lowering.cpp 29.6 KB
Newer Older
Paul's avatar
Paul committed
1

Paul's avatar
Paul committed
2
3
4
5
6
7
#include <migraphx/cpu/lowering.hpp>
#include <migraphx/instruction.hpp>
#include <migraphx/dfor.hpp>
#include <migraphx/operators.hpp>
#include <migraphx/shape_for_each.hpp>
#include <migraphx/iterator_for.hpp>
Paul's avatar
Paul committed
8
#include <migraphx/par_dfor.hpp>
Paul's avatar
Paul committed
9
#include <migraphx/cpu/gemm.hpp>
Paul's avatar
Paul committed
10
#include <unordered_map>
Paul's avatar
Paul committed
11
#include <utility>
Paul's avatar
Paul committed
12

Paul's avatar
Paul committed
13
namespace migraphx {
Paul's avatar
Paul committed
14
inline namespace MIGRAPHX_INLINE_NS {
Paul's avatar
Paul committed
15
16
17
18
19
20
21
22
namespace cpu {

template <typename T>
T zero(const T&)
{
    return T(0);
}

Khalique's avatar
Khalique committed
23
24
25
26
template <class T>
typename std::conditional_t<std::is_integral<T>{}, std::make_signed<T>, std::enable_if<true, T>>::
    type
    make_signed(T x)
Khalique's avatar
Khalique committed
27
28
29
30
{
    return x;
}

31
32
33
34
//
// cpu implemenataion of batch norm for inference
//
// inputs are:
35
36
37
38
// args[0] -> input data buffer
// args[1] -> mini batch mean
// args[2] -> mini batch variance
// args[3] -> gamma
Aditya Atluri's avatar
Aditya Atluri committed
39
// args[4] -> bias
40
41
42
//
// The equation to compute batch norm for inference is:
//
Aditya Atluri's avatar
Aditya Atluri committed
43
// output[i] = bias + gamma * (input[i] + mean) / sqrt(variance + epsilon)
44
45
46
47
48
//
// the input data format should be nchw
//
struct cpu_batch_norm_inference
{
49
    op::batch_norm_inference op;
50

51
52
    std::string name() const { return "cpu::batch_norm_inference"; }

Paul's avatar
Paul committed
53
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }
54

Paul's avatar
Paul committed
55
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
56
    {
57
58
        argument output{output_shape};

Aditya Atluri's avatar
Aditya Atluri committed
59
60
        double epsilon           = op.epsilon;
        auto input               = args[0];
Paul's avatar
Paul committed
61
62
63
64
        auto arg_gamma           = args[1];
        auto arg_bias            = args[2];
        auto mini_batch_mean     = args[3];
        auto mini_batch_variance = args[4];
65

66
        auto num_batch    = output_shape.lens()[0];
Aditya Atluri's avatar
Aditya Atluri committed
67
68
        auto num_channels = output_shape.lens()[1];
        auto image_height = output_shape.lens()[2];
69
        auto image_width  = output_shape.lens()[3];
Aditya Atluri's avatar
Aditya Atluri committed
70

71
        if(op.bn_mode == op::batch_norm_inference::spatial)
Scott Thornton's avatar
Scott Thornton committed
72
73
74
75
        {
            visit_all(output, input, mini_batch_mean, mini_batch_variance, arg_gamma, arg_bias)(
                [&](auto result, auto buffer, auto mean, auto variance, auto gamma, auto bias) {

Paul's avatar
Paul committed
76
                    par_dfor(num_batch, num_channels, image_height, image_width)(
Scott Thornton's avatar
Scott Thornton committed
77
                        [&](std::size_t n, std::size_t c, std::size_t h, std::size_t w) {
78
79
80
81
                            assert((variance[c] + epsilon) > 0);
                            result(n, c, h, w) = gamma[c] * (buffer(n, c, h, w) - mean[c]) /
                                                     std::sqrt(variance[c] + epsilon) +
                                                 bias[c];
Scott Thornton's avatar
Scott Thornton committed
82
83
                        });
                });
84
85
        }

86
        if(op.bn_mode == op::batch_norm_inference::per_activation)
Scott Thornton's avatar
Scott Thornton committed
87
        {
88
89
90
            visit_all(output, input, mini_batch_mean, mini_batch_mean, arg_gamma, arg_bias)(
                [&](auto result, auto buffer, auto mean, auto variance, auto gamma, auto bias) {

Paul's avatar
Paul committed
91
                    par_dfor(num_batch, num_channels, image_height, image_width)(
92
                        [&](std::size_t n, std::size_t c, std::size_t h, std::size_t w) {
Paul's avatar
Paul committed
93
                            assert((variance(c, h, w) + epsilon) > 0);
Scott Thornton's avatar
Scott Thornton committed
94
95
96
97
98
99
                            result(n, c, h, w) = gamma(c, h, w) *
                                                     (buffer(n, c, h, w) - mean(c, h, w)) /
                                                     std::sqrt(variance(c, h, w) + epsilon) +
                                                 bias(c, h, w);
                        });
                });
100
        }
101
102
103
104
105

        return output;
    }
};

