/kernels/CLNormalizePlanarYUVLayerKernel.cpp", - "src/core/CL/kernels/CLPadLayerKernel.cpp", - "src/core/CL/kernels/CLPriorBoxLayerKernel.cpp", - "src/core/CL/kernels/CLQLSTMLayerNormalizationKernel.cpp", - "src/core/CL/kernels/CLRangeKernel.cpp", - "src/core/CL/kernels/CLReductionOperationKernel.cpp", - "src/core/CL/kernels/CLRemapKernel.cpp", - "src/core/CL/kernels/CLReorgLayerKernel.cpp", - "src/core/CL/kernels/CLReverseKernel.cpp", - "src/core/CL/kernels/CLROIAlignLayerKernel.cpp", - "src/core/CL/kernels/CLROIPoolingLayerKernel.cpp", - "src/core/CL/kernels/CLSelectKernel.cpp", - "src/core/CL/kernels/CLSpaceToBatchLayerKernel.cpp", - "src/core/CL/kernels/CLSpaceToDepthLayerKernel.cpp", - "src/core/CL/kernels/CLStackLayerKernel.cpp", - "src/core/CL/kernels/CLStridedSliceKernel.cpp", - "src/core/CL/kernels/CLTileKernel.cpp", - "src/core/CL/kernels/CLWeightsReshapeKernel.cpp" - ] - } - } - }, - "cpu" : - { - "runtime" : - { - "all" : [ - "src/cpu/CpuContext.cpp", - "src/cpu/CpuQueue.cpp", - "src/cpu/CpuTensor.cpp" ], - "operators" : - { - "high_priority" : [ - "src/runtime/cpu/operators/CpuActivation.cpp", - "src/runtime/cpu/operators/CpuDepthwiseConv2d.cpp", - "src/runtime/cpu/operators/CpuDepthwiseConv2dAssemblyDispatch.cpp", - "src/runtime/cpu/operators/CpuDirectConv2d.cpp", - "src/runtime/cpu/operators/CpuPermute.cpp", - "src/runtime/cpu/operators/CpuPool2d.cpp" - ], - "internal" : [ - "src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.cpp" - ], - "all" : [ - "src/runtime/cpu/operators/CpuAdd.cpp", - "src/runtime/cpu/operators/CpuCast.cpp", - "src/runtime/cpu/operators/CpuConcatenate.cpp", - "src/runtime/cpu/operators/CpuConvertFullyConnectedWeights.cpp", - "src/runtime/cpu/operators/CpuCopy.cpp", - "src/runtime/cpu/operators/CpuDequantize.cpp", - "src/runtime/cpu/operators/CpuElementwise.cpp", - "src/runtime/cpu/operators/CpuElementwiseUnary.cpp", - "src/runtime/cpu/operators/CpuFill.cpp", - "src/runtime/cpu/operators/CpuFlatten.cpp", - "src/runtime/cpu/operators/CpuFloor.cpp", - "src/runtime/cpu/operators/CpuGemmDirectConv2d.cpp", - "src/runtime/cpu/operators/CpuMul.cpp", - "src/runtime/cpu/operators/CpuQuantize.cpp", - "src/runtime/cpu/operators/CpuReshape.cpp", - "src/runtime/cpu/operators/CpuScale.cpp", - "src/runtime/cpu/operators/CpuSoftmax.cpp", - "src/runtime/cpu/operators/CpuSub.cpp", - "src/runtime/cpu/operators/CpuTranspose.cpp" - ] - } - }, - "core" : - { - "kernels" : - { - "high_priority" : [ - "src/core/cpu/kernels/CpuActivationKernel.cpp", - "src/core/cpu/kernels/CpuDepthwiseConv2dNativeKernel.cpp", - "src/core/cpu/kernels/CpuDirectConv2dKernel.cpp", - "src/core/cpu/kernels/CpuDirectConv2dOutputStageKernel.cpp", - "src/core/cpu/kernels/CpuPermuteKernel.cpp", - "src/core/cpu/kernels/CpuPool2dKernel.cpp", - "src/core/cpu/kernels/CpuReshapeKernel.cpp" - ], - "all" : [ - "src/core/cpu/kernels/CpuAddKernel.cpp", - "src/core/cpu/kernels/CpuCastKernel.cpp", - "src/core/cpu/kernels/CpuConcatenateBatchKernel.cpp", - "src/core/cpu/kernels/CpuConcatenateDepthKernel.cpp", - "src/core/cpu/kernels/CpuConcatenateHeightKernel.cpp", - "src/core/cpu/kernels/CpuConcatenateWidthKernel.cpp", - "src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.cpp", - "src/core/cpu/kernels/CpuCopyKernel.cpp", - "src/core/cpu/kernels/CpuDequantizeKernel.cpp", - "src/core/cpu/kernels/CpuElementwiseKernel.cpp", - "src/core/cpu/kernels/CpuElementwiseUnaryKernel.cpp", - "src/core/cpu/kernels/CpuFillKernel.cpp", - "src/core/cpu/kernels/CpuFloorKernel.cpp", - "src/core/cpu/kernels/CpuMulKernel.cpp", - "src/core/cpu/kernels/CpuQuantizeKernel.cpp", - "src/core/cpu/kernels/CpuScaleKernel.cpp", - "src/core/cpu/kernels/CpuSoftmaxKernel.cpp", - "src/core/cpu/kernels/CpuSubKernel.cpp", - "src/core/cpu/kernels/CpuTransposeKernel.cpp" - ] - }, + "fp32": [ + "src/cpu/kernels/fuse_batch_normalization/generic/fp32.cpp", + "src/core/NEON/kernels/batchnormalization/impl/NEON/fp32.cpp", + "src/cpu/kernels/fuse_batch_normalization/nhwc/neon/fp32.cpp", + "src/cpu/kernels/fuse_batch_normalization/nchw/neon/fp32.cpp" + ] + }, + "sve": { + "fp16": [ "src/core/NEON/kernels/batchnormalization/impl/SVE/fp16.cpp" ], + "fp32": [ "src/core/NEON/kernels/batchnormalization/impl/SVE/fp32.cpp" ] + } + } + }, + "BatchToSpace": { + "files": { + "common": [ + "src/core/NEON/kernels/NEBatchToSpaceLayerKernel.cpp", + "src/runtime/NEON/functions/NEBatchToSpaceLayer.cpp" + ] + } + }, + "BitwiseAnd": { + "files": { + "common": [ + "src/core/NEON/kernels/NEBitwiseAndKernel.cpp", + "src/runtime/NEON/functions/NEBitwiseAnd.cpp" + ] + } + }, + "BitwiseNot": { + "files": { + "common": [ + "src/core/NEON/kernels/NEBitwiseNotKernel.cpp", + "src/runtime/NEON/functions/NEBitwiseNot.cpp" + ] + } + }, + "BitwiseOr": { + "files": { + "common": [ + "src/core/NEON/kernels/NEBitwiseOrKernel.cpp", + "src/runtime/NEON/functions/NEBitwiseOr.cpp" + ] + } + }, + "BitwiseXor": { + "files": { + "common": [ + "src/core/NEON/kernels/NEBitwiseXorKernel.cpp", + "src/runtime/NEON/functions/NEBitwiseXor.cpp" + ] + } + }, + "BoundingBoxTransform": { + "files": { + "common": [ + "src/core/NEON/kernels/NEBoundingBoxTransformKernel.cpp", + "src/runtime/NEON/functions/NEBoundingBoxTransform.cpp" + ], + "neon":{ + "common":["src/cpu/kernels/boundingboxtransform/generic/neon/impl.cpp"], + "fp32":["src/cpu/kernels/boundingboxtransform/generic/neon/fp32.cpp"], + "fp16":["src/cpu/kernels/boundingboxtransform/generic/neon/fp16.cpp"], + "qsymm16":["src/cpu/kernels/boundingboxtransform/generic/neon/qsymm16.cpp"] + } + } + }, + "Cast": { + "files": { + "common": [ + "src/cpu/operators/CpuCast.cpp", + "src/cpu/kernels/CpuCastKernel.cpp", + "src/runtime/NEON/functions/NECast.cpp" + ], + "neon":{ + "fp16":["src/cpu/kernels/cast/generic/neon/fp16.cpp"] + } + } + }, + "ChannelShuffle": { + "files": { + "common": [ + "src/core/NEON/kernels/NEChannelShuffleLayerKernel.cpp", + "src/runtime/NEON/functions/NEChannelShuffleLayer.cpp" + ] + } + }, + "Concatenate": { + "files": { + "common": [ + "src/cpu/operators/CpuConcatenate.cpp", + "src/cpu/kernels/CpuConcatenateWidthKernel.cpp", + "src/cpu/kernels/CpuConcatenateBatchKernel.cpp", + "src/cpu/kernels/CpuConcatenateDepthKernel.cpp", + "src/cpu/kernels/CpuConcatenateHeightKernel.cpp", + "src/runtime/NEON/functions/NEConcatenateLayer.cpp" + ] + } + }, + "Conv2d": { + "deps": [ + "Activation", + "ElementwiseBinary", + "FFT2D", + "Gemm", + "Mul", + "Pad", + "Permute", + "Reshape", + "Reverse", + "Slice" + ], + "files": { + "common": [ + "src/cpu/operators/CpuConv2d.cpp", + "src/cpu/operators/CpuDirectConv2d.cpp", + "src/cpu/operators/CpuGemmDirectConv2d.cpp", + "src/cpu/operators/CpuGemmConv2d.cpp", + "src/cpu/operators/CpuWinogradConv2d.cpp", + "src/cpu/operators/internal/CpuGemmAssemblyDispatch.cpp", + "src/cpu/kernels/CpuDirectConv2dKernel.cpp", + "src/cpu/kernels/CpuDirectConv2dOutputStageKernel.cpp", + "src/cpu/kernels/CpuWinogradConv2dKernel.cpp", + "src/cpu/kernels/CpuCol2ImKernel.cpp", + "src/cpu/kernels/CpuIm2ColKernel.cpp", + "src/cpu/kernels/CpuWeightsReshapeKernel.cpp", + "src/runtime/NEON/functions/NEConvolutionLayer.cpp", + "src/runtime/NEON/functions/NEDirectConvolutionLayer.cpp", + "src/runtime/NEON/functions/NEFFTConvolutionLayer.cpp", + "src/runtime/NEON/functions/NEGEMMConv2d.cpp", + "src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp", + "src/runtime/NEON/functions/NEWinogradConvolutionLayer.cpp" + ], + "neon": { + "common": [ + "src/core/NEON/kernels/convolution/common/padding.cpp", + "src/core/NEON/kernels/convolution/common/qasymm8.cpp", + "src/core/NEON/kernels/convolution/common/qsymm8.cpp", + "src/core/NEON/kernels/convolution/common/utils.cpp", + "src/core/NEON/kernels/convolution/winograd/input_transforms_fp32.