1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
|
/*
* Copyright (c) 2017-2019 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to
* deal in the Software without restriction, including without limitation the
* rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
* sell copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef __ARM_COMPUTE_NEGEMMLOWREDUCTIONKERNEL_H__
#define __ARM_COMPUTE_NEGEMMLOWREDUCTIONKERNEL_H__
#include "arm_compute/core/NEON/INEKernel.h"
namespace arm_compute
{
class ITensor;
/** Common interface for all NEON reduction kernels */
class INEGEMMLowpReductionKernel : public INEKernel
{
public:
/** Constructor */
INEGEMMLowpReductionKernel();
/** Prevent instances of this class from being copied (As this class contains pointers)*/
INEGEMMLowpReductionKernel(const INEGEMMLowpReductionKernel &) = delete;
/** Prevent instances of this class from being copied (As this class contains pointers)*/
INEGEMMLowpReductionKernel &operator=(const INEGEMMLowpReductionKernel &) = delete;
/** Allow instances of this class to be moved */
INEGEMMLowpReductionKernel(INEGEMMLowpReductionKernel &&) = default;
/** Allow instances of this class to be moved */
INEGEMMLowpReductionKernel &operator=(INEGEMMLowpReductionKernel &&) = default;
/** Initialise the kernel's input and output.
*
* @param[in] input Input tensor. Data type supported: QASYMM8/QASYMM8_SIGNED
* @param[out] output Output row-vector of sums of all the entries in each row/col of input tensor. Data type supported: S32
* @param[in] k Number of matrix A columns (or matrix B rows)
* @param[in] is_reshaped True if the input tensor has been reshaped
*/
virtual void configure(const ITensor *input, ITensor *output, int32_t k, bool is_reshaped) = 0;
protected:
const ITensor *_input;
ITensor *_output;
int32_t _k;
bool _is_reshaped;
};
/** NEON kernel used to compute the row-vectors of sums of all the entries in each row of Matrix A.
*
* @note This stage is needed to handle the offset of matrix product
* https://github.com/google/gemmlowp/blob/master/doc/low-precision.md
*/
class NEGEMMLowpMatrixAReductionKernel : public INEGEMMLowpReductionKernel
{
public:
const char *name() const override
{
return "NEGEMMLowpMatrixAReductionKernel";
}
/** Initialise the kernel's input and output.
*
* @param[in] mtx_a Input tensor. Data type supported: QASYMM8/QASYMM8_SIGNED
* @param[out] vector_sum_row Output row-vector of sums of all the entries in each row of mtx_a. Data type supported: S32
* @param[in] num_mtx_a_cols Number of matrix A columns
* @param[in] is_interleaved4x4 True if the matrix A has been interleaved4x4
*/
void configure(const ITensor *mtx_a, ITensor *vector_sum_row, int32_t num_mtx_a_cols, bool is_interleaved4x4) override;
/** Static function to check if given info will lead to a valid configuration of @ref NEGEMMLowpMatrixAReductionKernel
*
* @param[in] mtx_a Input tensor. Data type supported: QASYMM8/QASYMM8_SIGNED
* @param[in] vector_sum_row Output row-vector of sums of all the entries in each row of mtx_a. Data type supported: S32
* @param[in] num_mtx_a_cols Number of matrix A columns
* @param[in] is_interleaved4x4 True if the matrix A has been interleaved4x4
*
* @return a status
*/
static Status validate(const ITensorInfo *mtx_a, const ITensorInfo *vector_sum_row, int32_t num_mtx_a_cols, bool is_interleaved4x4);
// Inherited methods overridden:
void run(const Window &window, const ThreadInfo &info) override;
private:
/** Execution of the reduction kernel specialized on the input type
*
* @param[in] window Execution window
*/
template <typename T>
void run_internal(const Window &window);
};
/** NEON kernel used to compute the row-vectors of sums of all the entries in each column of Matrix B.
*
* @note This stage is needed to handle the offset of matrix product
* https://github.com/google/gemmlowp/blob/master/doc/low-precision.md
*/
class NEGEMMLowpMatrixBReductionKernel : public INEGEMMLowpReductionKernel
{
public:
const char *name() const override
{
return "NEGEMMLowpMatrixBReductionKernel";
}
/** Initialise the kernel's input and output.
*
* @param[in] mtx_b Input tensor. Data type supported: Data type supported: QASYMM8/QASYMM8_SIGNED
* @param[out] vector_sum_col Output row-vector of sums of all the entries in each column of mtx_b. Data type supported: S32
* @param[in] num_mtx_b_rows Number of matrix B rows
* @param[in] is_transposed1xW True if the input tensor is transposed 1xW
*/
void configure(const ITensor *mtx_b, ITensor *vector_sum_col, int32_t num_mtx_b_rows, bool is_transposed1xW) override;
/** Static function to check if given info will lead to a valid configuration of @ref NEGEMMLowpMatrixBReductionKernel
*
* @param[in] mtx_b Input tensor. Data type supported: Data type supported: QASYMM8/QASYMM8_SIGNED
* @param[in] vector_sum_col Output row-vector of sums of all the entries in each column of mtx_b. Data type supported: S32
* @param[in] num_mtx_b_rows Number of matrix B rows
* @param[in] is_transposed1xW True if the input tensor is transposed 1xW
*
* @return a status
*/
static Status validate(const ITensorInfo *mtx_b, const ITensorInfo *vector_sum_col, int32_t num_mtx_b_rows, bool is_transposed1xW);
// Inherited methods overridden:
void run(const Window &window, const ThreadInfo &info) override;
private:
/** Execution of the reduction kernel specialized on the input type
*
* @param[in] window Execution window
* @param[in] info Thread-related information
*/
template <typename T>
void run_internal(const Window &window, const ThreadInfo &info);
};
} // namespace arm_compute
#endif /* __ARM_COMPUTE_NEGEMMLOWREDUCTIONKERNEL_H__ */
|