Khalique's avatar
Khalique committed
106
struct cpu_lrn
Khalique's avatar
Khalique committed
107
{
Khalique's avatar
Khalique committed
108
    op::lrn op;
Khalique's avatar
Khalique committed
109

Khalique's avatar
Khalique committed
110
    std::string name() const { return "cpu::lrn"; }
Khalique's avatar
Khalique committed
111
112
113
114
115
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }
    argument compute(context&, shape output_shape, std::vector<argument> args) const
    {
        argument result{output_shape};
        visit_all(result, args[0])([&](auto output, auto input) {
Khalique's avatar
Khalique committed
116
117
118
119
            int n_batch         = output_shape.lens()[0];
            int channels        = output_shape.lens()[1];
            int height          = output_shape.lens()[2];
            int width           = output_shape.lens()[3];
Paul's avatar
Paul committed
120
            float alphaoverarea = op.alpha / float(op.size);
Khalique's avatar
Khalique committed
121
            int radius          = (op.size - 1) / 2;
Khalique's avatar
Khalique committed
122

123
            par_dfor(n_batch, height, width)([&](int b, int h, int w) {
Khalique's avatar
Khalique committed
124
                float scale = 0;
Khalique's avatar
Khalique committed
125
126
                dfor(channels)([&](int c) {
                    auto start = (c - radius) < 0 ? 0 : (c - radius);
Khalique's avatar
Khalique committed
127
                    auto end   = (c + radius) > channels ? channels : (c + radius);
Khalique's avatar
Khalique committed
128
129
                    for(auto k = start; k < end; ++k)
                    {
Khalique's avatar
Khalique committed
130
                        scale += std::pow(input(b, k, h, w), 2);
Khalique's avatar
Khalique committed
131
132
133
                    }
                    scale *= alphaoverarea;
                    scale += op.bias;
Khalique's avatar
Khalique committed
134
                    scale              = std::pow(scale, -op.beta);
Khalique's avatar
Khalique committed
135
136
137
138
139
140
141
142
                    output(b, c, h, w) = input(b, c, h, w) * scale;
                });
            });
        });
        return result;
    }
};

Paul's avatar
Paul committed
143
144
struct cpu_convolution
{
145
    op::convolution op;
Paul's avatar
Paul committed
146
147

    std::string name() const { return "cpu::convolution"; }
Paul's avatar
Paul committed
148
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }
Paul's avatar
Paul committed
149
150
151
152
    argument compute(context&, shape output_shape, std::vector<argument> args) const
    {
        argument result{output_shape};
        visit_all(result, args[0], args[1])([&](auto output, auto input, auto weights) {
Khalique's avatar
Khalique committed
153
            auto in   = input.get_shape().lens();
Khalique's avatar
Khalique committed
154
155
            auto in_h = in[2];
            auto in_w = in[3];
Paul's avatar
Paul committed
156

Khalique's avatar
Khalique committed
157
            auto wei   = weights.get_shape().lens();
Khalique's avatar
Khalique committed
158
159
160
161
            auto wei_n = wei[0];
            auto wei_c = wei[1];
            auto wei_h = wei[2];
            auto wei_w = wei[3];
Paul's avatar
Paul committed
162

Paul's avatar
Paul committed
163
            par_dfor(output_shape.lens()[0],
Paul's avatar
Paul committed
164
165
166
                     output_shape.lens()[1],
                     output_shape.lens()[2],
                     output_shape.lens()[3])(
Paul's avatar
Paul committed
167
                [&](std::size_t o, std::size_t w, std::size_t i, std::size_t j) {
Paul's avatar
Paul committed
168
169
170
                    const auto start_x  = i * op.stride[0] - op.padding[0];
                    const auto start_y  = j * op.stride[1] - op.padding[1];
                    const auto group_id = w / (wei_n / op.group);
Paul's avatar
Paul committed
171
172
173

                    double acc = 0;
                    dfor(wei_c, wei_h, wei_w)([&](std::size_t k, std::size_t x, std::size_t y) {
Paul's avatar
Paul committed
174
175
176
                        const auto in_x  = start_x + x;
                        const auto in_y  = start_y + y;
                        const auto in_ch = group_id * wei_c + k;
Paul's avatar
Paul committed
177
178
                        if(in_x >= 0 && in_x < in_h && in_y >= 0 && in_y < in_w)
                        {
Khalique's avatar
Khalique committed
179
                            acc += input(o, in_ch, in_x, in_y) * weights(w, k, x, y);
Paul's avatar
Paul committed
180
181
182
183
184
185
186
187
188
                        }
                    });
                    output(o, w, i, j) = acc;
                });
        });
        return result;
    }
};

Scott Thornton's avatar
Scott Thornton committed
189
190
struct cpu_im2col
{
191
    op::im2col op;
Scott Thornton's avatar
Scott Thornton committed
192

