tiny_dnn 1.0.0
A header only, dependency-free deep learning framework in C++11
Loading...
Searching...
No Matches
backend.h
1/*
2 Copyright (c) 2016, Taiga Nomi, Edgar Riba
3 All rights reserved.
4
5 Redistribution and use in source and binary forms, with or without
6 modification, are permitted provided that the following conditions are met:
7 * Redistributions of source code must retain the above copyright
8 notice, this list of conditions and the following disclaimer.
9 * Redistributions in binary form must reproduce the above copyright
10 notice, this list of conditions and the following disclaimer in the
11 documentation and/or other materials provided with the distribution.
12 * Neither the name of the <organization> nor the
13 names of its contributors may be used to endorse or promote products
14 derived from this software without specific prior written permission.
15
16 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY
17 EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
18 WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
19 DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY
20 DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
21 (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
22 LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
23 ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
25 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26*/
27#pragma once
28
29#include "tiny_dnn/layers/layer.h"
30#include "tiny_dnn/core/params/conv_params.h"
31#include "tiny_dnn/core/params/deconv_params.h"
32#include "tiny_dnn/core/params/maxpool_params.h"
33#include "tiny_dnn/core/params/fully_params.h"
34
35namespace tiny_dnn {
36namespace core {
37
38// TODO(edgar): remove this
39class context;
40
41enum class backend_t { internal, nnpack, libdnn, avx, opencl };
42
43inline std::ostream& operator << (std::ostream& os, backend_t type) {
44 switch (type) {
45 case backend_t::internal: os << "Internal"; break;
46 case backend_t::nnpack: os << "NNPACK"; break;
47 case backend_t::libdnn: os << "LibDNN"; break;
48 case backend_t::avx: os << "AVX"; break;
49 case backend_t::opencl: os << "OpenCL"; break;
50 default:
51 throw nn_error("Not supported ostream enum.");
52 break;
53 }
54 return os;
55}
56
57/*enum class Engine { OpenCL };*/
58
59inline backend_t default_engine() {
60#ifdef CNN_USE_AVX
61#if defined(__AVX__) || defined(__AVX2__)
62 return backend_t::avx;
63#endif
64#endif // CNN_USE_AVX
65 return backend_t::internal;
66}
67
68class backend {
69 public:
70 // context holds solution-dependent parameters
71 // context should be able to hold any types of structures (like boost::any)
72 explicit backend(context* ctx_ = nullptr) {}
73
74 // core math functions
75
76 virtual void conv2d(const std::vector<tensor_t*>& in_data,
77 std::vector<tensor_t*>& out_data) = 0;
78
79 virtual void conv2d_q(const std::vector<tensor_t*>& in_data,
80 std::vector<tensor_t*>& out_data) = 0;
81
82 virtual void conv2d_eq(const std::vector<tensor_t*>& in_data,
83 std::vector<tensor_t*>& out_data) = 0;
84
85 virtual void conv2d(const std::vector<tensor_t*>& in_data,
86 const std::vector<tensor_t*>& out_data,
87 std::vector<tensor_t*>& out_grad,
88 std::vector<tensor_t*>& in_grad) = 0;
89
90 virtual void conv2d_q(const std::vector<tensor_t*>& in_data,
91 const std::vector<tensor_t*>& out_data,
92 std::vector<tensor_t*>& out_grad,
93 std::vector<tensor_t*>& in_grad) = 0;
94
95 virtual void deconv2d(const std::vector<tensor_t*>& in_data,
96 std::vector<tensor_t*>& out_data) = 0;
97
98 virtual void deconv2d_q(const std::vector<tensor_t*>& in_data,
99 std::vector<tensor_t*>& out_data) = 0;
100
101 virtual void deconv2d_eq(const std::vector<tensor_t*>& in_data,
102 std::vector<tensor_t*>& out_data) = 0;
103
104 virtual void deconv2d(const std::vector<tensor_t*>& in_data,
105 const std::vector<tensor_t*>& out_data,
106 std::vector<tensor_t*>& out_grad,
107 std::vector<tensor_t*>& in_grad) = 0;
108
109 virtual void deconv2d_q(const std::vector<tensor_t*>& in_data,
110 const std::vector<tensor_t*>& out_data,
111 std::vector<tensor_t*>& out_grad,
112 std::vector<tensor_t*>& in_grad) = 0;
113
114 virtual void maxpool(const std::vector<tensor_t*>& in_data,
115 std::vector<tensor_t*>& out_data) = 0;
116
117 virtual void maxpool(const std::vector<tensor_t*>& in_data,
118 const std::vector<tensor_t*>& out_data,
119 std::vector<tensor_t*>& out_grad,
120 std::vector<tensor_t*>& in_grad) = 0;
121
122 virtual void fully(const std::vector<tensor_t*>& in_data,
123 std::vector<tensor_t*>& out_data) = 0;
124
125 virtual void fully_q(const std::vector<tensor_t*>& in_data,
126 std::vector<tensor_t*>& out_data) = 0;
127
128 virtual void fully_eq(const std::vector<tensor_t*>& in_data,
129 std::vector<tensor_t*>& out_data) = 0;
130
131 virtual void fully(const std::vector<tensor_t*>& in_data,
132 const std::vector<tensor_t*>& out_data,
133 std::vector<tensor_t*>& out_grad,
134 std::vector<tensor_t*>& in_grad) = 0;
135
136 virtual void fully_q(const std::vector<tensor_t*>& in_data,
137 const std::vector<tensor_t*>& out_data,
138 std::vector<tensor_t*>& out_grad,
139 std::vector<tensor_t*>& in_grad) = 0;
140
141 context* get_context() const { return ctx_; }
142
143 void set_layer(layerptr_t layer) { layer_ = layer; }
144
145 virtual backend_t type() const = 0;
146
147 protected:
148 context* ctx_;
149 layerptr_t layer_;
150};
151
152} // namespace core
153} // namespace tiny_dnn
Definition backend.h:68
Simple image utility class.
Definition image.h:94
base class of all kind of NN layers
Definition layer.h:62