forked from Tensor-Array/Tensor-Array
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathtensorbase.cc
More file actions
258 lines (222 loc) · 8.95 KB
/
tensorbase.cc
File metadata and controls
258 lines (222 loc) · 8.95 KB
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
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
/*
Copyright 2024 TensorArray-Creators
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 "data_type_wrapper.hh"
#ifndef TENSOR_CONTENT
#define TENSOR_CONTENT
#include "tensorbase.hh"
#undef TENSOR_CONTENT
#endif
#include <cstdio>
#include <cstring>
#define USING_DATA_TYPE_FLOAT() (float)(double)
#define USING_DATA_TYPE_SINT() (int8_t)(int16_t)(int32_t)(int64_t)
#define USING_DATA_TYPE_UINT() (uint8_t)(uint16_t)(uint32_t)(uint64_t)
#define USING_DATA_TYPE USING_DATA_TYPE_SINT() USING_DATA_TYPE_UINT() USING_DATA_TYPE_FLOAT()
#define LOOP(seq) END(A seq)
#define BODY(x) ADD_CODE(x)
#define A(x) BODY(x) B
#define B(x) BODY(x) A
#define A_END
#define B_END
#define END(...) END_(__VA_ARGS__)
#define END_(...) __VA_ARGS__##_END
namespace tensor_array
{
namespace value
{
class TensorBuf final : public TensorBase::TensorStorage
{
public:
TensorBuf(const std::type_info&, unsigned char, const unsigned int*, const void*, const devices::Device&, const devices::Device&);
TensorBuf(const TensorBuf&);
~TensorBuf() override;
const std::type_info& type() const override;
std::unique_ptr<TensorStorage> clone() const override;
std::initializer_list<unsigned int> dim_sizes() const override;
const void* data() const override;
size_t data_size() const override;
std::unique_ptr<TensorStorage> child_create(unsigned int) const override;
const devices::Device& get_device() const override;
private:
const void* const dat;
const unsigned int* const sizes;
const unsigned char dim;
const std::type_info& dtype;
const devices::Device device;
};
TensorBase::TensorBase(const std::type_info& dtype, const std::initializer_list<unsigned int>& shape_vec, const void* dat, const devices::Device& dev_other, const devices::Device& dev_this):
instance(std::make_unique<TensorBuf>(dtype, static_cast<unsigned char>(shape_vec.size()), shape_vec.begin(), dat, dev_other, dev_this))
{
}
TensorBase::TensorBase(const std::type_info& dtype, const std::vector<unsigned int>& shape_vec, const void* dat, const devices::Device& dev_other, const devices::Device& dev_this):
TensorBase(dtype, wrapper::initializer_wrapper<unsigned int>(shape_vec.begin().operator->(), shape_vec.end().operator->()), dat, dev_other, dev_this)
{
}
TensorBase::TensorBase(const TensorBase& other) :
instance(other.instance->clone())
{
}
TensorBase::TensorBase(const TensorBase& other, const devices::Device& other_device) :
TensorBase(other.type(), other.shape(), other.data(), other.get_device(), other_device)
{
}
TensorBase& TensorBase::operator=(const TensorBase& other)
{
TensorBase temp(other);
this->swap(temp);
return *this;
}
TensorBase TensorBase::operator[](unsigned int index) const
{
TensorBase temp;
temp.instance = this->instance->child_create(index);
return temp;
}
const std::type_info& TensorBase::type() const
{
return this->instance->type();
}
std::initializer_list<unsigned int> TensorBase::shape() const
{
return this->instance->dim_sizes();
}
const void* TensorBase::data() const
{
return this->instance->data();
}
size_t TensorBase::data_size() const
{
return this->instance->data_size();
}
const devices::Device& TensorBase::get_device() const
{
return this->instance->get_device();
}
TensorBase TensorBase::change_device(const devices::Device& dev) const
{
return TensorBase(*this, dev);
}
bool TensorBase::has_tensor() const
