onnx.cpp 38.4 KB
Newer Older
Paul's avatar
Paul committed
1
2
3
4
5
6
7
8
#include <google/protobuf/text_format.h>
#include <google/protobuf/io/zero_copy_stream_impl.h>
#include <onnx.pb.h>
#include <iostream>
#include <fstream>
#include <unordered_map>
#include <functional>
#include <array>
Paul's avatar
Paul committed
9
#include <utility>
10
#include <vector>
Paul's avatar
Paul committed
11

Paul's avatar
Paul committed
12
13
14
15
16
17
#include <migraphx/fallthrough.hpp>
#include <migraphx/program.hpp>
#include <migraphx/operators.hpp>
#include <migraphx/ranges.hpp>
#include <migraphx/instruction.hpp>
#include <migraphx/config.hpp>
18
#include <migraphx/onnx.hpp>
Paul's avatar
Paul committed
19
20

namespace migraphx {
Paul's avatar
Paul committed
21
inline namespace MIGRAPHX_INLINE_NS {
Paul's avatar
Paul committed
22
23
24
25
26

struct onnx_parser
{
    using attribute_map = std::unordered_map<std::string, onnx::AttributeProto>;
    using node_map      = std::unordered_map<std::string, onnx::NodeProto>;
Paul's avatar
Paul committed
27
    using op_func = std::function<instruction_ref(attribute_map, std::vector<instruction_ref>)>;
Paul's avatar
Paul committed
28
29
    node_map nodes;
    std::unordered_map<std::string, instruction_ref> instructions;
Scott Thornton's avatar
Scott Thornton committed
30
    program prog    = program();
31
    bool is_pytorch = false;
Paul's avatar
Paul committed
32
33

    std::unordered_map<std::string, op_func> ops;
34
    std::unordered_map<std::string, operation> map_actv_funcs;
Paul's avatar
Paul committed
35
36
37

    onnx_parser()
    {
Shucai Xiao's avatar
Shucai Xiao committed
38
        add_generic_op("MatMul", op::dot{});
Khalique's avatar
Khalique committed
39
        add_generic_op("Relu", op::relu{});
Khalique's avatar
Khalique committed
40
41
        add_generic_op("Sigmoid", op::sigmoid{});
        add_generic_op("Abs", op::abs{});
Shucai Xiao's avatar
Shucai Xiao committed
42
43
        add_generic_op("Exp", op::exp{});
        add_generic_op("Log", op::log{});
Khalique's avatar
Khalique committed
44
45
        // disable dropout for inference
        add_generic_op("Dropout", op::identity{});
Khalique's avatar
Khalique committed
46
        add_generic_op("Identity", op::identity{});
Shucai Xiao's avatar
Shucai Xiao committed
47
48
49
        add_generic_op("Sin", op::sin{});
        add_generic_op("Cos", op::cos{});
        add_generic_op("Tan", op::tan{});
50
51
        add_generic_op("Sinh", op::sinh{});
        add_generic_op("Cosh", op::cosh{});
52
        add_generic_op("Tanh", op::tanh{});
53
54
55
        add_generic_op("Asin", op::asin{});
        add_generic_op("Acos", op::acos{});
        add_generic_op("Atan", op::atan{});
Paul's avatar
Paul committed
56

Khalique's avatar
Khalique committed
57
58
59
60
61
        add_binary_op("Add", op::add{});
        add_binary_op("Div", op::div{});
        add_binary_op("Mul", op::mul{});
        add_binary_op("Sub", op::sub{});

Khalique's avatar
Khalique committed
62
63
64
        add_variadic_op("Sum", op::add{});
        add_variadic_op("Max", op::max{});
        add_variadic_op("Min", op::min{});
Paul's avatar
Paul committed
65

Khalique's avatar
Khalique committed
66
        add_mem_op("ImageScaler", &onnx_parser::parse_imagescaler);
67
        add_mem_op("LeakyRelu", &onnx_parser::parse_leaky_relu);
Khalique's avatar
Khalique committed
68
        add_mem_op("Elu", &onnx_parser::parse_elu);
Paul's avatar
Paul committed
69
70
        add_mem_op("Constant", &onnx_parser::parse_constant);
        add_mem_op("Conv", &onnx_parser::parse_conv);
Paul's avatar
Paul committed
71
72
        add_mem_op("MaxPool", &onnx_parser::parse_pooling);
        add_mem_op("AveragePool", &onnx_parser::parse_pooling);
73
74
        add_mem_op("GlobalMaxPool", &onnx_parser::parse_pooling);
        add_mem_op("GlobalAveragePool", &onnx_parser::parse_pooling);
Paul's avatar
Paul committed
75
        add_mem_op("Reshape", &onnx_parser::parse_reshape);
Paul's avatar
Paul committed
76
77
        add_mem_op("Flatten", &onnx_parser::parse_flatten);
        add_mem_op("Gemm", &onnx_parser::parse_gemm);
78
        add_mem_op("BatchNormalization", &onnx_parser::parse_batchnorm);
Paul's avatar
Paul committed
79
        add_mem_op("Softmax", &onnx_parser::parse_softmax);
80
81
82
        add_mem_op("Squeeze", &onnx_parser::parse_squeeze);
        add_mem_op("Unsqueeze", &onnx_parser::parse_unsqueeze);
        add_mem_op("Slice", &onnx_parser::parse_slice);
Scott Thornton's avatar
Scott Thornton committed
83
        add_mem_op("Concat", &onnx_parser::parse_concat);
84
85
86
        add_mem_op("Gather", &onnx_parser::parse_gather);
        add_mem_op("Shape", &onnx_parser::parse_shape);
        add_mem_op("ConstantFill", &onnx_parser::parse_constant_fill);
Khalique's avatar
Khalique committed
87
        add_mem_op("Transpose", &onnx_parser::parse_transpose);
Shucai Xiao's avatar
Shucai Xiao committed
88
        add_mem_op("RNN", &onnx_parser::parse_rnn);
89
90
91
92
93
94
95

