diff options
Diffstat (limited to 'arm_compute/runtime/NEON/functions')
66 files changed, 824 insertions, 359 deletions
diff --git a/arm_compute/runtime/NEON/functions/NEActivationLayer.h b/arm_compute/runtime/NEON/functions/NEActivationLayer.h index 9992de2af8..5584fdc783 100644 --- a/arm_compute/runtime/NEON/functions/NEActivationLayer.h +++ b/arm_compute/runtime/NEON/functions/NEActivationLayer.h @@ -24,10 +24,9 @@ #ifndef ARM_COMPUTE_NEACTIVATIONLAYER_H #define ARM_COMPUTE_NEACTIVATIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/IRuntimeContext.h" #include <memory> @@ -102,5 +101,5 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} // namespace arm_computes +} // namespace arm_compute #endif /* ARM_COMPUTE_NEACTIVATIONLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEAddMulAdd.h b/arm_compute/runtime/NEON/functions/NEAddMulAdd.h index e5e85542f8..6c65c055dd 100644 --- a/arm_compute/runtime/NEON/functions/NEAddMulAdd.h +++ b/arm_compute/runtime/NEON/functions/NEAddMulAdd.h @@ -81,19 +81,28 @@ public: * @param[in] act_info (Optional) Activation layer information in case of a fused activation. * */ - void configure(ITensor *input1, ITensor *input2, ITensor *bn_mul, ITensor *bn_add, - ITensor *add_output, ITensor *final_output, - ConvertPolicy policy, const ActivationLayerInfo &act_info); + void configure(ITensor *input1, + ITensor *input2, + ITensor *bn_mul, + ITensor *bn_add, + ITensor *add_output, + ITensor *final_output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info); /** Static function to check if given info will lead to a valid configuration of @ref NEAddMulAdd * * Similar to @ref NEAddMulAdd::configure() except the arguments are @ref ITensorInfo * instead of @ref ITensor * * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, - const ITensorInfo *bn_mul, const ITensorInfo *bn_add, - const ITensorInfo *add_output, const ITensorInfo *final_output, - ConvertPolicy policy, const ActivationLayerInfo &act_info); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *bn_mul, + const ITensorInfo *bn_add, + const ITensorInfo *add_output, + const ITensorInfo *final_output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h b/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h index b0d710d517..73a43dbc44 100644 --- a/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h +++ b/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h @@ -27,6 +27,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -74,7 +75,11 @@ public: * @param[in] policy Policy to use to handle overflow. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(const ITensor *input1, const ITensor *input2, ITensor *output, ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const ITensor *input1, + const ITensor *input2, + ITensor *output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEArithmeticAddition * * @param[in] input1 First tensor input info. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/QSYMM16/F16/S32/F32 @@ -85,7 +90,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h b/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h index 6fbe9ad450..3e4f6356c5 100644 --- a/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h +++ b/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h @@ -80,7 +80,11 @@ public: * @param[in] policy Policy to use to handle overflow. Convert policy cannot be WRAP if datatype is quantized. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(const ITensor *input1, const ITensor *input2, ITensor *output, ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const ITensor *input1, + const ITensor *input2, + ITensor *output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEArithmeticSubtraction * * @param[in] input1 First tensor input. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/S32/F16/F32 @@ -91,7 +95,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h b/arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h index ec00fbdbf2..99e2dcadbb 100644 --- a/arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h +++ b/arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h @@ -81,7 +81,13 @@ public: * @param[in] epsilon (Optional) Small value to avoid division with zero. Default value is 0.001f. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Only RELU, BOUNDED_RELU and LU_BOUNDED_RELU supported. */ - void configure(ITensor *input, ITensor *output, const ITensor *mean, const ITensor *var, const ITensor *beta = nullptr, const ITensor *gamma = nullptr, float epsilon = 0.001f, + void configure(ITensor *input, + ITensor *output, + const ITensor *mean, + const ITensor *var, + const ITensor *beta = nullptr, + const ITensor *gamma = nullptr, + float epsilon = 0.001f, ActivationLayerInfo act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEBatchNormalizationLayer * @@ -98,10 +104,14 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, - const ITensorInfo *mean, const ITensorInfo *var, - const ITensorInfo *beta = nullptr, const ITensorInfo *gamma = nullptr, - float epsilon = 0.001f, ActivationLayerInfo act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const ITensorInfo *mean, + const ITensorInfo *var, + const ITensorInfo *beta = nullptr, + const ITensorInfo *gamma = nullptr, + float epsilon = 0.001f, + ActivationLayerInfo act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -109,5 +119,5 @@ public: private: std::unique_ptr<NEBatchNormalizationLayerKernel> _norm_kernel; /**< Batch normalization layer kernel */ }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_NEBATCHNORMALIZATIONLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h b/arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h index b33ba435a8..ebed0bea29 100644 --- a/arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h +++ b/arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEBATCHTOSPACELAYER_H #define ARM_COMPUTE_NEBATCHTOSPACELAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/NEON/INESimpleFunctionNoBorder.h" namespace arm_compute @@ -77,7 +76,11 @@ public: * @param[out] output Tensor output. Data types supported: same as @p input * @param[in] crop_info Specifies how the output shape is cropped after batch to space is performed */ - void configure(const ITensor *input, int32_t block_shape_x, int32_t block_shape_y, ITensor *output, const CropInfo &crop_info = CropInfo{}); + void configure(const ITensor *input, + int32_t block_shape_x, + int32_t block_shape_y, + ITensor *output, + const CropInfo &crop_info = CropInfo{}); /** Static function to check if given info will lead to a valid configuration of @ref CLBatchToSpaceLayer * * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All. @@ -99,7 +102,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, int32_t block_shape_x, int32_t block_shape_y, const ITensorInfo *output, const CropInfo &crop_info = CropInfo{}); + static Status validate(const ITensorInfo *input, + int32_t block_shape_x, + int32_t block_shape_y, + const ITensorInfo *output, + const CropInfo &crop_info = CropInfo{}); }; } // namespace arm_compute #endif /* ARM_COMPUTE_NEBATCHTOSPACELAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEBoundingBoxTransform.h b/arm_compute/runtime/NEON/functions/NEBoundingBoxTransform.h index 2a196a2de5..aa41fc0df2 100644 --- a/arm_compute/runtime/NEON/functions/NEBoundingBoxTransform.h +++ b/arm_compute/runtime/NEON/functions/NEBoundingBoxTransform.h @@ -57,7 +57,8 @@ public: * * @note Only single image prediction is supported. Height and Width (and scale) of the image will be contained in the BoundingBoxTransformInfo struct. */ - void configure(const ITensor *boxes, ITensor *pred_boxes, const ITensor *deltas, const BoundingBoxTransformInfo &info); + void + configure(const ITensor *boxes, ITensor *pred_boxes, const ITensor *deltas, const BoundingBoxTransformInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref NEBoundingBoxTransform * @@ -71,7 +72,10 @@ public: * * @return a Status */ - static Status validate(const ITensorInfo *boxes, const ITensorInfo *pred_boxes, const ITensorInfo *deltas, const BoundingBoxTransformInfo &info); + static Status validate(const ITensorInfo *boxes, + const ITensorInfo *pred_boxes, + const ITensorInfo *deltas, + const BoundingBoxTransformInfo &info); }; } // namespace arm_compute #endif /* ARM_COMPUTE_NEBOUNDINGBOXTRANSFORM_H */ diff --git a/arm_compute/runtime/NEON/functions/NECast.h b/arm_compute/runtime/NEON/functions/NECast.h index 821249c142..43cae777f6 100644 --- a/arm_compute/runtime/NEON/functions/NECast.h +++ b/arm_compute/runtime/NEON/functions/NECast.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NECAST_H #define ARM_COMPUTE_NECAST_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> diff --git a/arm_compute/runtime/NEON/functions/NEConcatenateLayer.h b/arm_compute/runtime/NEON/functions/NEConcatenateLayer.h index dd1c709d76..1600f85488 100644 --- a/arm_compute/runtime/NEON/functions/NEConcatenateLayer.h +++ b/arm_compute/runtime/NEON/functions/NEConcatenateLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NECONCATENATELAYER_H #define ARM_COMPUTE_NECONCATENATELAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -87,7 +86,8 @@ public: * * @return a status */ - static Status validate(const std::vector<const ITensorInfo *> &inputs_vector, const ITensorInfo *output, size_t axis); + static Status + validate(const std::vector<const ITensorInfo *> &inputs_vector, const ITensorInfo *output, size_t axis); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEConv3D.h b/arm_compute/runtime/NEON/functions/NEConv3D.h index 2a3c5351b0..525f37f3e7 100644 --- a/arm_compute/runtime/NEON/functions/NEConv3D.h +++ b/arm_compute/runtime/NEON/functions/NEConv3D.h @@ -24,11 +24,10 @@ #ifndef ARM_COMPUTE_NECONV3D_H #define ARM_COMPUTE_NECONV3D_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/ITensorInfo.h" #include "arm_compute/core/Types.h" #include "arm_compute/runtime/FunctionDescriptors.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -76,14 +75,19 @@ public: * @param[out] output Destination tensor. 4 lower dimensions represent a single output [OFM, width, height, depth], while the rest represent batch of outputs. * @param[in] conv_info Contains padding, stride, acitvation information described in @ref Conv3dInfo. */ - void configure(ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const Conv3dInfo &conv_info); + void configure( + ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const Conv3dInfo &conv_info); /** Static function to check if given info will lead to a valid configuration * * Similar to NEConv3D::configure() * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const Conv3dInfo &conv_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const Conv3dInfo &conv_info); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEConvertFullyConnectedWeights.h b/arm_compute/runtime/NEON/functions/NEConvertFullyConnectedWeights.h index a892d3036b..dc6b22d717 100644 --- a/arm_compute/runtime/NEON/functions/NEConvertFullyConnectedWeights.h +++ b/arm_compute/runtime/NEON/functions/NEConvertFullyConnectedWeights.