megenginelite-sys 1.8.2

A safe megenginelite wrapper in Rust
Documentation 
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

/**
 * \file dnn/src/cuda/deformable_conv/opr_impl.h
 * MegEngine is Licensed under the Apache License, Version 2.0 (the "License")
 *
 * Copyright (c) 2014-2021 Megvii Inc. All rights reserved.
 *
 * Unless required by applicable law or agreed to in writing,
 * software distributed under the License is distributed on an
 * "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or
 * implied.
 */
#pragma once

#include "megdnn/oprs/nn.h"

namespace megdnn {
namespace cuda {

class DeformableConvForwardImpl : public DeformableConvForward {
public:
    using DeformableConvForward::DeformableConvForward;

    void exec(
            _megdnn_tensor_in im, _megdnn_tensor_in filter, _megdnn_tensor_in offset,
            _megdnn_tensor_in mask, _megdnn_tensor_out dst,
            _megdnn_workspace workspace) override;

    size_t get_workspace_in_bytes(
            const TensorLayout& im, const TensorLayout& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& dst) override;

    Algorithm* get_algorithm_heuristic(
            const TensorLayout& im, const CanonizedFilterMeta& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& dst, size_t workspace_limit_in_bytes,
            const AlgoAttribute& positive_attr, const AlgoAttribute& negative_attr);

    const char* get_algorithm_set_name() const override;

    class AlgoBase;
    class AlgoMatmul;

    class AlgoPack;

    static const AlgoPack& algo_pack() { return sm_algo_pack; }
    Algorithm* get_algorithm_from_desc(const AlgorithmDesc& desc) override;

protected:
    std::vector<Algorithm*> get_all_algorithms(
            const TensorLayout& im, const TensorLayout& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& dst) override;
    std::vector<Algorithm*> get_all_algorithms_safe(
            const TensorLayout& im, const TensorLayout& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& dst) override;

    Algorithm* get_algorithm_heuristic(
            const TensorLayout& im, const TensorLayout& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& dst, size_t workspace_limit_in_bytes,
            const AlgoAttribute& positive_attr,
            const AlgoAttribute& negative_attr) override;

private:
    static AlgoPack sm_algo_pack;
};

class DeformableConvBackwardFilterImpl : public DeformableConvBackwardFilter {
public:
    using DeformableConvBackwardFilter::DeformableConvBackwardFilter;

    void exec(
            _megdnn_tensor_in im, _megdnn_tensor_in offset, _megdnn_tensor_in mask,
            _megdnn_tensor_in out_grad, _megdnn_tensor_out filter_grad,
            _megdnn_workspace workspace) override;

    Algorithm* get_algorithm_heuristic(
            const TensorLayout& im, const TensorLayout& offset,
            const TensorLayout& mask, const TensorLayout& out_grad,
            const CanonizedFilterMeta& filter_grad, size_t workspace_limit_in_bytes,
            const AlgoAttribute& positive_attr, const AlgoAttribute& negative_attr);

    size_t get_workspace_in_bytes(
            const TensorLayout& im, const TensorLayout& offset,
            const TensorLayout& mask, const TensorLayout& out_grad,
            const TensorLayout& filter_grad) override;

    const char* get_algorithm_set_name() const override;

    class AlgoBase;
    class AlgoMatmul;

    class AlgoPack;

    static const AlgoPack& algo_pack() { return sm_algo_pack; }
    Algorithm* get_algorithm_from_desc(const AlgorithmDesc& desc) override;

protected:
    std::vector<Algorithm*> get_all_algorithms(
            const TensorLayout& im, const TensorLayout& offset,
            const TensorLayout& mask, const TensorLayout& out_grad,
            const TensorLayout& filter_grad) override;
    std::vector<Algorithm*> get_all_algorithms_safe(
            const TensorLayout& im, const TensorLayout& offset,
            const TensorLayout& mask, const TensorLayout& out_grad,
            const TensorLayout& filter_grad) override;

    Algorithm* get_algorithm_heuristic(
            const TensorLayout& im, const TensorLayout& offset,
            const TensorLayout& mask, const TensorLayout& out_grad,
            const TensorLayout& filter_grad, size_t workspace_limit_in_bytes,
            const AlgoAttribute& positive_attr,
            const AlgoAttribute& negative_attr) override;

private:
    static AlgoPack sm_algo_pack;
};

class DeformableConvBackwardDataImpl : public DeformableConvBackwardData {
public:
    using DeformableConvBackwardData::DeformableConvBackwardData;

    void exec(
            _megdnn_tensor_in im, _megdnn_tensor_in filter, _megdnn_tensor_in offset,
            _megdnn_tensor_in mask, _megdnn_tensor_in out_grad,
            _megdnn_tensor_out im_grad, _megdnn_tensor_out offset_grad,
            _megdnn_tensor_out mask_grad, _megdnn_workspace workspace) override;

    Algorithm* get_algorithm_heuristic(
            const TensorLayout& im, const CanonizedFilterMeta& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& out_grad, const TensorLayout& im_grad,
            const TensorLayout& offset_grad, const TensorLayout& mask_grad,
            size_t workspace_limit_in_bytes, const AlgoAttribute& positive_attr,
            const AlgoAttribute& negative_attr);

    size_t get_workspace_in_bytes(
            const TensorLayout& im, const TensorLayout& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& out_grad, const TensorLayout& im_grad,
            const TensorLayout& offset_grad, const TensorLayout& mask_grad) override;

    const char* get_algorithm_set_name() const override;

    class AlgoBase;
    class AlgoMatmul;

    class AlgoPack;

    static const AlgoPack& algo_pack() { return sm_algo_pack; }
    Algorithm* get_algorithm_from_desc(const AlgorithmDesc& desc) override;

protected:
    std::vector<Algorithm*> get_all_algorithms(
            const TensorLayout& im, const TensorLayout& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& out_grad, const TensorLayout& im_grad,
            const TensorLayout& offset_grad, const TensorLayout& mask_grad) override;

    std::vector<Algorithm*> get_all_algorithms_safe(
            const TensorLayout& im, const TensorLayout& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& out_grad, const TensorLayout& im_grad,
            const TensorLayout& offset_grad, const TensorLayout& mask_grad) override;

    Algorithm* get_algorithm_heuristic(
            const TensorLayout& im, const TensorLayout& filter,
            const TensorLayout& offset, const TensorLayout& mask,
            const TensorLayout& out_grad, const TensorLayout& im_grad,
            const TensorLayout& offset_grad, const TensorLayout& mask_grad,
            size_t workspace_limit_in_bytes, const AlgoAttribute& positive_attr,
            const AlgoAttribute& negative_attr) override;

private:
    static AlgoPack sm_algo_pack;
};

}  // namespace cuda
}  // namespace megdnn

// vim: syntax=cpp.doxygen