        // init the activation function map
        init_actv_func();
    }

    void init_actv_func()
    {
96
97
98
99
100
        map_actv_funcs.insert(std::make_pair("tanh", op::tanh{}));
        map_actv_funcs.insert(std::make_pair("relu", op::relu{}));
        map_actv_funcs.insert(std::make_pair("sigmoid", op::sigmoid{}));
        map_actv_funcs.insert(std::make_pair("leakyrelu", op::leaky_relu{}));
        map_actv_funcs.insert(std::make_pair("elu", op::elu{}));
Paul's avatar
Paul committed
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
    }

    template <class F>
    void add_op(std::string name, F f)
    {
        ops.emplace(name, f);
    }

    template <class F>
    void add_mem_op(std::string name, F f)
    {
        ops.emplace(name, [=](auto&&... xs) {
            return std::mem_fn(f)(*this, name, std::forward<decltype(xs)>(xs)...);
        });
    }
Khalique's avatar
Khalique committed
116

117
    template <class T>
Khalique's avatar
Khalique committed
118
    void add_binary_op(std::string name, T x)
119
120
    {
        ops.emplace(name, [this, x](attribute_map attributes, std::vector<instruction_ref> args) {
Scott Thornton's avatar
Scott Thornton committed
121
            if(args.size() != 2)
Paul's avatar
Paul committed
122
                MIGRAPHX_THROW("binary operators should have 2 operands");
123
124
125
126
127
128
129
130
131
132
133
134
            if(contains(attributes, "broadcast"))
            {
                uint64_t broadcasted = parse_value(attributes.at("broadcast")).at<uint64_t>();
                if(broadcasted != 0)
                {
                    uint64_t axis = (contains(attributes, "axis"))
                                        ? parse_value(attributes.at("axis")).at<uint64_t>()
                                        : 0;
                    auto l =
                        prog.add_instruction(op::broadcast{axis, args[0]->get_shape()}, args[1]);
                    return prog.add_instruction(x, args[0], l);
                }
135
                return prog.add_instruction(x, args);
136
            }
Paul's avatar
Paul committed
137
            else
138
            {
Khalique's avatar
Khalique committed
139
                return add_broadcastable_binary_op(args[0], args[1], x);
140
141
142
143
            }
        });
    }

Khalique's avatar
Khalique committed
144
145
146
147
148
    template <class T>
    instruction_ref add_broadcastable_binary_op(instruction_ref arg0, instruction_ref arg1, T x)
    {
        if(arg0->get_shape() != arg1->get_shape())
        {
Khalique's avatar
Khalique committed
149
150
151
152
153
154
155
156
157
158
159
160
161
            // Example:
            // s0 = (3,2,4,5) and s1 = (2,1,1)
            //
            // In this case we need to broadcast (:,1,1) portion of
            // s1 plus broadcast the 1st dimension of s1
            // giving output_lens = (3,2,4,5)
            //
            // Another example:
            // s0 = (3,2,1,5) and s1 = (2,7,5)
            // In this case we need to broadcast the (:,:,1:,:) axis
            // of s0 plus the 1st dimension of s1 giving
            // output_lens = (3,2,7,5)
            //
Khalique's avatar
Khalique committed
162
163
164
165
166
167
168
169
            // Get lengths for both arguments
            const std::vector<std::size_t>* s0 = &arg0->get_shape().lens();
            const std::vector<std::size_t>* s1 = &arg1->get_shape().lens();

            // Make sure s0 is the smaller size
            if(s0->size() > s1->size())
                std::swap(s0, s1);

Khalique's avatar
Khalique committed
170
            std::vector<std::size_t> output_lens(*s1);
Khalique's avatar
Khalique committed
171
172
            auto offset = s1->size() - s0->size();
            std::transform(s0->begin(),
Khalique's avatar
Khalique committed
173
174
175
176
                           s0->end(),
                           s1->begin() + offset,
                           output_lens.begin() + offset,
                           [](auto a, auto b) { return std::max(a, b); });
Khalique's avatar
Khalique committed
177
178
179
180
181
182
183
184
185

            auto l0 = prog.add_instruction(op::multibroadcast{output_lens}, arg0);
            auto l1 = prog.add_instruction(op::multibroadcast{output_lens}, arg1);
            return prog.add_instruction(x, l0, l1);
        }
        else
        {
            return prog.add_instruction(x, {arg0, arg1});
        }
186
187
    }

Paul's avatar
Paul committed
188
    template <class T>
Paul's avatar
Paul committed
189
190
    void add_generic_op(std::string name, T x)
    {
Paul's avatar
Paul committed
191
        ops.emplace(name, [this, x](attribute_map, std::vector<instruction_ref> args) {
Paul's avatar
Paul committed
192
193
194
195
            return prog.add_instruction(x, args);
        });
    }

Khalique's avatar
Khalique committed
196
    template <class T>
Khalique's avatar
Khalique committed
197
    void add_variadic_op(std::string name, T x)
Khalique's avatar
Khalique committed
198
    {
Khalique's avatar
Khalique committed
199
200
        ops.emplace(name, [this, x](attribute_map, std::vector<instruction_ref> args) {
            return std::accumulate(std::next(args.begin()),
Khalique's avatar
Khalique committed
201
202
203
204
205
                                   args.end(),
                                   args.front(),
                                   [this, x](instruction_ref a, instruction_ref b) {
                                       return add_broadcastable_binary_op(a, b, x);
                                   });
Khalique's avatar
Khalique committed
206
        });
Khalique's avatar
Khalique committed
207
208
    }

Paul's avatar
Paul committed
209
    instruction_ref
Paul's avatar
Paul committed
210
    parse_softmax(const std::string&, const attribute_map&, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
211
212
    {
        auto dims = args.front()->get_shape().lens();
Scott Thornton's avatar
Scott Thornton committed
213
214
        auto r =
            prog.add_instruction(op::reshape{{long(dims[0]), long(dims[1]), 1, 1}}, args.front());
215
216
        auto s = prog.add_instruction(op::softmax{}, r);
        return prog.add_instruction(op::reshape{{long(dims[0]), long(dims[1])}}, s);
Paul's avatar
Paul committed
217
218
    }