Scott Thornton's avatar
Scott Thornton committed
193
194
    static std::string name() { return "cpu::im2col"; }
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }
Scott Thornton's avatar
Scott Thornton committed
195

wsttiger's avatar
wsttiger committed
196
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
Scott Thornton's avatar
Scott Thornton committed
197
    {
Scott Thornton's avatar
Scott Thornton committed
198
        argument result{output_shape};
Scott Thornton's avatar
Scott Thornton committed
199
        auto input_shape   = args[0].get_shape();
Scott Thornton's avatar
Scott Thornton committed
200
201
        auto weights_shape = args[1].get_shape();
        visit_all(result, args[0])([&](auto col, auto input) {
Scott Thornton's avatar
Scott Thornton committed
202
203
            const std::size_t& height   = input_shape.lens()[2];
            const std::size_t& width    = input_shape.lens()[3];
Scott Thornton's avatar
Scott Thornton committed
204
205
206
            const std::size_t& channels = weights_shape.lens()[1];
            const std::size_t& kernel_h = weights_shape.lens()[2];
            const std::size_t& kernel_w = weights_shape.lens()[3];
Scott Thornton's avatar
Scott Thornton committed
207
208
            const std::size_t& pad_h    = op.padding[0];
            const std::size_t& pad_w    = op.padding[1];
Scott Thornton's avatar
Scott Thornton committed
209
210
211
            const std::size_t& stride_h = op.stride[0];
            const std::size_t& stride_w = op.stride[1];

Paul's avatar
Paul committed
212
213
            auto kdiv2_h = kernel_h / 2;
            auto kdiv2_w = kernel_w / 2;
Scott Thornton's avatar
Scott Thornton committed
214
            // calculate output sizes
Scott Thornton's avatar
Scott Thornton committed
215
216
            const std::size_t col_height = (height - kernel_h + 2 * pad_h) / stride_h + 1;
            const std::size_t col_width  = (width - kernel_w + 2 * pad_w) / stride_w + 1;
wsttiger's avatar
wsttiger committed
217
            // account for padding for the starting position of the input pixels
Scott Thornton's avatar
Scott Thornton committed
218
            std::size_t iinput = kdiv2_h - pad_h;
wsttiger's avatar
wsttiger committed
219
            // loop over output pixels (ioutput, joutput)
Scott Thornton's avatar
Scott Thornton committed
220
221
222
223
224
225
226
227
            for(std::size_t ioutput = 0; ioutput < col_height; ioutput++, iinput += stride_h)
            {
                std::size_t jinput = kdiv2_w - pad_w;
                for(std::size_t joutput = 0; joutput < col_width; joutput++, jinput += stride_w)
                {
                    // compute linear index for output
                    std::size_t ldx = ioutput * col_width + joutput;
                    std::size_t p   = 0;
wsttiger's avatar
wsttiger committed
228
229
230
                    dfor(channels,
                         kernel_h,
                         kernel_w)([&](std::size_t c, std::size_t koffset, std::size_t loffset) {
Paul's avatar
Paul committed
231
232
                        auto idx    = iinput + koffset - kdiv2_h;
                        auto jdx    = jinput + loffset - kdiv2_w;
wsttiger's avatar
wsttiger committed
233
234
235
236
237
                        col(ldx, p) = ((idx >= 0) && (idx < height) && (jdx >= 0) && (jdx < width))
                                          ? input(0, c, idx, jdx)
                                          : 0;
                        p++;
                    });
Scott Thornton's avatar
Scott Thornton committed
238
239
                }
            }
Scott Thornton's avatar
Scott Thornton committed
240
        });
Scott Thornton's avatar
Scott Thornton committed
241
242
243
244
        return result;
    }
};

Paul's avatar
Paul committed
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
struct max_pool
{
    static std::string name() { return "max"; }
    static double start() { return std::numeric_limits<double>::lowest(); }

    static double apply(double x, double y)
    {
        double m = std::max(x, y);
        return (m);
    }

    static double final(double x, double) { return (x); }
};

struct avg_pool
{
    static std::string name() { return "average"; }
    static double start() { return 0.0; }

    static double apply(double x, double y) { return x + y; }

    static double final(double x, double y) { return x / y; }
};

template <class Op>
struct cpu_pooling
{
272
    op::pooling op;
Paul's avatar
Paul committed
273
274

    std::string name() const { return "cpu::pooling_" + Op::name(); }
Paul's avatar
Paul committed
275
276
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
Paul's avatar
Paul committed
277
278
279
280
281
282
283
    {
        argument result{output_shape};
        visit_all(result, args[0])([&](auto output, auto input) {
            using type = typename decltype(output)::value_type;
            auto in_h  = input.get_shape().lens()[2];
            auto in_w  = input.get_shape().lens()[3];