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms_fp32.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms_fp32.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_fp32.cpp", + "src/core/NEON/kernels/convolution/winograd/input_transforms/a64_fp32_6x6.cpp", + "src/core/NEON/kernels/convolution/winograd/input_transforms/arm_fp32_1x8.cpp", + "src/core/NEON/kernels/convolution/winograd/input_transforms/arm_fp32_4x4.cpp", + "src/core/NEON/kernels/convolution/winograd/input_transforms/arm_fp32_6x6.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms/arm_fp32_1x2_1x7.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms/arm_fp32_1x4_1x5.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms/arm_fp32_1x6_1x3.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms/arm_fp32_2x2_3x3.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms/arm_fp32_2x2_5x5.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms/arm_fp32_4x4_3x3.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms/arm_fp32_2x2_3x3.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms/arm_fp32_2x2_5x5.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms/arm_fp32_4x4_3x3.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms/cpp_fp32_1x2_1x7.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms/cpp_fp32_1x4_1x5.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms/cpp_fp32_1x6_1x3.cpp", + "src/cpu/kernels/directconv2d/nhwc/neon/impl.cpp", + "src/cpu/kernels/directconv2d/nhwc/neon/qasymm8.cpp", + "src/cpu/kernels/directconv2d/nchw/all.cpp" - "sve" : - { - "all" : [ - "src/core/cpu/kernels/add/sve/impl.cpp", - "src/core/cpu/kernels/softmax/impl/sve/impl.cpp", - "src/core/cpu/kernels/elementwise/sve/elementwise.cpp", - "src/core/cpu/kernels/elementwise/sve/elementwise_unary.cpp" - ], - "fp32" : [ - "src/core/cpu/kernels/activation/sve/fp32.cpp", - "src/core/cpu/kernels/scale/sve/fp32.cpp", - "src/core/NEON/kernels/batchnormalization/impl/SVE/fp32.cpp" - ], - "fp16" : [ - "src/core/cpu/kernels/activation/sve/fp16.cpp", - "src/core/cpu/kernels/scale/sve/fp16.cpp", - "src/core/NEON/kernels/batchnormalization/impl/SVE/fp16.cpp" - ], - "qsymm16" : [ - "src/core/cpu/kernels/activation/sve/qsymm16.cpp", - "src/core/cpu/kernels/add/sve/qsymm16.cpp" - ], - "qasymm8" : [ - "src/core/cpu/kernels/activation/sve/qasymm8.cpp", - "src/core/cpu/kernels/add/sve/qasymm8.cpp", - "src/core/cpu/kernels/scale/sve/qasymm8.cpp" - ], - "qasymm8_signed" : [ - "src/core/cpu/kernels/activation/sve/qasymm8_signed.cpp", - "src/core/cpu/kernels/add/sve/qasymm8_signed.cpp", - "src/core/cpu/kernels/scale/sve/qasymm8_signed.cpp" - ], - "integer" : [ - "src/core/cpu/kernels/add/sve/integer.cpp", - "src/core/cpu/kernels/scale/sve/integer.cpp" - ] - }, + ], + "fp32": [ + "src/cpu/kernels/directconv2d/nhwc/neon/fp32.cpp" + ], + "fp16": [ + "src/cpu/kernels/directconv2d/nchw/fp16.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_fp16.cpp", + "src/core/NEON/kernels/convolution/winograd/input_transforms_fp16.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms_fp16.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms_fp16.cpp", + "src/core/NEON/kernels/convolution/winograd/input_transforms/a64_fp16_6x6.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms/a64_fp16_4x4_3x3.cpp", + "src/core/NEON/kernels/convolution/winograd/weight_transforms/a64_fp16_4x4_3x3.cpp", + "src/cpu/kernels/directconv2d/nhwc/neon/fp16.cpp" + ] + }, + "sve": { + "common": [ + "src/core/NEON/kernels/convolution/winograd/input_transforms/sve_fp32_6x6.cpp", + "src/core/NEON/kernels/convolution/winograd/input_transforms/sme_fp32_mla_6x6.cpp", + "src/core/NEON/kernels/convolution/winograd/output_transforms/sme_fp32_mopa_4x4_3x3.cpp" + ] + } + } + }, + "Copy": { + "files": { + "common": [ + "src/cpu/operators/CpuCopy.cpp", + "src/cpu/kernels/CpuCopyKernel.cpp", + "src/runtime/NEON/functions/NECopy.cpp" + ] + } + }, + "CropResize": { + "deps": [ "Scale" ], + "files": { + "common": [ + "src/core/NEON/kernels/NECropKernel.cpp", + "src/runtime/NEON/functions/NECropResize.cpp" + ], + "neon": { + "fp32": [ "src/cpu/kernels/crop/generic/neon/fp32.cpp" ], + "fp16": [ "src/cpu/kernels/crop/generic/neon/fp16.cpp" ], + "integer": [ "src/cpu/kernels/crop/generic/neon/integer.cpp" ] + } + } + }, + "Deconv2d": { + "deps": [ "Conv2d", "Reverse", "Transpose"], + "files": { + "common": [ + "src/runtime/NEON/functions/NEDeconvolutionLayer.cpp" + ] + } + }, + "DepthConvert": { + "deps": [ "Cast"], + "files": { + "common": [ + "src/runtime/NEON/functions/NEDepthConvertLayer.cpp" + ] + } + }, + "DepthToSpace": { + "files": { + "common": [ + "src/core/NEON/kernels/NEDepthToSpaceLayerKernel.cpp", + "src/runtime/NEON/functions/NEDepthToSpaceLayer.cpp", + "src/cpu/kernels/depth_to_space/nhwc/any/impl.cpp", + "src/cpu/kernels/depth_to_space/nchw/any/impl.cpp" + ] + } + }, + "DepthwiseConv2d": { + "deps": [ "Activation", "Permute" ], + "files": { + "common": [ + "src/cpu/operators/CpuDepthwiseConv2d.cpp", + "src/cpu/operators/CpuDepthwiseConv2dAssemblyDispatch.cpp", + "src/cpu/kernels/CpuDepthwiseConv2dNativeKernel.cpp", + "src/cpu/kernels/internal/CpuDepthwiseConv2dAssemblyWrapperKernel.cpp", + "src/runtime/NEON/functions/NEDepthwiseConvolutionLayer.cpp" + ], + "neon": { + "common": [ + "src/core/NEON/kernels/convolution/common/padding.cpp", + "src/core/NEON/kernels/convolution/common/qasymm8.cpp", + "src/core/NEON/kernels/convolution/common/qsymm8.cpp", + "src/core/NEON/kernels/convolution/common/utils.cpp", + "src/core/NEON/kernels/arm_conv/addressing.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_common.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_strategies_common.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_u8q.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_u8s8u8q.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_s8q.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp32.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_packed_to_nhwc_3x3_s2_with_multiplier_output3x3_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_packed_to_nhwc_5x5_s1_with_multiplier_output2x4_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8q_packed_to_nhwc_3x3_s2_with_multiplier_output2x4_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8q_packed_to_nhwc_5x5_s1_with_multiplier_output4x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8q_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_packed_to_nhwc_3x3_s2_with_multiplier_output2x4_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_packed_to_nhwc_5x5_s1_with_multiplier_output4x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8s8u8q_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_generic_output9_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8q_nhwc_3x3_s1_output2x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8q_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8q_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8q_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8q_nhwc_generic_output9_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_s8qs_nhwc_3x3_s1_output2x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_nhwc_3x3_s1_output2x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8q_nhwc_generic_output9_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8s8u8q_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8s8u8q_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8s8u8q_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8s8u8q_nhwc_generic_output9_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8qa_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8qa_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_u8qa_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/interleaves/a64_s8q_3x3_dot.