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
153
|
/*
* Copyright (c) 2018-2019 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.
*/
#pragma once
#include "depthwise.hpp"
#include "qasymm8.hpp"
namespace depthwise
{
namespace nck = neon_convolution_kernels;
template <
unsigned int OutputTileRows, unsigned int OutputTileCols,
unsigned int KernelRows, unsigned int KernelCols,
unsigned int StrideRows, unsigned int StrideCols
>
class QAsymm8DepthwiseConvolution : public DepthwiseConvolutionBase<
OutputTileRows, OutputTileCols,
KernelRows, KernelCols,
StrideRows, StrideCols,
uint8_t, int32_t, uint8_t,
QAsymm8DepthwiseConvolution<OutputTileRows, OutputTileCols, KernelRows, KernelCols, StrideRows, StrideCols>
>
{
using Base = DepthwiseConvolutionBase<
OutputTileRows, OutputTileCols,
KernelRows, KernelCols,
StrideRows, StrideCols,
uint8_t, int32_t, uint8_t,
QAsymm8DepthwiseConvolution<OutputTileRows, OutputTileCols, KernelRows, KernelCols, StrideRows, StrideCols>
>;
friend Base;
using InputType = typename Base::InputType;
using OutputType = typename Base::OutputType;
public:
QAsymm8DepthwiseConvolution(
int n_batches, int n_input_rows, int n_input_cols, int n_channels,
nck::ActivationFunction activation,
const qasymm8::QAsymm8Params& weight_quantisation,
const qasymm8::QAsymm8Params& input_quantisation,
const qasymm8::QAsymm8Params& output_quantisation,
unsigned int padding_top,
unsigned int padding_left,
unsigned int padding_bottom,
unsigned int padding_right
);
QAsymm8DepthwiseConvolution(
int n_batches, int n_input_rows, int n_input_cols, int n_channels,
int n_output_rows, int n_output_cols,
nck::ActivationFunction activation,
const qasymm8::QAsymm8Params& weight_quantisation,
const qasymm8::QAsymm8Params& input_quantisation,
const qasymm8::QAsymm8Params& output_quantisation,
unsigned int padding_top,
unsigned int padding_left,
unsigned int padding_bottom,
unsigned int padding_right
);
QAsymm8DepthwiseConvolution(
int n_batches, int n_input_rows, int n_input_cols, int n_channels,
nck::ActivationFunction activation,
const qasymm8::QAsymm8Params& weight_quantisation,
const qasymm8::QAsymm8Params& input_quantisation,
const qasymm8::QAsymm8Params& output_quantisation,
const qasymm8::QAsymm8RescaleParams& rescale_parameters,
unsigned int padding_top,
unsigned int padding_left,
unsigned int padding_bottom,
unsigned int padding_right
);
QAsymm8DepthwiseConvolution(
int n_batches, int n_input_rows, int n_input_cols, int n_channels,
int n_output_rows, int n_output_cols,
nck::ActivationFunction activation,
const qasymm8::QAsymm8Params& weight_quantisation,
const qasymm8::QAsymm8Params& input_quantisation,
const qasymm8::QAsymm8Params& output_quantisation,
const qasymm8::QAsymm8RescaleParams& rescale_parameters,
unsigned int padding_top,
unsigned int padding_left,
unsigned int padding_bottom,
unsigned int padding_right
);
protected:
static nck::ActivationFunction get_activation_fn(
nck::ActivationFunction activation,
const qasymm8::QAsymm8Params& output_quantisation
);
uint8_t _input_padding_value(void) const;
void _pack_params(
void *buffer,
const void *weights,
unsigned int weight_row_stride,
unsigned int weight_col_stride,
const void *biases=nullptr
) const;
template <nck::ActivationFunction Activation>
void execute_tile(
int n_channels,
const void* packed_params,
const uint8_t* inptr,
unsigned int in_row_stride,
unsigned int in_col_stride,
uint8_t* outptr,
unsigned int out_row_stride,
unsigned int out_col_stride
);
template <nck::ActivationFunction Activation>
void execute_tile(
int n_channels,
const void* packed_params,
const uint8_t* inptrs[Base::inner_tile_rows][Base::inner_tile_cols],
uint8_t* outptrs[Base::output_tile_rows][Base::output_tile_cols]
);
private:
// Quantization parameters
const qasymm8::QAsymm8Params _weights_quant, _inputs_quant, _output_quant;
const qasymm8::QAsymm8RescaleParams rescale_parameters;
};
} // namespace depthwise
|