Paul's avatar
Paul committed
284
            par_dfor(output_shape.lens()[0],
Paul's avatar
Paul committed
285
286
287
                     output_shape.lens()[1],
                     output_shape.lens()[2],
                     output_shape.lens()[3])(
Paul's avatar
Paul committed
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
                [&](std::size_t o, std::size_t w, std::size_t i, std::size_t j) {
                    const int start_x0 = i * op.stride[0] - op.padding[0];
                    const int start_y0 = j * op.stride[1] - op.padding[1];

                    const int hend = std::min(start_x0 + op.lengths[0], in_h);
                    const int wend = std::min(start_y0 + op.lengths[1], in_w);

                    const int start_x = std::max(start_x0, 0);
                    const int start_y = std::max(start_y0, 0);

                    const int w_h       = (hend - start_x);
                    const int w_w       = (wend - start_y);
                    const int pool_size = std::max(w_h * w_w, 1);

                    double acc = Op::start();
                    dfor(w_h, w_w)([&](int x, int y) {
                        const int in_x = start_x + x;
                        const int in_y = start_y + y;
                        if(in_x >= 0 && in_x < in_h && in_y >= 0 && in_y < in_w)
                        {
                            acc = Op::apply(acc, input(o, w, in_x, in_y));
                        }
                    });
                    output(o, w, i, j) = type(Op::final(acc, pool_size));
                });
        });
        return result;
    }
};

struct cpu_contiguous
{
320
    op::contiguous op;
Paul's avatar
Paul committed
321
    std::string name() const { return "cpu::contiguous"; }
Paul's avatar
Paul committed
322
323
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
Paul's avatar
Paul committed
324
    {
Paul's avatar
Paul committed
325
        return op.compute(output_shape, std::move(args));
Paul's avatar
Paul committed
326
327
328
    }
};

Khalique's avatar
Khalique committed
329
struct cpu_pad
330
{
Khalique's avatar
Khalique committed
331
332
    op::pad op;
    std::string name() const { return "cpu::contiguous"; }
333
334
335
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
    {
Khalique's avatar
Khalique committed
336
        assert(output_shape.standard());
337
        argument result{output_shape};
Khalique's avatar
Khalique committed
338
        result.visit([&](auto output) { std::fill(output.begin(), output.end(), op.value); });
Khalique's avatar
Khalique committed
339
340

        visit_all(result, args[0])([&](auto output, auto input) {
341
            shape_for_each(input.get_shape(), [&](const auto& idx) {
Khalique's avatar
Khalique committed
342
                std::vector<std::size_t> new_idx(idx.size());
Khalique's avatar
Khalique committed
343
344
345
346
                std::transform(
                    idx.begin(), idx.end(), op.pads.begin(), new_idx.begin(), [](auto i, auto j) {
                        return i + j;
                    });
Khalique's avatar
Khalique committed
347
                output(new_idx.begin(), new_idx.end()) = input(idx.begin(), idx.end());
348
            });
Khalique's avatar
Khalique committed
349
350
        });

351
352
353
354
355
356
357
358
359
360
361
        return result;
    }
};

struct cpu_concat
{
    op::concat op;
    std::string name() const { return "cpu::concat"; }
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
    {
Paul's avatar
Paul committed
362
        return op.compute(output_shape, std::move(args));
363
364
365
    }
};

Paul's avatar
Paul committed
366
367
struct cpu_gemm
{
Shucai Xiao's avatar
Shucai Xiao committed
368
369
    op::dot op;
    std::string name() const { return "cpu::dot"; }
Shucai Xiao's avatar
Shucai Xiao committed
370
371
    shape compute_shape(const std::vector<shape>& inputs) const
    {
Shucai Xiao's avatar
Shucai Xiao committed
372
373
374
375
376
        if(inputs.size() == 3)
        {
            auto c_shape = inputs.at(2);
            check_shapes{{c_shape}}.not_broadcasted();
        }
Shucai Xiao's avatar
Shucai Xiao committed
377
        return op.compute_shape(inputs);
Shucai Xiao's avatar
Shucai Xiao committed
378
    }
Paul's avatar
Paul committed
379

Paul's avatar
Paul committed
380
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
Paul's avatar
Paul committed
381
382
    {
        argument result{output_shape};
Shucai Xiao's avatar
Shucai Xiao committed
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
        // 3 inputs, it is alpha * A * B + beta * C, then
        // A and B are matrics, and C is broadcastable to A * B
        if(args.size() == 3)
        {
            // no need to consider the value of args[2]
            if(op.beta == 0.0f)
            {
                result.visit([&](auto output) { std::fill(output.begin(), output.end(), 0); });
            }
            else
            {
                visit_all(result, args[2])([&](auto output, auto input) {
                    std::copy(input.begin(), input.end(), output.begin());
                });
            }

            migemm(result, args[0], args[1], op.alpha, op.beta);

            return result;
        }

        // 2 input arguments
        migemm(result, args[0], args[1], op.alpha, 0.0f);

Paul's avatar
Paul committed
407
408
409
410
        return result;
    }
};

411
412
413
414
415
416
417
418
struct cpu_gather
{
    op::gather op;
    std::string name() const { return "cpu::gather"; }
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }

    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
    {
Shucai Xiao's avatar
Shucai Xiao committed
419
        return op.compute(output_shape, std::move(args));
420
421
422
    }
};

