1 /* 2 * Copyright (c) 2019-2020 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_CLGEMMLOWPMATRIXMULTIPLYNATIVEKERNEL_H 25 #define ARM_COMPUTE_CLGEMMLOWPMATRIXMULTIPLYNATIVEKERNEL_H 26 27 #include "src/core/CL/ICLKernel.h" 28 29 namespace arm_compute 30 { 31 class ICLTensor; 32 33 /** OpenCL kernel to multiply matrices with QASYMM8/QASYMM8_SIGNED data type */ 34 class CLGEMMLowpMatrixMultiplyNativeKernel : public ICLKernel 35 { 36 public: 37 /** Default Constructor */ 38 CLGEMMLowpMatrixMultiplyNativeKernel(); 39 /** Prevent instances of this class from being copied (As this class contains pointers) */ 40 CLGEMMLowpMatrixMultiplyNativeKernel(const CLGEMMLowpMatrixMultiplyNativeKernel &) = delete; 41 /** Prevent instances of this class from being copied (As this class contains pointers) */ 42 CLGEMMLowpMatrixMultiplyNativeKernel &operator=(const CLGEMMLowpMatrixMultiplyNativeKernel &) = delete; 43 /** Allow instances of this class to be moved */ 44 CLGEMMLowpMatrixMultiplyNativeKernel(CLGEMMLowpMatrixMultiplyNativeKernel &&) = default; 45 /** Allow instances of this class to be moved */ 46 CLGEMMLowpMatrixMultiplyNativeKernel &operator=(CLGEMMLowpMatrixMultiplyNativeKernel &&) = default; 47 /** Initialise the kernel's input and output. 48 * 49 * @param[in] input0 Input tensor containing the LHS matrix. Data type supported: QASYMM8/QASYMM8_SIGNED 50 * @param[in] input1 Input tensor containing the RHS matrix. Data type supported: same as @p input0 51 * @param[out] output Output tensor to store the result of matrix multiplication. Data type supported: S32 52 * @param[in] lhs_info LHS matrix information used to retrieve the number of rows to be processed by each thread 53 * lhs_info.m0: 2,3,4,5,6,7,8 54 * lhs_info.k0: 2,3,4,8,16 55 * @param[in] rhs_info RHS matrix information used to retrieve the number of columns to be processed by each thread 56 * rhs_info.n0: 2,3,4,8,16 57 * rhs_info.k0: same as lhs_info.k0 58 * @param[in] gemm_info GEMM information used to retrieve the original dimensions of the input matrices 59 */ 60 void configure(const ICLTensor *input0, const ICLTensor *input1, ICLTensor *output, const GEMMLHSMatrixInfo &lhs_info, const GEMMRHSMatrixInfo &rhs_info, const GEMMReshapeInfo &gemm_info); 61 /** Initialise the kernel's input and output. 62 * 63 * @param[in] compile_context The compile context to be used. 64 * @param[in] input0 Input tensor containing the LHS matrix. Data type supported: QASYMM8/QASYMM8_SIGNED 65 * @param[in] input1 Input tensor containing the RHS matrix. Data type supported: same as @p input0 66 * @param[out] output Output tensor to store the result of matrix multiplication. Data type supported: S32 67 * @param[in] lhs_info LHS matrix information used to retrieve the number of rows to be processed by each thread 68 * lhs_info.m0: 2,3,4,5,6,7,8 69 * lhs_info.k0: 2,3,4,8,16 70 * @param[in] rhs_info RHS matrix information used to retrieve the number of columns to be processed by each thread 71 * rhs_info.n0: 2,3,4,8,16 72 * rhs_info.k0: same as lhs_info.k0 73 * @param[in] gemm_info GEMM information used to retrieve the original dimensions of the input matrices 74 */ 75 void configure(const CLCompileContext &compile_context, const ICLTensor *input0, const ICLTensor *input1, ICLTensor *output, const GEMMLHSMatrixInfo &lhs_info, const GEMMRHSMatrixInfo &rhs_info, 76 const GEMMReshapeInfo &gemm_info); 77 /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMLowpMatrixMultiplyNativeKernel 78 * 79 * @param[in] input0 Input tensor info for the LHS matrix. Data type supported: QASYMM8/QASYMM8_SIGNED 80 * @param[in] input1 Input tensor info for the RHS matrix. Data type supported: same as @p input0 81 * @param[in] output Output tensor info. Data type supported: S32 82 * @param[in] lhs_info LHS matrix information used to retrieve the number of rows to be processed by each thread 83 * lhs_info.m0: 2,3,4,5,6,7,8 84 * lhs_info.k0: 2,3,4,8,16 85 * @param[in] rhs_info RHS matrix information used to retrieve the number of columns to be processed by each thread 86 * rhs_info.n0: 2,3,4,8,16 87 * rhs_info.k0: same as lhs_info.k0 88 * @param[in] gemm_info GEMM information used to retrieve the original dimensions of the input matrices 89 * 90 * @return a status 91 */ 92 static Status validate(const ITensorInfo *input0, const ITensorInfo *input1, const ITensorInfo *output, const GEMMLHSMatrixInfo &lhs_info, const GEMMRHSMatrixInfo &rhs_info, 93 const GEMMReshapeInfo &gemm_info); 94 95 // Inherited methods overridden: 96 void run(const Window &window, cl::CommandQueue &queue) override; 97 98 private: 99 const ICLTensor *_input0; 100 const ICLTensor *_input1; 101 ICLTensor *_output; 102 bool _slide_matrix_b; 103 bool _reinterpret_input_as_3d; 104 bool _reinterpret_output_as_3d; 105 bool _use_dummy_work_items; 106 }; 107 } // namespace arm_compute 108 #endif /*ARM_COMPUTE_CLGEMMLOWPMATRIXMULTIPLYNATIVEKERNEL_H*/ 109