Paul's avatar
Paul committed
219
    instruction_ref
Paul's avatar
Paul committed
220
    parse_conv(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
221
    {
222
        op::convolution op;
Paul's avatar
Paul committed
223
224
        if(contains(attributes, "pads"))
        {
Scott Thornton's avatar
Scott Thornton committed
225
            if(contains(attributes, "auto_pad"))
226
            {
Paul's avatar
Paul committed
227
                MIGRAPHX_THROW("auto_pad and padding cannot be specified simultaneously");
228
229
230
            }
            std::vector<std::size_t> padding(4);
            copy(attributes["pads"].ints(), padding.begin());
Scott Thornton's avatar
Scott Thornton committed
231
            if(padding.size() != 4)
232
            {
Paul's avatar
Paul committed
233
                MIGRAPHX_THROW("padding should have 4 values");
234
            }
Scott Thornton's avatar
Scott Thornton committed
235
            if(padding[0] != padding[2] || padding[1] != padding[3])
236
            {
Paul's avatar
Paul committed
237
                MIGRAPHX_THROW("migraphx does not support asymetric padding");
238
239
240
            }
            op.padding[0] = padding[0];
            op.padding[1] = padding[1];
Paul's avatar
Paul committed
241
        }
Paul's avatar
Paul committed
242
243
244
245
246
247
248
249
        if(contains(attributes, "strides"))
        {
            copy(attributes["strides"].ints(), op.stride.begin());
        }
        if(contains(attributes, "dilations"))
        {
            copy(attributes["dilations"].ints(), op.dilation.begin());
        }
Scott Thornton's avatar
Scott Thornton committed
250
        if(contains(attributes, "auto_pad"))
251
252
        {
            auto s = attributes["auto_pad"].s();
Scott Thornton's avatar
Scott Thornton committed
253
            if(contains(attributes, "pads") and to_upper(s) != "NOTSET")
254
            {
Paul's avatar
Paul committed
255
                MIGRAPHX_THROW("auto_pad and padding cannot be specified simultaneously");
256
257
            }

wsttiger's avatar
fixes  
wsttiger committed
258
            if(s.find("SAME") != std::string::npos)
259
260
261
262
            {
                op.padding_mode = op::convolution::same;
            }
        }
Khalique's avatar
Khalique committed
263
264
265
266
        if(contains(attributes, "group"))
        {
            op.group = parse_value(attributes.at("group")).at<int>();
        }
Paul's avatar
Paul committed
267
268
269
270
        if(args.size() == 3)
        {
            uint64_t axis = 1;
            auto l1       = prog.add_instruction(op, args[0], args[1]);
Scott Thornton's avatar
Scott Thornton committed
271
            auto l2       = prog.add_instruction(op::broadcast{axis, l1->get_shape()}, args[2]);
272
            return prog.add_instruction(op::add{}, l1, l2);
Paul's avatar
Paul committed
273
        }
Paul's avatar
Paul committed
274
275
        return prog.add_instruction(op, args);
    }
Paul's avatar
Paul committed
276

Paul's avatar
Paul committed
277
278
279
    instruction_ref parse_pooling(const std::string& name,
                                  attribute_map attributes,
                                  std::vector<instruction_ref> args)
Paul's avatar
Paul committed
280
    {
Khalique's avatar
Khalique committed
281
282
        op::pooling op{ends_with(name, "MaxPool") ? "max" : "average"};
        if(starts_with(name, "Global"))
283
        {
Khalique's avatar
Khalique committed
284
285
            auto lens  = args.front()->get_shape().lens();
            op.lengths = {lens[2], lens[3]};
286
        }
Paul's avatar
Paul committed
287
288
        if(contains(attributes, "pads"))
        {
289
290
            std::vector<std::size_t> padding(4);
            copy(attributes["pads"].ints(), padding.begin());
Scott Thornton's avatar
Scott Thornton committed
291
            if(padding.size() != 4)
292
            {
Paul's avatar
Paul committed
293
                MIGRAPHX_THROW("padding should have 4 values");
294
            }
Scott Thornton's avatar
Scott Thornton committed
295
            if(padding[0] != padding[2] || padding[1] != padding[3])
296
            {
Paul's avatar
Paul committed
297
                MIGRAPHX_THROW("migraphx does not support asymetric padding");
298
299
300
            }
            op.padding[0] = padding[0];
            op.padding[1] = padding[1];
Paul's avatar
Paul committed
301
302
303
304
305
306
307
308
309
        }
        if(contains(attributes, "strides"))
        {
            copy(attributes["strides"].ints(), op.stride.begin());
        }
        if(contains(attributes, "kernel_shape"))
        {
            copy(attributes["kernel_shape"].ints(), op.lengths.begin());
        }
Scott Thornton's avatar
Scott Thornton committed
310
        if(contains(attributes, "auto_pad"))
311
312
        {
            auto s = attributes["auto_pad"].s();
Scott Thornton's avatar
Scott Thornton committed
313
            if(to_upper(s) != "NOTSET")
314
            {
Paul's avatar
Paul committed
315
                MIGRAPHX_THROW("auto_pad is not supported for pooling");
316
317
318
            }
        }

Paul's avatar
Paul committed
319
        return prog.add_instruction(op, std::move(args));
Paul's avatar
Paul committed
320
321
    }

Paul's avatar
Paul committed
322
    instruction_ref
Paul's avatar
Paul committed
323
    parse_reshape(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
324
    {
325
        op::reshape op;
Paul's avatar
Paul committed
326
327
328
329
330
331
332
        if(args.size() == 1)
        {
            literal s = parse_value(attributes.at("shape"));
            s.visit([&](auto v) { copy(v, std::back_inserter(op.dims)); });
        }
        if(args.size() == 2)
        {
Paul's avatar
Paul committed
333
            literal s = args[1]->get_literal();
Paul's avatar
Paul committed
334
            s.visit([&](auto v) { copy(v, std::back_inserter(op.dims)); });
Paul's avatar
Paul committed
335
        }
Paul's avatar
Paul committed
336
337
338
        return prog.add_instruction(op, args[0]);
    }