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NECONVERTFULLYCONNECTEDWEIGHTS_H #define ARM_COMPUTE_NECONVERTFULLYCONNECTEDWEIGHTS_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" namespace arm_compute { @@ -66,7 +65,8 @@ public: * @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer). * @param[in] data_layout The data layout the weights have been trained in. */ - void configure(const ITensor *input, ITensor *output, const TensorShape &original_input_shape, DataLayout data_layout); + void + configure(const ITensor *input, ITensor *output, const TensorShape &original_input_shape, DataLayout data_layout); /** Static function to check if given info will lead to a valid configuration of @ref NEConvertFullyConnectedWeights * * @param[in] input Source weights tensor info to convert. Must be 2 dimensional. Data types supported: All. @@ -76,7 +76,10 @@ public: * * @return A Status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const TensorShape &original_input_shape, DataLayout data_layout); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const TensorShape &original_input_shape, + DataLayout data_layout); // Inherited methods overriden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h index 4dd76d082b..cdf0f652e1 100644 --- a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h @@ -24,11 +24,10 @@ #ifndef ARM_COMPUTE_NECONVOLUTIONLAYER_H #define ARM_COMPUTE_NECONVOLUTIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/ITensorInfo.h" #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/MemoryGroup.h" #include <memory> @@ -119,8 +118,16 @@ public: * available which may introduce a drop of accuracy as well. Default is false * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution. num_groups != 1 is not supported */ - void configure(ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info, const WeightsInfo &weights_info = WeightsInfo(), - const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false, unsigned int num_groups = 1); + void configure(ITensor *input, + const ITensor *weights, + const ITensor *biases, + ITensor *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false, + unsigned int num_groups = 1); /** Static function to check if given info will lead to a valid configuration of @ref NEConvolutionLayer * * @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM], @@ -143,9 +150,16 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - const WeightsInfo &weights_info = WeightsInfo(), const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false, - unsigned int num_groups = 1); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false, + unsigned int num_groups = 1); /** Static function to check if given info will return the convolution called by @ref NEConvolutionLayer * * @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM], @@ -165,8 +179,14 @@ public: * * @return the Convolution Method Hint */ - static ConvolutionMethod get_convolution_method(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *output, const PadStrideInfo &conv_info, - const WeightsInfo &weights_info = WeightsInfo(), const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + static ConvolutionMethod get_convolution_method(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); // Inherited methods overridden: void run() override; void prepare() override; diff --git a/arm_compute/runtime/NEON/functions/NECopy.h b/arm_compute/runtime/NEON/functions/NECopy.h index ee02c259f4..840c03e968 100644 --- a/arm_compute/runtime/NEON/functions/NECopy.h +++ b/arm_compute/runtime/NEON/functions/NECopy.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NECOPY_H #define ARM_COMPUTE_NECOPY_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> diff --git a/arm_compute/runtime/NEON/functions/NECropResize.h b/arm_compute/runtime/NEON/functions/NECropResize.h index 143bbbc6f1..f806762158 100644 --- a/arm_compute/runtime/NEON/functions/NECropResize.h +++ b/arm_compute/runtime/NEON/functions/NECropResize.h @@ -75,8 +75,13 @@ public: * @param[in] method The policy to be used when resizing image. Default is bilinear. * @param[in] extrapolation_value Value to be used for values outside of the image for cropping and resizing. Default is 0. */ - void configure(const ITensor *input, const ITensor *boxes, const ITensor *box_ind, ITensor *output, Coordinates2D crop_size, - InterpolationPolicy method = InterpolationPolicy::BILINEAR, float extrapolation_value = 0); + void configure(const ITensor *input, + const ITensor *boxes, + const ITensor *box_ind, + ITensor *output, + Coordinates2D crop_size, + InterpolationPolicy method = InterpolationPolicy::BILINEAR, + float extrapolation_value = 0); /** Static function to check if given info will lead to a valid configuration of @ref NESlice * @@ -96,8 +101,13 @@ public: * * @return A status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *boxes, const ITensorInfo *box_ind, const ITensorInfo *output, - Coordinates2D crop_size, InterpolationPolicy method, float extrapolation_value); + static Status validate(const ITensorInfo *input, + const ITensorInfo *boxes, + const ITensorInfo *box_ind, + const ITensorInfo *output, + Coordinates2D crop_size, + InterpolationPolicy method, + float extrapolation_value); void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEDeconvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEDeconvolutionLayer.h index cdc3a636b0..aabe42f928 100644 --- a/arm_compute/runtime/NEON/functions/NEDeconvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDeconvolutionLayer.h @@ -24,15 +24,14 @@ #ifndef ARM_COMPUTE_NEDECONVOLUTIONLAYER_H #define ARM_COMPUTE_NEDECONVOLUTIONLAYER_H -#include "arm_compute/runtime/CPP/functions/CPPUpsample.h" -#include "arm_compute/runtime/NEON/functions/NEConvolutionLayer.h" -#include "arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h" -#include "arm_compute/runtime/NEON/functions/NEReverse.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/CPP/functions/CPPUpsample.h" #include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/IMemoryManager.h" #include "arm_compute/runtime/MemoryGroup.h" +#include "arm_compute/runtime/NEON/functions/NEConvolutionLayer.h" +#include "arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h" +#include "arm_compute/runtime/NEON/functions/NEReverse.h" #include "arm_compute/runtime/Tensor.h" #include <memory> @@ -117,7 +116,13 @@ public: * the GEMM convolution. * */ - void configure(ITensor *input, const ITensor *weights, const ITensor *bias, ITensor *output, const PadStrideInfo &info, bool enable_fast_math = false, const WeightsInfo &weights_info = WeightsInfo()); + void configure(ITensor *input, + const ITensor *weights, + const ITensor *bias, + ITensor *output, + const PadStrideInfo &info, + bool enable_fast_math = false, + const WeightsInfo &weights_info = WeightsInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEDeconvolutionLayer * * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. @@ -134,8 +139,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *bias, const ITensorInfo *output, const PadStrideInfo &info, - bool enable_fast_math = false, const WeightsInfo &weights_info = WeightsInfo()); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *bias, + const ITensorInfo *output, + const PadStrideInfo &info, + bool enable_fast_math = false, + const WeightsInfo &weights_info = WeightsInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h b/arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h index eb0724ae12..7bfdfbd13d 100644 --- a/arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEDEPTHCONVERT_H #define ARM_COMPUTE_NEDEPTHCONVERT_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -84,7 +83,8 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, ConvertPolicy policy, uint32_t shift = 0); + static Status + validate(const ITensorInfo *input, const ITensorInfo *output, ConvertPolicy policy, uint32_t shift = 0); // Inherited methods overridden void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEDepthToSpaceLayer.h b/arm_compute/runtime/NEON/functions/NEDepthToSpaceLayer.h index b9bdcd1f11..c7df29a704 100644 --- a/arm_compute/runtime/NEON/functions/NEDepthToSpaceLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDepthToSpaceLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEDEPTHTOSPACELAYER_H #define ARM_COMPUTE_NEDEPTHTOSPACELAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/NEON/INESimpleFunctionNoBorder.h" namespace arm_compute diff --git a/arm_compute/runtime/NEON/functions/NEDepthwiseConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEDepthwiseConvolutionLayer.h index 6f2ec8cddb..6ad5aa7bfa 100644 --- a/arm_compute/runtime/NEON/functions/NEDepthwiseConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDepthwiseConvolutionLayer.h @@ -28,6 +28,7 @@ #include "arm_compute/runtime/MemoryGroup.h" #include "arm_compute/runtime/NEON/functions/NEActivationLayer.h" #include "arm_compute/runtime/NEON/functions/NEPermute.h" + #include <memory> namespace arm_compute @@ -80,8 +81,14 @@ public: * @param[in] act_info (Optional) Activation layer information in case of a fused activation. * @param[in] dilation (Optional) Dilation, in elements, across x and y. Defaults to (1, 1). */ - void configure(ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, const ActivationLayerInfo &act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + void configure(ITensor *input, + const ITensor *weights, + const ITensor *biases, + ITensor *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); /** Static function to check if given info will lead to a valid configuration of @ref NEDepthwiseConvolutionLayer * @@ -98,8 +105,14 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, const ActivationLayerInfo &act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); // Inherited methods overriden: void run() override; @@ -127,9 +140,11 @@ private: /** Default move constructor */ NEDepthwiseConvolutionLayerOptimizedInternal(NEDepthwiseConvolutionLayerOptimizedInternal &&) = default; /** Prevent instances of this class from being copied (As this class contains pointers) */ - NEDepthwiseConvolutionLayerOptimizedInternal &operator=(const NEDepthwiseConvolutionLayerOptimizedInternal &) = delete; + NEDepthwiseConvolutionLayerOptimizedInternal & + operator=(const NEDepthwiseConvolutionLayerOptimizedInternal &) = delete; /** Default move assignment operator */ - NEDepthwiseConvolutionLayerOptimizedInternal &operator=(NEDepthwiseConvolutionLayerOptimizedInternal &&) = default; + NEDepthwiseConvolutionLayerOptimizedInternal & + operator=(NEDepthwiseConvolutionLayerOptimizedInternal &&) = default; /** Default destructor */ ~NEDepthwiseConvolutionLayerOptimizedInternal() = default; /** Initialize the function's source, destination, kernels and border_size. @@ -144,8 +159,14 @@ private: * @param[in] act_info (Optional) Activation layer information in case of a fused activation. * @param[in] dilation (Optional) Dilation, in elements, across x and y. Defaults to (1, 1). */ - void configure(ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, const ActivationLayerInfo &act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + void configure(ITensor *input, + const ITensor *weights, + const ITensor *biases, + ITensor *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); /** Static function to check if given info will lead to a valid configuration of @ref NEDepthwiseConvolutionLayer3x3 * @@ -161,8 +182,14 @@ private: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, const ActivationLayerInfo &act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); // Inherited methods overriden: void run() override; @@ -207,8 +234,14 @@ private: * @param[in] act_info (Optional) Activation layer information in case of a fused activation. * @param[in] dilation (Optional) Dilation, in elements, across x and y. Defaults to (1, 1). */ - void configure(ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, const ActivationLayerInfo &act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + void configure(ITensor *input, + const ITensor *weights, + const ITensor *biases, + ITensor *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); /** Static function to check if given info will lead to a valid configuration of @ref NEDepthwiseConvolutionLayerGeneric * @@ -225,8 +258,14 @@ private: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, const ActivationLayerInfo &act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); // Inherited methods overriden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEDetectionPostProcessLayer.h b/arm_compute/runtime/NEON/functions/NEDetectionPostProcessLayer.h index 2affa8d49e..7a94833d10 100644 --- a/arm_compute/runtime/NEON/functions/NEDetectionPostProcessLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDetectionPostProcessLayer.h @@ -24,13 +24,12 @@ #ifndef ARM_COMPUTE_NE_DETECTION_POSTPROCESS_H #define ARM_COMPUTE_NE_DETECTION_POSTPROCESS_H -#include "arm_compute/runtime/NEON/INESimpleFunction.h" - #include "arm_compute/core/Types.h" #include "arm_compute/runtime/CPP/functions/CPPDetectionPostProcessLayer.h" #include "arm_compute/runtime/IMemoryManager.h" #include "arm_compute/runtime/MemoryGroup.h" #include "arm_compute/runtime/NEON/functions/NEDequantizationLayer.h" +#include "arm_compute/runtime/NEON/INESimpleFunction.h" #include "arm_compute/runtime/Tensor.h" #include <map> @@ -78,8 +77,14 @@ public: * * @note Output contains all the detections. Of those, only the ones selected by the valid region are valid. */ - void configure(const ITensor *input_box_encoding, const ITensor *input_score, const ITensor *input_anchors, - ITensor *output_boxes, ITensor *output_classes, ITensor *output_scores, ITensor *num_detection, DetectionPostProcessLayerInfo info = DetectionPostProcessLayerInfo()); + void configure(const ITensor *input_box_encoding, + const ITensor *input_score, + const ITensor *input_anchors, + ITensor *output_boxes, + ITensor *output_classes, + ITensor *output_scores, + ITensor *num_detection, + DetectionPostProcessLayerInfo info = DetectionPostProcessLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEDetectionPostProcessLayer * * @param[in] input_box_encoding The bounding box input tensor info. Data types supported: QASYMM8/QASYMM8_SIGNED/F32. @@ -93,8 +98,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input_box_encoding, const ITensorInfo *input_class_score, const ITensorInfo *input_anchors, - ITensorInfo *output_boxes, ITensorInfo *output_classes, ITensorInfo *output_scores, ITensorInfo *num_detection, + static Status validate(const ITensorInfo *input_box_encoding, + const ITensorInfo *input_class_score, + const ITensorInfo *input_anchors, + ITensorInfo *output_boxes, + ITensorInfo *output_classes, + ITensorInfo *output_scores, + ITensorInfo *num_detection, DetectionPostProcessLayerInfo info = DetectionPostProcessLayerInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h index 8db7e6596b..3ae3b2a15c 100644 --- a/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDirectConvolutionLayer.h @@ -85,7 +85,12 @@ public: * @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ITensor *input, const ITensor *weights, const ITensor *bias, ITensor *output, const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ITensor *input, + const ITensor *weights, + const ITensor *bias, + ITensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEDirectConvolutionLayer * * @note: DirectConvolution only works in the following configurations: @@ -106,7 +111,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *bias, const ITensorInfo *output, const PadStrideInfo &conv_info, + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *bias, + const ITensorInfo *output, + const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: diff --git a/arm_compute/runtime/NEON/functions/NEElementwiseOperations.h b/arm_compute/runtime/NEON/functions/NEElementwiseOperations.h index bfcd221e17..ebf2277d1f 100644 --- a/arm_compute/runtime/NEON/functions/NEElementwiseOperations.h +++ b/arm_compute/runtime/NEON/functions/NEElementwiseOperations.h @@ -73,7 +73,10 @@ public: * @param[out] output Output tensor. Data types supported: Same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(ITensor *input1, ITensor *input2, ITensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ITensor *input1, + ITensor *input2, + ITensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref cpu::kernels::CpuArithmeticKernel for max * * @param[in] input1 First tensor input info. Data types supported: QASYMM8/QASYMM8_SIGNED/S16/F16/S32/F32. @@ -83,7 +86,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -133,7 +139,10 @@ public: * @param[out] output Output tensor. Data types supported: Same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(ITensor *input1, ITensor *input2, ITensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ITensor *input1, + ITensor *input2, + ITensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref cpu::kernels::CpuArithmeticKernel for min * * @param[in] input1 First tensor input info. Data types supported: QASYMM8/QASYMM8_SIGNED/S16/F16/S32/F32. @@ -143,7 +152,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -193,7 +205,10 @@ public: * @param[out] output Output tensor. Data types supported: Same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(ITensor *input1, ITensor *input2, ITensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ITensor *input1, + ITensor *input2, + ITensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref cpu::kernels::CpuArithmeticKernel for squared difference * * @param[in] input1 First tensor input info. Data types supported: QASYMM8/QASYMM8_SIGNED/S16/F16/S32/F32. @@ -203,7 +218,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -249,7 +267,10 @@ public: * @param[out] output Output tensor. Data types supported: Same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(ITensor *input1, ITensor *input2, ITensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ITensor *input1, + ITensor *input2, + ITensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref cpu::kernels::CpuArithmeticKernel for division * * @param[in] input1 First tensor input info. Data types supported: F16/F32. @@ -259,7 +280,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -306,7 +330,10 @@ public: * @param[out] output Output tensor. Data types supported: Same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(ITensor *input1, ITensor *input2, ITensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ITensor *input1, + ITensor *input2, + ITensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref cpu::kernels::CpuArithmeticKernel for power * * @param[in] input1 First tensor input info. Data types supported: F16/F32. @@ -316,7 +343,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -377,7 +407,8 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ComparisonOperation op); + static Status + validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ComparisonOperation op); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEFFT1D.h b/arm_compute/runtime/NEON/functions/NEFFT1D.h index 9654b1e604..99c6fd4eb4 100644 --- a/arm_compute/runtime/NEON/functions/NEFFT1D.h +++ b/arm_compute/runtime/NEON/functions/NEFFT1D.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEFFT1D_H #define ARM_COMPUTE_NEFFT1D_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/runtime/FunctionDescriptors.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/MemoryGroup.h" #include "arm_compute/runtime/Tensor.h" diff --git a/arm_compute/runtime/NEON/functions/NEFFT2D.h b/arm_compute/runtime/NEON/functions/NEFFT2D.h index 57f38d1942..cefd3df17a 100644 --- a/arm_compute/runtime/NEON/functions/NEFFT2D.h +++ b/arm_compute/runtime/NEON/functions/NEFFT2D.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEFFT2D_H #define ARM_COMPUTE_NEFFT2D_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/runtime/FunctionDescriptors.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/MemoryGroup.h" #include "arm_compute/runtime/NEON/functions/NEFFT1D.h" #include "arm_compute/runtime/Tensor.h" diff --git a/arm_compute/runtime/NEON/functions/NEFFTConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEFFTConvolutionLayer.h index c5f4d45b6b..84bfe6b02f 100644 --- a/arm_compute/runtime/NEON/functions/NEFFTConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEFFTConvolutionLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEFFTCONVOLUTIONLAYER_H #define ARM_COMPUTE_NEFFTCONVOLUTIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/NEON/functions/NEActivationLayer.h" #include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h" #include "arm_compute/runtime/NEON/functions/NEFFT2D.h" @@ -94,8 +93,13 @@ public: * @param[in] act_info (Optional) Activation layer information in case of a fused activation. * @param[in] enable_fast_math (Optional) Enable fast math computation. Unused for CPU backend. */ - void configure(ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + void configure(ITensor *input, + const ITensor *weights, + const ITensor *biases, + ITensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); /** Static function to check if given info will lead to a valid configuration of @ref NEFFTConvolutionLayer * * @note: This function only works with any square kernel size and unit strides for both NCHW and NHWC data layout @@ -113,8 +117,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEFill.h b/arm_compute/runtime/NEON/functions/NEFill.h index e923ce33e1..1829c71fef 100644 --- a/arm_compute/runtime/NEON/functions/NEFill.h +++ b/arm_compute/runtime/NEON/functions/NEFill.h @@ -24,10 +24,9 @@ #ifndef ARM_COMPUTE_NEFILL_H #define ARM_COMPUTE_NEFILL_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/PixelValue.h" #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> diff --git a/arm_compute/runtime/NEON/functions/NEFillBorder.h b/arm_compute/runtime/NEON/functions/NEFillBorder.h index ab77c28839..44b1d4a62b 100644 --- a/arm_compute/runtime/NEON/functions/NEFillBorder.h +++ b/arm_compute/runtime/NEON/functions/NEFillBorder.h @@ -27,6 +27,7 @@ #include "arm_compute/core/PixelValue.h" #include "arm_compute/core/Types.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -57,7 +58,10 @@ public: * @param[in] border_mode Strategy to use for borders. * @param[in] constant_border_value (Optional) Constant value to use for borders if border_mode is set to CONSTANT. */ - void configure(ITensor *input, unsigned int border_width, BorderMode border_mode, const PixelValue &constant_border_value = PixelValue()); + void configure(ITensor *input, + unsigned int border_width, + BorderMode border_mode, + const PixelValue &constant_border_value = PixelValue()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEFloor.h b/arm_compute/runtime/NEON/functions/NEFloor.h index 4d47b068db..77ac484bab 100644 --- a/arm_compute/runtime/NEON/functions/NEFloor.h +++ b/arm_compute/runtime/NEON/functions/NEFloor.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEFLOOR_H #define ARM_COMPUTE_NEFLOOR_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> diff --git a/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h b/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h index 05b7ce3735..885f8430cf 100644 --- a/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h +++ b/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h @@ -28,7 +28,6 @@ #include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/IMemoryManager.h" #include "arm_compute/runtime/IWeightsManager.h" - #include "arm_compute/runtime/NEON/functions/NETranspose.h" #include "arm_compute/runtime/Tensor.h" @@ -88,7 +87,8 @@ class NEFullyConnectedLayer : public IFunction { public: /** Constructor */ - NEFullyConnectedLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr, IWeightsManager *weights_manager = nullptr); + NEFullyConnectedLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr, + IWeightsManager *weights_manager = nullptr); /** Prevent instances of this class from being copied (As this class contains pointers) */ NEFullyConnectedLayer(const NEFullyConnectedLayer &) = delete; /** Prevent instances of this class from being moved (As this class contains pointers) */ @@ -126,16 +126,24 @@ public: * @param[in] fc_info (Optional) Fully connected layer additional info * @param[in] weights_info (Optional) Stores neccessary compute information when weights are already reshaped */ - void configure(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, - FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo(), const WeightsInfo &weights_info = WeightsInfo()); + void configure(const ITensor *input, + const ITensor *weights, + const ITensor *biases, + ITensor *output, + FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo(), + const WeightsInfo &weights_info = WeightsInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEFullyConnectedLayer * * Similar to @ref NEFullyConnectedLayer::configure() * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, - FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo(), const WeightsInfo &weights_info = WeightsInfo()); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo(), + const WeightsInfo &weights_info = WeightsInfo()); /** Static function that queries whether fixed-format kernel exists for a given problem description * @@ -149,8 +157,13 @@ public: * * @return a status */ - static Status has_opt_impl(arm_compute::WeightFormat &expected_weight_format, const ITensorInfo *input, const ITensorInfo *weights, - const ITensorInfo *biases, const ITensorInfo *output, const FullyConnectedLayerInfo &fc_info, const WeightsInfo &weights_info); + static Status has_opt_impl(arm_compute::WeightFormat &expected_weight_format, + const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const FullyConnectedLayerInfo &fc_info, + const WeightsInfo &weights_info); //Inherited methods override void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEFuseBatchNormalization.h b/arm_compute/runtime/NEON/functions/NEFuseBatchNormalization.h index 3dd7f49044..f53b3de7f6 100644 --- a/arm_compute/runtime/NEON/functions/NEFuseBatchNormalization.h +++ b/arm_compute/runtime/NEON/functions/NEFuseBatchNormalization.h @@ -75,9 +75,16 @@ public: * @param[in] epsilon (Optional) Batch normalization layer epsilon parameter. Defaults to 0.001f. * @param[in] fbn_type (Optional) Fused batch normalization type. Defaults to Convolution. */ - void configure(const ITensor *input_weights, const ITensor *bn_mean, const ITensor *bn_var, ITensor *fused_weights, ITensor *fused_bias, - const ITensor *input_bias = nullptr, const ITensor *bn_beta = nullptr, const ITensor *bn_gamma = nullptr, - float epsilon = 0.001f, FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); + void configure(const ITensor *input_weights, + const ITensor *bn_mean, + const ITensor *bn_var, + ITensor *fused_weights, + ITensor *fused_bias, + const ITensor *input_bias = nullptr, + const ITensor *bn_beta = nullptr, + const ITensor *bn_gamma = nullptr, + float epsilon = 0.001f, + FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); /** Static function to check if given info will lead to a valid configuration of @ref NEFuseBatchNormalization * * @param[in] input_weights Input weights tensor info for convolution or depthwise convolution layer. Data type supported: F16/F32. Data layout supported: NCHW, NHWC @@ -95,10 +102,16 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input_weights, const ITensorInfo *bn_mean, const ITensorInfo *bn_var, - const ITensorInfo *fused_weights, const ITensorInfo *fused_bias, - const ITensorInfo *input_bias = nullptr, const ITensorInfo *bn_beta = nullptr, const ITensorInfo *bn_gamma = nullptr, - float epsilon = 0.001f, FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); + static Status validate(const ITensorInfo *input_weights, + const ITensorInfo *bn_mean, + const ITensorInfo *bn_var, + const ITensorInfo *fused_weights, + const ITensorInfo *fused_bias, + const ITensorInfo *input_bias = nullptr, + const ITensorInfo *bn_beta = nullptr, + const ITensorInfo *bn_gamma = nullptr, + float epsilon = 0.001f, + FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEGEMM.h b/arm_compute/runtime/NEON/functions/NEGEMM.h index c6ff2dfb92..29650a5eca 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMM.h +++ b/arm_compute/runtime/NEON/functions/NEGEMM.h @@ -78,14 +78,26 @@ public: * @param[in] gemm_info (Optional) Specifies if the matrix A and/or matrix B have been reshaped and * if the reshape of matrix B should happen only for the first run */ - void configure(const ITensor *a, const ITensor *b, const ITensor *c, ITensor *d, float alpha, float beta, const GEMMInfo &gemm_info = GEMMInfo()); + void configure(const ITensor *a, + const ITensor *b, + const ITensor *c, + ITensor *d, + float alpha, + float beta, + const GEMMInfo &gemm_info = GEMMInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEGEMM. * * Similar to @ref NEGEMM::configure() * * @return a status */ - static Status validate(const ITensorInfo *a, const ITensorInfo *b, const ITensorInfo *c, const ITensorInfo *output, float alpha, float beta, const GEMMInfo &gemm_info = GEMMInfo()); + static Status validate(const ITensorInfo *a, + const ITensorInfo *b, + const ITensorInfo *c, + const ITensorInfo *output, + float alpha, + float beta, + const GEMMInfo &gemm_info = GEMMInfo()); /** Static function that queries whether there exists fixed-format kernel and if it exists it will return in the first argument in what format * weights are expected to be reshaped as defined by WeightFormat class. Apart from the first argument the rest of the arguments are the same @@ -93,8 +105,14 @@ public: * * @return a status */ - static Status has_opt_impl(arm_compute::WeightFormat &expected_weight_format, const ITensorInfo *a, const ITensorInfo *b, const ITensorInfo *c, const ITensorInfo *output, - float alpha, float beta, const GEMMInfo &gemm_info = GEMMInfo()); + static Status has_opt_impl(arm_compute::WeightFormat &expected_weight_format, + const ITensorInfo *a, + const ITensorInfo *b, + const ITensorInfo *c, + const ITensorInfo *output, + float alpha, + float beta, + const GEMMInfo &gemm_info = GEMMInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEGEMMConv2d.h b/arm_compute/runtime/NEON/functions/NEGEMMConv2d.h index 53ceb6d978..d1c5a1c9b3 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMConv2d.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMConv2d.h @@ -86,7 +86,8 @@ public: * Data types supported: Same as @p input. * @param[in] info Convolution layer descriptor */ - void configure(ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const Conv2dInfo &info); + void + configure(ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const Conv2dInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref NEGEMMConv2d * * @param[in] input Source tensor info. 3 lower dimensions represent a single input [width, height, IFM], @@ -102,7 +103,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const Conv2dInfo &info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const Conv2dInfo &info); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h index 72309e464e..3e84c3e2cf 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h @@ -24,8 +24,6 @@ #ifndef ARM_COMPUTE_NEGEMMCONVOLUTIONLAYER_H #define ARM_COMPUTE_NEGEMMCONVOLUTIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" #include "arm_compute/runtime/IFunction.h" @@ -49,7 +47,8 @@ class NEGEMMConvolutionLayer : public IFunction { public: /** Constructor */ - NEGEMMConvolutionLayer(const std::shared_ptr<IMemoryManager> &memory_manager = nullptr, IWeightsManager *weights_manager = nullptr); + NEGEMMConvolutionLayer(const std::shared_ptr<IMemoryManager> &memory_manager = nullptr, + IWeightsManager *weights_manager = nullptr); /** Prevent instances of this class from being copied (As this class contains pointers) */ NEGEMMConvolutionLayer(const NEGEMMConvolutionLayer &) = delete; /** Prevent instances of this class from being moved (As this class contains non movable objects) */ @@ -95,8 +94,16 @@ public: * available which may introduce a drop of accuracy as well. Default is false * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution. num_groups != 1 is not supported */ - void configure(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info, const WeightsInfo &weights_info = WeightsInfo(), - const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false, unsigned int num_groups = 1); + void configure(const ITensor *input, + const ITensor *weights, + const ITensor *biases, + ITensor *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false, + unsigned int num_groups = 1); /** Static function to check if given info will lead to a valid configuration of @ref NEGEMMConvolutionLayer * * @param[in] input Source tensor info. 3 lower dimensions represent a single input [width, height, IFM], @@ -119,9 +126,16 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - const WeightsInfo &weights_info = WeightsInfo(), const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), - bool enable_fast_math = false, unsigned int num_groups = 1); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false, + unsigned int num_groups = 1); /** Static function to check if there is an optimized version of * GEMM available for the input parameters. @@ -178,10 +192,16 @@ public: * * @return a Status */ - static Status has_opt_impl(arm_compute::WeightFormat &expected_weight_format, const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *dst, - const PadStrideInfo &conv_info, - const WeightsInfo &weights_info = WeightsInfo(), const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), - bool enable_fast_math = false); + static Status has_opt_impl(arm_compute::WeightFormat &expected_weight_format, + const ITensorInfo *src, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *dst, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); // Inherited methods overridden: void run() override; void prepare() override; diff --git a/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h b/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h index addb13cdfa..824c4443ad 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h @@ -47,7 +47,8 @@ class NEGEMMLowpMatrixMultiplyCore : public IFunction { public: /** Constructor */ - NEGEMMLowpMatrixMultiplyCore(std::shared_ptr<IMemoryManager> memory_manager = nullptr, IWeightsManager *weights_manager = nullptr); + NEGEMMLowpMatrixMultiplyCore(std::shared_ptr<IMemoryManager> memory_manager = nullptr, + IWeightsManager *weights_manager = nullptr); /** Prevent instances of this class from being copied (As this class contains pointers) */ NEGEMMLowpMatrixMultiplyCore(const NEGEMMLowpMatrixMultiplyCore &) = delete; /** Default move constructor */ @@ -96,14 +97,19 @@ public: * @param[in] gemm_info (Optional) Specifies if the matrix A and/or matrix B have been reshaped and * if the reshape of matrix B should be executed only for the first run */ - void configure(const ITensor *a, const ITensor *b, const ITensor *c, ITensor *output, const GEMMInfo &gemm_info = GEMMInfo()); + void configure( + const ITensor *a, const ITensor *b, const ITensor *c, ITensor *output, const GEMMInfo &gemm_info = GEMMInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEGEMMLowpMatrixMultiplyCore * * Similar to @ref NEGEMMLowpMatrixMultiplyCore::configure() * * @return a status */ - static Status validate(const ITensorInfo *a, const ITensorInfo *b, const ITensorInfo *c, const ITensorInfo *output, const GEMMInfo &gemm_info = GEMMInfo()); + static Status validate(const ITensorInfo *a, + const ITensorInfo *b, + const ITensorInfo *c, + const ITensorInfo *output, + const GEMMInfo &gemm_info = GEMMInfo()); // Inherited methods overridden void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEGEMMLowpOutputStage.h b/arm_compute/runtime/NEON/functions/NEGEMMLowpOutputStage.h index 232344e5c2..0d932bb4af 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMLowpOutputStage.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMLowpOutputStage.h @@ -89,7 +89,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *bias, const ITensorInfo *output, const GEMMLowpOutputStageInfo &info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *bias, + const ITensorInfo *output, + const GEMMLowpOutputStageInfo &info); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEGenerateProposalsLayer.h b/arm_compute/runtime/NEON/functions/NEGenerateProposalsLayer.h index 3b683382ec..0f294fde22 100644 --- a/arm_compute/runtime/NEON/functions/NEGenerateProposalsLayer.h +++ b/arm_compute/runtime/NEON/functions/NEGenerateProposalsLayer.h @@ -95,7 +95,12 @@ public: * @note Only single image prediction is supported. Height and Width (and scale) of the image will be contained in the @ref GenerateProposalsInfo struct. * @note Proposals contains all the proposals. Of those, only the first num_valid_proposals are valid. */ - void configure(const ITensor *scores, const ITensor *deltas, const ITensor *anchors, ITensor *proposals, ITensor *scores_out, ITensor *num_valid_proposals, + void configure(const ITensor *scores, + const ITensor *deltas, + const ITensor *anchors, + ITensor *proposals, + ITensor *scores_out, + ITensor *num_valid_proposals, const GenerateProposalsInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref NEGenerateProposalsLayer @@ -112,7 +117,11 @@ public: * * @return a Status */ - static Status validate(const ITensorInfo *scores, const ITensorInfo *deltas, const ITensorInfo *anchors, const ITensorInfo *proposals, const ITensorInfo *scores_out, + static Status validate(const ITensorInfo *scores, + const ITensorInfo *deltas, + const ITensorInfo *anchors, + const ITensorInfo *proposals, + const ITensorInfo *scores_out, const ITensorInfo *num_valid_proposals, const GenerateProposalsInfo &info); diff --git a/arm_compute/runtime/NEON/functions/NEInstanceNormalizationLayer.h b/arm_compute/runtime/NEON/functions/NEInstanceNormalizationLayer.h index bb0697072b..0bc57be09e 100644 --- a/arm_compute/runtime/NEON/functions/NEInstanceNormalizationLayer.h +++ b/arm_compute/runtime/NEON/functions/NEInstanceNormalizationLayer.h @@ -89,7 +89,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, float gamma = 1.0f, float beta = 0.0f, float epsilon = 1e-12f); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + float gamma = 1.0f, + float beta = 0.0f, + float epsilon = 1e-12f); // Inherited methods overridden: void run() override; @@ -103,5 +107,5 @@ private: Tensor _permuted_input; Tensor _permuted_output; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_NEINSTANCENORMALIZATIONLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEL2NormalizeLayer.h b/arm_compute/runtime/NEON/functions/NEL2NormalizeLayer.h index 7f1a5e785e..8502cee5d2 100644 --- a/arm_compute/runtime/NEON/functions/NEL2NormalizeLayer.h +++ b/arm_compute/runtime/NEON/functions/NEL2NormalizeLayer.h @@ -97,5 +97,5 @@ private: std::unique_ptr<NEL2NormalizeLayerKernel> _normalize_kernel; Tensor _sumsq; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_NEL2NORMALIZELAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NELSTMLayer.h b/arm_compute/runtime/NEON/functions/NELSTMLayer.h index 4272215486..629c5d10a0 100644 --- a/arm_compute/runtime/NEON/functions/NELSTMLayer.h +++ b/arm_compute/runtime/NEON/functions/NELSTMLayer.h @@ -25,6 +25,7 @@ #define ARM_COMPUTE_NELSTMLAYER_H #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/common/LSTMParams.h" #include "arm_compute/runtime/NEON/functions/NEActivationLayer.h" #include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h" #include "arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h" @@ -35,7 +36,6 @@ #include "arm_compute/runtime/NEON/functions/NEMeanStdDevNormalizationLayer.h" #include "arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h" #include "arm_compute/runtime/NEON/functions/NETranspose.h" -#include "arm_compute/runtime/common/LSTMParams.h" namespace arm_compute { @@ -104,13 +104,26 @@ public: * @param[in] projection_threshold The clipping threshold for the output from the projection layer, such that values are bound within [-proj_clip, proj_clip]. * If set to 0.0 then clipping is disabled. */ - void configure(const ITensor *input, - const ITensor *input_to_forget_weights, const ITensor *input_to_cell_weights, const ITensor *input_to_output_weights, - const ITensor *recurrent_to_forget_weights, const ITensor *recurrent_to_cell_weights, const ITensor *recurrent_to_output_weights, - const ITensor *forget_gate_bias, const ITensor *cell_bias, const ITensor *output_gate_bias, - const ITensor *output_state_in, const ITensor *cell_state_in, - ITensor *scratch_buffer, ITensor *output_state_out, ITensor *cell_state_out, ITensor *output, - const LSTMParams<ITensor> &lstm_params, const ActivationLayerInfo &activation_info, float cell_threshold = 0.f, float projection_threshold = 0.f); + void configure(const ITensor *input, + const ITensor *input_to_forget_weights, + const ITensor *input_to_cell_weights, + const ITensor *input_to_output_weights, + const ITensor *recurrent_to_forget_weights, + const ITensor *recurrent_to_cell_weights, + const ITensor *recurrent_to_output_weights, + const ITensor *forget_gate_bias, + const ITensor *cell_bias, + const ITensor *output_gate_bias, + const ITensor *output_state_in, + const ITensor *cell_state_in, + ITensor *scratch_buffer, + ITensor *output_state_out, + ITensor *cell_state_out, + ITensor *output, + const LSTMParams<ITensor> &lstm_params, + const ActivationLayerInfo &activation_info, + float cell_threshold = 0.f, + float projection_threshold = 0.f); /** Static function to check if given info will lead to a valid configuration of @ref NELSTMLayer * @@ -151,13 +164,26 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, - const ITensorInfo *input_to_forget_weights, const ITensorInfo *input_to_cell_weights, const ITensorInfo *input_to_output_weights, - const ITensorInfo *recurrent_to_forget_weights, const ITensorInfo *recurrent_to_cell_weights, const ITensorInfo *recurrent_to_output_weights, - const ITensorInfo *forget_gate_bias, const ITensorInfo *cell_bias, const ITensorInfo *output_gate_bias, - const ITensorInfo *output_state_in, const ITensorInfo *cell_state_in, - const ITensorInfo *scratch_buffer, const ITensorInfo *output_state_out, const ITensorInfo *cell_state_out, const ITensorInfo *output, - const LSTMParams<ITensorInfo> &lstm_params, const ActivationLayerInfo &activation_info, float cell_threshold = 0.f, float projection_threshold = 0.f); + static Status validate(const ITensorInfo *input, + const ITensorInfo *input_to_forget_weights, + const ITensorInfo *input_to_cell_weights, + const ITensorInfo *input_to_output_weights, + const ITensorInfo *recurrent_to_forget_weights, + const ITensorInfo *recurrent_to_cell_weights, + const ITensorInfo *recurrent_to_output_weights, + const ITensorInfo *forget_gate_bias, + const ITensorInfo *cell_bias, + const ITensorInfo *output_gate_bias, + const ITensorInfo *output_state_in, + const ITensorInfo *cell_state_in, + const ITensorInfo *scratch_buffer, + const ITensorInfo *output_state_out, + const ITensorInfo *cell_state_out, + const ITensorInfo *output, + const LSTMParams<ITensorInfo> &lstm_params, + const ActivationLayerInfo &activation_info, + float cell_threshold = 0.f, + float projection_threshold = 0.f); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NELSTMLayerQuantized.h b/arm_compute/runtime/NEON/functions/NELSTMLayerQuantized.h index bcb89d997d..ae951669b3 100644 --- a/arm_compute/runtime/NEON/functions/NELSTMLayerQuantized.h +++ b/arm_compute/runtime/NEON/functions/NELSTMLayerQuantized.h @@ -25,6 +25,7 @@ #define ARM_COMPUTE_NELSTMLAYERQUANTIZED_H #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/common/LSTMParams.h" #include "arm_compute/runtime/NEON/functions/NEActivationLayer.h" #include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h" #include "arm_compute/runtime/NEON/functions/NEConcatenateLayer.h" @@ -38,8 +39,6 @@ #include "arm_compute/runtime/NEON/functions/NESlice.h" #include "arm_compute/runtime/NEON/functions/NETranspose.h" -#include "arm_compute/runtime/common/LSTMParams.h" - namespace arm_compute { // Forward declarations @@ -104,11 +103,22 @@ public: * @param[out] output_state_out Destination tensor. Output is a 2D tensor with dimensions [output_size, batch_size].Data types supported: Same as @p input. */ void configure(const ITensor *input, - const ITensor *input_to_input_weights, const ITensor *input_to_forget_weights, const ITensor *input_to_cell_weights, const ITensor *input_to_output_weights, - const ITensor *recurrent_to_input_weights, const ITensor *recurrent_to_forget_weights, const ITensor *recurrent_to_cell_weights, const ITensor *recurrent_to_output_weights, - const ITensor *input_gate_bias, const ITensor *forget_gate_bias, const ITensor *cell_bias, const ITensor *output_gate_bias, - ITensor *cell_state_in, const ITensor *output_state_in, - ITensor *cell_state_out, ITensor *output_state_out); + const ITensor *input_to_input_weights, + const ITensor *input_to_forget_weights, + const ITensor *input_to_cell_weights, + const ITensor *input_to_output_weights, + const ITensor *recurrent_to_input_weights, + const ITensor *recurrent_to_forget_weights, + const ITensor *recurrent_to_cell_weights, + const ITensor *recurrent_to_output_weights, + const ITensor *input_gate_bias, + const ITensor *forget_gate_bias, + const ITensor *cell_bias, + const ITensor *output_gate_bias, + ITensor *cell_state_in, + const ITensor *output_state_in, + ITensor *cell_state_out, + ITensor *output_state_out); /** Static function to check if given info will lead to a valid configuration of @ref NELSTMLayer * @@ -133,11 +143,22 @@ public: * @return a status */ static Status validate(const ITensorInfo *input, - const ITensorInfo *input_to_input_weights, const ITensorInfo *input_to_forget_weights, const ITensorInfo *input_to_cell_weights, const ITensorInfo *input_to_output_weights, - const ITensorInfo *recurrent_to_input_weights, const ITensorInfo *recurrent_to_forget_weights, const ITensorInfo *recurrent_to_cell_weights, const ITensorInfo *recurrent_to_output_weights, - const ITensorInfo *input_gate_bias, const ITensorInfo *forget_gate_bias, const ITensorInfo *cell_bias, const ITensorInfo *output_gate_bias, - const ITensorInfo *cell_state_in, const ITensorInfo *output_state_in, - const ITensorInfo *cell_state_out, const ITensorInfo *output_state_out); + const ITensorInfo *input_to_input_weights, + const ITensorInfo *input_to_forget_weights, + const ITensorInfo *input_to_cell_weights, + const ITensorInfo *input_to_output_weights, + const ITensorInfo *recurrent_to_input_weights, + const ITensorInfo *recurrent_to_forget_weights, + const ITensorInfo *recurrent_to_cell_weights, + const ITensorInfo *recurrent_to_output_weights, + const ITensorInfo *input_gate_bias, + const ITensorInfo *forget_gate_bias, + const ITensorInfo *cell_bias, + const ITensorInfo *output_gate_bias, + const ITensorInfo *cell_state_in, + const ITensorInfo *output_state_in, + const ITensorInfo *cell_state_out, + const ITensorInfo *output_state_out); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEMatMul.h b/arm_compute/runtime/NEON/functions/NEMatMul.h index e961f860c1..414fc2f3fd 100644 --- a/arm_compute/runtime/NEON/functions/NEMatMul.h +++ b/arm_compute/runtime/NEON/functions/NEMatMul.h @@ -27,6 +27,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -48,7 +49,7 @@ public: }; private: - bool _fast_math{ false }; + bool _fast_math{false}; }; // Forward declarations @@ -96,7 +97,12 @@ public: * @param[in] settings Contains flags for function level settings i.e fast math * @param[in] act_info (Optional) Contains activation function and lower and upper bound values for bounded activation functions. */ - void configure(ITensor *lhs, ITensor *rhs, ITensor *dst, const MatMulInfo &info, const CpuMatMulSettings &settings, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ITensor *lhs, + ITensor *rhs, + ITensor *dst, + const MatMulInfo &info, + const CpuMatMulSettings &settings, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEMatMul * * @param[in] lhs Left-hand side tensor info. Data types supported: F16/F32/QASYMM8_SIGNED/QASYMM8. @@ -108,7 +114,11 @@ public: * * @return Status */ - static Status validate(const ITensorInfo *lhs, const ITensorInfo *rhs, const ITensorInfo *dst, const MatMulInfo &info, const CpuMatMulSettings &settings, + static Status validate(const ITensorInfo *lhs, + const ITensorInfo *rhs, + const ITensorInfo *dst, + const MatMulInfo &info, + const CpuMatMulSettings &settings, const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden @@ -118,5 +128,5 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} +} // namespace arm_compute #endif /* ACL_ARM_COMPUTE_RUNTIME_NEON_FUNCTIONS_NEMATMUL */ diff --git a/arm_compute/runtime/NEON/functions/NEMaxUnpoolingLayer.h b/arm_compute/runtime/NEON/functions/NEMaxUnpoolingLayer.h index 2f77540e1e..e00fc4544f 100644 --- a/arm_compute/runtime/NEON/functions/NEMaxUnpoolingLayer.h +++ b/arm_compute/runtime/NEON/functions/NEMaxUnpoolingLayer.h @@ -26,6 +26,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -86,7 +87,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *indices, const ITensorInfo *output, const PoolingLayerInfo &pool_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *indices, + const ITensorInfo *output, + const PoolingLayerInfo &pool_info); // Inherited methods overridden: void run() override; @@ -96,5 +100,5 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_NEMAXUNPOOLINGLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NENormalizationLayer.h b/arm_compute/runtime/NEON/functions/NENormalizationLayer.h index fbe000445c..27e3fa674e 100644 --- a/arm_compute/runtime/NEON/functions/NENormalizationLayer.h +++ b/arm_compute/runtime/NEON/functions/NENormalizationLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NENORMALIZATIONLAYER_H #define ARM_COMPUTE_NENORMALIZATIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/IMemoryManager.h" #include "arm_compute/runtime/MemoryGroup.h" #include "arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h" @@ -88,16 +87,17 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const NormalizationLayerInfo &norm_info); + static Status + validate(const ITensorInfo *input, const ITensorInfo *output, const NormalizationLayerInfo &norm_info); // Inherited methods overridden: void run() override; private: - MemoryGroup _memory_group; /**< Function memory group */ - std::unique_ptr<NENormalizationLayerKernel> _norm_kernel; /**< Normalization layer kernel */ - NEPixelWiseMultiplication _multiply_f; /**< Pixel multiplication function */ - Tensor _input_squared; /**< The intermediate buffer which stores results of squaring input */ + MemoryGroup _memory_group; /**< Function memory group */ + std::unique_ptr<NENormalizationLayerKernel> _norm_kernel; /**< Normalization layer kernel */ + NEPixelWiseMultiplication _multiply_f; /**< Pixel multiplication function */ + Tensor _input_squared; /**< The intermediate buffer which stores results of squaring input */ }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_NENORMALIZATIONLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEPadLayer.h b/arm_compute/runtime/NEON/functions/NEPadLayer.h index 4aa6725496..494b1c0641 100644 --- a/arm_compute/runtime/NEON/functions/NEPadLayer.h +++ b/arm_compute/runtime/NEON/functions/NEPadLayer.h @@ -24,14 +24,14 @@ #ifndef ARM_COMPUTE_NEPADLAYER_H #define ARM_COMPUTE_NEPADLAYER_H +#include "arm_compute/core/Types.h" #include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/NEON/functions/NEConcatenateLayer.h" #include "arm_compute/runtime/NEON/functions/NECopy.h" #include "arm_compute/runtime/NEON/functions/NEStridedSlice.h" #include "arm_compute/runtime/SubTensor.h" - -#include "arm_compute/core/Types.h" #include "arm_compute/runtime/Tensor.h" + #include <memory> namespace arm_compute @@ -82,7 +82,11 @@ public: * @param[in] mode (Optional) Controls whether the padding should be filled with @p constant_value using CONSTANT, * or reflect the input, either including the border values (SYMMETRIC) or not (REFLECT). */ - void configure(ITensor *input, ITensor *output, const PaddingList &padding, const PixelValue constant_value = PixelValue(), const PaddingMode mode = PaddingMode::CONSTANT); + void configure(ITensor *input, + ITensor *output, + const PaddingList &padding, + const PixelValue constant_value = PixelValue(), + const PaddingMode mode = PaddingMode::CONSTANT); /** Static function to check if given info will lead to a valid configuration of @ref NEPadLayer. * * @param[in] input Source tensor info. Data types supported: All. @@ -95,7 +99,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const PaddingList &padding, const PixelValue constant_value = PixelValue(), const PaddingMode mode = PaddingMode::CONSTANT); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const PaddingList &padding, + const PixelValue constant_value = PixelValue(), + const PaddingMode mode = PaddingMode::CONSTANT); // Inherited methods overridden: void run() override; @@ -109,7 +117,10 @@ private: * specifies the front and the end padding in the i-th dimension. * @param[in] constant_value Constant value to be used for the padding */ - void configure_constant_mode(ITensor *input, ITensor *output, const PaddingList &padding, const PixelValue constant_value); + void configure_constant_mode(ITensor *input, + ITensor *output, + const PaddingList &padding, + const PixelValue constant_value); /** Configure functions for when reflect or symmetric padding is used. * * @param[in] input Source tensor. Data types supported: All. diff --git a/arm_compute/runtime/NEON/functions/NEPermute.h b/arm_compute/runtime/NEON/functions/NEPermute.h index c863fde0ac..2cef64764d 100644 --- a/arm_compute/runtime/NEON/functions/NEPermute.h +++ b/arm_compute/runtime/NEON/functions/NEPermute.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEPERMUTE_H #define ARM_COMPUTE_NEPERMUTE_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> diff --git a/arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h b/arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h index 634e8e0c39..3d81bf6087 100644 --- a/arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h +++ b/arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h @@ -95,7 +95,12 @@ public: * @param[in] rounding_policy Rounding policy. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(const ITensor *input1, const ITensor *input2, ITensor *output, float scale, ConvertPolicy overflow_policy, RoundingPolicy rounding_policy, + void configure(const ITensor *input1, + const ITensor *input2, + ITensor *output, + float scale, + ConvertPolicy overflow_policy, + RoundingPolicy rounding_policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEPixelWiseMultiplication * @@ -122,7 +127,12 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, float scale, ConvertPolicy overflow_policy, RoundingPolicy rounding_policy, + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + float scale, + ConvertPolicy overflow_policy, + RoundingPolicy rounding_policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: @@ -158,7 +168,10 @@ public: * @param[out] output The output tensor. Data types supported: same as @p input1. Number of channels: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - void configure(ITensor *input1, ITensor *input2, ITensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ITensor *input1, + ITensor *input2, + ITensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref NEComplexPixelWiseMultiplication * * @param[in] input1 An input tensor info. Data types supported: F32. Number of channels supported: 2 (complex tensor). @@ -166,7 +179,10 @@ public: * @param[in] output The output tensor info. Data types supported: same as @p input1. Number of channels supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Currently not supported. */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEPooling3dLayer.h b/arm_compute/runtime/NEON/functions/NEPooling3dLayer.h index 4c5eb58e05..09251f2a5f 100644 --- a/arm_compute/runtime/NEON/functions/NEPooling3dLayer.h +++ b/arm_compute/runtime/NEON/functions/NEPooling3dLayer.h @@ -92,5 +92,5 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_NEPOOLING3DLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEPoolingLayer.h b/arm_compute/runtime/NEON/functions/NEPoolingLayer.h index 9147ad9687..768ad0d818 100644 --- a/arm_compute/runtime/NEON/functions/NEPoolingLayer.h +++ b/arm_compute/runtime/NEON/functions/NEPoolingLayer.h @@ -91,7 +91,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const PoolingLayerInfo &pool_info, const ITensorInfo *indices = nullptr); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const PoolingLayerInfo &pool_info, + const ITensorInfo *indices = nullptr); // Inherited methods overridden: void run() override; @@ -100,5 +103,5 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_NEPOOLINGLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEPriorBoxLayer.h b/arm_compute/runtime/NEON/functions/NEPriorBoxLayer.h index 38e0c9f3ad..858e3299af 100644 --- a/arm_compute/runtime/NEON/functions/NEPriorBoxLayer.h +++ b/arm_compute/runtime/NEON/functions/NEPriorBoxLayer.h @@ -62,7 +62,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const PriorBoxLayerInfo &info); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const PriorBoxLayerInfo &info); }; } // namespace arm_compute #endif /* ARM_COMPUTE_NEPRIORBOXLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEQLSTMLayer.h b/arm_compute/runtime/NEON/functions/NEQLSTMLayer.h index 185d821ec0..009a4e0911 100644 --- a/arm_compute/runtime/NEON/functions/NEQLSTMLayer.h +++ b/arm_compute/runtime/NEON/functions/NEQLSTMLayer.h @@ -25,6 +25,7 @@ #define ARM_COMPUTE_NEQLSTMLAYER_H #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/common/LSTMParams.h" #include "arm_compute/runtime/NEON/functions/NEActivationLayer.h" #include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h" #include "arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h" @@ -35,7 +36,6 @@ #include "arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h" #include "arm_compute/runtime/NEON/functions/NEQuantizationLayer.h" #include "arm_compute/runtime/NEON/functions/NETranspose.h" -#include "arm_compute/runtime/common/LSTMParams.h" #include <memory> @@ -130,12 +130,21 @@ public: * projection_threshold (Optional) The clipping threshold for the output from the projection layer, such that values are bound within * [-proj_clip, proj_clip]. If set to 0.0 then clipping is disabled. */ - void configure(const ITensor *input, - const ITensor *input_to_forget_weights, const ITensor *input_to_cell_weights, const ITensor *input_to_output_weights, - const ITensor *recurrent_to_forget_weights, const ITensor *recurrent_to_cell_weights, const ITensor *recurrent_to_output_weights, - const ITensor *forget_gate_bias, const ITensor *cell_bias, const ITensor *output_gate_bias, - const ITensor *cell_state_in, ITensor *output_state_in, - ITensor *cell_state_out, ITensor *output_state_out, ITensor *output, + void configure(const ITensor *input, + const ITensor *input_to_forget_weights, + const ITensor *input_to_cell_weights, + const ITensor *input_to_output_weights, + const ITensor *recurrent_to_forget_weights, + const ITensor *recurrent_to_cell_weights, + const ITensor *recurrent_to_output_weights, + const ITensor *forget_gate_bias, + const ITensor *cell_bias, + const ITensor *output_gate_bias, + const ITensor *cell_state_in, + ITensor *output_state_in, + ITensor *cell_state_out, + ITensor *output_state_out, + ITensor *output, const LSTMParams<ITensor> &lstm_params); /** Static function to check if given info will lead to a valid configuration of @ref NEQLSTMLayer @@ -180,12 +189,21 @@ public: * [-proj_clip, proj_clip]. If set to 0.0 then clipping is disabled. * @return a status */ - static Status validate(const ITensorInfo *input, - const ITensorInfo *input_to_forget_weights, const ITensorInfo *input_to_cell_weights, const ITensorInfo *input_to_output_weights, - const ITensorInfo *recurrent_to_forget_weights, const ITensorInfo *recurrent_to_cell_weights, const ITensorInfo *recurrent_to_output_weights, - const ITensorInfo *forget_gate_bias, const ITensorInfo *cell_bias, const ITensorInfo *output_gate_bias, - const ITensorInfo *cell_state_in, const ITensorInfo *output_state_in, - const ITensorInfo *cell_state_out, const ITensorInfo *output_state_out, const ITensorInfo *output, + static Status validate(const ITensorInfo *input, + const ITensorInfo *input_to_forget_weights, + const ITensorInfo *input_to_cell_weights, + const ITensorInfo *input_to_output_weights, + const ITensorInfo *recurrent_to_forget_weights, + const ITensorInfo *recurrent_to_cell_weights, + const ITensorInfo *recurrent_to_output_weights, + const ITensorInfo *forget_gate_bias, + const ITensorInfo *cell_bias, + const ITensorInfo *output_gate_bias, + const ITensorInfo *cell_state_in, + const ITensorInfo *output_state_in, + const ITensorInfo *cell_state_out, + const ITensorInfo *output_state_out, + const ITensorInfo *output, const LSTMParams<ITensorInfo> &lstm_params); // Inherited methods overridden: @@ -218,10 +236,17 @@ private: * @param[in] mm_res_info Tensor info to be used to initialize output stage result tensor. * */ - void configure_mm(NEGEMMLowpMatrixMultiplyCore &mm, NEGEMMLowpOutputStage &outstage, GEMMLowpOutputStageInfo &gemmlowp_info, - const ITensor *mm_input, const ITensor *mm_weights, const ITensor *bias, Tensor *mm_res, - Tensor *outstage_res, float gemmlowp_scale, - const TensorInfo &mm_res_info, const TensorInfo &outstage_tensor_info); + void configure_mm(NEGEMMLowpMatrixMultiplyCore &mm, + NEGEMMLowpOutputStage &outstage, + GEMMLowpOutputStageInfo &gemmlowp_info, + const ITensor *mm_input, + const ITensor *mm_weights, + const ITensor *bias, + Tensor *mm_res, + Tensor *outstage_res, + float gemmlowp_scale, + const TensorInfo &mm_res_info, + const TensorInfo &outstage_tensor_info); MemoryGroup _memory_group; @@ -230,8 +255,8 @@ private: { static constexpr uint32_t max_dimension_supported = 2; - ITensor *_src{ nullptr }; - ITensor *_dst{ nullptr }; + ITensor *_src{nullptr}; + ITensor *_dst{nullptr}; size_t _row_size{}; Window _window{}; @@ -335,19 +360,16 @@ private: NECopy _copy_output; // Tensor pointers - const ITensor *_input_to_input_weights - { - nullptr - }; - const ITensor *_recurrent_to_input_weights{ nullptr }; - const ITensor *_projection_bias{ nullptr }; - const ITensor *_input_to_forget_weights{ nullptr }; - const ITensor *_input_to_cell_weights{ nullptr }; - const ITensor *_input_to_output_weights{ nullptr }; - const ITensor *_recurrent_to_forget_weights{ nullptr }; - const ITensor *_recurrent_to_cell_weights{ nullptr }; - const ITensor *_recurrent_to_output_weights{ nullptr }; - const ITensor *_projection_weights{ nullptr }; + const ITensor *_input_to_input_weights{nullptr}; + const ITensor *_recurrent_to_input_weights{nullptr}; + const ITensor *_projection_bias{nullptr}; + const ITensor *_input_to_forget_weights{nullptr}; + const ITensor *_input_to_cell_weights{nullptr}; + const ITensor *_input_to_output_weights{nullptr}; + const ITensor *_recurrent_to_forget_weights{nullptr}; + const ITensor *_recurrent_to_cell_weights{nullptr}; + const ITensor *_recurrent_to_output_weights{nullptr}; + const ITensor *_projection_weights{nullptr}; std::array<const ITensor *, _layer_norm_count> _layer_norm_weights{}; std::array<const ITensor *, _layer_norm_count> _layer_norm_bias{}; @@ -382,66 +404,66 @@ private: return _layer_norms[getGateIndex(g)]; } - void configure_layer_norm(LayerNormGate g, const ITensor *in); + void configure_layer_norm(LayerNormGate g, const ITensor *in); static Status validate_layer_norm(const ITensorInfo &in, const ITensorInfo &weight, const ITensorInfo &bias); // Temporary tensors - Tensor _input_to_forget_weights_f32{ nullptr }; - Tensor _input_to_forget_weights_symm8{ nullptr }; + Tensor _input_to_forget_weights_f32{nullptr}; + Tensor _input_to_forget_weights_symm8{nullptr}; - Tensor _input_to_forget_weights_transposed{ nullptr }; - Tensor _input_to_cell_weights_transposed{ nullptr }; - Tensor _input_to_output_weights_transposed{ nullptr }; - Tensor _input_to_input_weights_transposed{ nullptr }; - Tensor _recurrent_to_forget_weights_transposed{ nullptr }; - Tensor _recurrent_to_cell_weights_transposed{ nullptr }; - Tensor _recurrent_to_output_weights_transposed{ nullptr }; - Tensor _recurrent_to_input_weights_transposed{ nullptr }; - Tensor _projection_weights_transposed{ nullptr }; - Tensor _input_to_input_eff_bias{ nullptr }; - Tensor _recurrent_to_input_eff_bias{ nullptr }; - Tensor _input_to_forget_eff_bias{ nullptr }; - Tensor _recurrent_to_forget_eff_bias{ nullptr }; - Tensor _input_to_cell_eff_bias{ nullptr }; - Tensor _recurrent_to_cell_eff_bias{ nullptr }; - Tensor _input_to_output_eff_bias{ nullptr }; - Tensor _recurrent_to_output_eff_bias{ nullptr }; - Tensor _projection_reduction_res{ nullptr }; - Tensor _projection_eff_bias{ nullptr }; - Tensor _mm_input_to_forget_res{ nullptr }; - Tensor _mm_recurrent_to_forget_res{ nullptr }; - Tensor _mul_cell_to_forget_res{ nullptr }; - Tensor _input_to_forget_outstage_res{ nullptr }; - Tensor _cell_to_forget_outstage_res{ nullptr }; - Tensor _recurrent_to_forget_outstage_res{ nullptr }; - Tensor _forget_gate{ nullptr }; - Tensor _mm_input_to_cell_res{ nullptr }; - Tensor _input_to_cell_outstage_res{ nullptr }; - Tensor _mm_recurrent_to_cell_res{ nullptr }; - Tensor _recurrent_to_cell_outstage_res{ nullptr }; - Tensor _cell_gate{ nullptr }; - Tensor _mul_input_cell_res{ nullptr }; - Tensor _mm_input_to_input_res{ nullptr }; - Tensor _input_to_input_outstage_res{ nullptr }; - Tensor _mm_recurrent_to_input_res{ nullptr }; - Tensor _mul_cell_to_input_res{ nullptr }; - Tensor _cell_to_input_outstage_res{ nullptr }; - Tensor _recurrent_to_input_outstage_res{ nullptr }; - Tensor _input_gate{ nullptr }; - Tensor _mm_input_to_output_res{ nullptr }; - Tensor _input_to_output_outstage_res{ nullptr }; - Tensor _mm_recurrent_to_output_res{ nullptr }; - Tensor _mul_cell_to_output_res{ nullptr }; - Tensor _cell_to_output_outstage_res{ nullptr }; - Tensor _recurrent_to_output_outstage_res{ nullptr }; - Tensor _output_gate{ nullptr }; - Tensor _hidden_mul_res{ nullptr }; - Tensor _hidden_gate{ nullptr }; - Tensor _mm_projection_res{ nullptr }; - Tensor _projection_outstage_res{ nullptr }; - Tensor _projection_out_res{ nullptr }; - Tensor _projection_accumulate_res{ nullptr }; - Tensor _ones{ nullptr }; + Tensor _input_to_forget_weights_transposed{nullptr}; + Tensor _input_to_cell_weights_transposed{nullptr}; + Tensor _input_to_output_weights_transposed{nullptr}; + Tensor _input_to_input_weights_transposed{nullptr}; + Tensor _recurrent_to_forget_weights_transposed{nullptr}; + Tensor _recurrent_to_cell_weights_transposed{nullptr}; + Tensor _recurrent_to_output_weights_transposed{nullptr}; + Tensor _recurrent_to_input_weights_transposed{nullptr}; + Tensor _projection_weights_transposed{nullptr}; + Tensor _input_to_input_eff_bias{nullptr}; + Tensor _recurrent_to_input_eff_bias{nullptr}; + Tensor _input_to_forget_eff_bias{nullptr}; + Tensor _recurrent_to_forget_eff_bias{nullptr}; + Tensor _input_to_cell_eff_bias{nullptr}; + Tensor _recurrent_to_cell_eff_bias{nullptr}; + Tensor _input_to_output_eff_bias{nullptr}; + Tensor _recurrent_to_output_eff_bias{nullptr}; + Tensor _projection_reduction_res{nullptr}; + Tensor _projection_eff_bias{nullptr}; + Tensor _mm_input_to_forget_res{nullptr}; + Tensor _mm_recurrent_to_forget_res{nullptr}; + Tensor _mul_cell_to_forget_res{nullptr}; + Tensor _input_to_forget_outstage_res{nullptr}; + Tensor _cell_to_forget_outstage_res{nullptr}; + Tensor _recurrent_to_forget_outstage_res{nullptr}; + Tensor _forget_gate{nullptr}; + Tensor _mm_input_to_cell_res{nullptr}; + Tensor _input_to_cell_outstage_res{nullptr}; + Tensor _mm_recurrent_to_cell_res{nullptr}; + Tensor _recurrent_to_cell_outstage_res{nullptr}; + Tensor _cell_gate{nullptr}; + Tensor _mul_input_cell_res{nullptr}; + Tensor _mm_input_to_input_res{nullptr}; + Tensor _input_to_input_outstage_res{nullptr}; + Tensor _mm_recurrent_to_input_res{nullptr}; + Tensor _mul_cell_to_input_res{nullptr}; + Tensor _cell_to_input_outstage_res{nullptr}; + Tensor _recurrent_to_input_outstage_res{nullptr}; + Tensor _input_gate{nullptr}; + Tensor _mm_input_to_output_res{nullptr}; + Tensor _input_to_output_outstage_res{nullptr}; + Tensor _mm_recurrent_to_output_res{nullptr}; + Tensor _mul_cell_to_output_res{nullptr}; + Tensor _cell_to_output_outstage_res{nullptr}; + Tensor _recurrent_to_output_outstage_res{nullptr}; + Tensor _output_gate{nullptr}; + Tensor _hidden_mul_res{nullptr}; + Tensor _hidden_gate{nullptr}; + Tensor _mm_projection_res{nullptr}; + Tensor _projection_outstage_res{nullptr}; + Tensor _projection_out_res{nullptr}; + Tensor _projection_accumulate_res{nullptr}; + Tensor _ones{nullptr}; std::array<Tensor, _layer_norm_count> _layer_norm_output{}; inline Tensor &get_layer_norm_output(LayerNormGate g) @@ -449,15 +471,15 @@ private: return _layer_norm_output[getGateIndex(g)]; } - bool _is_prepared{ false }; - bool _has_cifg{ false }; - bool _has_cell_clipping{ false }; - bool _has_projection{ false }; - bool _has_projection_clipping{ false }; - bool _has_peephole{ false }; - bool _has_layer_norm{ false }; - bool _projection_tensor_copy_required{ false }; - bool _convert_input_to_forget_weights_to_qsymm8{ false }; + bool _is_prepared{false}; + bool _has_cifg{false}; + bool _has_cell_clipping{false}; + bool _has_projection{false}; + bool _has_projection_clipping{false}; + bool _has_peephole{false}; + bool _has_layer_norm{false}; + bool _projection_tensor_copy_required{false}; + bool _convert_input_to_forget_weights_to_qsymm8{false}; }; } // namespace arm_compute #endif /* ARM_COMPUTE_NEQLSTMLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NERNNLayer.h b/arm_compute/runtime/NEON/functions/NERNNLayer.h index 667d3144ac..af7f464ac9 100644 --- a/arm_compute/runtime/NEON/functions/NERNNLayer.h +++ b/arm_compute/runtime/NEON/functions/NERNNLayer.h @@ -72,7 +72,13 @@ public: * @param[in,out] hidden_state Output tensor of shape [num_units, batch_size]. Data types supported: Same as @p input * @param[in] info Activation layer parameter. */ - void configure(const ITensor *input, const ITensor *weights, const ITensor *recurrent_weights, const ITensor *bias, ITensor *hidden_state, ITensor *output, ActivationLayerInfo &info); + void configure(const ITensor *input, + const ITensor *weights, + const ITensor *recurrent_weights, + const ITensor *bias, + ITensor *hidden_state, + ITensor *output, + ActivationLayerInfo &info); /** Initialize the function * * @param[in] input Input is a 2-D tensor of shape [input_size, batch_size]. Data types supported: F16/F32 @@ -85,7 +91,12 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *recurrent_weights, const ITensorInfo *bias, const ITensorInfo *hidden_state, const ITensorInfo *output, + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *recurrent_weights, + const ITensorInfo *bias, + const ITensorInfo *hidden_state, + const ITensorInfo *output, const ActivationLayerInfo &info); // Inherited methods overridden: diff --git a/arm_compute/runtime/NEON/functions/NEROIAlignLayer.h b/arm_compute/runtime/NEON/functions/NEROIAlignLayer.h index ea1af4daea..b06ebe899d 100644 --- a/arm_compute/runtime/NEON/functions/NEROIAlignLayer.h +++ b/arm_compute/runtime/NEON/functions/NEROIAlignLayer.h @@ -77,7 +77,10 @@ public: * * @return a Status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *rois, ITensorInfo *output, const ROIPoolingLayerInfo &pool_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *rois, + ITensorInfo *output, + const ROIPoolingLayerInfo &pool_info); }; } // namespace arm_compute #endif /* ARM_COMPUTE_NEROIALIGNLAYER_H */ diff --git a/arm_compute/runtime/NEON/functions/NEROIPoolingLayer.h b/arm_compute/runtime/NEON/functions/NEROIPoolingLayer.h index 2992b3eb95..929111ad4b 100644 --- a/arm_compute/runtime/NEON/functions/NEROIPoolingLayer.h +++ b/arm_compute/runtime/NEON/functions/NEROIPoolingLayer.h @@ -26,6 +26,7 @@ #include "arm_compute/core/IArray.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -73,7 +74,8 @@ public: * @note The z dimensions of @p output tensor and @p input tensor must be the same. * @note The fourth dimension of @p output tensor must be the same as the number of elements in @p rois array. */ - void configure(const ITensor *input, const ITensor *rois, const ITensor *output, const ROIPoolingLayerInfo &pool_info); + void + configure(const ITensor *input, const ITensor *rois, const ITensor *output, const ROIPoolingLayerInfo &pool_info); // Inherited methods overridden: void run() override; @@ -91,7 +93,10 @@ public: * @note The fourth dimension of @p output tensor must be the same as the number of elements in @p rois array. * @return a Status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *rois, const ITensorInfo *output, const ROIPoolingLayerInfo &pool_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *rois, + const ITensorInfo *output, + const ROIPoolingLayerInfo &pool_info); private: std::unique_ptr<NEROIPoolingLayerKernel> _roi_kernel; diff --git a/arm_compute/runtime/NEON/functions/NERange.h b/arm_compute/runtime/NEON/functions/NERange.h index cb14c8fdde..609456a4ef 100644 --- a/arm_compute/runtime/NEON/functions/NERange.h +++ b/arm_compute/runtime/NEON/functions/NERange.h @@ -26,6 +26,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute diff --git a/arm_compute/runtime/NEON/functions/NEReduceMean.h b/arm_compute/runtime/NEON/functions/NEReduceMean.h index caaee8284a..5b8d8cdf2b 100644 --- a/arm_compute/runtime/NEON/functions/NEReduceMean.h +++ b/arm_compute/runtime/NEON/functions/NEReduceMean.