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
180
181
182
183
|
/*
* 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 "nnkit/support/onnx/Runner.h"
#include "nnkit/support/onnx/Status.h"
#include <stdex/Memory.h>
#include <cassert>
namespace nnkit
{
namespace support
{
namespace onnx
{
Runner::Runner(const std::string &path) : _allocator(stdex::make_unique<Allocator>())
{
Status status;
status = OrtCreateEnv(ORT_LOGGING_LEVEL_WARNING, "nnkit", &_env);
assert(!status.isError());
status = OrtCreateSession(_env, path.c_str(), nullptr, &_session);
assert(!status.isError());
}
Runner::~Runner(void)
{
if (_session)
{
OrtReleaseSession(_session);
}
if (_env)
{
OrtReleaseEnv(_env);
}
}
void Runner::prepareInputs(void)
{
Status status;
assert(_inputs == nullptr);
size_t num_input_nodes;
status = OrtSessionGetInputCount(_session, &num_input_nodes);
status.throwOnError();
_inputs = stdex::make_unique<TensorSet>(_allocator.get(), num_input_nodes);
for (size_t i = 0; i < num_input_nodes; ++i)
{
char *input_name;
status = OrtSessionGetInputName(_session, i, _allocator.get(), &input_name);
status.throwOnError();
assert(input_name != nullptr);
std::string name{input_name};
_allocator->Free(input_name);
OrtTypeInfo *typeinfo;
status = OrtSessionGetInputTypeInfo(_session, i, &typeinfo);
status.throwOnError();
const OrtTensorTypeAndShapeInfo *tensor_info = OrtCastTypeInfoToTensorInfo(typeinfo);
ONNXTensorElementDataType type = OrtGetTensorElementType(tensor_info);
uint32_t num_dims = OrtGetNumOfDimensions(tensor_info);
std::vector<size_t> dims(num_dims);
OrtGetDimensions(tensor_info, (int64_t *)dims.data(), num_dims);
// NOTE To run OnnxRuntime, the total size of input tensor must be fixed.
// In the present code, the unknown shape that is -1 is arbitrarily changed to 1.
//
// TODO Add user argument related to unknown shape
//
for (uint32_t j = 0; j < num_dims; ++j)
{
if (dims[j] == -1)
{
dims[j] = 1;
}
}
OrtReleaseTypeInfo(typeinfo);
_inputs->set(i, name, type, dims);
}
}
void Runner::prepareOutputs(void)
{
Status status;
assert(_outputs == nullptr);
size_t num_output_nodes;
status = OrtSessionGetOutputCount(_session, &num_output_nodes);
status.throwOnError();
_outputs = stdex::make_unique<TensorSet>(_allocator.get(), num_output_nodes);
for (size_t i = 0; i < num_output_nodes; ++i)
{
char *output_name;
status = OrtSessionGetOutputName(_session, i, _allocator.get(), &output_name);
status.throwOnError();
assert(output_name != nullptr);
std::string name{output_name};
_allocator->Free(output_name);
OrtTypeInfo *typeinfo;
status = OrtSessionGetOutputTypeInfo(_session, i, &typeinfo);
status.throwOnError();
const OrtTensorTypeAndShapeInfo *tensor_info = OrtCastTypeInfoToTensorInfo(typeinfo);
ONNXTensorElementDataType type = OrtGetTensorElementType(tensor_info);
uint32_t num_dims = OrtGetNumOfDimensions(tensor_info);
std::vector<size_t> dims(num_dims);
OrtGetDimensions(tensor_info, (int64_t *)dims.data(), num_dims);
// NOTE To run OnnxRuntime, the total size of output tensor must be fixed.
// In the present code, the unknown shape that is -1 is arbitrarily changed to 1.
//
// TODO Add user argument related to unknown shape
//
for (uint32_t j = 0; j < num_dims; ++j)
{
if (dims[j] == -1)
{
dims[j] = 1;
}
}
OrtReleaseTypeInfo(typeinfo);
_outputs->set(i, name, type, dims);
}
}
void Runner::run(void)
{
Status status;
auto pinput_names = _inputs->names();
std::vector<const char *> input_names(pinput_names.size());
for (size_t i = 0; i < pinput_names.size(); ++i)
{
input_names[i] = pinput_names[i].c_str();
}
auto poutput_names = _outputs->names();
std::vector<const char *> output_names(poutput_names.size());
for (size_t i = 0; i < poutput_names.size(); ++i)
{
output_names[i] = poutput_names[i].c_str();
}
status = OrtRun(_session, NULL, input_names.data(), _inputs->tensors().data(), _inputs->size(),
output_names.data(), _outputs->size(), _outputs->mutable_tensors().data());
status.throwOnError();
}
} // namespace onnx
} // namespace support
} // namespace nnkit
|