Paul's avatar
Paul committed
339
    instruction_ref
Paul's avatar
Paul committed
340
    parse_flatten(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
341
    {
342
        uint64_t axis = 1;
Paul's avatar
Paul committed
343
344
345
346
        if(contains(attributes, "axis"))
        {
            axis = parse_value(attributes.at("axis")).at<int>();
        }
347
        return prog.add_instruction(op::flatten{axis}, args[0]);
Paul's avatar
Paul committed
348
349
    }

350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
    instruction_ref
    parse_squeeze(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        op::squeeze op;
        literal s = parse_value(attributes.at("axes"));
        s.visit([&](auto v) { copy(v, std::back_inserter(op.axes)); });
        return prog.add_instruction(op, args[0]);
    }

    instruction_ref
    parse_unsqueeze(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        op::unsqueeze op;
        literal s = parse_value(attributes.at("axes"));
        s.visit([&](auto v) { copy(v, std::back_inserter(op.axes)); });
        return prog.add_instruction(op, args[0]);
    }

Scott Thornton's avatar
Scott Thornton committed
368
369
370
371
372
373
374
    instruction_ref
    parse_concat(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        std::size_t axis = parse_value(attributes.at("axis")).at<int>();
        op::concat op{axis};
        return prog.add_instruction(op, std::move(args));
    }
375

376
377
378
379
380
381
382
383
    instruction_ref
    parse_gather(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        std::size_t axis = 0;
        if(contains(attributes, "axis"))
        {
            axis = parse_value(attributes.at("axis")).at<int>();
        }
384
        op::gather op{axis};
385
386
387
        return prog.add_instruction(op, std::move(args));
    }

388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
    instruction_ref
    parse_slice(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        op::slice op;
        if(contains(attributes, "axes"))
        {
            literal s = parse_value(attributes.at("axes"));
            s.visit([&](auto v) { copy(v, std::back_inserter(op.axes)); });
        }
        {
            literal s = parse_value(attributes.at("ends"));
            s.visit([&](auto v) { copy(v, std::back_inserter(op.ends)); });
        }
        {
            literal s = parse_value(attributes.at("starts"));
            s.visit([&](auto v) { copy(v, std::back_inserter(op.starts)); });
        }
        return prog.add_instruction(op, args[0]);
    }

Paul's avatar
Paul committed
408
409
410
    instruction_ref parse_constant(const std::string&,
                                   attribute_map attributes,
                                   const std::vector<instruction_ref>&)
Paul's avatar
Paul committed
411
412
413
414
    {
        literal v = parse_value(attributes.at("value"));
        return prog.add_literal(v);
    }
Paul's avatar
Paul committed
415

Paul's avatar
Paul committed
416
    instruction_ref
Paul's avatar
Paul committed
417
    parse_gemm(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Paul's avatar
Paul committed
418
419
    {
        float alpha = 1.0f;
Khalique's avatar
Khalique committed
420
        float beta  = 1.0f;
Paul's avatar
Paul committed
421
422
423
424
425
426
427
428
        bool transa = false;
        bool transb = false;
        if(contains(attributes, "alpha"))
        {
            alpha = parse_value(attributes.at("alpha")).at<float>();
        }
        if(contains(attributes, "beta"))
        {
429
            beta = parse_value(attributes.at("beta")).at<float>();
Paul's avatar
Paul committed
430
431
432
433
434
435
436
437
438
439
        }
        if(contains(attributes, "transA"))
        {
            transa = parse_value(attributes.at("transA")).at<bool>();
        }
        if(contains(attributes, "transB"))
        {
            transb = parse_value(attributes.at("transB")).at<bool>();
        }
        std::vector<int64_t> perm = {1, 0};
440
441
        auto l1 = (transa) ? prog.add_instruction(op::transpose{perm}, args[0]) : args[0];
        auto l2 = (transb) ? prog.add_instruction(op::transpose{perm}, args[1]) : args[1];
Paul's avatar
Paul committed
442
443
        if(args.size() == 3)
        {
Khalique's avatar
Khalique committed
444
            if(beta != 0.f)
445
            {
Khalique's avatar
Khalique committed
446
                auto l3 = prog.add_instruction(op::dot{alpha}, l1, l2);
Khalique's avatar
Khalique committed
447
                auto l4 = args[2];
Khalique's avatar
Khalique committed
448
                if(l4->get_shape().scalar()) // ignore args[2] (no C value added to alpha*A*B)
Khalique's avatar
Khalique committed
449
                    return l3;
Khalique's avatar
Khalique committed
450
                if(beta != 1.f)
Khalique's avatar
Khalique committed
451
452
                {
                    auto beta_val = prog.add_literal(beta);
Khalique's avatar
Khalique committed
453
454
                    auto l5 = prog.add_instruction(op::scalar{args[2]->get_shape()}, beta_val);
                    l4      = prog.add_instruction(op::mul{}, args[2], l5);
Khalique's avatar
Khalique committed
455
456
                }
                return add_broadcastable_binary_op(l3, l4, op::add{});
457
            }
Paul's avatar
Paul committed
458
        }
Shucai Xiao's avatar
Shucai Xiao committed
459
        return prog.add_instruction(op::dot{alpha, beta}, l1, l2);
Paul's avatar
Paul committed
460
461
    }

462
    instruction_ref
Paul's avatar
Paul committed
463
    parse_batchnorm(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
464
    {
Scott Thornton's avatar
Scott Thornton committed
465
466
        float epsilon                                     = 1e-5f;
        float momentum                                    = 0.9f;
467
        op::batch_norm_inference::bn_infer_mode_t bn_mode = op::batch_norm_inference::spatial;
Scott Thornton's avatar
Scott Thornton committed
468
        bool is_test                                      = false;
469
470
471
472
473
474
        if(contains(attributes, "epsilon"))
        {
            epsilon = parse_value(attributes.at("epsilon")).at<float>();
        }
        if(contains(attributes, "momentum"))
        {
475
            momentum = parse_value(attributes.at("momentum")).at<float>();
476
477
478
        }
        if(contains(attributes, "is_test"))
        {
wsttiger's avatar
wsttiger committed
479
            is_test = parse_value(attributes.at("is_test")).at<uint64_t>() > 0;
480
481
482
        }
        if(contains(attributes, "spatial"))
        {
483
            bn_mode = (parse_value(attributes.at("spatial")).at<uint64_t>() > 0)
484
485
                          ? op::batch_norm_inference::spatial
                          : op::batch_norm_inference::per_activation;
486
        }
Paul's avatar
Paul committed
487
        (void)is_test;
Paul's avatar
Paul committed
488
        op::batch_norm_inference op{epsilon, momentum, bn_mode};
Paul's avatar
Paul committed
489
        return prog.add_instruction(op, std::move(args));
490
491
    }

