From d9cdf1402fb7e1231f56c1d5549639b423e4e323 Mon Sep 17 00:00:00 2001 From: Michele Di Giorgio Date: Fri, 2 Jul 2021 15:17:08 +0100 Subject: Port NEFullyConnectedLayer to memory injecting interface Resolves: COMPMID-4501 Change-Id: Ib61b3d06974009e501b3fb86467735427e13a94a Signed-off-by: Michele Di Giorgio Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5931 Comments-Addressed: Arm Jenkins Tested-by: Arm Jenkins Reviewed-by: Georgios Pinitas --- src/runtime/cpu/operators/CpuFullyConnected.h | 145 ++++++++++++++++++++++++++ 1 file changed, 145 insertions(+) create mode 100644 src/runtime/cpu/operators/CpuFullyConnected.h (limited to 'src/runtime/cpu/operators/CpuFullyConnected.h') diff --git a/src/runtime/cpu/operators/CpuFullyConnected.h b/src/runtime/cpu/operators/CpuFullyConnected.h new file mode 100644 index 0000000000..954a7b7ffc --- /dev/null +++ b/src/runtime/cpu/operators/CpuFullyConnected.h @@ -0,0 +1,145 @@ +/* + * Copyright (c) 2021 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef ARM_COMPUTE_CPU_FULLY_CONNECTED_H +#define ARM_COMPUTE_CPU_FULLY_CONNECTED_H + +#include "src/runtime/cpu/ICpuOperator.h" + +#include "arm_compute/core/TensorInfo.h" + +#include + +namespace arm_compute +{ +namespace cpu +{ +// Forward declarations +class CpuConvertFullyConnectedWeights; +class CpuFlatten; +class CpuGemm; +class CpuGemmLowpMatrixMultiplyCore; +namespace kernels +{ +class CpuTransposeKernel; +} // namespace kernels +/** Basic function to compute a Fully Connected layer. This function calls the following kernels: + * -# @ref kernels::CpuIm2ColKernel (called when the input comes from a convolutional layer) + * -# @ref kernels::CpuTransposeKernel (if @p are_weights_reshaped is set to false and transpose_weights is set to true ) (called once) + * -# @ref CpuGemm or @ref CpuGemmLowpMatrixMultiplyCore (if quantized asymmetric) + * -# @ref kernels::CpuGemmMatrixAdditionKernel or @ref CpuGemmLowpOutputStage (if quantized asymmetric) (if @p biases is not equal to nullptr) + * + * @note The fully connected layer accepts "weights" tensors only with 2 dimensions. + */ +class CpuFullyConnected : public ICpuOperator +{ +public: + /** Constructor */ + CpuFullyConnected(); + /** Destructor */ + ~CpuFullyConnected(); + /** Set the input and output tensors. + * + * Valid data layouts: + * - NHWC + * - NCHW + * + * Valid data type configurations: + * |src0 |src1 |src2 |dst | + * |:--------------|:------------------|:------|:--------------| + * |F16 |F16 |F16 |F16 | + * |F32 |F32 |F32 |F32 | + * |QASYMM8 |QASYMM8 |S32 |QASYMM8 | + * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED | + * + * @param[in] src Source tensor info. Data type supported: QASYMM8/QASYMM8_SIGNED/F16/F32. + * @param[in] weights Weights tensor info. The weights must be 2 dimensional. + * If this function is called after a Convolution Layer, the (transposed) weights will have as many rows as the product of the first 3 input's dimensions. + * If it is called after another FullyConnected Layer, the (transposed) weights will have as many rows as the input's first dimension. + * Data type supported: Same as @p src. + * @param[in] biases Bias tensor info. Can be nullptr. Data type supported: Same as @p weights, S32 if @p weights is QASYMM8/QASYMM8_SIGNED. + * @param[out] dst Destination tensor info. Its shape should be equal to the output of a matrix multiplication between: + * - The output of im2col on the input and the (transposed) 2D weights, if the function is called after a Convolution Layer + * - The input tensor and the (transposed) 2D weights, if the function is called after another FullyConnected Layer. + * Data type supported: Same as @p src. + * @param[in] fc_info (Optional) Fully connected layer additional info + */ + void configure(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, ITensorInfo *dst, + FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo()); + /** Static function to check if given info will lead to a valid configuration of @ref CpuFullyConnected + * + * Similar to @ref CpuFullyConnected + * + * @return a status + */ + static Status validate(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *dst, + FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo()); + + //Inherited methods override + void run(ITensorPack &tensors) override; + void prepare(ITensorPack &tensors) override; + experimental::MemoryRequirements workspace() const override; + +private: + void configure_fc_fc(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, ITensorInfo *dst, const ActivationLayerInfo &act); + void configure_conv_fc(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, ITensorInfo *dst, const ActivationLayerInfo &act); + void configure_mm(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, ITensorInfo *dst, const ActivationLayerInfo &act); + + enum AuxTensorIdx + { + AsmGemmWorkspace = 0, + Pretranspose, + GemmTemp1, // Both CpuGemm and CpuGemmLowpMatrixMultiplyCore + GemmTemp2, // Both CpuGemm and CpuGemmLowpMatrixMultiplyCore + GemmTemp3, // Both CpuGemm and CpuGemmLowpMatrixMultiplyCore + GemmTemp4, // CpuGemmLowpMatrixMultiplyCore only + GemmTemp5, // CpuGemmLowpMatrixMultiplyCore only + GemmTemp6, // CpuGemmLowpMatrixMultiplyCore only + GemmTemp7, // CpuGemmLowpMatrixMultiplyCore only + TransposedWeights, + ConvertedWeights, + FlattenedSrc, + Count + }; + + std::unique_ptr _flatten; + std::unique_ptr _convert_weights; + std::unique_ptr _transpose_weights; + std::unique_ptr _mm_gemm; + std::unique_ptr _mm_gemmlowp; + + TensorInfo _flattened_src; + TensorInfo _converted_weights; + TensorInfo _reshaped_weights; + + experimental::MemoryRequirements _aux_mem; + + bool _are_weights_converted; + bool _are_weights_reshaped; + bool _is_fc_after_conv; + bool _is_quantized_asymmetric; + bool _is_prepared; +}; +} // namespace cpu +} // namespace arm_compute +#endif /* ARM_COMPUTE_CPU_FULLY_CONNECTED_H */ -- cgit v1.2.1