diff options
Diffstat (limited to 'filelist.json')
| -rw-r--r-- | filelist.json | 2060 |
1 files changed, 1769 insertions, 291 deletions
diff --git a/filelist.json b/filelist.json index e30426bd19..0e17944e1d 100644 --- a/filelist.json +++ b/filelist.json @@ -1,297 +1,1775 @@ { - "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" + ], + "c_api": { + "cpu": [ + "src/c/AclContext.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" + ] + }, + "gpu": { + "high_priority": [ + "Activation", + "DepthwiseConv2d", + "DirectConv2d", + "Permute", + "Pool2d", + "Reshape" + ], + "operators": { + "Activation": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClActivation.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClActivationKernel.cpp" + ] + } + }, + "Add": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClAdd.cpp" + ] + } + }, + "Cast": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClCast.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClCastKernel.cpp" + ] + } + }, + "Concatenate": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClConcatenate.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClWidthConcatenate4TensorsKernel.cpp", + "src/core/gpu/cl/kernels/ClWidthConcatenate2TensorsKernel.cpp", + "src/core/gpu/cl/kernels/ClWidthConcatenateKernel.cpp", + "src/core/gpu/cl/kernels/ClHeightConcatenateKernel.cpp", + "src/core/gpu/cl/kernels/ClDepthConcatenateKernel.cpp", + "src/core/gpu/cl/kernels/ClBatchConcatenateKernel.cpp" + ] + } + }, + "DirectConv2d": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClDirectConv2d.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClDirectConv2dKernel.cpp" + ] + } + }, + "ConvertFullyConnectedWeights": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClConvertFullyConnectedWeights.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClConvertFullyConnectedWeightsKernel.cpp" + ] + } + }, + "Permute": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClPermute.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClPermuteKernel.cpp" + ] + } + }, + "Pool2d": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClPool2d.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClPool2dKernel.cpp" + ] + } + }, + "PRelu": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClPRelu.cpp" + ] + } + }, + "Reshape": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClReshape.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClReshapeKernel.cpp" + ] + } + }, + "Copy": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClCopy.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClCopyKernel.cpp" + ] + } + }, + "Crop": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClCrop.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClCropKernel.cpp" + ] + } + }, + "Dequantize": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClDequantize.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClDequantizeKernel.cpp" + ] + } + }, + "Elementwise": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClElementwiseOperations.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClElementwiseKernel.cpp" + ] + } + }, + "ElementwiseUnary": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClElementwiseUnary.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.cpp" + ] + } + }, + "Fill": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClFill.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClFillKernel.cpp" + ] + } + }, + "Flatten": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClFlatten.cpp" + ] + } + }, + "Floor": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClFloor.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClFloorKernel.cpp" + ] + } + }, + "GEMM": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClGemm.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClGemmMatrixMultiplyNativeKernel.cpp", + "src/core/gpu/cl/kernels/gemm/ClGemmHelpers.cpp", + "src/core/gpu/cl/kernels/gemm/native/ClGemmDefaultConfigNativeBifrost.cpp", + "src/core/gpu/cl/kernels/gemm/native/ClGemmDefaultConfigNativeMidgard.cpp", + "src/core/gpu/cl/kernels/gemm/native/ClGemmDefaultConfigNativeValhall.cpp", + "src/core/gpu/cl/kernels/ClGemmMatrixMultiplyReshapedKernel.cpp", + "src/core/gpu/cl/kernels/gemm/reshaped/ClGemmDefaultConfigReshapedBifrost.cpp", + "src/core/gpu/cl/kernels/gemm/reshaped/ClGemmDefaultConfigReshapedValhall.cpp", + "src/core/gpu/cl/kernels/ClGemmMatrixMultiplyReshapedOnlyRhsKernel.cpp", + "src/core/gpu/cl/kernels/gemm/reshaped_only_rhs/ClGemmDefaultConfigReshapedRhsOnlyBifrost.cpp", + "src/core/gpu/cl/kernels/gemm/reshaped_only_rhs/ClGemmDefaultConfigReshapedRhsOnlyValhall.cpp", + "src/core/gpu/cl/kernels/ClGemmMatrixMultiplyKernel.cpp", + "src/core/gpu/cl/kernels/ClGemmReshapeRhsMatrixKernel.cpp", + "src/core/gpu/cl/kernels/ClGemmReshapeLhsMatrixKernel.cpp" + ] + } + }, + "Mul": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClMul.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClMulKernel.cpp" + ] + } + }, + "Quantize": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClQuantize.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClQuantizeKernel.cpp" + ] + } + }, + "Scale": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClScale.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClScaleKernel.cpp" + ] + } + }, + "Softmax": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClSoftmax.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClSoftmaxKernel.cpp" + ] + } + }, + "Sub": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClSub.cpp" + ] + } + }, + "Transpose": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClTranspose.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClTransposeKernel.cpp" + ] + } + }, + "GenerateProposals": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGenerateProposalsLayerKernel.cpp" + ] + } + }, + "ArgMinMax": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLArgMinMaxLayerKernel.cpp" + ] + } + }, + "BatchNormalization": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLBatchNormalizationLayerKernel.cpp" + ] + } + }, + "BatchToSpace": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLBatchToSpaceLayerKernel.cpp" + ] + } + }, + "Bitwise": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLBitwiseKernel.cpp" + ] + } + }, + "BoundingBoxTransform": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLBoundingBoxTransformKernel.cpp" + ] + } + }, + "ChannelShuffleLayer": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLChannelShuffleLayerKernel.cpp" + ] + } + }, + "GEMMConv2d": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLCol2ImKernel.cpp", + "src/core/CL/kernels/CLIm2ColKernel.cpp" + ] + } + }, + "Comparison": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLComparisonKernel.cpp" + ] + } + }, + "DeconvolutionLayerUpsample": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLDeconvolutionLayerUpsampleKernel.cpp" + ] + } + }, + "DeconvolutionReshapeOutput": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLDeconvolutionReshapeOutputKernel.cpp" + ] + } + }, + "DepthToSpace": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLDepthToSpaceLayerKernel.cpp" + ] + } + }, + "DepthwiseConvolutionLayer3x3NCHW": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLDepthwiseConvolutionLayer3x3NCHWKernel.cpp" + ] + } + }, + "DepthwiseConvolutionLayer3x3NHWC": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLDepthwiseConvolutionLayer3x3NHWCKernel.cpp" + ] + } + }, + "DepthwiseConvolutionLayerNative": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLDepthwiseConvolutionLayerNativeKernel.cpp" + ] + } + }, + "FFTDigitReverse": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLFFTDigitReverseKernel.cpp" + ] + } + }, + "FFTRadixStage": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLFFTRadixStageKernel.cpp" + ] + } + }, + "FFTScale": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLFFTScaleKernel.cpp" + ] + } + }, + "FillBorder": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLFillBorderKernel.cpp" + ] + } + }, + "FuseBatchNormalization": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLFuseBatchNormalizationKernel.cpp" + ] + } + }, + "Gather": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGatherKernel.cpp" + ] + } + }, + "GEMMLowpMatrixMultiplyNative": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpMatrixMultiplyNativeKernel.cpp" + ] + } + }, + "GEMMLowpMatrixMultiplyReshaped": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.cpp" + ] + } + }, + "GEMMLowpMatrixMultiplyReshapedOnlyRHS": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel.cpp" + ] + } + }, + "GEMMLowpOffsetContribution": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpOffsetContributionKernel.cpp" + ] + } + }, + "GEMMLowpOffsetContributionOutputStage": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpOffsetContributionOutputStageKernel.cpp" + ] + } + }, + "GEMMLowpQuantizeDownInt32ScaleByFixedPoint": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ScaleByFixedPointKernel.cpp" + ] + } + }, + "GEMMLowpQuantizeDownInt32ScaleByFloat": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ScaleByFloatKernel.cpp" + ] + } + }, + "GEMMLowpQuantizeDownInt32Scale": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ScaleKernel.cpp" + ] + } + }, + "GEMMLowpReduction": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLGEMMLowpReductionKernel.cpp" + ] + } + }, + "InstanceNormalization": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLInstanceNormalizationLayerKernel.cpp" + ] + } + }, + "L2