492
493
494
495
    instruction_ref parse_leaky_relu(const std::string&,
                                     attribute_map attributes,
                                     std::vector<instruction_ref> args)
    {
Khalique's avatar
Khalique committed
496
        float alpha = 0.01; // default alpha val for leaky relu
497
498
499
500
501
502
503
504
        if(contains(attributes, "alpha"))
        {
            alpha = parse_value(attributes.at("alpha")).at<float>();
        }
        op::leaky_relu op{alpha};
        return prog.add_instruction(op, args.front());
    }

Khalique's avatar
Khalique committed
505
506
    instruction_ref
    parse_elu(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Khalique's avatar
Khalique committed
507
508
509
510
511
512
513
514
515
516
    {
        float alpha = 1.0; // default alpha val for elu
        if(contains(attributes, "alpha"))
        {
            alpha = parse_value(attributes.at("alpha")).at<float>();
        }
        op::elu op{alpha};
        return prog.add_instruction(op, args.front());
    }

Khalique's avatar
Khalique committed
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
    instruction_ref parse_imagescaler(const std::string&,
                                      attribute_map attributes,
                                      std::vector<instruction_ref> args)
    {
        float scale = 1.0;
        std::vector<float> bias{};
        if(contains(attributes, "scale"))
        {
            scale = parse_value(attributes.at("scale")).at<float>();
        }

        if(contains(attributes, "bias"))
        {
            auto&& bias_floats = attributes["bias"].floats();
            bias               = std::vector<float>(bias_floats.begin(), bias_floats.end());
        }
        auto input_shape = args.front()->get_shape();
Khalique's avatar
Khalique committed
534

Khalique's avatar
Khalique committed
535
536
        auto scale_val = prog.add_literal(scale);
        auto bias_vals = prog.add_literal(
Paul's avatar
Paul committed
537
            migraphx::literal{migraphx::shape{migraphx::shape::float_type, {bias.size()}}, bias});
Khalique's avatar
Khalique committed
538

Paul's avatar
Paul committed
539
540
        auto scale_tensor = prog.add_instruction(migraphx::op::scalar{input_shape}, scale_val);
        auto img_scaled   = prog.add_instruction(migraphx::op::mul{}, args.front(), scale_tensor);
Paul's avatar
Paul committed
541
        auto bias_bcast = prog.add_instruction(migraphx::op::broadcast{1, input_shape}, bias_vals);
Paul's avatar
Paul committed
542
        return prog.add_instruction(migraphx::op::add{}, img_scaled, bias_bcast);
Khalique's avatar
Khalique committed
543
    }
Khalique's avatar
Khalique committed
544

Khalique's avatar
Khalique committed
545
546
    instruction_ref
    parse_transpose(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
Khalique's avatar
Khalique committed
547
548
549
550
551
552
553
    {
        std::vector<int64_t> perm{};
        if(contains(attributes, "perm"))
        {
            auto&& perm_vals = attributes["perm"].ints();
            perm             = std::vector<int64_t>(perm_vals.begin(), perm_vals.end());
        }
Paul's avatar
Paul committed
554
        return prog.add_instruction(migraphx::op::transpose{perm}, args.front());
Khalique's avatar
Khalique committed
555
556
    }

557
558
559
    // Use a literal instruction to replace the shape since, output of
    // shape operator are literals in migraphx
    instruction_ref
Shucai Xiao's avatar
Shucai Xiao committed
560
    parse_shape(const std::string&, const attribute_map&, std::vector<instruction_ref> args)
561
562
    {
        if(args.size() != 1)
563
            MIGRAPHX_THROW("Shape: operator should have 1 operand");
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
        std::vector<std::size_t> arg_shape = args[0]->get_shape().lens();
        std::vector<int64_t> vec_shape(arg_shape.size());
        migraphx::shape s(migraphx::shape::int64_type, {arg_shape.size()});
        std::transform(arg_shape.begin(), arg_shape.end(), vec_shape.begin(), [](auto i) {
            return int64_t(i);
        });
        return prog.add_literal(migraphx::literal{s, vec_shape});
    }

    // Use a literal instruction to replace the constantFill operator. In RNN, input shape
    // and value are fixed, so no need to do the actual computation for the constantFill
    // operator
    instruction_ref parse_constant_fill(const std::string&,
                                        attribute_map attributes,
                                        std::vector<instruction_ref> args)
    {
        int input_as_shape = 0;
        int dtype          = 1;
        float value        = 0.0f;

        if(contains(attributes, "dtype"))
        {
            dtype = parse_value(attributes.at("dtype")).at<int>();
        }
        migraphx::shape::type_t type = get_type(dtype);

        if(contains(attributes, "input_as_shape"))
        {
            input_as_shape = parse_value(attributes.at("input_as_shape")).at<int>();
        }

        if(contains(attributes, "value"))
        {
            value = parse_value(attributes.at("value")).at<float>();
        }

Shucai Xiao's avatar
Shucai Xiao committed
600
601
        if(contains(attributes, "extra_shape"))
        {
602
            MIGRAPHX_THROW("ConstantFill: cannot handle extra shape attribute");
603
604
        }

605
606
        if(input_as_shape == 1)
        {
Shucai Xiao's avatar
Shucai Xiao committed
607
            if(args.size() != 1)
608
            {
609
                MIGRAPHX_THROW("ConstantFill: need an input argument as output shape");
610
611
            }