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/interleaves/a64_u8q_3x3_dot.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/interleaves/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/interleaves/generic_quantized_dot_product.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/premultiply.cpp", + "src/cpu/kernels/depthwiseconv2d/generic/neon/impl.cpp" + ], + "fp16":[ + "src/cpu/kernels/depthwiseconv2d/generic/neon/fp16.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp16_nhwc_generic_output9_mla_depthfirst/generic.cpp" + ], + "fp32":["src/cpu/kernels/depthwiseconv2d/generic/neon/fp32.cpp"], + "qasymm8":["src/cpu/kernels/depthwiseconv2d/generic/neon/qasymm8.cpp"], + "qasymm8_signed":["src/cpu/kernels/depthwiseconv2d/generic/neon/qasymm8_signed.cpp"] + }, + "sve": { + "common": [ + "src/core/NEON/kernels/arm_conv/depthwise/interleaves/sve_s8q_3x3_dot.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/interleaves/sve_u8q_3x3_dot.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_nhwc_generic_output9_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_packed_to_nhwc_3x3_s2_with_multiplier_output3x3_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_packed_to_nhwc_5x5_s1_with_multiplier_output2x4_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_fp32_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_s8q_nhwc_3x3_s1_output2x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_s8q_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_s8q_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_s8q_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_s8q_packed_to_nhwc_3x3_s2_with_multiplier_output2x4_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_s8q_packed_to_nhwc_5x5_s1_with_multiplier_output4x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_s8qs_nhwc_3x3_s1_output2x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8q_nhwc_3x3_s1_output2x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8q_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8q_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8q_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8q_packed_to_nhwc_3x3_s2_with_multiplier_output2x4_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8q_packed_to_nhwc_5x5_s1_with_multiplier_output4x2_dot_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8s8u8q_nhwc_3x3_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8s8u8q_nhwc_3x3_s2_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sve_u8s8u8q_nhwc_5x5_s1_output2x2_mla_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp16_nhwc_5x5_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_nhwc_3x3_s1_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_nhwc_3x3_s1_output3x3_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_nhwc_3x3_s1_output4x4_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_direct.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_nhwc_3x3_s2_output2x2_mla_depthfirst/generic_indirect.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_planar_3x3_s1_4rows_mla_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_planar_3x3_s2_4rows_mla_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_planar_5x5_s1_4rows_mla_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32_planar_5x5_s2_4rows_mla_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32bf16fp32_planar_3x3_s1_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32bf16fp32_planar_3x3_s2_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32bf16fp32_planar_5x5_s1_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_fp32bf16fp32_planar_5x5_s2_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_s8q_planar_3x3_s1_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_s8q_planar_3x3_s2_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_s8q_planar_5x5_s1_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_s8q_planar_5x5_s2_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_u8q_planar_3x3_s1_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_u8q_planar_3x3_s2_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_u8q_planar_5x5_s1_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_u8q_planar_5x5_s2_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_u8s8u8q_planar_3x3_s1_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_u8s8u8q_planar_3x3_s2_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_u8s8u8q_planar_5x5_s1_4rows_dot_za/generic.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/kernels/sme2_u8s8u8q_planar_5x5_s2_4rows_dot_za/generic.cpp" + ] + } + } + }, + "Dequantize": { + "files": { + "common": [ + "src/cpu/operators/CpuDequantize.cpp", + "src/cpu/kernels/CpuDequantizeKernel.cpp", + "src/runtime/NEON/functions/NEDequantizationLayer.cpp" + ] + } + }, + "DetectionPostProcess": { + "deps": [ "Dequantize" ], + "files": { + "common" : [ "src/runtime/NEON/functions/NEDetectionPostProcessLayer.cpp" ] + } + }, + "Conv3d": { + "deps": [ + "Activation" + ], + "files": { + "common": [ + "src/cpu/operators/CpuDirectConv3d.cpp", + "src/cpu/kernels/CpuDirectConv3dKernel.cpp", + "src/runtime/NEON/functions/NEConv3D.cpp" + ] + } + }, + "ElementwiseBinary": { + "files": { + "common": [ + "src/cpu/operators/CpuElementwise.cpp", + "src/cpu/kernels/CpuElementwiseKernel.cpp", + "src/runtime/NEON/functions/NEElementwiseOperations.cpp" + ], + "neon":{ + "fp32": ["src/cpu/kernels/elementwise_binary/generic/neon/fp32.cpp"], + "fp16": ["src/cpu/kernels/elementwise_binary/generic/neon/fp16.cpp"], + "integer": ["src/cpu/kernels/elementwise_binary/generic/neon/integer.cpp"], + "qasymm8": ["src/cpu/kernels/elementwise_binary/generic/neon/qasymm8.cpp"], + "qasymm8_signed": ["src/cpu/kernels/elementwise_binary/generic/neon/qasymm8_signed.cpp"] + }, + "sve": { + "common": ["src/cpu/kernels/elementwise_binary/generic/sve/impl.cpp" ], + "integer": ["src/cpu/kernels/elementwise_binary/generic/sve/integer.cpp"], + "fp32": ["src/cpu/kernels/elementwise_binary/generic/sve/fp32.cpp"], + "fp16": ["src/cpu/kernels/elementwise_binary/generic/sve/fp16.cpp"] - "neon": - { - "nchw" : [ - "src/core/cpu/kernels/pooling/neon/nchw/all.cpp" - ], - "fp32" : [ - "src/core/cpu/kernels/activation/neon/fp32.cpp", - "src/core/cpu/kernels/floor/neon/fp32.cpp", - "src/core/cpu/kernels/pooling/neon/fp32.cpp", - "src/core/NEON/kernels/batchnormalization/impl/NEON/fp32.cpp" - ], - "fp16" : [ - "src/core/cpu/kernels/activation/neon/fp16.cpp", - "src/core/cpu/kernels/floor/neon/fp16.cpp", - "src/core/cpu/kernels/pooling/neon/fp16.cpp", - "src/core/cpu/kernels/scale/neon/fp16.cpp", - "src/core/NEON/kernels/batchnormalization/impl/NEON/fp16.cpp" - ], - "qsymm16" : [ - "src/core/cpu/kernels/activation/neon/qsymm16.cpp", - "src/core/cpu/kernels/add/neon/qsymm16.cpp", - "src/core/cpu/kernels/sub/neon/qsymm16.cpp" + }, + "sve2":{ + "qasymm8": ["src/cpu/kernels/elementwise_binary/generic/sve2/qasymm8.cpp"], + "qasymm8_signed": ["src/cpu/kernels/elementwise_binary/generic/sve2/qasymm8_signed.cpp"] + } + } + }, + "ElementwiseUnary":{ + "deps": [ "LUT" ], + "files": { + "common": [ + "src/cpu/operators/CpuElementwiseUnary.cpp", + "src/cpu/kernels/CpuElementwiseUnaryKernel.cpp", + "src/runtime/NEON/functions/NEElementwiseUnaryLayer.cpp" + ], + "neon": { + "integer": ["src/cpu/kernels/elementwise_unary/generic/neon/integer.cpp"], + "fp32": ["src/cpu/kernels/elementwise_unary/generic/neon/fp32.cpp"], + "fp16": ["src/cpu/kernels/elementwise_unary/generic/neon/fp16.cpp"], + "qasymm8": [ + "src/cpu/kernels/elementwise_unary/generic/neon/q8.cpp", + "src/cpu/kernels/elementwise_unary/generic/neon/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/cpu/kernels/elementwise_unary/generic/neon/q8.cpp", + "src/cpu/kernels/elementwise_unary/generic/neon/qasymm8_signed.cpp" + ] + }, + "sve": { + "common": ["src/cpu/kernels/elementwise_unary/generic/sve/impl.cpp" ], + "integer": ["src/cpu/kernels/elementwise_unary/generic/sve/integer.cpp"], + "fp32": ["src/cpu/kernels/elementwise_unary/generic/sve/fp32.cpp"], + "fp16": ["src/cpu/kernels/elementwise_unary/generic/sve/fp16.cpp"] + }, + "sve2": { + "qasymm8": ["src/cpu/kernels/elementwise_unary/generic/sve2/q8.cpp"], + "qasymm8_signed": ["src/cpu/kernels/elementwise_unary/generic/sve2/q8.cpp"] + } + } + }, + "FFT1D": { + "deps": [ "Reduction" ], + "files": { + "common": [ + "src/core/NEON/kernels/NEFFTDigitReverseKernel.cpp", + "src/core/NEON/kernels/NEFFTRadixStageKernel.cpp", + "src/core/NEON/kernels/NEFFTScaleKernel.cpp", + "src/runtime/NEON/functions/NEFFT1D.cpp" + ] + } + }, + "FFT2D": { + "deps": [ "FFT1D" ], + "files": { + "common": [ + "src/runtime/NEON/functions/NEFFT2D.cpp" + ] + } + }, + "Fill": { + "files": { + "common": [ + "src/cpu/operators/CpuFill.cpp", + "src/cpu/kernels/CpuFillKernel.cpp", + "src/runtime/NEON/functions/NEFill.cpp" + ] + } + }, + "Flatten": { + "deps: ": [ "Reshape" ], + "files": { + "common": [ + "src/cpu/operators/CpuFlatten.cpp", + "src/runtime/NEON/functions/NEFlattenLayer.cpp" + ] + } + }, + "Floor": { + "files": { + "common": [ + "src/cpu/operators/CpuFloor.cpp", + "src/cpu/kernels/CpuFloorKernel.cpp", + "src/runtime/NEON/functions/NEFloor.cpp" + ], + "neon": { + "fp32": [ "src/cpu/kernels/floor/neon/fp32.cpp" ], + "fp16": [ "src/cpu/kernels/floor/neon/fp16.cpp" ] + } + } + }, + "FullyConnected": { + "deps": [ "Flatten", "Gemm", "Transpose"], + "files": { + "common": [ + "src/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.cpp", + "src/cpu/operators/CpuConvertFullyConnectedWeights.cpp", + "src/cpu/operators/CpuFullyConnected.cpp", + "src/runtime/NEON/functions/NEConvertFullyConnectedWeights.cpp", + "src/runtime/NEON/functions/NEFullyConnectedLayer.cpp" + ] + } + }, + "Gather": { + "files": { + "common": [ + "src/core/NEON/kernels/NEGatherKernel.cpp", + "src/runtime/NEON/functions/NEGather.cpp" + ] + } + }, + "Gemm": { + "deps": [ "Quantize", "Add"], + "files": { + "common": [ + "src/cpu/kernels/CpuConvertQuantizedSignednessKernel.cpp", + "src/cpu/kernels/CpuGemmMatrixAdditionKernel.cpp", + "src/cpu/kernels/CpuGemmMatrixMultiplyKernel.cpp", + "src/cpu/kernels/CpuGemmTranspose1xWKernel.cpp", + "src/cpu/kernels/CpuGemmInterleave4x4Kernel.cpp", + "src/cpu/kernels/CpuGemmLowpQuantizeDownInt32ScaleKernel.cpp", + "src/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToInt16ScaleByFixedPointKernel.cpp", + "src/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToInt8ScaleByFixedPointKernel.cpp", + "src/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel.cpp", + "src/cpu/kernels/CpuGemmLowpMatrixMultiplyKernel.cpp", + "src/cpu/kernels/CpuGemmLowpMatrixReductionKernel.cpp", + "src/cpu/kernels/CpuGemmLowpOffsetContributionOutputStageKernel.cpp", + "src/cpu/kernels/CpuGemmLowpOffsetContributionKernel.cpp", + "src/cpu/operators/CpuGemm.cpp", + "src/cpu/operators/CpuGemmLowpOutputStage.cpp", + "src/cpu/operators/CpuGemmLowpMatrixMultiplyCore.cpp", + "src/runtime/NEON/functions/NEGEMM.cpp", + "src/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.cpp", + "src/runtime/NEON/functions/NEGEMMLowpOutputStage.cpp" + ], + "neon": { + "common": [ + "src/core/NEON/kernels/arm_gemm/gemm_bf16.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_fp32.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_bf16bf16.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_int16.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_int8.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_s8fp32.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_qint8.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_quint8.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_uint16.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp", + "src/core/NEON/kernels/arm_gemm/interleave-8way.cpp", + "src/core/NEON/kernels/arm_gemm/interleave_indirect.cpp", + "src/core/NEON/kernels/arm_gemm/mergeresults-fp16.cpp", + "src/core/NEON/kernels/arm_gemm/mergeresults.cpp", + "src/core/NEON/kernels/arm_gemm/misc.cpp", + "src/core/NEON/kernels/arm_gemm/quantized.cpp", + "src/core/NEON/kernels/arm_gemm/rowsum_indirect_s8.cpp", + "src/core/NEON/kernels/arm_gemm/rowsum_indirect_u8.cpp", + "src/core/NEON/kernels/arm_gemm/transform.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_s16_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_s8_4x4/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_s8_8x12/a55r1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_s8_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_s8_8x12/x1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_u16_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_u8_4x4/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_u8_8x12/a55r1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_u8_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_gemm_u8_8x12/x1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hgemm_8x24/a55r1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hgemm_8x24/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hgemm_8x24/x1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_bf16fp32_dot_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_bf16fp32_mmla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp16_mla_6x32/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp16_mla_6x32/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp32_mla_4x24/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp32_mla_4x24/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp32_mla_6x16/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp32_mla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp32_mla_8x4/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp32_mla_8x4/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp32bf16fp32_mmla_4x24/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_fp32bf16fp32_mmla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8qa_dot_4x16/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8qa_dot_4x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8qa_mmla_4x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8qs_dot_6x16/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8qs_dot_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8qs_mmla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8s32_dot_6x16/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8s32_dot_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_s8s32_mmla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8qa_dot_4x16/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8qa_dot_4x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8qa_mmla_4x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8u32_dot_6x16/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8u32_dot_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8u32_mmla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_dot_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_dot_8x12/x1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_mmla_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_mmla_8x12/a510.