Normalize": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLL2NormalizeLayerKernel.cpp" + ] + } + }, + "LogicalNot": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClLogicalNot.cpp" + ] + } + }, + "MaxUnpooling": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLMaxUnpoolingLayerKernel.cpp" + ] + } + }, + "MeanStdDevNormalization": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLMeanStdDevNormalizationKernel.cpp" + ] + } + }, + "MinMax": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLMinMaxLayerKernel.cpp" + ] + } + }, + "Normalization": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLNormalizationLayerKernel.cpp" + ] + } + }, + "NormalizePlanarYUV": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLNormalizePlanarYUVLayerKernel.cpp" + ] + } + }, + "Pad": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLPadLayerKernel.cpp" + ] + } + }, + "PriorBox": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLPriorBoxLayerKernel.cpp" + ] + } + }, + "QLSTMLayerNormalization": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLQLSTMLayerNormalizationKernel.cpp" + ] + } + }, + "Range": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLRangeKernel.cpp" + ] + } + }, + "ReductionOperation": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLReductionOperationKernel.cpp" + ] + } + }, + "Remap": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLRemapKernel.cpp" + ] + } + }, + "Reorg": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLReorgLayerKernel.cpp" + ] + } + }, + "Reverse": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLReverseKernel.cpp" + ] + } + }, + "ROIAlign": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLROIAlignLayerKernel.cpp" + ] + } + }, + "ROIPooling": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLROIPoolingLayerKernel.cpp" + ] + } + }, + "Select": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLSelectKernel.cpp" + ] + } + }, + "SpaceToBatch": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLSpaceToBatchLayerKernel.cpp" + ] + } + }, + "SpaceToDepth": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLSpaceToDepthLayerKernel.cpp" + ] + } + }, + "Stack": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLStackLayerKernel.cpp" + ] + } + }, + "StridedSlice": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLStridedSliceKernel.cpp" + ] + } + }, + "Tile": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLTileKernel.cpp" + ] + } + }, + "WeightsReshape": { + "files": { + "kernel": [ + "src/core/CL/kernels/CLWeightsReshapeKernel.cpp" + ] + } + }, + "WinogradConv2d": { + "files": { + "operator": [ + "src/runtime/gpu/cl/operators/ClWinogradConv2d.cpp" + ], + "kernel": [ + "src/core/gpu/cl/kernels/ClWinogradFilterTransformKernel.cpp", + "src/core/gpu/cl/kernels/ClWinogradInputTransformKernel.cpp", + "src/core/gpu/cl/kernels/ClWinogradOutputTransformKernel.cpp" + ] + } + } + } + }, + "cpu": { + "all": [ + "src/cpu/CpuContext.cpp", + "src/cpu/CpuQueue.cpp", + "src/cpu/CpuTensor.cpp" + ], + "high_priority": [ + "Activation", + "DepthwiseConv2d", + "DirectConv2d", + "Permute", + "Pool2d", + "Reshape" + ], + "operators": { + "Activation": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuActivation.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuActivationKernel.cpp" + ], + "sve": { + "fp32": [ + "src/core/cpu/kernels/activation/sve/fp32.cpp" + ], + "fp16": [ + "src/core/cpu/kernels/activation/sve/fp16.cpp" + ], + "qsymm16": [ + "src/core/cpu/kernels/activation/sve/qsymm16.cpp" + ], + "qasymm8": [ + "src/core/cpu/kernels/activation/sve/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/core/cpu/kernels/activation/sve/qasymm8_signed.cpp" + ] + }, + "neon": { + "fp32": [ + "src/core/cpu/kernels/activation/neon/fp32.cpp" + ], + "fp16": [ + "src/core/cpu/kernels/activation/neon/fp16.cpp" + ], + "qsymm16": [ + "src/core/cpu/kernels/activation/neon/qsymm16.cpp" + ], + "qasymm8": [ + "src/core/cpu/kernels/activation/neon/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/core/cpu/kernels/activation/neon/qasymm8_signed.cpp" + ] + } + } + }, + "Add": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuAdd.