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
|
/*
* Copyright (c) 2019 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 "SubLayer.h"
#include <cker/operation/BinaryArithmeticOps.h>
#include "OperationUtils.h"
namespace neurun
{
namespace backend
{
namespace cpu
{
namespace kernel
{
void SubLayer::subFloat32()
{
float output_activation_min, output_activation_max;
CalculateActivationRangeFloat(_activation, &output_activation_min, &output_activation_max);
nnfw::cker::BinaryArithmeticOpParam op_params;
op_params.float_activation_max = output_activation_max;
op_params.float_activation_min = output_activation_min;
const std::function<float(const float &, const float &)> fn = [](const float &a, const float &b) {
return a - b;
};
if (!HaveSameShapes(&_lhsDescr, &_rhsDescr))
{
nnfw::cker::BroadcastBinaryArithmeticOpSlow(
op_params, convertToExtendedCkerShape(_lhsDescr), _lhsData.f,
convertToExtendedCkerShape(_rhsDescr), _rhsData.f, convertToExtendedCkerShape(_outputDescr),
_outputData.f, fn);
return;
}
nnfw::cker::BinaryArithmeticOp(op_params, convertTensorDescriptorToCkerShape(_lhsDescr),
_lhsData.f, convertTensorDescriptorToCkerShape(_rhsDescr),
_rhsData.f, convertTensorDescriptorToCkerShape(_outputDescr),
_outputData.f, fn);
}
void SubLayer::subQuant8()
{
int32_t output_activation_min, output_activation_max;
CalculateActivationRangeUint8(_activation, _outputDescr, &output_activation_min,
&output_activation_max);
// nnfw::cker::SubParam op_params;
// op_params.quantized_activation_max = output_activation_max;
// op_params.quantized_activation_min = output_activation_min;
// cker quant8 sub is not implemented yet
throw std::runtime_error{"NYI"};
}
void SubLayer::configure(uint8_t *lhsData, const TensorDescriptor &lhsDescr, uint8_t *rhsData,
const TensorDescriptor &rhsDescr, const ir::Activation activation,
uint8_t *outputData, const TensorDescriptor &outputDescr)
{
_lhsData.u8 = lhsData;
_lhsDescr = lhsDescr;
_rhsData.u8 = rhsData;
_rhsDescr = rhsDescr;
_inputType = lhsDescr.type;
_activation = activation;
_outputData.u8 = outputData;
_outputDescr = outputDescr;
}
void SubLayer::run()
{
if (_inputType == OperandType::FLOAT32)
{
subFloat32();
}
else if (_inputType == OperandType::QUANT8_ASYMM)
{
subQuant8();
}
}
} // namespace kernel
} // namespace cpu
} // namespace backend
} // namespace neurun
|
