blob: 77d7494a7f6007f595be82c2f45f4471548965d5 [file] [log] [blame]
Gian Marco Iodice5ba5e092018-12-06 17:13:09 +00001/*
2 * Copyright (c) 2018 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_CLGEMMRESHAPELHSMATRIXKERNEL_H__
25#define __ARM_COMPUTE_CLGEMMRESHAPELHSMATRIXKERNEL_H__
26
27#include "arm_compute/core/CL/ICLKernel.h"
28
29namespace arm_compute
30{
31class ICLTensor;
32
33/** OpenCL kernel to reshape the LHS matrix when performing the matrix multiplication.
34 * In particular, this function splits the input matrix in blocks of size M0xK0 (defined through GEMMLHSInfo) and
35 * stores each one in the output matrix unrolling the values
36 */
37class CLGEMMReshapeLHSMatrixKernel : public ICLKernel
38{
39public:
40 /** Default constructor */
41 CLGEMMReshapeLHSMatrixKernel();
42 /** Prevent instances of this class from being copied (As this class contains pointers) */
43 CLGEMMReshapeLHSMatrixKernel(const CLGEMMReshapeLHSMatrixKernel &) = delete;
44 /** Prevent instances of this class from being copied (As this class contains pointers) */
45 CLGEMMReshapeLHSMatrixKernel &operator=(const CLGEMMReshapeLHSMatrixKernel &) = delete;
46 /** Allow instances of this class to be moved */
47 CLGEMMReshapeLHSMatrixKernel(CLGEMMReshapeLHSMatrixKernel &&) = default;
48 /** Allow instances of this class to be moved */
49 CLGEMMReshapeLHSMatrixKernel &operator=(CLGEMMReshapeLHSMatrixKernel &&) = default;
50 /** Initialise the kernel's input and output.
51 *
52 * @param[in] input Input tensor. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32
53 * @param[out] output Output tensor. Data type supported: same as @p input
54 * @param[in] lhs_info LHS matrix information to be used for reshaping. This object contains all the necessary
55 * information to reshape the input tensor. Only the following values are supported:
56 * lhs_info.m0: 2,3,4,5,6,7,8
57 * lhs_info.k0: 2,4,8,16
58 * lhs_info.v0: greater than 0
59 * lhs_info.transpose: false
60 * lhs_info.interleave: true, false
61 * @param[in] reinterpret_input_as_3d (Optional) True if the input has to be reinterpreted as 3D tensor
62 */
63 void configure(const ICLTensor *input, ICLTensor *output, const GEMMLHSMatrixInfo &lhs_info, bool reinterpret_input_as_3d = false);
64 /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMReshapeLHSMatrixKernel
65 *
66 * @param[in] input Input tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32
67 * @param[in] output Output tensor info which stores the interleaved matrix. Data type supported: same as @p input.
68 * @param[in] lhs_info LHS matrix information to be used for reshaping. This object contains all the necessary
69 * information to reshape the input tensor. Only the following values are supported:
70 * lhs_info.m0: 2,3,4,5,6,7,8
71 * lhs_info.k0: 2,4,8,16
72 * lhs_info.v0: greater than 0
73 * lhs_info.transpose: false
74 * lhs_info.interleave: true, false
75 * @param[in] reinterpret_input_as_3d True if the input has to be reinterpreted as 3D tensor
76 *
77 * @return a status
78 */
79 static Status validate(const ITensorInfo *input, const ITensorInfo *output, const GEMMLHSMatrixInfo &lhs_info, bool reinterpret_input_as_3d);
80
81 // Inherited methods overridden
82 void run(const Window &window, cl::CommandQueue &queue) override;
83
84private:
85 const ICLTensor *_input;
86 ICLTensor *_output;
87 bool _reinterpret_input_as_3d;
88};
89} // namespace arm_compute
90#endif /* __ARM_COMPUTE_CLGEMMRESHAPELHSMATRIXKERNEL_H__ */