Shucai Xiao's avatar
Shucai Xiao committed
612
613
            if(contains(attributes, "shape"))
            {
614
                MIGRAPHX_THROW("ConstantFill: cannot set the shape argument and pass in an input "
Shucai Xiao's avatar
Shucai Xiao committed
615
                               "at the same time");
616
617
            }

618
619
620
            migraphx::argument in = args[0]->eval();
            if(in.empty())
            {
621
                MIGRAPHX_THROW("ConstantFill: cannot handle dynamic shape as input");
622
            }
623

624
625
626
            std::vector<std::size_t> dims;
            in.visit([&](auto input) { dims.assign(input.begin(), input.end()); });
            migraphx::shape s(type, dims);
627
628
            std::vector<float> values(s.elements(), value);
            return prog.add_literal(migraphx::literal(s, values));
629
630
631
        }
        else if(input_as_shape == 0)
        {
Shucai Xiao's avatar
Shucai Xiao committed
632
633
            if(!contains(attributes, "shape"))
            {
634
                MIGRAPHX_THROW("ConstantFill: attribute output shape is needed");
635
636
637
            }

            literal ls = parse_value(attributes.at("shape"));
638
            std::vector<std::size_t> dims;
Shucai Xiao's avatar
Shucai Xiao committed
639
            ls.visit([&](auto s) { dims.assign(s.begin(), s.end()); });
640
            migraphx::shape s{type, dims};
641
642
            std::vector<float> values(s.elements(), value);
            return prog.add_literal(migraphx::literal(s, values));
643
644
645
        }
        else
        {
646
            MIGRAPHX_THROW("ConstantFill: wrong value of attribute input_as_shape");
647
648
649
        }
    }

Shucai Xiao's avatar
Shucai Xiao committed
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
    instruction_ref
    parse_rnn(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
    {
        migraphx::shape input_shape = args[0]->get_shape();
        migraphx::shape w_shape     = args[1]->get_shape();
        std::size_t hidden_size     = w_shape.lens()[1];

        if(contains(attributes, "hidden_size"))
        {
            hidden_size = parse_value(attributes.at("hidden_size")).at<int>();
        }
        else
        {
            MIGRAPHX_THROW("RNN: hidden size attribute missing");
        }

        // Handling of direction to be added later
        std::string direction{"forward"};
        if(contains(attributes, "direction"))
        {
            direction = attributes.at("direction").s();
        }

        op::rnn::rnn_direction_t dirct = op::rnn::forward;
        if(direction == "bidirectional")
        {
            dirct = op::rnn::bidirectional;
        }
        else if(direction == "reverse")
        {
            dirct = op::rnn::reverse;
        }

683
684
685
686
687
        std::vector<std::string> vec_names{"tanh"};
        if(contains(attributes, "activations"))
        {
            auto names = attributes.at("activations").strings();
            vec_names.clear();
Shucai Xiao's avatar
Shucai Xiao committed
688
            for_each(names.begin(), names.end(), [&](auto& fn) { vec_names.push_back(fn); });
689
690
        }

Shucai Xiao's avatar
Shucai Xiao committed
691
        for_each(vec_names.begin(), vec_names.end(), [&](auto& fn) {
692
693
694
695
696
697
698
699
700
            if(map_actv_funcs.count(fn) == 0)
            {
                MIGRAPHX_THROW("RNN: activation function " + fn + " not supported");
            }
        });

        // bidirectional should have two activation functions
        // if only one actv function is provides, we use it in both
        // forward and reverse direction
Shucai Xiao's avatar
Shucai Xiao committed
701
        if(dirct == op::rnn::bidirectional)
702
        {
Shucai Xiao's avatar
Shucai Xiao committed
703
            if(vec_names.size() == 1)
704
705
706
707
708
709
            {
                vec_names.push_back(vec_names.at(0));
            }
        }

        std::vector<operation> vec_actv_funcs;
Shucai Xiao's avatar
Shucai Xiao committed
710
        for_each(vec_names.begin(), vec_names.end(), [&](auto& fn) {
711
712
            vec_actv_funcs.push_back(map_actv_funcs[fn]);
        });
Shucai Xiao's avatar
Shucai Xiao committed
713

Shucai Xiao's avatar
Shucai Xiao committed
714
715
716
717
718
719
720
        // To be added later
        float clip = 0.0;
        if(contains(attributes, "clip"))
        {
            clip = parse_value(attributes.at("clip")).at<float>();
        }

721
        return prog.add_instruction(op::rnn{hidden_size, vec_actv_funcs, dirct, clip},
722
                                    std::move(args));
Shucai Xiao's avatar
Shucai Xiao committed
723
724
    }

Paul's avatar
Paul committed
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
    void parse_from(std::istream& is)
    {
        onnx::ModelProto model;
        if(model.ParseFromIstream(&is))
        {
            if(model.has_graph())
            {
                this->parse_graph(model.graph());
            }
        }
        else
        {
            throw std::runtime_error("Failed reading");
        }
    }

    void parse_graph(const onnx::GraphProto& graph)
    {
        nodes = get_nodes(graph);
744
745
746
747
748
        std::unordered_map<std::string, onnx::TensorProto> initializer_data;
        for(auto&& f : graph.initializer())
        {
            initializer_data[f.name()] = f;
        }
Paul's avatar
Paul committed
749
750
751
        for(auto&& input : graph.input())
        {
            const std::string& name = input.name();
752
753
754
755
756
757
758
759
760
761
762
763
            // Does the input have an initializer?
            if(contains(initializer_data, name))
            {
                auto t             = initializer_data[name];
                instructions[name] = prog.add_literal(parse_tensor(t));
            }
            else
            {
                // TODO: Get shape of input parameter
                shape s            = parse_type(input.type());
                instructions[name] = prog.add_parameter(name, s);
            }
Paul's avatar
Paul committed
764
765
766
        }
        for(auto&& p : nodes)
        {
767
            this->parse_node(get_name(p.second));
Paul's avatar
Paul committed
768
769
770
        }
    }