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_s8s32_mmla_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_s8s32_mmla_8x12/a510.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_u8u32_mmla_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_u8u32_mmla_8x12/a510.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x12/a53.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x12/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x12/a55r1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x12/x1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x6/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_fp32_mla_6x4/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_fp32_mla_8x4/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_s8s32_dot_6x4/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_s8s32_dot_6x4/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_s8s32_dot_8x4/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_s8s32_dot_8x4/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_u8u32_dot_6x4/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_u8u32_dot_6x4/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_u8u32_dot_8x4/a55.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_smallK_hybrid_u8u32_dot_8x4/generic.cpp", + "src/cpu/kernels/gemm_matrix_mul/generic/neon/impl.cpp", + "src/cpu/kernels/gemm_matrix_add/generic/neon/impl.cpp" + ], + "fp32":["src/cpu/kernels/gemm_matrix_mul/generic/neon/fp32.cpp", + "src/cpu/kernels/gemm_matrix_add/generic/neon/fp32.cpp"], + "fp16":["src/cpu/kernels/gemm_matrix_mul/generic/neon/fp16.cpp", + "src/cpu/kernels/gemm_matrix_add/generic/neon/fp16.cpp"], + "estate32": [ + "src/core/NEON/kernels/arm_gemm/kernels/a32_sgemm_8x6/a53.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a32_sgemm_8x6/a55r1.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a32_sgemm_8x6/generic.cpp" + ], + "estate64": [ + "src/core/NEON/kernels/arm_gemm/kernels/a64_sgemv_pretransposed/generic.cpp" + ], + "fixed_format_kernels": [ + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffhybrid_bf16fp32_mmla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffhybrid_fp16_mla_6x32/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffhybrid_fp32_mla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffhybrid_fp32bf16fp32_mmla_4x24/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffhybrid_fp32bf16fp32_mmla_6x16/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_bf16fp32_dot_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_bf16fp32_mmla_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_fp16_mla_8x24/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_fp32_mla_8x12/generic.cpp" + ] + }, + "sve": { + "common": [ + "src/core/NEON/kernels/arm_gemm/interleave_indirect-sve.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_gemv_bf16fp32_dot_16VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_gemv_fp16fp32fp16_dot_16VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_gemv_fp32_mla_16VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_gemv_fp32bf16fp32_dot_16VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_gemv_s8qa_dot_16VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_gemv_u8qa_dot_16VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_bf16fp32_mopa_1VLx4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_bf16fp32_mopa_2VLx2VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_bf16fp32_mopa_4VLx1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_fp16fp32fp16_mopa_1VLx4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_fp16fp32fp16_mopa_2VLx2VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_fp16fp32fp16_mopa_4VLx1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_fp32_mopa_1VLx4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_fp32_mopa_2VLx2VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_fp32_mopa_4VLx1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8q_mopa_1VLx4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8q_mopa_2VLx2VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8q_mopa_4VLx1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8qfp32_mopa_1VLx4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8qfp32_mopa_2VLx2VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8qfp32_mopa_4VLx1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8s32_mopa_1VLx4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8s32_mopa_2VLx2VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_s8s32_mopa_4VLx1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_u8q_mopa_1VLx4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_u8q_mopa_2VLx2VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sme2_interleaved_nomerge_u8q_mopa_4VLx1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_bf16fp32_dot_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_bf16fp32_mmla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_fp16_mla_6x4VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_fp16_mla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_fp32_mla_6x4VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_fp32_mla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_fp32_mla_8x1VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_fp32_mla_8x1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_fp32bf16fp32_mmla_4x6VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_fp32bf16fp32_mmla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8qa_dot_4x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8qa_mmla_4x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8qs_dot_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8qs_mmla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8s32_dot_6x4VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8s32_dot_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8s32_mmla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8qa_dot_4x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8qa_mmla_4x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8u32_dot_6x4VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8u32_dot_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8u32_mmla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_dot_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_bf16fp32_mmla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_fp16_mla_8x3VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_fp16_mla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_fp32_mla_8x3VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_fp32_mla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_fp32_mmla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_s8s32_dot_8x3VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_s8s32_dot_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_s8s32_mmla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_u8u32_dot_8x3VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_u8u32_dot_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_u8u32_mmla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/mergeresults-sve.cpp", + "src/core/NEON/kernels/arm_gemm/transform-sve.cpp", + "src/core/NEON/kernels/arm_gemm/misc-sve.cpp" + ], + "fixed_format_kernels": [ + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_bf16fp32_mmla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp16_mla_6x4VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp16_mla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32_mla_6x4VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32_mla_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_bf16fp32_mmla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_fp16_mla_8x3VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_fp16_mla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_fp32_mla_8x3VL/a64fx.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_fp32_mla_8x3VL/generic.cpp" + ] + } + } + }, + "GenerateProposals": { + "deps": [ "BoundingBoxTransform", "Dequantize", "Pad", "Permute", "Quantize", "Reshape" ], + "files": { + "common": [ + "src/core/NEON/kernels/NEGenerateProposalsLayerKernel.cpp", + "src/runtime/NEON/functions/NEGenerateProposalsLayer.cpp" + ], + "neon":{ + "common":["src/cpu/kernels/genproposals/generic/neon/impl.cpp"], + "fp16":["src/cpu/kernels/genproposals/generic/neon/fp16.cpp"], + "fp32":["src/cpu/kernels/genproposals/generic/neon/fp32.cpp"], + "qsymm16":["src/cpu/kernels/genproposals/generic/neon/qsymm16.cpp"] + } + } + }, + "InstanceNormalize": { + "deps": [ "Permute", "Reduction" ], + "files": { + "common": [ + "src/core/NEON/kernels/NEInstanceNormalizationLayerKernel.cpp", + "src/runtime/NEON/functions/NEInstanceNormalizationLayer.cpp" + ], + "neon":{ + "common":["src/cpu/kernels/instancenorm/generic/neon/impl.cpp"], + "fp16":["src/cpu/kernels/instancenorm/generic/neon/fp16.cpp"], + "fp32":["src/cpu/kernels/instancenorm/generic/neon/fp32.cpp"] + } + } + }, + "L2Normalize": { + "deps": [ "Reduction" ], + "files": { + "common": [ + "src/core/NEON/kernels/NEL2NormalizeLayerKernel.cpp", + "src/runtime/NEON/functions/NEL2NormalizeLayer.cpp" + ], + "neon":{ + "fp32":["src/cpu/kernels/l2normlayer/generic/neon/fp32.cpp"], + "fp16":["src/cpu/kernels/l2normlayer/generic/neon/fp16.cpp"] + } + } + }, + "Logical": { + "files": { + "common": [ + "src/core/NEON/kernels/NELogicalKernel.cpp", + "src/runtime/NEON/functions/NELogical.cpp" + ] + } + }, + "LUT": { + "files": { + "common": [], + "neon":{ + "qasymm8": ["src/cpu/kernels/lut/generic/neon/u8.cpp"], + "qasymm8_signed": ["src/cpu/kernels/lut/generic/neon/u8.cpp"] + }, + "sve": { + "fp16": ["src/cpu/kernels/lut/generic/sve/u16.cpp"], + "qasymm16": ["src/cpu/kernels/lut/generic/sve/u16.cpp"], + "qasymm16_signed": ["src/cpu/kernels/lut/generic/sve/u16.cpp"] + }, + "sve2": { + "qasymm8": ["src/cpu/kernels/lut/generic/sve2/u8.cpp"], + "qasymm8_signed": ["src/cpu/kernels/lut/generic/sve2/u8.cpp"] + } + } + }, + "LSTM": { + "deps": [ + "Activation", + "Concatenate", + "Copy", + "Dequantize", + "ElementwiseBinary", + "Fill", + "FullyConnected", + "Gemm", + "MeanStdDevNormalize", + "Mul", + "Quantize", + "Slice", + "Transpose" + ], + "files": { + "common": [ + "src/core/NEON/kernels/NEQLSTMLayerNormalizationKernel.cpp", + "src/runtime/NEON/functions/NELSTMLayer.cpp", + "src/runtime/NEON/functions/NELSTMLayerQuantized.cpp", + "src/runtime/NEON/functions/NEQLSTMLayer.cpp" + ] + } + }, + "MaxUnpool2d": { + "deps": [ "Fill" ], + "files": { + "common": [ + "src/cpu/kernels/CpuMaxUnpoolingLayerKernel.cpp", + "src/runtime/NEON/functions/NEMaxUnpoolingLayer.cpp", + "src/cpu/operators/CpuMaxUnpooling.cpp" + ], + "neon":{ + "fp32":["src/cpu/kernels/maxunpool/generic/neon/fp32.cpp"], + "fp16":["src/cpu/kernels/maxunpool/generic/neon/fp16.cpp"], + "qasymm8":["src/cpu/kernels/maxunpool/generic/neon/qasymm8.cpp"], + "qasymm8_signed":[ "src/cpu/kernels/maxunpool/generic/neon/qasymm8_signed.cpp"] + } + } + }, + "Mean": { + "deps" : [ "Reduction" ], + "files": { + "common": [ "src/runtime/NEON/functions/NEReduceMean.cpp" ] + } + }, + "MeanStdDevNormalize": { + "files": { + "common": [ + "src/core/NEON/kernels/NEMeanStdDevNormalizationKernel.cpp", + "src/runtime/NEON/functions/NEMeanStdDevNormalizationLayer.cpp" + ], + "neon":{ + "common":["src/cpu/kernels/meanstddevnorm/generic/neon/impl.cpp"], + "fp32":["src/cpu/kernels/meanstddevnorm/generic/neon/fp32.cpp"], + "fp16":["src/cpu/kernels/meanstddevnorm/generic/neon/fp16.cpp"], + "qasymm8":["src/cpu/kernels/meanstddevnorm/generic/neon/qasymm8.cpp"] + } + } + }, + "MatMul" : { + "deps": [ "Transpose" ], + "files": { + "common": [ + "src/cpu/operators/CpuMatMul.cpp", + "src/runtime/NEON/functions/NEMatMul.cpp" + ] + } + }, + "Mul": { + "files": { + "common": [ + "src/cpu/operators/CpuMul.cpp", + "src/cpu/kernels/CpuMulKernel.cpp", + "src/runtime/NEON/functions/NEPixelWiseMultiplication.cpp" + ], + "neon":{ + "fp16":["src/cpu/kernels/mul/generic/neon/fp16.cpp"], + "fp32":["src/cpu/kernels/mul/generic/neon/fp32.cpp"] + } + } + }, + "Normalize": { + "deps": [ "Mul" ], + "files": { + "common": [ + "src/core/NEON/kernels/NENormalizationLayerKernel.cpp", + "src/runtime/NEON/functions/NENormalizationLayer.cpp" + ], + "neon":{ + "fp16":["src/cpu/kernels/norm_layer/generic/neon/fp16.cpp"], + "fp32":["src/cpu/kernels/norm_layer/generic/neon/fp32.cpp"] + } + } + }, + "Pad": { + "deps": [ "Concatenate", "Copy", "StridedSlice" ], + "files": { + "common": [ + "src/core/NEON/kernels/NEPadLayerKernel.cpp", + "src/runtime/NEON/functions/NEPadLayer.cpp" + ] + } + }, + "Permute": { + "files": { + "common": [ + "src/cpu/operators/CpuPermute.cpp", + "src/cpu/kernels/CpuPermuteKernel.cpp", + "src/runtime/NEON/functions/NEPermute.cpp" + ] + } + }, + "Pool2d": { + "files": { + "common": [ + "src/cpu/operators/CpuPool2d.cpp", + "src/cpu/kernels/CpuPool2dKernel.cpp", + "src/cpu/kernels/internal/CpuPool2dAssemblyWrapperKernel.cpp", + "src/runtime/NEON/functions/NEPoolingLayer.cpp" + ], + "neon": { + "common": [ + "src/core/NEON/kernels/arm_conv/pooling/kernels/cpp_nhwc_1x1_stride_any_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/pooling_fp32.cpp", + "src/core/NEON/kernels/arm_conv/pooling/pooling_s8.cpp", + "src/core/NEON/kernels/arm_conv/pooling/pooling_s8q.cpp", + "src/core/NEON/kernels/arm_conv/pooling/pooling_u8.cpp", + "src/core/NEON/kernels/arm_conv/pooling/pooling_u8q.