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuAddKernel.cpp" + ], + "sve": { + "all": [ + "src/core/cpu/kernels/add/sve/impl.cpp" + ], + "qsymm16": [ + "src/core/cpu/kernels/add/sve/qsymm16.cpp" + ], + "qasymm8": [ + "src/core/cpu/kernels/add/sve/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/core/cpu/kernels/add/sve/qasymm8_signed.cpp" + ], + "integer": [ + "src/core/cpu/kernels/add/sve/integer.cpp" + ] + }, + "neon": { + "qsymm16": [ + "src/core/cpu/kernels/add/neon/qsymm16.cpp" + ], + "qasymm8": [ + "src/core/cpu/kernels/add/neon/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/core/cpu/kernels/add/neon/qasymm8_signed.cpp" + ], + "integer": [ + "src/core/cpu/kernels/add/neon/integer.cpp" + ] + } + } + }, + "BatchNorm": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEBatchNormalizationLayerKernel.cpp" + ], + "sve": { + "fp32": [ + "src/core/NEON/kernels/batchnormalization/impl/SVE/fp32.cpp" + ], + "fp16": [ + "src/core/NEON/kernels/batchnormalization/impl/SVE/fp16.cpp" + ] + }, + "neon": { + "fp32": [ + "src/core/NEON/kernels/batchnormalization/impl/NEON/fp32.cpp" + ], + "fp16": [ + "src/core/NEON/kernels/batchnormalization/impl/NEON/fp16.cpp" + ] + } + } + }, + "BatchToSpace": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEBatchToSpaceLayerKernel.cpp" + ] + } + }, + "BitwiseAnd": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEBitwiseAndKernel.cpp" + ] + } + }, + "BitwiseNot": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEBitwiseNotKernel.cpp" + ] + } + }, + "BitwiseOr": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEBitwiseOrKernel.cpp" + ] + } + }, + "BitwiseXor": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEBitwiseXorKernel.cpp" + ] + } + }, + "BoundingBoxTransform": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEBoundingBoxTransformKernel.cpp" + ] + } + }, + "ChannelShuffleLayer": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEChannelShuffleLayerKernel.cpp" + ] + } + }, + "Col2Im": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NECol2ImKernel.cpp" + ] + } + }, + "Cast": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuCast.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuCastKernel.cpp" + ] + } + }, + "Concatenate": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuConcatenate.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuConcatenateWidthKernel.cpp", + "src/core/cpu/kernels/CpuConcatenateBatchKernel.cpp", + "src/core/cpu/kernels/CpuConcatenateDepthKernel.cpp", + "src/core/cpu/kernels/CpuConcatenateHeightKernel.cpp" + ] + } + }, + "ConvertFullyConnectedWeights": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuConvertFullyConnectedWeights.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuConvertFullyConnectedWeightsKernel.cpp" + ] + } + }, + "ConvertQuantizedSignedness": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEConvertQuantizedSignednessKernel.cpp" + ] + } + }, + "Copy": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuCopy.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuCopyKernel.cpp" + ] + } + }, + "Crop": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NECropKernel.cpp" + ] + } + }, + "DepthwiseConv2d": { + "deps": [ + "Activation", + "Permute" + ], + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuDepthwiseConv2d.cpp", + "src/runtime/cpu/operators/CpuDepthwiseConv2dAssemblyDispatch.cpp", + "src/core/cpu/kernels/internal/CpuDepthwiseConv2dAssemblyWrapperKernel.cpp" + ], + "kernel": [ + "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/cpu/kernels/CpuDepthwiseConv2dNativeKernel.cpp" + ], + "sve": { + "all": [ + "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.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_output2x2_mla_depthfirst_strided/generic.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" + ] + }, + "neon": { + "estate64": [ + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp16.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_fp32.cpp", + "src/core/NEON/kernels/arm_conv/depthwise/depthwise_s8q.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/interleaves/8b_mla.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/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_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", + "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_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_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_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_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_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_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_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_u8s8u8q_packed_to_nhwc_generic_with_multiplier_output2x8_mla_depthfirst/generic.cpp" + ] + } + } + }, + "DepthToSpaceLayer": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEDepthToSpaceLayerKernel.cpp" + ] + } + }, + "Dequantize": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuDequantize.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuDequantizeKernel.cpp" + ] + } + }, + "DirectConv2d": { + "deps": [ + "Activation", + "FillBorder" + ], + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuDirectConv2d.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuDirectConv2dKernel.cpp", + "src/core/cpu/kernels/CpuDirectConv2dOutputStageKernel.cpp" + ] + } + }, + "Elementwise": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuElementwise.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuElementwiseKernel.cpp" + ], + "sve": { + "all": [ + "src/core/cpu/kernels/elementwise/sve/elementwise.cpp" + ] + } + } + }, + "ElementwiseUnary": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuElementwiseUnary.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuElementwiseUnaryKernel.cpp" + ], + "sve": { + "all": [ + "src/core/cpu/kernels/elementwise/sve/elementwise_unary.cpp" + ] + } + } + }, + "FFT1D": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEFFTDigitReverseKernel.cpp", + "src/core/NEON/kernels/NEFFTRadixStageKernel.cpp", + "src/core/NEON/kernels/NEFFTScaleKernel.cpp" + ] + } + }, + "FillBorder": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEFillBorderKernel.cpp" + ] + } + }, + "Flatten": { + "deps: ": [ + "Reshape" + ], + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuFlatten.cpp" + ] + } + }, + "Fill": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuFill.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuFillKernel.cpp" + ] + } + }, + "Floor": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuFloor.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuFloorKernel.cpp" + ], + "neon": { + "fp32": [ + "src/core/cpu/kernels/floor/neon/fp32.cpp" + ], + "fp16": [ + "src/core/cpu/kernels/floor/neon/fp16.cpp" + ] + } + } + }, + "FuseBatchNormalization": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEFuseBatchNormalizationKernel.cpp" + ] + } + }, + "GEMM": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp", + "src/core/NEON/kernels/NEGEMMMatrixMultiplyKernel.cpp", + "src/core/NEON/kernels/NEGEMMTranspose1xWKernel.cpp", + "src/core/NEON/kernels/NEGEMMInterleave4x4Kernel.cpp" + ] + } + }, + "GEMMLowp": { + "deps": [ + "GemmAssemblyDispatch" + ], + "files": { + "operator" : ["src/runtime/cpu/operators/CpuGemmLowpOutputStage.cpp"], + "kernel": [ + "src/core/NEON/kernels/NEGEMMLowpMatrixMultiplyKernel.cpp", + "src/core/cpu/kernels/CpuGemmLowpQuantizeDownInt32ScaleKernel.cpp", + "src/core/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToInt16ScaleByFixedPointKernel.cpp", + "src/core/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToInt8ScaleByFixedPointKernel.cpp", + "src/core/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel.cpp", + "src/core/NEON/kernels/NEGEMMLowpReductionKernel.cpp", + "src/core/NEON/kernels/NEGEMMLowpOffsetContributionOutputStageKernel.cpp", + "src/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.cpp" + ] + } + }, + "GEMMConvolution": { + "deps": [ + "Activation", + "Col2Im", + "Reshape", + "Im2Col", + "GEMMLowpOffsetContributionOutputStage", + "ConvertQuantizedSignedness" + ], + "files": { + "kernel": [ + "src/core/NEON/kernels/NEWeightsReshapeKernel.cpp" + ] + } + }, + "GemmAssemblyDispatch": { + "files": { + "operator": [ + "src/runtime/cpu/operators/internal/CpuGemmAssemblyDispatch.cpp" + ], + "kernel": [ + "src/core/NEON/kernels/arm_gemm/gemm_fp16.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_bf16.cpp", + "src/core/NEON/kernels/arm_gemm/gemm_fp32.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_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/mergeresults.cpp", + "src/core/NEON/kernels/arm_gemm/interleave_indirect.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" + ], + "neon": { + "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_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_gemv_fp32_mla_32/generic.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_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_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_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_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_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_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_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_interleaved_bf16fp32_dot_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_mmla_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_s8s32_mmla_8x12/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/a64_interleaved_u8u32_mmla_8x12/generic.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_sgemv_pretransposed/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" + ] + }, + "sve": { + "all": [ + "src/core/NEON/kernels/arm_gemm/mergeresults-sve.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_gemv_fp32_mla_8VL/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_fp16_mla_6x4VL/generic.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/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_s8qs_dot_6x4VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8s32_dot_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_u8u32_dot_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/generic.