{
return static_cast<bool>(this->instance);
}
void TensorBase::swap(TensorBase& other)
{
this->instance.swap(other.instance);
}
void TensorBase::save(const char* dir) const
{
if (static_cast<bool>(this->instance))
{
if (std::FILE* tensor_file = std::fopen(dir, "wb"))
{
std::initializer_list<unsigned int> list = this->shape();
std::size_t temp_save;
std::size_t total_dim_size = 1;
for (auto& it : list)
total_dim_size *= it;
datatype::DataType temp_type = datatype::warp_type(this->type());
unsigned char temp_dim_size = list.size();
temp_save = std::fwrite(&temp_type, sizeof(datatype::DataType), 1, tensor_file);
temp_save = std::fwrite(&temp_dim_size, sizeof(char), 1, tensor_file);
temp_save = std::fwrite(list.begin(), sizeof(unsigned int), list.size(), tensor_file);
temp_save = std::fwrite(this->data(), get_sizeof_type(this->type()), total_dim_size, tensor_file);
std::fclose(tensor_file);
}
}
}
const unsigned int* create_arr_dim_sizes(size_t other_dim, const unsigned int* other_sizes)
{
unsigned int* temp_sizes = new unsigned int[other_dim];
if (other_sizes)
std::memcpy(temp_sizes, other_sizes, other_dim * sizeof(unsigned int));
return temp_sizes;
}
void* create_data_1(size_t other_data_size, const void* other_data, const devices::Device& dev_other, const devices::Device& dev_this)
{
void* temp_sizes = operator new(other_data_size, dev_this);
if (other_data)
devices::device_memcpy(temp_sizes, dev_this, other_data, dev_other, other_data_size);
else
devices::device_memset(temp_sizes, dev_this, 0, other_data_size);
return temp_sizes;
}
std::size_t get_sizeof_type(const std::type_info& t_info)
{
return dynamic_type_size[t_info];
}
unsigned long long get_buffer_size0(const std::type_info& dtype, unsigned char dim, const unsigned int* sizes)
{
unsigned long long temp_size = 1;
for (unsigned char i = 0; i < dim; i++)
temp_size *= sizes[i];
return temp_size * get_sizeof_type(dtype);
}
TensorBuf::TensorBuf(const std::type_info& dtype, unsigned char dim, const unsigned int* sizes, const void* dat, const devices::Device& dev_other, const devices::Device& dev_this) :
dtype(dtype),
dim(dim),
sizes(create_arr_dim_sizes(dim, sizes)),
dat(create_data_1(get_buffer_size0(dtype, dim, sizes), dat, dev_other, dev_this)),
device(dev_this)
{
if (dtype == typeid(void)) throw 0;
}
TensorBuf::TensorBuf(const TensorBuf& other) :
TensorBuf(other.dtype, other.dim, other.sizes, other.dat, other.device, other.device)
{
}
TensorBuf::~TensorBuf()
{
delete[] this->sizes;
operator delete(const_cast<void*>(this->dat), this->device);
}
const std::type_info& TensorBuf::type() const
{
return this->dtype;
}
std::unique_ptr<TensorBase::TensorStorage> TensorBuf::clone() const
{
return std::make_unique<TensorBuf>(*this);
}
std::initializer_list<unsigned int> TensorBuf::dim_sizes() const
{
return wrapper::initializer_wrapper<unsigned int>(this->sizes, this->sizes + this->dim);
}
const void* TensorBuf::data() const
{
return this->dat;
}
size_t TensorBuf::data_size() const
{
return get_buffer_size0(this->dtype, this->dim, this->sizes);
}
std::unique_ptr<TensorBase::TensorStorage> TensorBuf::child_create(unsigned int index) const
{
if (this->dim == 0) throw std::exception();
const void* data_ptr = reinterpret_cast<const void*>(reinterpret_cast<std::size_t>(this->dat) + index * (this->data_size() / this->sizes[0]));
return std::make_unique<TensorBuf>(this->dtype, this->dim - 1, &this->sizes[1], data_ptr, this->device, this->device);
}
const devices::Device& TensorBuf::get_device() const
{
return this->device;
}
}
}
#undef LOOP
#undef BODY
#undef A
#undef B
#undef A_END
#undef B_END
#undef END
#undef END_