Paul's avatar
Paul committed
423
424
425
426
427
428
429
430
431
432
433
434
435
436
struct identity_op
{
    std::string name() const { return "cpu::identity"; }
    auto fcn() const
    {
        return [](auto x) { return x; };
    }
};

struct abs_op
{
    std::string name() const { return "cpu::abs"; }
    auto fcn() const
    {
Khalique's avatar
Khalique committed
437
        return [](auto x) { return std::abs(make_signed(x)); };
Paul's avatar
Paul committed
438
439
440
441
442
443
444
445
446
447
448
449
    }
};

struct exp_op
{
    std::string name() const { return "cpu::exp"; }
    auto fcn() const
    {
        return [](auto x) { return std::exp(x); };
    }
};

Shucai Xiao's avatar
Shucai Xiao committed
450
451
452
453
454
455
456
457
458
struct log_op
{
    std::string name() const { return "cpu::log"; }
    auto fcn() const
    {
        return [](auto x) { return std::log(x); };
    }
};

Paul's avatar
Paul committed
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
struct sin_op
{
    std::string name() const { return "cpu::sin"; }
    auto fcn() const
    {
        return [](auto x) { return std::sin(x); };
    }
};

struct cos_op
{
    std::string name() const { return "cpu::cos"; }
    auto fcn() const
    {
        return [](auto x) { return std::cos(x); };
    }
};

struct tan_op
{
    std::string name() const { return "cpu::tan"; }
    auto fcn() const
    {
        return [](auto x) { return std::tan(x); };
    }
};

struct asin_op
{
    std::string name() const { return "cpu::asin"; }
    auto fcn() const
    {
        return [](auto x) { return std::asin(x); };
    }
};

struct acos_op
{
    std::string name() const { return "cpu::acos"; }
    auto fcn() const
    {
        return [](auto x) { return std::acos(x); };
    }
};

struct atan_op
{
    std::string name() const { return "cpu::atan"; }
    auto fcn() const
    {
        return [](auto x) { return std::atan(x); };
    }
};

513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
struct sinh_op
{
    std::string name() const { return "cpu::sinh"; }
    auto fcn() const
    {
        return [](auto x) { return std::sinh(x); };
    }
};

struct cosh_op
{
    std::string name() const { return "cpu::cosh"; }
    auto fcn() const
    {
        return [](auto x) { return std::cosh(x); };
    }
};

Paul's avatar
Paul committed
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
struct tanh_op
{
    std::string name() const { return "cpu::tanh"; }
    auto fcn() const
    {
        return [](auto x) { return std::tanh(x); };
    }
};

struct sigmoid_op
{
    std::string name() const { return "cpu::sigmoid"; }
    auto fcn() const
    {
        return [](auto x) { return 1.f / (1.f + std::exp(-x)); };
    }
};

struct neg_op
{
    std::string name() const { return "cpu::neg"; }
    auto fcn() const
    {
        return [](auto x) { return -x; };
    }
};

struct relu_op
{
    std::string name() const { return "cpu::relu"; }
    auto fcn() const
    {
Paul's avatar
Paul committed
563
        return [](auto x) { return std::max(decltype(x){0}, x); };
Paul's avatar
Paul committed
564
565
566
    }
};

Khalique's avatar
Khalique committed
567
568
569
570
571
572
573
574
575
576
577
struct leaky_relu_op
{
    op::leaky_relu op;
    std::string name() const { return "cpu::leaky_relu"; }
    auto fcn() const
    {
        auto& a = op.alpha;
        return [a](auto x) { return x > 0 ? x : x * a; };
    }
};

Khalique's avatar
Khalique committed
578
579
580
581
582
583
584
585
586
587
588
struct elu_op
{
    op::elu op;
    std::string name() const { return "cpu::elu"; }
    auto fcn() const
    {
        auto& a = op.alpha;
        return [a](auto x) { return x > 0 ? x : a * std::expm1(x); };
    }
};

Paul's avatar
Paul committed
589
590
591
592
593
template <typename Op>
struct cpu_unary
{
    Op op;
    std::string name() const { return op.name(); }
Shucai Xiao's avatar
Shucai Xiao committed
594
    shape compute_shape(const std::vector<shape>& inputs) const
595
    {
Shucai Xiao's avatar
Shucai Xiao committed
596
597
598
        check_shapes{inputs}.has(1);
        auto s = inputs.at(0);
        if(s.packed())
599
        {
Shucai Xiao's avatar
Shucai Xiao committed
600
            return s;
601
602
603
        }
        else
        {
Shucai Xiao's avatar
Shucai Xiao committed
604
            return {s.type(), s.lens()};
605
606
607
        }
    }

Paul's avatar
Paul committed
608
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
Paul's avatar
Paul committed
609
610
611
612
    {
        argument result{output_shape};
        result.visit([&](auto output) {
            args[0].visit([&](auto input) {
Shucai Xiao's avatar
Shucai Xiao committed
613
                if(input.get_shape().standard())
614
615
616
617
618
619
620
621
622
                {
                    std::transform(input.begin(), input.end(), output.begin(), op.fcn());
                }
                else
                {
                    shape_for_each(output.get_shape(), [&](const auto& idx) {
                        output(idx.begin(), idx.end()) = op.fcn()(input(idx.begin(), idx.end()));
                    });
                }
Paul's avatar
Paul committed
623
624
            });
        });
625

