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
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
|
/*
* Copyright (c) 2020 Samsung Electronics Co., Ltd. All Rights Reserved
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "QuantizationUtils.h"
#include <luci/Log.h>
#include <iostream>
#include <cmath>
namespace luci
{
void compute_sym_scale_zp(float min, float max, float &scaling_factor, int64_t &zp,
float &nudged_min, float &nudged_max)
{
assert(min != max);
const int32_t kMaxScale = std::numeric_limits<int16_t>::max();
const int32_t kMinScale = -kMaxScale;
const double qmin_double = kMinScale;
const double qmax_double = kMaxScale;
const double rmin = std::fmin(0, min);
const double rmax = std::fmax(0, max);
double scale_factor_from_min_side{0};
double scale_factor_from_max_side{0};
if ((qmin_double * rmin) > 0)
scale_factor_from_min_side = rmin / qmin_double;
if ((qmax_double * rmax) > 0)
scale_factor_from_max_side = rmax / qmax_double;
scaling_factor = scale_factor_from_min_side > scale_factor_from_max_side
? scale_factor_from_min_side
: scale_factor_from_max_side;
zp = 0;
nudged_min = static_cast<float>(qmin_double * scaling_factor);
nudged_max = static_cast<float>(qmax_double * scaling_factor);
}
void compute_asym_scale_zp(float min, float max, float &scaling_factor, int64_t &zp,
float &nudged_min, float &nudged_max)
{
LOGGER(l);
assert(min <= max);
const int32_t kMinScale = 0;
const int32_t kMaxScale = 255;
const double qmin_double = kMinScale;
const double qmax_double = kMaxScale;
const double rmin = std::fmin(0, min);
const double rmax = std::fmax(0, max);
double scale = (rmax - rmin) / (qmax_double - qmin_double);
double zero_point_double = 0;
uint8_t nudged_zero_point = 0;
if (scale == 0)
{
WARN(l) << "The minimum and maximum values are the same." << std::endl;
if (min >= 0 && max >= 0)
zero_point_double = kMinScale;
else
zero_point_double = kMaxScale;
}
else
zero_point_double = qmin_double - rmin / scale;
if (zero_point_double <= qmin_double)
{
assert(min >= 0 && max >= 0);
nudged_zero_point = kMinScale;
scale = max / (qmax_double - qmin_double);
if (min > 0 && max > 0)
WARN(l) << "The minimum and maximum values are all positive." << std::endl;
}
else if (zero_point_double >= qmax_double)
{
assert(min < 0 && max < 0);
nudged_zero_point = kMaxScale;
scale = -min / (qmax_double - qmin_double);
WARN(l) << "The minimum and maximum values are all negative." << std::endl;
}
else
{
assert(min < 0 && max >= 0);
nudged_zero_point = static_cast<uint8_t>(std::round(zero_point_double));
}
// protect scale from being very low due to overflow
if (scale < 1e-5)
{
scale = 1e-5;
nudged_zero_point = static_cast<uint8_t>(std::round(qmin_double - rmin / scale));
}
nudged_min = static_cast<float>((qmin_double - nudged_zero_point) * scale);
nudged_max = static_cast<float>((qmax_double - nudged_zero_point) * scale);
scaling_factor = scale;
zp = nudged_zero_point;
}
bool get_channel_dim_index(CircleConst *node, loco::TensorShape &dimension, int &channel_dim_index)
{
auto succs = loco::succs(node);
if (succs.size() != 1) // assume weights is used by only one node
return false;
for (auto out : succs)
{
auto conv = dynamic_cast<CircleConv2D *>(out);
auto dw_conv = dynamic_cast<CircleDepthwiseConv2D *>(out);
auto tw_conv = dynamic_cast<CircleTransposeConv *>(out);
auto fc = dynamic_cast<CircleFullyConnected *>(out);
// Refer to https://github.com/Samsung/ONE/pull/2448.
if ((conv != nullptr && conv->filter() == node) ||
(tw_conv != nullptr && tw_conv->filter() == node)) // OHWI
{
assert(node->rank() == 4);
dimension.dim(0).set(node->dim(0).value());
dimension.dim(1).set(node->dim(1).value());
dimension.dim(2).set(node->dim(2).value());
dimension.dim(3).set(node->dim(3).value());
channel_dim_index = 0; // Set channel_dim_index based on "O"
return true;
}
else if (dw_conv != nullptr && dw_conv->filter() == node) // IHWC
{
assert(node->rank() == 4);
dimension.dim(0).set(node->dim(0).value());
dimension.dim(1).set(node->dim(1).value());
dimension.dim(2).set(node->dim(2).value());
dimension.dim(3).set(node->dim(3).value());
channel_dim_index = 3; // Set channel_dim_index based on "C"
return true;
}
else if (fc != nullptr && fc->weights() == node) // OI
{
assert(node->rank() == 2);
dimension.dim(0).set(node->dim(0).value());
dimension.dim(1).set(1); // Set FC layer like CONV
dimension.dim(2).set(1);
dimension.dim(3).set(node->dim(1).value());
channel_dim_index = 0; // Set channel_dim_index based on "O"
return true;
}
else
{
// node does not support channle-wise quantization
assert(false);
}
}
return false;
}
uint32_t cal_offset(loco::TensorShape &dimension, uint32_t *indices)
{
return indices[0] * dimension.dim(1).value() * dimension.dim(2).value() *
dimension.dim(3).value() +
indices[1] * dimension.dim(2).value() * dimension.dim(3).value() +
indices[2] * dimension.dim(3).value() + indices[3];
}
} // namespace luci
|