Paul's avatar
Paul committed
771
    void parse_node(const std::string& name)
Paul's avatar
Paul committed
772
    {
Paul's avatar
Paul committed
773
        if(name.empty())
Paul's avatar
Paul committed
774
            MIGRAPHX_THROW("Onnx node must have a name");
Paul's avatar
Paul committed
775
776
777
778
779
780
        if(instructions.count(name) == 0)
        {
            auto&& node = nodes.at(name);
            std::vector<instruction_ref> args;
            for(auto&& input : node.input())
            {
781
782
783
784
785
786
787
788
789
790
                // For RNN, LSTM, and GRU operators, one of the input arguments
                // is prim::Undefined, and it is ignored by protobuf. We use a
                // hack to ignore this argument for these three operators
                std::string op_type = node.op_type();
                if((op_type == "RNN" || op_type == "LSTM" || op_type == "GRU") &&
                   input.empty() == true)
                {
                    continue;
                }

Paul's avatar
Paul committed
791
792
                if(nodes.count(input) > 0)
                {
793
                    auto&& iname = get_name(nodes.at(input));
Paul's avatar
Paul committed
794
                    assert(name != iname);
Paul's avatar
Paul committed
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
                    this->parse_node(iname);
                    args.push_back(instructions.at(iname));
                }
                else
                {
                    args.push_back(instructions.at(input));
                }
            }
            if(ops.count(node.op_type()) == 0)
            {
                instructions[name] = prog.add_instruction(unknown{node.op_type()}, args);
            }
            else
            {
                instructions[name] = ops[node.op_type()](get_attributes(node), args);
            }
        }
    }

    static attribute_map get_attributes(const onnx::NodeProto& node)
    {
        std::unordered_map<std::string, onnx::AttributeProto> result;
        for(auto&& attr : node.attribute())
        {
            result[attr.name()] = attr;
        }
        return result;
    }

824
825
826
827
    static std::string get_name(const onnx::NodeProto& node)
    {
        if(node.name().empty())
        {
Paul's avatar
Paul committed
828
            std::string generated = "migraphx_unnamed_node";
Paul's avatar
Paul committed
829
830
831
832
            return std::accumulate(node.output().begin(),
                                   node.output().end(),
                                   generated,
                                   [](auto x, auto y) { return x + "_" + y; });
833
834
835
836
        }
        return node.name();
    }

Paul's avatar
Paul committed
837
838
839
840
841
    static node_map get_nodes(const onnx::GraphProto& graph)
    {
        std::unordered_map<std::string, onnx::NodeProto> result;
        for(auto&& node : graph.node())
        {
842
            result[get_name(node)] = node;
Paul's avatar
Paul committed
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
            for(auto&& output : node.output())
            {
                result[output] = node;
            }
        }
        return result;
    }

    template <class T>
    static literal from_repeated(shape::type_t t, const T& r)
    {
        std::size_t size = r.size();
        return literal{{t, {size}}, r.begin(), r.end()};
    }

    static literal parse_value(const onnx::AttributeProto& attr)
    {
        switch(attr.type())
        {
        case onnx::AttributeProto::UNDEFINED: return {};
        case onnx::AttributeProto::FLOAT: return literal{attr.f()};
        case onnx::AttributeProto::INT: return literal{attr.i()};
        case onnx::AttributeProto::STRING: return {};
        case onnx::AttributeProto::TENSOR: return parse_tensor(attr.t());
        case onnx::AttributeProto::GRAPH: return {};
Paul's avatar
Paul committed
868
        case onnx::AttributeProto::FLOATS: return from_repeated(shape::float_type, attr.floats());
Paul's avatar
Paul committed
869
870
871
872
873
        case onnx::AttributeProto::INTS: return from_repeated(shape::int64_type, attr.ints());
        case onnx::AttributeProto::STRINGS: return {};
        case onnx::AttributeProto::TENSORS: return {};
        case onnx::AttributeProto::GRAPHS: return {};
        }
Paul's avatar
Paul committed
874
        MIGRAPHX_THROW("Invalid attribute type");
Paul's avatar
Paul committed
875
876
877
878
879
    }