Paul's avatar
Paul committed
626
627
628
629
630
631
632
        return result;
    }
};

struct softmax2d
{
    std::string name() const { return "cpu::softmax2d"; }
Paul's avatar
Paul committed
633
634
    shape compute_shape(const std::vector<shape>& inputs) const { return inputs.front(); }
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
Paul's avatar
Paul committed
635
636
637
638
639
640
641
642
643
644
    {
        argument result{output_shape};
        visit_all(result, args[0])([&](auto output, auto input) {
            using value_type = typename decltype(input)::value_type;
            auto nb          = input.get_shape().lens()[0];
            auto nc          = input.get_shape().lens()[1];
            auto nh          = input.get_shape().lens()[2];
            auto nw          = input.get_shape().lens()[3];
            dfor(nb, nh, nw)([&](std::size_t b, std::size_t i, std::size_t j) {
                value_type cmax = std::numeric_limits<value_type>::lowest();
Paul's avatar
Paul committed
645
                for(std::size_t c = 0; c < nc; c++)
Paul's avatar
Paul committed
646
647
648
                {
                    cmax = std::max(cmax, input(b, c, i, j));
                }
Paul's avatar
Paul committed
649
                for(std::size_t c = 0; c < nc; c++)
Paul's avatar
Paul committed
650
651
652
653
                {
                    output(b, c, i, j) = std::exp(input(b, c, i, j) - cmax);
                }
                value_type sum = value_type(0);
Paul's avatar
Paul committed
654
                for(std::size_t c = 0; c < nc; c++)
Paul's avatar
Paul committed
655
656
657
                {
                    sum += output(b, c, i, j);
                }
Paul's avatar
Paul committed
658
                for(std::size_t c = 0; c < nc; c++)
Paul's avatar
Paul committed
659
660
661
662
663
664
665
666
667
                {
                    output(b, c, i, j) = output(b, c, i, j) / sum;
                }
            });
        });
        return result;
    }
};

Shucai Xiao's avatar
Shucai Xiao committed
668
669
670
671
672
673
struct cpu_logsoftmax
{
    op::logsoftmax op;
    std::string name() const { return "cpu::logsoftmax"; }
    shape compute_shape(const std::vector<shape>& inputs) const { return op.compute_shape(inputs); }

Shucai Xiao's avatar
Shucai Xiao committed
674
    template <typename T>
Shucai Xiao's avatar
Shucai Xiao committed
675
676
    std::size_t compute_batch_index(const T& idx, shape& batch_shape, int axis) const
    {
Shucai Xiao's avatar
Shucai Xiao committed
677
        if(axis == 0)
678
679
680
681
682
683
        {
            return 0;
        }
        else
        {
            std::vector<std::size_t> batch_idx(idx.begin(), idx.begin() + axis);
Shucai Xiao's avatar
Shucai Xiao committed
684
            return batch_shape.index(batch_idx.begin(), batch_idx.end());
685
        }
Shucai Xiao's avatar
Shucai Xiao committed
686
687
688
689
690
691
    }

    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
    {
        argument result{output_shape};
        auto lens = output_shape.lens();
692
        std::vector<std::size_t> batch_lens{};
Shucai Xiao's avatar
Shucai Xiao committed
693
        if(op.axis == 0)
694
695
696
        {
            batch_lens.push_back(1);
        }
Shucai Xiao's avatar
Shucai Xiao committed
697
        else
698
699
700
        {
            batch_lens.insert(batch_lens.begin(), lens.begin(), lens.begin() + op.axis);
        }
Shucai Xiao's avatar
Shucai Xiao committed
701
702
703
        shape batch_shape{migraphx::shape::uint32_type, batch_lens};
        visit_all(result, args[0])([&](auto output, auto input) {
            using value_type = typename decltype(input)::value_type;
Shucai Xiao's avatar
Shucai Xiao committed
704
705
            std::vector<value_type> batch_max(batch_shape.elements(),
                                              std::numeric_limits<value_type>::lowest());
Shucai Xiao's avatar
Shucai Xiao committed
706
            shape_for_each(output_shape, [&](auto idx) {
707
                auto index       = this->compute_batch_index(idx, batch_shape, op.axis);
Shucai Xiao's avatar
Shucai Xiao committed
708
709
710
711
                batch_max[index] = std::max(batch_max[index], input(idx.begin(), idx.end()));
            });

            shape_for_each(output_shape, [&](auto idx) {
Shucai Xiao's avatar
Shucai Xiao committed
712
                auto index = this->compute_batch_index(idx, batch_shape, op.axis);
Shucai Xiao's avatar
Shucai Xiao committed
713
714
715
716
717
                output(idx.begin(), idx.end()) = input(idx.begin(), idx.end()) - batch_max[index];
            });

            std::vector<value_type> batch_sum(batch_shape.elements(), value_type(0));
            shape_for_each(output_shape, [&](auto idx) {
718
                auto index = this->compute_batch_index(idx, batch_shape, op.axis);
Shucai Xiao's avatar
Shucai Xiao committed
719
720
721
                batch_sum[index] += std::exp(output(idx.begin(), idx.end()));
            });