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NEON_REDUCE_MEAN_H #define ARM_COMPUTE_NEON_REDUCE_MEAN_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/MemoryGroup.h" #include "arm_compute/runtime/NEON/functions/NEReductionOperation.h" #include "arm_compute/runtime/NEON/functions/NEReshapeLayer.h" @@ -81,7 +80,8 @@ public: * * @return A status */ - static Status validate(const ITensorInfo *input, const Coordinates &reduction_axis, bool keep_dims, const ITensorInfo *output); + static Status + validate(const ITensorInfo *input, const Coordinates &reduction_axis, bool keep_dims, const ITensorInfo *output); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEReductionOperation.h b/arm_compute/runtime/NEON/functions/NEReductionOperation.h index 533c10adcf..f5391a6d0e 100644 --- a/arm_compute/runtime/NEON/functions/NEReductionOperation.h +++ b/arm_compute/runtime/NEON/functions/NEReductionOperation.h @@ -25,9 +25,9 @@ #define ARM_COMPUTE_NEREDUCTIONOPERATION_H #include "arm_compute/runtime/IFunction.h" - #include "arm_compute/runtime/NEON/functions/NEReshapeLayer.h" #include "arm_compute/runtime/Tensor.h" + #include <memory> namespace arm_compute @@ -88,7 +88,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, unsigned int axis, ReductionOperation op, bool keep_dims = true); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + unsigned int axis, + ReductionOperation op, + bool keep_dims = true); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NEReorderLayer.h b/arm_compute/runtime/NEON/functions/NEReorderLayer.h index eb777f1925..e3fa7b9c16 100644 --- a/arm_compute/runtime/NEON/functions/NEReorderLayer.h +++ b/arm_compute/runtime/NEON/functions/NEReorderLayer.h @@ -66,7 +66,10 @@ public: * @param[in] input_wf WeightFormat of input. * @param[in] output_wf WeightFormat of output. */ - void configure(const ITensor *input, ITensor *output, arm_compute::WeightFormat input_wf, arm_compute::WeightFormat output_wf); + void configure(const ITensor *input, + ITensor *output, + arm_compute::WeightFormat input_wf, + arm_compute::WeightFormat output_wf); /** Static function to check if given info will lead to a valid configuration of @ref NEReorderLayer * @@ -74,7 +77,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, arm_compute::WeightFormat input_wf, arm_compute::WeightFormat output_wf); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + arm_compute::WeightFormat input_wf, + arm_compute::WeightFormat output_wf); // Inherited methods overridden: void run() override; @@ -85,4 +91,4 @@ private: } // namespace arm_compute #endif /* ACL_ARM_COMPUTE_RUNTIME_NEON_FUNCTIONS_NEREORDERLAYER */ -#endif // defined(__aarch64__)
\ No newline at end of file +#endif // defined(__aarch64__) diff --git a/arm_compute/runtime/NEON/functions/NEReverse.h b/arm_compute/runtime/NEON/functions/NEReverse.h index f58eb2373f..e03e415068 100644 --- a/arm_compute/runtime/NEON/functions/NEReverse.h +++ b/arm_compute/runtime/NEON/functions/NEReverse.h @@ -24,9 +24,8 @@ #ifndef ACL_ARM_COMPUTE_RUNTIME_NEON_FUNCTIONS_NEREVERSE_H #define ACL_ARM_COMPUTE_RUNTIME_NEON_FUNCTIONS_NEREVERSE_H -#include "arm_compute/runtime/NEON/INESimpleFunctionNoBorder.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/NEON/INESimpleFunctionNoBorder.h" namespace arm_compute { @@ -68,7 +67,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *axis, const bool use_inverted_axis = false); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const ITensorInfo *axis, + const bool use_inverted_axis = false); }; } // namespace arm_compute #endif // ACL_ARM_COMPUTE_RUNTIME_NEON_FUNCTIONS_NEREVERSE_H diff --git a/arm_compute/runtime/NEON/functions/NEScale.h b/arm_compute/runtime/NEON/functions/NEScale.h index 0920ff3802..72dfa3bda4 100644 --- a/arm_compute/runtime/NEON/functions/NEScale.h +++ b/arm_compute/runtime/NEON/functions/NEScale.h @@ -24,10 +24,9 @@ #ifndef ARM_COMPUTE_NESCALEIMAGE_H #define ARM_COMPUTE_NESCALEIMAGE_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/KernelDescriptors.h" #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> diff --git a/arm_compute/runtime/NEON/functions/NESlice.h b/arm_compute/runtime/NEON/functions/NESlice.h index ac79a5c633..70a688d3b0 100644 --- a/arm_compute/runtime/NEON/functions/NESlice.h +++ b/arm_compute/runtime/NEON/functions/NESlice.h @@ -85,7 +85,8 @@ public: * * @return A status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); + static Status + validate(const ITensorInfo *input, const ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); // Inherited methods overridden: void run() override; @@ -129,7 +130,8 @@ public: * * @return A status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); + static Status + validate(const ITensorInfo *input, const ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); }; } // namespace experimental } // namespace arm_compute diff --git a/arm_compute/runtime/NEON/functions/NESpaceToBatchLayer.h b/arm_compute/runtime/NEON/functions/NESpaceToBatchLayer.h index ad8c1467d0..5dee61a4a8 100644 --- a/arm_compute/runtime/NEON/functions/NESpaceToBatchLayer.h +++ b/arm_compute/runtime/NEON/functions/NESpaceToBatchLayer.h @@ -24,9 +24,9 @@ #ifndef ARM_COMPUTE_NESPACETOBATCHLAYER_H #define ARM_COMPUTE_NESPACETOBATCHLAYER_H +#include "arm_compute/core/Types.h" #include "arm_compute/runtime/IFunction.h" -#include "arm_compute/core/Types.h" #include <memory> namespace arm_compute @@ -82,7 +82,12 @@ public: * @param[in] padding_right The padding at the end of every dimension of the output tensor. * @param[out] output Tensor output. Data types supported: same as @p input */ - void configure(const ITensor *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left, const Size2D &padding_right, ITensor *output); + void configure(const ITensor *input, + const int block_shape_x, + const int block_shape_y, + const Size2D &padding_left, + const Size2D &padding_right, + ITensor *output); /** Static function to check if given info will lead to a valid configuration of @ref NESpaceToBatchLayer * * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All. @@ -92,7 +97,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *block_shape, const ITensorInfo *paddings, const ITensorInfo *output); + static Status validate(const ITensorInfo *input, + const ITensorInfo *block_shape, + const ITensorInfo *paddings, + const ITensorInfo *output); /** Static function to check if given info will lead to a valid configuration of @ref NESpaceToBatchLayer (Static block shape and paddings) * * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All. @@ -104,7 +112,12 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left, const Size2D &padding_right, const ITensorInfo *output); + static Status validate(const ITensorInfo *input, + const int block_shape_x, + const int block_shape_y, + const Size2D &padding_left, + const Size2D &padding_right, + const ITensorInfo *output); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/NEON/functions/NESplit.h b/arm_compute/runtime/NEON/functions/NESplit.h index 206f299c06..36358a7094 100644 --- a/arm_compute/runtime/NEON/functions/NESplit.h +++ b/arm_compute/runtime/NEON/functions/NESplit.h @@ -26,7 +26,6 @@ #include "arm_compute/core/ITensor.h" #include "arm_compute/core/Types.h" - #include "arm_compute/runtime/CPP/functions/CPPSplit.h" #include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/NEON/functions/NESlice.h" diff --git a/arm_compute/runtime/NEON/functions/NEStridedSlice.h b/arm_compute/runtime/NEON/functions/NEStridedSlice.h index 4b14d946f6..fa1113ffec 100644 --- a/arm_compute/runtime/NEON/functions/NEStridedSlice.h +++ b/arm_compute/runtime/NEON/functions/NEStridedSlice.h @@ -71,9 +71,14 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - void configure(const ITensor *input, ITensor *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + void configure(const ITensor *input, + ITensor *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); /** Static function to check if given info will lead to a valid configuration of @ref NEStridedSlice * @@ -89,9 +94,14 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); // Inherited methods overridden: void run() override; @@ -121,9 +131,14 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - void configure(const ITensorInfo *input, ITensorInfo *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + void configure(const ITensorInfo *input, + ITensorInfo *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); /** Static function to check if given info will lead to a valid configuration of @ref NEStridedSlice * @@ -139,9 +154,14 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); }; } // namespace experimental } // namespace arm_compute diff --git a/arm_compute/runtime/NEON/functions/NETile.h b/arm_compute/runtime/NEON/functions/NETile.h index 915e5aa1da..001a0a4128 100644 --- a/arm_compute/runtime/NEON/functions/NETile.h +++ b/arm_compute/runtime/NEON/functions/NETile.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NETILE_H #define ARM_COMPUTE_NETILE_H -#include "arm_compute/runtime/NEON/INESimpleFunctionNoBorder.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/NEON/INESimpleFunctionNoBorder.h" namespace arm_compute { diff --git a/arm_compute/runtime/NEON/functions/NETranspose.h b/arm_compute/runtime/NEON/functions/NETranspose.h index 581fe74309..5d2d1f1b01 100644 --- a/arm_compute/runtime/NEON/functions/NETranspose.h +++ b/arm_compute/runtime/NEON/functions/NETranspose.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_NETRANSPOSE_H #define ARM_COMPUTE_NETRANSPOSE_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -83,4 +82,4 @@ private: std::unique_ptr<Impl> _impl; }; } // namespace arm_compute -#endif /* ARM_COMPUTE_NETRANSPOSE_H */
\ No newline at end of file +#endif /* ARM_COMPUTE_NETRANSPOSE_H */ diff --git a/arm_compute/runtime/NEON/functions/NEUnstack.h b/arm_compute/runtime/NEON/functions/NEUnstack.h index 079fee5b9e..e1af96d08d 100644 --- a/arm_compute/runtime/NEON/functions/NEUnstack.h +++ b/arm_compute/runtime/NEON/functions/NEUnstack.h @@ -26,7 +26,6 @@ #include "arm_compute/core/Types.h" #include "arm_compute/runtime/IFunction.h" - #include "arm_compute/runtime/NEON/functions/NEStridedSlice.h" #include <memory> diff --git a/arm_compute/runtime/NEON/functions/NEWinogradConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEWinogradConvolutionLayer.h index f6f0185e7d..7f4e354362 100644 --- a/arm_compute/runtime/NEON/functions/NEWinogradConvolutionLayer.h +++ b/arm_compute/runtime/NEON/functions/NEWinogradConvolutionLayer.h @@ -24,10 +24,9 @@ #ifndef ARM_COMPUTE_NEWINOGRADCONVOLUTIONLAYER_H #define ARM_COMPUTE_NEWINOGRADCONVOLUTIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/Tensor.h" #include <memory> @@ -87,8 +86,13 @@ public: * @param[in] enable_fast_math (Optional) Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation * available which may introduce a drop of accuracy as well. Default is false */ - void configure(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo(), - bool enable_fast_math = false); + void configure(const ITensor *input, + const ITensor *weights, + const ITensor *biases, + ITensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); // Inherited methods overridden: void run() override; @@ -100,8 +104,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); private: struct Impl; |