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_fp32_nhwc_avg_3x3_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_fp32_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_fp32_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_fp32_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_s8_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_s8_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_s8_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_s8q_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_s8q_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_u8_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_u8_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_u8_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_u8q_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_u8q_nhwc_max_generic_depthfirst/generic.cpp" + ], + "nchw": [ "src/cpu/kernels/pool2d/neon/nchw/all.cpp" ], + "fp16": [ + "src/cpu/kernels/pool2d/neon/fp16.cpp", + "src/core/NEON/kernels/arm_conv/pooling/pooling_fp16.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_fp16_nhwc_avg_3x3_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_fp16_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_fp16_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/a64_fp16_nhwc_max_generic_depthfirst/generic.cpp" + ], + "fp32": [ "src/cpu/kernels/pool2d/neon/fp32.cpp" ], + "qasymm8":[ "src/cpu/kernels/pool2d/neon/qasymm8.cpp" ], + "qasymm8_signed":["src/cpu/kernels/pool2d/neon/qasymm8_signed.cpp"] + }, + "sve": { + "common": [ + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_fp16_nhwc_avg_3x3_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_fp16_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_fp16_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_fp16_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_fp32_nhwc_avg_3x3_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_fp32_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_fp32_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_fp32_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_s8_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_s8_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_s8_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_s8q_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_s8q_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_u8_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_u8_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_u8_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_u8q_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sve_u8q_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_fp16_nhwc_avg_3x3_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_fp16_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_fp16_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_fp16_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_fp32_nhwc_avg_3x3_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_fp32_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_fp32_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_fp32_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_s8_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_s8_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_s8_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_s8q_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_s8q_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_u8_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_u8_nhwc_max_2x2_s1_output2x2_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_u8_nhwc_max_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_u8q_nhwc_avg_generic_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/sme_u8q_nhwc_max_generic_depthfirst/generic.cpp" + ] + } + } + }, + "Pool3d": { + "files": { + "common": [ + "src/cpu/operators/CpuPool3d.cpp", + "src/cpu/kernels/CpuPool3dKernel.cpp", + "src/runtime/NEON/functions/NEPooling3dLayer.cpp" + ], + "neon": { + "fp16": [ "src/cpu/kernels/pool3d/neon/fp16.cpp" ], + "fp32": [ "src/cpu/kernels/pool3d/neon/fp32.cpp" ], + "qasymm8": [ "src/cpu/kernels/pool3d/neon/qasymm8.cpp" ], + "qasymm8_signed": [ "src/cpu/kernels/pool3d/neon/qasymm8_signed.cpp" ] + } + } + }, + "PRelu": { + "deps": [ "ElementwiseBinary" ], + "files": { + "common": [ + "src/runtime/NEON/functions/NEPReluLayer.cpp" + ] + } + }, + "PriorBox": { + "files": { + "common": [ + "src/core/NEON/kernels/NEPriorBoxLayerKernel.cpp", + "src/runtime/NEON/functions/NEPriorBoxLayer.cpp" + ] + } + }, + "Quantize": { + "files": { + "common": [ + "src/cpu/operators/CpuQuantize.cpp", + "src/cpu/kernels/CpuQuantizeKernel.cpp", + "src/runtime/NEON/functions/NEQuantizationLayer.cpp" + ] + } + }, + "Range": { + "files": { + "common": [ + "src/core/NEON/kernels/NERangeKernel.cpp", + "src/runtime/NEON/functions/NERange.cpp" + ], + "neon": { + "fp32": [ "src/cpu/kernels/range/generic/neon/fp32.cpp" ], + "fp16": [ "src/cpu/kernels/range/generic/neon/fp16.cpp" ], + "integer": [ "src/cpu/kernels/range/generic/neon/integer.cpp" ] + } + } + }, + "Reduction":{ + "deps": [ "Reshape" ], + "files": { + "common": [ + "src/core/NEON/kernels/NEReductionOperationKernel.cpp", + "src/runtime/NEON/functions/NEReductionOperation.cpp" + ] + } + }, + "Reorg": { + "files": { + "common": [ + "src/core/NEON/kernels/NEReorgLayerKernel.cpp", + "src/runtime/NEON/functions/NEReorgLayer.cpp" + ] + } + }, + "Reshape": { + "files": { + "common": [ + "src/cpu/operators/CpuReshape.cpp", + "src/cpu/kernels/CpuReshapeKernel.cpp", + "src/runtime/NEON/functions/NEReshapeLayer.cpp" + ] + } + }, + "Reorder": { + "files": { + "common": [ + "src/runtime/NEON/functions/NEReorderLayer.cpp", + "src/core/NEON/kernels/NEReorderKernel.cpp" + ] + } + }, + "Reverse": { + "files": { + "common": [ + "src/core/NEON/kernels/NEReverseKernel.cpp", + "src/runtime/NEON/functions/NEReverse.cpp" + ] + } + }, + "RNN": { + "deps": [ "Activation", "Add", "FullyConnected", "Gemm"], + "files": { + "common": [ "src/runtime/NEON/functions/NERNNLayer.cpp" ] + } + }, + "ROIAlign": { + "files": { + "common": [ + "src/core/NEON/kernels/NEROIAlignLayerKernel.cpp", + "src/runtime/NEON/functions/NEROIAlignLayer.cpp" + ], + "neon":{ + "fp32":["src/cpu/kernels/roialign/generic/neon/fp32.cpp"], + "fp16":["src/cpu/kernels/roialign/generic/neon/fp16.cpp"], + "qasymm8":["src/cpu/kernels/roialign/generic/neon/qasymm8.cpp"], + "qasymm8_signed":["src/cpu/kernels/roialign/generic/neon/qasymm8_signed.cpp"] + } + } + }, + "ROIPool2d": { + "files": { + "common": [ + "src/core/NEON/kernels/NEROIPoolingLayerKernel.cpp", + "src/runtime/NEON/functions/NEROIPoolingLayer.cpp" + ] + } + }, + "Scale": { + "files": { + "common": [ + "src/cpu/operators/CpuScale.cpp", + "src/cpu/kernels/CpuScaleKernel.cpp", + "src/runtime/NEON/functions/NEScale.cpp" + ], + "sve": { + "fp16": [ "src/cpu/kernels/scale/sve/fp16.cpp" ], + "fp32": [ "src/cpu/kernels/scale/sve/fp32.cpp" ], + "integer": [ "src/cpu/kernels/scale/sve/integer.cpp" ], + "qasymm8": [ "src/cpu/kernels/scale/sve/qasymm8.cpp" ], + "qasymm8_signed": [ "src/cpu/kernels/scale/sve/qasymm8_signed.cpp" ] + }, + "neon": { + "fp16": [ "src/cpu/kernels/scale/neon/fp16.cpp" ], + "integer": [ "src/cpu/kernels/scale/neon/integer.cpp" ], + "qasymm8": [ "src/cpu/kernels/scale/neon/qasymm8.cpp", "src/cpu/kernels/scale/neon/integer.cpp" ], + "qasymm8_signed": [ "src/cpu/kernels/scale/neon/qasymm8_signed.cpp", "src/cpu/kernels/scale/neon/integer.cpp" ] + } + } + }, + "Select": { + "files": { + "common": [ + "src/core/NEON/kernels/NESelectKernel.cpp", + "src/runtime/NEON/functions/NESelect.cpp" + ], + "neon": { + "fp32": [ "src/cpu/kernels/select/generic/neon/fp32.cpp" ], + "fp16": [ "src/cpu/kernels/select/generic/neon/fp16.cpp" ], + "integer": [ "src/cpu/kernels/select/generic/neon/integer.cpp" ] + } + } + }, + "Slice": { + "deps": [ "StridedSlice" ], + "files": { + "common": [ "src/runtime/NEON/functions/NESlice.cpp" ] + } + }, + "Softmax": { + "deps": [ + "Permute" + ], + "files": { + "common": [ + "src/cpu/operators/CpuSoftmax.cpp", + "src/cpu/kernels/CpuSoftmaxKernel.cpp", + "src/runtime/NEON/functions/NESoftmaxLayer.cpp" + ], + "neon":{ + "common":["src/cpu/kernels/softmax/generic/neon/impl.cpp"], + "fp32": ["src/cpu/kernels/softmax/generic/neon/fp32.cpp"], + "fp16": ["src/cpu/kernels/softmax/generic/neon/fp16.cpp"], + "qasymm8":[ "src/cpu/kernels/softmax/generic/neon/qasymm8.cpp"], + "qasymm8_signed":["src/cpu/kernels/softmax/generic/neon/qasymm8_signed.cpp"] + }, + "sve": { + "common": [ "src/cpu/kernels/softmax/generic/sve/impl.cpp" ] + }, + "sve2":{ + "common" :["src/cpu/kernels/softmax/generic/sve2/impl.cpp"], + "fp32" :["src/cpu/kernels/softmax/generic/sme2/fp32.cpp"], + "fp16" :["src/cpu/kernels/softmax/generic/sme2/fp16.cpp"] + } + } + }, + "SpaceToBatch": { + "files": { + "common": [ + "src/core/NEON/kernels/NESpaceToBatchLayerKernel.cpp", + "src/runtime/NEON/functions/NESpaceToBatchLayer.cpp" + ] + } + }, + "SpaceToDepth": { + "files": { + "common": [ + "src/core/NEON/kernels/NESpaceToDepthLayerKernel.cpp", + "src/runtime/NEON/functions/NESpaceToDepthLayer.cpp" + ] + } + }, + "Split": { + "deps": [ "StridedSlice" ], + "files": { + "common": [ + "src/runtime/NEON/functions/NESplit.cpp" + ] + } + }, + "Stack": { + "files": { + "common": [ + "src/core/NEON/kernels/NEStackLayerKernel.cpp", + "src/runtime/NEON/functions/NEStackLayer.cpp" + ] + } + }, + "StridedSlice": { + "files": { + "common": [ + "src/core/NEON/kernels/NEStridedSliceKernel.cpp", + "src/runtime/NEON/functions/NEStridedSlice.cpp" + ] + } + }, + "Sub": { + "files": { + "common": [ + "src/cpu/operators/CpuSub.cpp", + "src/cpu/kernels/CpuSubKernel.cpp", + "src/runtime/NEON/functions/NEArithmeticSubtraction.cpp" + ], + "neon": { + "fp16":["src/cpu/kernels/sub/neon/fp16.cpp"], + "qasymm8": ["src/cpu/kernels/sub/neon/qasymm8.cpp"], + "qasymm8_signed": ["src/cpu/kernels/sub/neon/qasymm8_signed.cpp"], + "qsymm16": ["src/cpu/kernels/sub/neon/qsymm16.cpp"] + } + } + }, - ], - "qasymm8" : [ - "src/core/cpu/kernels/activation/neon/qasymm8.cpp", - "src/core/cpu/kernels/add/neon/qasymm8.cpp", - "src/core/cpu/kernels/pooling/neon/qasymm8.cpp", - "src/core/cpu/kernels/scale/neon/qasymm8.cpp", - "src/core/cpu/kernels/sub/neon/qasymm8.cpp" - ], - "qasymm8_signed" : [ - "src/core/cpu/kernels/activation/neon/qasymm8_signed.cpp", - "src/core/cpu/kernels/add/neon/qasymm8_signed.cpp", - "src/core/cpu/kernels/pooling/neon/qasymm8_signed.cpp", - "src/core/cpu/kernels/scale/neon/qasymm8_signed.cpp", - "src/core/cpu/kernels/sub/neon/qasymm8_signed.cpp" - ], - "integer" : [ - "src/core/cpu/kernels/sub/neon/integer.cpp", - "src/core/cpu/kernels/add/neon/integer.cpp" - ] - } + "Tile": { + "files": { + "common": [ + "src/core/NEON/kernels/NETileKernel.cpp", + "src/runtime/NEON/functions/NETile.cpp" + ] } + }, + "Transpose": { + "files": { + "common": [ + "src/cpu/kernels/CpuTransposeKernel.cpp", + "src/cpu/operators/CpuTranspose.cpp", + "src/runtime/NEON/functions/NETranspose.cpp" + ] + } + }, + "Unstack": { + "deps": [ "StridedSlice" ], + "files": { + "common": [ "src/runtime/NEON/functions/NEUnstack.cpp" ] + } + } + } + }, + "experimental": { + "dynamic_fusion": { + "common": [ + "src/dynamic_fusion/runtime/gpu/cl/ClKernelRuntime.cpp", + "src/dynamic_fusion/runtime/gpu/cl/ClWorkloadRuntime.cpp", + "src/dynamic_fusion/sketch/attributes/CastAttributes.cpp", + "src/dynamic_fusion/sketch/attributes/ClampAttributes.cpp", + "src/dynamic_fusion/sketch/attributes/Conv2dAttributes.cpp", + "src/dynamic_fusion/sketch/attributes/DepthwiseConv2dAttributes.cpp", + "src/dynamic_fusion/sketch/attributes/MatMulAttributes.cpp", + "src/dynamic_fusion/sketch/attributes/Pool2dAttributes.cpp", + "src/dynamic_fusion/sketch/attributes/ResizeAttributes.cpp", + "src/dynamic_fusion/sketch/attributes/SoftmaxAttributes.cpp", + "src/dynamic_fusion/sketch/attributes/ReshapeAttributes.cpp", + "src/dynamic_fusion/sketch/gpu/GpuKernelComponentGraph.cpp", + "src/dynamic_fusion/sketch/gpu/GpuKernelComponentGroup.cpp", + "src/dynamic_fusion/sketch/gpu/GpuKernelComponentStream.cpp", + "src/dynamic_fusion/sketch/gpu/GpuLogicalKernel.cpp", + "src/dynamic_fusion/sketch/gpu/GpuOperatorGroup.cpp", + "src/dynamic_fusion/sketch/gpu/GpuWorkloadContext.cpp", + "src/dynamic_fusion/sketch/gpu/GpuWorkloadSketch.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentActivation.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentCast.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDepthwiseConv2d.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDirectConv2d.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentMatMul.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentPool2d.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentElementwiseBinary.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentReshape.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentResize.cpp", + "src/dynamic_fusion/sketch/gpu/components/cl/ClComponentStore.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuAdd.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuCast.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuClamp.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuConv2d.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuDepthwiseConv2d.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuMatMul.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuMul.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuReshape.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuPool2d.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuOutput.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuResize.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuSigmoid.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuSoftmax.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuSub.cpp", + "src/dynamic_fusion/sketch/gpu/operators/GpuTanh.cpp", + "src/dynamic_fusion/sketch/gpu/operators/internal/GpuElementwiseBinaryCommon.cpp" + ], + "ckw_driver": [ + "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwComponentArgument.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwDriver.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwActivation.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwCast.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwDepthwiseConv2d.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwDirectConv2d.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwElementwiseBinary.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwPool2d.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwMatMul.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwResize.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/GpuCkwStore.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwScopedKernelWriter.cpp", + "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwVariableTable.cpp", + "src/dynamic_fusion/runtime/gpu/cl/ckw_driver/GpuCkwKernelArgumentsHelpers.cpp" + ] } -}
\ No newline at end of file + } +} |