Shucai Xiao's avatar
Shucai Xiao committed
722
            for(std::size_t i = 0; i < batch_sum.size(); ++i)
Shucai Xiao's avatar
Shucai Xiao committed
723
724
725
726
727
            {
                batch_sum[i] = std::log(batch_sum[i]);
            }

            shape_for_each(output_shape, [&](auto idx) {
728
                auto index = this->compute_batch_index(idx, batch_shape, op.axis);
729
                output(idx.begin(), idx.end()) -= batch_sum[index];
Shucai Xiao's avatar
Shucai Xiao committed
730
731
732
733
734
735
736
            });
        });

        return result;
    }
};

Paul's avatar
Paul committed
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
struct add_op
{
    std::string name() const { return "add"; }
    auto fcn() const
    {
        return [](auto x, auto y) { return x + y; };
    }
};

struct sub_op
{
    std::string name() const { return "sub"; }
    auto fcn() const
    {
        return [](auto x, auto y) { return x - y; };
    }
};

struct mul_op
{
    std::string name() const { return "mul"; }
    auto fcn() const
    {
        return [](auto x, auto y) { return x * y; };
    }
};

struct div_op
{
    std::string name() const { return "div"; }
    auto fcn() const
    {
        return [](auto x, auto y) { return x / y; };
    }
};

Khalique's avatar
Khalique committed
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
struct max_op
{
    std::string name() const { return "max"; }
    auto fcn() const
    {
        return [](auto x, auto y) { return std::max(x, y); };
    }
};

struct min_op
{
    std::string name() const { return "min"; }
    auto fcn() const
    {
        return [](auto x, auto y) { return std::min(x, y); };
    }
};

Paul's avatar
Paul committed
791
792
793
794
template <typename Op>
struct cpu_binary
{
    Op op;
Shucai Xiao's avatar
Shucai Xiao committed
795
    std::string name() const { return "cpu::" + op.name(); }
Shucai Xiao's avatar
Shucai Xiao committed
796
    shape compute_shape(const std::vector<shape>& inputs) const
797
    {
Shucai Xiao's avatar
Shucai Xiao committed
798
799
800
801
        check_shapes{inputs}.has(2).same_type().same_dims();
        auto s0 = inputs.at(0);
        auto s1 = inputs.at(1);
        if(s0 == s1 and s0.packed())
802
        {
Shucai Xiao's avatar
Shucai Xiao committed
803
            return s0;
804
805
806
        }
        else
        {
Shucai Xiao's avatar
Shucai Xiao committed
807
            return {s0.type(), s0.lens()};
808
809
        }
    }
Shucai Xiao's avatar
Shucai Xiao committed
810

Paul's avatar
Paul committed
811
    argument compute(context&, const shape& output_shape, std::vector<argument> args) const
Paul's avatar
Paul committed
812
813
814
    {
        argument result{output_shape};
        visit_all(result, args[0], args[1])([&](auto output, auto input1, auto input2) {
815
816
            auto s1 = input1.get_shape();
            auto s2 = input2.get_shape();
Shucai Xiao's avatar
Shucai Xiao committed
817
            if(s1 == s2 and s1.standard())
Paul's avatar
Paul committed
818
819
820
821
822
823
824
825
826
827
828
829
            {
                std::transform(
                    input1.begin(), input1.end(), input2.begin(), output.begin(), op.fcn());
            }
            else
            {
                shape_for_each(output.get_shape(), [&](const auto& idx) {
                    output(idx.begin(), idx.end()) =
                        op.fcn()(input1(idx.begin(), idx.end()), input2(idx.begin(), idx.end()));
                });
            }
        });
830

Paul's avatar
Paul committed
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
        return result;
    }
};

struct cpu_apply
{
    program* prog;
    std::unordered_map<std::string, std::function<void(instruction_ref)>> apply_map{};

    template <class T>
    auto simple_op()
    {
        return [this](instruction_ref ins) { apply_simple_op<T>(ins); };
    }

    template <class T, class Op>
    auto extend_op()
    {
        return [this](instruction_ref ins) { apply_extend_op<T, Op>(ins); };
    }

