diff options
Diffstat (limited to 'arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h')
| -rw-r--r-- | arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h | 56 |
1 files changed, 42 insertions, 14 deletions
diff --git a/arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h b/arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h index d8710a461f..3e8929c5ad 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2019-2020 ARM Limited. + * Copyright (c) 2019-2021 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -24,9 +24,10 @@ #ifndef ARM_COMPUTE_CLGEMMDECONVOLUTIONLAYER_H #define ARM_COMPUTE_CLGEMMDECONVOLUTIONLAYER_H -#include "arm_compute/core/CL/kernels/CLDeconvolutionReshapeOutputKernel.h" #include "arm_compute/runtime/CL/CLTensor.h" #include "arm_compute/runtime/CL/functions/CLConvolutionLayer.h" +#include "arm_compute/runtime/CL/functions/CLGEMM.h" +#include "arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h" #include "arm_compute/runtime/CL/functions/CLGEMMLowpOutputStage.h" #include "arm_compute/runtime/CL/functions/CLPermute.h" #include "arm_compute/runtime/CL/functions/CLReshapeLayer.h" @@ -40,6 +41,7 @@ namespace arm_compute { +class CLDeconvolutionReshapeOutputKernel; class ICLTensor; /** Function to run the deconvolution layer through a call to GEMM. * @@ -89,8 +91,21 @@ public: CLGEMMDeconvolutionLayer &operator=(const CLGEMMDeconvolutionLayer &) = delete; /** Default move assignment operator */ CLGEMMDeconvolutionLayer &operator=(CLGEMMDeconvolutionLayer &&) = default; + /** Default desctructor */ + ~CLGEMMDeconvolutionLayer(); /** Set the input, weights, biases and output tensors. * + * Valid data layouts: + * - NHWC + * + * 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,out] input Input tensor. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. * Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32. Data layout supported: NHWC * @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data type supported: Same as @p input. Data layout supported: same as @p input. @@ -98,7 +113,11 @@ public: * @param[out] output Output tensor. The output has the same number of dimensions as the @p input. Data layout supported: same as @p input. * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, this is described in @ref PadStrideInfo. This function supports only stride_x = weights.width && stride_y = weights.height. Moreover, padding is not supported. */ - void configure(const ICLTensor *input, const ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &deconv_info); + void configure(const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *bias, + ICLTensor *output, + const PadStrideInfo &deconv_info); /** Set the input, weights, biases and output tensors. * * @param[in] compile_context The compile context to be used. @@ -109,7 +128,12 @@ public: * @param[out] output Output tensor. The output has the same number of dimensions as the @p input. Data layout supported: same as @p input. * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, this is described in @ref PadStrideInfo. This function supports only stride_x = weights.width && stride_y = weights.height. Moreover, padding is not supported. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &deconv_info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *bias, + ICLTensor *output, + const PadStrideInfo &deconv_info); /** Static function to check if given info will lead to a valid configuration of @ref CLDeconvolutionLayer * * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. @@ -121,7 +145,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *bias, const ITensorInfo *output, const PadStrideInfo &deconv_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *bias, + const ITensorInfo *output, + const PadStrideInfo &deconv_info); // Inherited methods overridden: void run() override; @@ -130,15 +158,15 @@ public: private: MemoryGroup _memory_group; - CLGEMM _mm_gemm; - CLGEMMLowpMatrixMultiplyCore _mm_gemmlowp; - CLGEMMLowpOutputStage _gemmlowp_output_stage; - CLPermute _permute_input_to_nhwc; - CLPermute _permute_weights_to_nhwc; - CLReshapeLayer _reshape_weights; - CLTranspose _transpose_weights; - CLDeconvolutionReshapeOutputKernel _deconv_reshape; - CLSlice _slice_gemm; + CLGEMM _mm_gemm; + CLGEMMLowpMatrixMultiplyCore _mm_gemmlowp; + CLGEMMLowpOutputStage _gemmlowp_output_stage; + CLPermute _permute_input_to_nhwc; + CLPermute _permute_weights_to_nhwc; + CLReshapeLayer _reshape_weights; + CLTranspose _transpose_weights; + std::unique_ptr<CLDeconvolutionReshapeOutputKernel> _deconv_reshape; + CLSlice _slice_gemm; CLTensor _gemmlowp_final; CLTensor _reshaped_weights; |