blob: aed9ae286f842524ce9151cd3117939f6666d04f (
plain)
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
|
/*
* Copyright (c) 2018 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.
*/
/**
* @file SimpleArithmeticAddition.h
* @brief This file contains SimpleArithmeticAddition class
* @ingroup COM_AI_RUNTIME
*/
#ifndef __SIMPLE_ARITHMETIC_ADDITION_H__
#define __SIMPLE_ARITHMETIC_ADDITION_H__
#include "internal/arm_compute.h"
#include <arm_compute/core/ITensor.h>
/**
* @brief Class to run SimpleArithmeticAddition Layer
*/
class SimpleArithmeticAddition : public ::arm_compute::IFunction
{
public:
SimpleArithmeticAddition(void) : _lhs(nullptr), _rhs(nullptr), _out(nullptr)
{
// DO NOTHING
}
/**
* @brief Configure the layer
* @param[in] lhs Lefthand-side operand
* @param[in] rhs Righthand-side operand
* @param[in] out The destination tensor(Result operand)
* @return N/A
*/
void configure(::arm_compute::ITensor *lhs, ::arm_compute::ITensor *rhs,
::arm_compute::ITensor *out)
{
_lhs = lhs;
_rhs = rhs;
_out = out;
}
public:
/**
* @brief Run the operation. Must be called after configure().
* @return N/A
*/
void run(void) override
{
if (::internal::arm_compute::isGpuMode())
{
auto &q = ::arm_compute::CLScheduler::get().queue();
CAST_CL(_lhs)->map(q);
CAST_CL(_rhs)->map(q);
CAST_CL(_out)->map(q);
}
arm_compute::Window window;
window.use_tensor_dimensions(_out->info()->tensor_shape());
execute_window_loop(window, [this](const arm_compute::Coordinates &id) {
// NOTE Must be two input tensors of identical type
// Must be output tensor of the same type as input0.
assert(_lhs->info()->data_type() == _rhs->info()->data_type());
assert(_lhs->info()->data_type() == _out->info()->data_type());
switch (_lhs->info()->data_type())
{
case ::arm_compute::DataType::F32:
{
const auto lhs_value = *reinterpret_cast<float *>(_lhs->ptr_to_element(id));
const auto rhs_value = *reinterpret_cast<float *>(_rhs->ptr_to_element(id));
*reinterpret_cast<float *>(_out->ptr_to_element(id)) = lhs_value + rhs_value;
break;
}
case ::arm_compute::DataType::S32:
{
const auto lhs_value = *reinterpret_cast<int32_t *>(_lhs->ptr_to_element(id));
const auto rhs_value = *reinterpret_cast<int32_t *>(_rhs->ptr_to_element(id));
*reinterpret_cast<int32_t *>(_out->ptr_to_element(id)) = lhs_value + rhs_value;
break;
}
case ::arm_compute::DataType::U32:
{
const auto lhs_value = *reinterpret_cast<uint32_t *>(_lhs->ptr_to_element(id));
const auto rhs_value = *reinterpret_cast<uint32_t *>(_rhs->ptr_to_element(id));
*reinterpret_cast<uint32_t *>(_out->ptr_to_element(id)) = lhs_value + rhs_value;
break;
}
case ::arm_compute::DataType::QASYMM8:
{
const auto lhs_value = *reinterpret_cast<uint8_t *>(_lhs->ptr_to_element(id));
const auto rhs_value = *reinterpret_cast<uint8_t *>(_rhs->ptr_to_element(id));
// How to handle with overflow?
*reinterpret_cast<uint8_t *>(_out->ptr_to_element(id)) = lhs_value + rhs_value;
break;
}
default:
throw std::runtime_error("Not supported, yet");
break;
}
});
if (::internal::arm_compute::isGpuMode())
{
auto &q = ::arm_compute::CLScheduler::get().queue();
CAST_CL(_out)->unmap(q);
CAST_CL(_rhs)->unmap(q);
CAST_CL(_lhs)->unmap(q);
}
}
private:
::arm_compute::ITensor *_lhs;
::arm_compute::ITensor *_rhs;
::arm_compute::ITensor *_out;
};
#endif // __SIMPLE_ARITHMETIC_ADDITION_H__
|