    static literal parse_tensor(const onnx::TensorProto& t)
    {
        std::vector<std::size_t> dims(t.dims().begin(), t.dims().end());
Khalique's avatar
Khalique committed
880
        // in case of scalar constants in onnx file, use dims=1 to fill initializer data
881
        if(dims.empty())
Khalique's avatar
Khalique committed
882
883
884
        {
            dims = {1};
        }
885
886
        if(t.has_raw_data())
        {
wsttiger's avatar
wsttiger committed
887
            const std::string& s = t.raw_data();
Scott Thornton's avatar
Scott Thornton committed
888
889
890
891
892
893
894
895
896
897
898
899
            switch(t.data_type())
            {
            case onnx::TensorProto::UNDEFINED: throw std::runtime_error("");
            case onnx::TensorProto::FLOAT: return literal{{shape::float_type, dims}, s.data()};
            case onnx::TensorProto::UINT8: throw std::runtime_error("");
            case onnx::TensorProto::INT8: return literal{{shape::int32_type, dims}, s.data()};
            case onnx::TensorProto::UINT16: return literal{{shape::int32_type, dims}, s.data()};
            case onnx::TensorProto::INT16: return literal{{shape::int32_type, dims}, s.data()};
            case onnx::TensorProto::INT32: return literal{{shape::int32_type, dims}, s.data()};
            case onnx::TensorProto::INT64: return literal{{shape::int64_type, dims}, s.data()};
            case onnx::TensorProto::STRING: throw std::runtime_error("");
            case onnx::TensorProto::BOOL: return literal{{shape::int32_type, dims}, s.data()};
Paul's avatar
Paul committed
900
            case onnx::TensorProto::FLOAT16: return literal{{shape::half_type, dims}, s.data()};
Scott Thornton's avatar
Scott Thornton committed
901
902
903
904
905
906
            case onnx::TensorProto::DOUBLE: return literal{{shape::double_type, dims}, s.data()};
            case onnx::TensorProto::UINT32: throw std::runtime_error("");
            case onnx::TensorProto::UINT64: throw std::runtime_error("");
            case onnx::TensorProto::COMPLEX64: throw std::runtime_error("");
            case onnx::TensorProto::COMPLEX128: throw std::runtime_error("");
            }
Paul's avatar
Paul committed
907
            MIGRAPHX_THROW("Invalid tensor type");
908
        }
Paul's avatar
Paul committed
909
910
911
912
        switch(t.data_type())
        {
        case onnx::TensorProto::UNDEFINED: throw std::runtime_error("");
        case onnx::TensorProto::FLOAT:
Paul's avatar
Paul committed
913
            return literal{{shape::float_type, dims}, t.float_data().begin(), t.float_data().end()};
Paul's avatar
Paul committed
914
915
        case onnx::TensorProto::UINT8: throw std::runtime_error("");
        case onnx::TensorProto::INT8:
Paul's avatar
Paul committed
916
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
917
        case onnx::TensorProto::UINT16:
Paul's avatar
Paul committed
918
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
919
        case onnx::TensorProto::INT16:
Paul's avatar
Paul committed
920
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
921
        case onnx::TensorProto::INT32:
Paul's avatar
Paul committed
922
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
923
        case onnx::TensorProto::INT64:
Paul's avatar
Paul committed
924
            return literal{{shape::int64_type, dims}, t.int64_data().begin(), t.int64_data().end()};
Paul's avatar
Paul committed
925
926
        case onnx::TensorProto::STRING: throw std::runtime_error("");
        case onnx::TensorProto::BOOL:
Paul's avatar
Paul committed
927
            return literal{{shape::int32_type, dims}, t.int32_data().begin(), t.int32_data().end()};
Paul's avatar
Paul committed
928
929
        case onnx::TensorProto::FLOAT16:
            return literal{{shape::half_type, dims}, t.float_data().begin(), t.float_data().end()};
Paul's avatar
Paul committed
930
931
932
933
934
935
936
937
        case onnx::TensorProto::DOUBLE:
            return literal{
                {shape::double_type, dims}, t.double_data().begin(), t.double_data().end()};
        case onnx::TensorProto::UINT32: throw std::runtime_error("");
        case onnx::TensorProto::UINT64: throw std::runtime_error("");
        case onnx::TensorProto::COMPLEX64: throw std::runtime_error("");
        case onnx::TensorProto::COMPLEX128: throw std::runtime_error("");
        }
Paul's avatar
Paul committed
938
        MIGRAPHX_THROW("Invalid tensor type");
Paul's avatar
Paul committed
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
    }

    static shape parse_type(const onnx::TypeProto& t)
    {
        shape::type_t shape_type{};
        switch(t.tensor_type().elem_type())
        {
        case onnx::TensorProto::UNDEFINED:
            break; // throw std::runtime_error("Unsupported type UNDEFINED");
        case onnx::TensorProto::FLOAT: shape_type = shape::float_type; break;
        case onnx::TensorProto::UINT8:
            break; // throw std::runtime_error("Unsupported type UINT8");
        case onnx::TensorProto::INT8: shape_type = shape::int8_type; break;
        case onnx::TensorProto::UINT16: shape_type = shape::uint16_type; break;
        case onnx::TensorProto::INT16: shape_type = shape::int16_type; break;
        case onnx::TensorProto::INT32: shape_type = shape::int32_type; break;
        case onnx::TensorProto::INT64: shape_type = shape::int64_type; break;
        case onnx::TensorProto::STRING:
            break; // throw std::runtime_error("Unsupported type STRING");
        case onnx::TensorProto::BOOL:
            break; // throw std::runtime_error("Unsupported type BOOL");
Paul's avatar
Paul committed
960
        case onnx::TensorProto::FLOAT16: shape_type = shape::half_type; break;
Paul's avatar
Paul committed
961
962
963
964
965
966
967
968
969
        case onnx::TensorProto::DOUBLE: shape_type = shape::double_type; break;
        case onnx::TensorProto::UINT32: shape_type = shape::uint32_type; break;
        case onnx::TensorProto::UINT64: shape_type = shape::uint64_type; break;
        case onnx::TensorProto::COMPLEX64:
            break; // throw std::runtime_error("Unsupported type COMPLEX64");
        case onnx::TensorProto::COMPLEX128:
            break; // throw std::runtime_error("Unsupported type COMPLEX128");
        }
        std::vector<std::size_t> dims;
Paul's avatar
Paul committed
970
        auto&& tensor_dims = t.tensor_type().shape().dim();
971
972
973
974
975
976
977
978
979
980
981
        std::transform(tensor_dims.begin(),
                       tensor_dims.end(),
                       std::back_inserter(dims),
                       [](auto&& d) -> std::size_t {
                           if(not d.has_dim_value())
                           {
                               long default_batch_size = 1; // FIXME
                               return default_batch_size;
                           }
                           return d.dim_value();
                       });
Paul's avatar
Paul committed
982
983
        return {shape_type, dims};
    }
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005

    shape::type_t get_type(int dtype)
    {
        switch(dtype)
        {
        case 1: return shape::float_type;
        case 2: return shape::uint8_type;
        case 3: return shape::int8_type;
        case 4: return shape::uint16_type;
        case 5: return shape::int16_type;
        case 6: return shape::int32_type;
        case 7: return shape::int64_type;
        case 10: return shape::half_type;
        case 11: return shape::double_type;
        case 12: return shape::uint32_type;
        case 13: return shape::uint64_type;
        default:
        {
            MIGRAPHX_THROW("Prototensor data type " + std::to_string(dtype) + " not supported");
        }
        }
    }
Paul's avatar
Paul committed
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
};

program parse_onnx(const std::string& name)
{
    std::fstream input(name.c_str(), std::ios::in | std::ios::binary);
    onnx_parser parser;
#ifndef NDEBUG
    // Log the program when it can't be parsed
    try
    {
        parser.parse_from(input);
    }
    catch(...)
    {
        std::cerr << parser.prog << std::endl;
        throw;
    }
#else
    parser.parse_from(input);
#endif
    return std::move(parser.prog);
}

Paul's avatar
Paul committed
1029
} // namespace MIGRAPHX_INLINE_NS
Paul's avatar
Paul committed
1030
} // namespace migraphx