1 /* 2 * Copyright (c) 2017-2020 Arm Limited. 3 * 4 * SPDX-License-Identifier: MIT 5 * 6 * Permission is hereby granted, free of charge, to any person obtaining a copy 7 * of this software and associated documentation files (the "Software"), to 8 * deal in the Software without restriction, including without limitation the 9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or 10 * sell copies of the Software, and to permit persons to whom the Software is 11 * furnished to do so, subject to the following conditions: 12 * 13 * The above copyright notice and this permission notice shall be included in all 14 * copies or substantial portions of the Software. 15 * 16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 22 * SOFTWARE. 23 */ 24 #ifndef ARM_COMPUTE_CLCOL2IMKERNEL_H 25 #define ARM_COMPUTE_CLCOL2IMKERNEL_H 26 27 #include "src/core/CL/ICLKernel.h" 28 29 namespace arm_compute 30 { 31 class ICLTensor; 32 33 /** Interface for the col2im reshaping kernel. 34 * 35 * Rearranges each matrix column into image blocks. It's the inverse operation of @ref CLIm2ColKernel. 36 * 37 * For example, a vector of 9 elements can be reshaped to a block(image) of 3x3: 38 * 39 * @f[ 40 * \left( \begin{array}{ccccccccc} 41 * a0 & a1 & a2 & a3 & a4 & a5 & a6 & a7 & a8 \\ 42 * \end{array} \right) 43 * \rightarrow 44 * \left( \begin{array}{ccc} 45 * a0 & a1 & a2 \\ 46 * a3 & a4 & a5 \\ 47 * a6 & a7 & a8 \\ 48 * \end{array} \right) 49 * @f] 50 */ 51 class CLCol2ImKernel : public ICLKernel 52 { 53 public: 54 /** Default constructor */ 55 CLCol2ImKernel(); 56 /** Prevent instances of this class from being copied (As this class contains pointers) */ 57 CLCol2ImKernel(const CLCol2ImKernel &) = delete; 58 /** Prevent instances of this class from being copied (As this class contains pointers) */ 59 CLCol2ImKernel &operator=(const CLCol2ImKernel &) = delete; 60 /** Allow instances of this class to be moved */ 61 CLCol2ImKernel(CLCol2ImKernel &&) = default; 62 /** Allow instances of this class to be moved */ 63 CLCol2ImKernel &operator=(CLCol2ImKernel &&) = default; 64 /** Default destructor */ 65 ~CLCol2ImKernel() = default; 66 /** Set the input and output of the kernel. 67 * 68 * @param[in] input The input tensor to convert. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32 69 * @param[out] output The output tensor. 3 lower dimensions represent a single output [width, height, OFM], 70 * while the rest represent batch of outputs. Data types supported: Same as @p input. Data layout: NCHW 71 * @param[in] convolved_dims Output convolved dimensions. 72 * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution 73 */ 74 void configure(const ICLTensor *input, ICLTensor *output, const Size2D &convolved_dims, unsigned int num_groups = 1); 75 /** Set the input and output of the kernel. 76 * 77 * @param[in] compile_context The compile context to be used. 78 * @param[in] input The input tensor to convert. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32 79 * @param[out] output The output tensor. 3 lower dimensions represent a single output [width, height, OFM], 80 * while the rest represent batch of outputs. Data types supported: Same as @p input. Data layout: NCHW 81 * @param[in] convolved_dims Output convolved dimensions. 82 * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution 83 */ 84 void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const Size2D &convolved_dims, unsigned int num_groups = 1); 85 /** Static function to check if given info will lead to a valid configuration of @ref CLCol2ImKernel 86 * 87 * @param[in] input The input tensor to convert. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32 88 * @param[in] output The output tensor. 3 lower dimensions represent a single output [width, height, OFM], 89 * while the rest represent batch of outputs. Data types supported: Same as @p input. Data layout: NCHW 90 * @param[in] convolved_dims Output convolved dimensions. 91 * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution 92 * 93 * @return a status 94 */ 95 static Status validate(const ITensorInfo *input, const ITensorInfo *output, const Size2D &convolved_dims, unsigned int num_groups = 1); 96 97 // Inherited methods overridden: 98 void run(const Window &window, cl::CommandQueue &queue) override; 99 100 public: 101 const ICLTensor *_input; 102 ICLTensor *_output; 103 Size2D _convolved_dims; 104 }; 105 } // namespace arm_compute 106 #endif /*ARM_COMPUTE_CLCOL2IMKERNEL_H */ 107