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/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/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_interleaved_u8u32_mmla_8x3VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_smallK_hybrid_fp32_mla_8x1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_smallK_hybrid_s8s32_dot_8x1VL/generic.cpp", + "src/core/NEON/kernels/arm_gemm/kernels/sve_smallK_hybrid_u8u32_dot_8x1VL/generic.cpp" + ] + } + } + }, + "GemmDirectConv2d": { + "deps": [ + "Activation", + "GemmAssemblyDispatch", + "Permute" + ], + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuGemmDirectConv2d.cpp" + ] + } + }, + "Mul": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuMul.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuMulKernel.cpp" + ] + } + }, + "Quantize": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuQuantize.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuQuantizeKernel.cpp" + ] + } + }, + "Reshape": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuReshape.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuReshapeKernel.cpp" + ] + } + }, + "Gather": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEGatherKernel.cpp" + ] + } + }, + "GenerateProposalsLayer": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEGenerateProposalsLayerKernel.cpp" + ] + } + }, + "Im2Col": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEIm2ColKernel.cpp" + ] + } + }, + "InstanceNormalization": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEInstanceNormalizationLayerKernel.cpp" + ] + } + }, + "L2Normalize": { + "deps": [ + "Reduction" + ], + "files": { + "kernel": [ + "src/core/NEON/kernels/NEL2NormalizeLayerKernel.cpp" + ] + } + }, + "Logical": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NELogicalKernel.cpp" + ] + } + }, + "MaxUnpooling": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEMaxUnpoolingLayerKernel.cpp" + ] + } + }, + "MeanStdDevNormalization": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEMeanStdDevNormalizationKernel.cpp" + ] + } + }, + "MinMax": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEMinMaxLayerKernel.cpp" + ] + } + }, + "Normalization": { + "deps": [ + "PixelWiseMultiplication" + ], + "files": { + "kernel": [ + "src/core/NEON/kernels/NENormalizationLayerKernel.cpp" + ] + } + }, + "Pad": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEPadLayerKernel.cpp" + ] + } + }, + "Permute": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuPermute.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuPermuteKernel.cpp" + ] + } + }, + "Pool2d": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuPool2d.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuPool2dKernel.cpp", + "src/core/cpu/kernels/internal/CpuPool2dAssemblyWrapperKernel.cpp", + "src/core/NEON/kernels/arm_conv/pooling/kernels/cpp_nhwc_1x1_stride_any_depthfirst/generic.cpp", + "src/core/NEON/kernels/arm_conv/pooling/pooling_fp16.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" + ], + "neon": { + "nchw": [ + "src/core/cpu/kernels/pooling/neon/nchw/all.cpp" + ], + "fp32": [ + "src/core/cpu/kernels/pooling/neon/fp32.cpp" + ], + "fp16": [ + "src/core/cpu/kernels/pooling/neon/fp16.cpp" + ], + "qasymm8": [ + "src/core/cpu/kernels/pooling/neon/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/core/cpu/kernels/pooling/neon/qasymm8_signed.cpp" + ], + "estate64": [ + "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", + "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" + ] + }, + "sve": { + "all": [ + "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_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_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" + ] + } + } + }, + "PriorBox": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEPriorBoxLayerKernel.cpp" + ] + } + }, + "QLSTMLayerNormalization": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEQLSTMLayerNormalizationKernel.cpp" + ] + } + }, + "Range": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NERangeKernel.cpp" + ] + } + }, + "ReductionOperation": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEReductionOperationKernel.cpp" + ] + } + }, + "Remap": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NERemapKernel.cpp" + ] + } + }, + "Reorg": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEReorgLayerKernel.cpp" + ] + } + }, + "Reverse": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEReverseKernel.cpp" + ] + } + }, + "ROIAlign": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEROIAlignLayerKernel.cpp" + ] + } + }, + "ROIPooling": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEROIPoolingLayerKernel.cpp" + ] + } + }, + "Select": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NESelectKernel.cpp" + ] + } + }, + "SpaceToBatch": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NESpaceToBatchLayerKernel.cpp" + ] + } + }, + "SpaceToDepth": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NESpaceToDepthLayerKernel.cpp" + ] + } + }, + "Stack": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEStackLayerKernel.cpp" + ] + } + }, + "StridedSlice": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NEStridedSliceKernel.cpp" + ] + } + }, + "Scale": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuScale.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuScaleKernel.cpp" + ], + "sve": { + "fp32": [ + "src/core/cpu/kernels/scale/sve/fp32.cpp" + ], + "fp16": [ + "src/core/cpu/kernels/scale/sve/fp16.cpp" + ], + "qasymm8": [ + "src/core/cpu/kernels/scale/sve/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/core/cpu/kernels/scale/sve/qasymm8_signed.cpp" + ], + "integer": [ + "src/core/cpu/kernels/scale/sve/integer.cpp" + ] + }, + "neon": { + "fp16": [ + "src/core/cpu/kernels/scale/neon/fp16.cpp" + ], + "qasymm8": [ + "src/core/cpu/kernels/scale/neon/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/core/cpu/kernels/scale/neon/qasymm8_signed.cpp" + ], + "integer": [ + "src/core/cpu/kernels/scale/neon/integer.cpp" + ] + } + } + }, + "Softmax": { + "deps": [ + "Permute" + ], + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuSoftmax.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuSoftmaxKernel.cpp" + ], + "sve": { + "all": [ + "src/core/cpu/kernels/softmax/impl/sve/impl.cpp" + ] + } + } + }, + "Sub": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuSub.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuSubKernel.cpp" + ], + "neon": { + "qsymm16": [ + "src/core/cpu/kernels/sub/neon/qsymm16.cpp" + ], + "qasymm8": [ + "src/core/cpu/kernels/sub/neon/qasymm8.cpp" + ], + "qasymm8_signed": [ + "src/core/cpu/kernels/sub/neon/qasymm8_signed.cpp" + ], + "integer": [ + "src/core/cpu/kernels/sub/neon/integer.cpp" + ] + } + } + }, + "Transpose": { + "files": { + "operator": [ + "src/runtime/cpu/operators/CpuTranspose.cpp" + ], + "kernel": [ + "src/core/cpu/kernels/CpuTransposeKernel.cpp" + ] + } + }, + "Tile": { + "files": { + "kernel": [ + "src/core/NEON/kernels/NETileKernel.cpp" + ] + } + }, + "WinogradConvolution": { + "deps": [ + "Activation", + "Permute" ], - "gpu": [ - "src/c/cl/AclOpenClExt.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/CpuGemmLowpOutputStage.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/CpuGemmLowpQuantizeDownInt32ScaleKernel.cpp", - "src/core/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToInt16ScaleByFixedPointKernel.cpp", - "src/core/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToInt8ScaleByFixedPointKernel.cpp", - "src/core/cpu/kernels/CpuGemmLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel.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" - ] - }, - - "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" - ] - }, - - "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" - - ], - "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" - ] - } + "files": { + "kernel": [ + "src/core/NEON/kernels/NEWinogradConvolutionLayerKernel.cpp", + "src/core/NEON/kernels/convolution/winograd/padding.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/input_1x8_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/input_4x4_fp16_fp16_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/input_4x4_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/input_6x6_fp16_fp16_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/input_6x6_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/output_2_7_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/output_2x2_3x3_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/output_2x2_5x5_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/output_4_5_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/output_4x4_3x3_fp16_fp16_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/output_4x4_3x3_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/output_6_3_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/weights_2_7_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/weights_2x2_3x3_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/weights_2x2_5x5_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/weights_4_5_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/weights_4x4_3x3_fp16_fp16_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/weights_4x4_3x3_fp32_fp32_integers.cpp", + "src/core/NEON/kernels/convolution/winograd/winograd_transforms/weights_6_3_fp32_fp32_integers.cpp" + ] } + } } + } }
\ No newline at end of file |