• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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_NELOCALLYCONNECTEDMATRIXMULTIPLYKERNEL_H
25 #define ARM_COMPUTE_NELOCALLYCONNECTEDMATRIXMULTIPLYKERNEL_H
26 
27 #include "src/core/NEON/INEKernel.h"
28 
29 namespace arm_compute
30 {
31 class ITensor;
32 
33 /** NEON kernel to multiply each row of first tensor with low 2 dimensions of second tensor. */
34 class NELocallyConnectedMatrixMultiplyKernel : public INEKernel
35 {
36 public:
name()37     const char *name() const override
38     {
39         return "NELocallyConnectedMatrixMultiplyKernel";
40     }
41     /** Default constructor */
42     NELocallyConnectedMatrixMultiplyKernel();
43     /** Prevent instances of this class from being copied (As this class contains pointers) */
44     NELocallyConnectedMatrixMultiplyKernel(const NELocallyConnectedMatrixMultiplyKernel &) = delete;
45     /** Prevent instances of this class from being copied (As this class contains pointers) */
46     NELocallyConnectedMatrixMultiplyKernel &operator=(const NELocallyConnectedMatrixMultiplyKernel &) = delete;
47     /** Allow instances of this class to be moved */
48     NELocallyConnectedMatrixMultiplyKernel(NELocallyConnectedMatrixMultiplyKernel &&) = default;
49     /** Allow instances of this class to be moved */
50     NELocallyConnectedMatrixMultiplyKernel &operator=(NELocallyConnectedMatrixMultiplyKernel &&) = default;
51     /** Default destructor */
52     ~NELocallyConnectedMatrixMultiplyKernel() = default;
53     /** Initialise the kernel's input and output
54      *
55      * @param[in]  input0 First input tensor. Data types supported: F16, F32
56      * @param[in]  input1 Second input tensor containing the Matrix B. Data type supported: same as @p input0
57      * @param[out] output Output tensor to store the result of matrix multiplication. Data type supported: same as @p input0
58      */
59     void configure(const ITensor *input0, const ITensor *input1, ITensor *output);
60     /** Static function to check if given info will lead to a valid configuration of @ref NELocallyConnectedMatrixMultiplyKernel
61      *
62      * @param[in] input0 First input tensor info. Data types supported: F16, F32
63      * @param[in] input1 Second input tensor info. Data type supported: same as @p input0
64      * @param[in] output Output tensor info. Data type supported: same as @p input0
65      *
66      * @return a status
67      */
68     static Status validate(const ITensorInfo *input0, const ITensorInfo *input1, const ITensorInfo *output);
69 
70     // Inherited methods overridden:
71     void run(const Window &window, const ThreadInfo &info) override;
72 
73 private:
74     const ITensor *_input0;
75     const ITensor *_input1;
76     ITensor       *_output;
77 };
78 } // namespace arm_compute
79 #endif /* ARM_COMPUTE_NELOCALLYCONNECTEDMATRIXMULTIPLYKERNEL_H */
80