    void init()
    {
854
855
        apply_map["im2col"]      = extend_op<cpu_im2col, op::im2col>();
        apply_map["convolution"] = extend_op<cpu_convolution, op::convolution>();
856
        apply_map["dot"]         = extend_op<cpu_gemm, op::dot>();
Aditya Atluri's avatar
Aditya Atluri committed
857
        apply_map["batch_norm_inference"] =
858
            extend_op<cpu_batch_norm_inference, op::batch_norm_inference>();
Khalique's avatar
Khalique committed
859
        apply_map["lrn"]        = extend_op<cpu_lrn, op::lrn>();
860
        apply_map["contiguous"] = extend_op<cpu_contiguous, op::contiguous>();
Khalique's avatar
Khalique committed
861
        apply_map["pad"]        = extend_op<cpu_pad, op::pad>();
Scott Thornton's avatar
Scott Thornton committed
862
        apply_map["concat"]     = extend_op<cpu_concat, op::concat>();
863
        apply_map["gather"]     = extend_op<cpu_gather, op::gather>();
Shucai Xiao's avatar
Shucai Xiao committed
864
        apply_map["logsoftmax"] = extend_op<cpu_logsoftmax, op::logsoftmax>();
Khalique's avatar
Khalique committed
865
        apply_map["leaky_relu"] = extend_op<cpu_unary<leaky_relu_op>, op::leaky_relu>();
Khalique's avatar
Khalique committed
866
        apply_map["elu"]        = extend_op<cpu_unary<elu_op>, op::elu>();
wsttiger's avatar
wsttiger committed
867
        apply_map["identity"]   = simple_op<cpu_unary<identity_op>>();
Khalique's avatar
Khalique committed
868
        apply_map["abs"]        = simple_op<cpu_unary<abs_op>>();
869
870
        apply_map["sinh"]       = simple_op<cpu_unary<sinh_op>>();
        apply_map["cosh"]       = simple_op<cpu_unary<cosh_op>>();
wsttiger's avatar
wsttiger committed
871
872
873
        apply_map["tanh"]       = simple_op<cpu_unary<tanh_op>>();
        apply_map["sigmoid"]    = simple_op<cpu_unary<sigmoid_op>>();
        apply_map["exp"]        = simple_op<cpu_unary<exp_op>>();
Shucai Xiao's avatar
Shucai Xiao committed
874
        apply_map["log"]        = simple_op<cpu_unary<log_op>>();
wsttiger's avatar
wsttiger committed
875
876
877
878
        apply_map["neg"]        = simple_op<cpu_unary<neg_op>>();
        apply_map["sin"]        = simple_op<cpu_unary<sin_op>>();
        apply_map["cos"]        = simple_op<cpu_unary<cos_op>>();
        apply_map["tan"]        = simple_op<cpu_unary<tan_op>>();
879
880
881
        apply_map["asin"]       = simple_op<cpu_unary<asin_op>>();
        apply_map["acos"]       = simple_op<cpu_unary<acos_op>>();
        apply_map["atan"]       = simple_op<cpu_unary<atan_op>>();
Khalique's avatar
Khalique committed
882
        apply_map["relu"]       = simple_op<cpu_unary<relu_op>>();
wsttiger's avatar
wsttiger committed
883
884
885
886
        apply_map["add"]        = simple_op<cpu_binary<add_op>>();
        apply_map["sub"]        = simple_op<cpu_binary<sub_op>>();
        apply_map["mul"]        = simple_op<cpu_binary<mul_op>>();
        apply_map["div"]        = simple_op<cpu_binary<div_op>>();
Khalique's avatar
Khalique committed
887
888
        apply_map["max"]        = simple_op<cpu_binary<max_op>>();
        apply_map["min"]        = simple_op<cpu_binary<min_op>>();
Paul's avatar
Paul committed
889
890
891
892
893
894
895
896
897

        apply_map["softmax"] = simple_op<softmax2d>();
    }

    void apply()
    {
        init();
        for(auto it : iterator_for(*prog))
        {
Khalique's avatar
Khalique committed
898
            if(it->name() == "pooling")
Paul's avatar
Paul committed
899
900
901
            {
                apply_pooling(it);
            }
Paul's avatar
Paul committed
902
            else if(apply_map.count(it->name()) > 0)
Paul's avatar
Paul committed
903
            {
Paul's avatar
Paul committed
904
                apply_map.at(it->name())(it);
Paul's avatar
Paul committed
905
906
907
908
909
910
911
            }
        }
    }

    template <class T>
    void apply_simple_op(instruction_ref ins)
    {
Paul's avatar
Paul committed
912
        prog->replace_instruction(ins, T{}, ins->inputs());
Paul's avatar
Paul committed
913
914
915
916
917
    }

    template <class T, class Op>
    void apply_extend_op(instruction_ref ins)
    {
918
        auto&& op = any_cast<Op>(ins->get_operator());
Paul's avatar
Paul committed
919
        prog->replace_instruction(ins, T{op}, ins->inputs());
Paul's avatar
Paul committed
920
921
922
923
    }

    void apply_pooling(instruction_ref ins)
    {
924
        auto&& op = any_cast<op::pooling>(ins->get_operator());
Paul's avatar
Paul committed
925
        if(op.mode == "max")
Paul's avatar
Paul committed
926
            prog->replace_instruction(ins, cpu_pooling<max_pool>{op}, ins->inputs());
Paul's avatar
Paul committed
927
        else if(op.mode == "average")
Paul's avatar
Paul committed
928
            prog->replace_instruction(ins, cpu_pooling<avg_pool>{op}, ins->inputs());
Paul's avatar
Paul committed
929
930
931
    }
};

Shucai Xiao's avatar
Shucai Xiao committed
932
void lowering::apply(program& p) const { cpu_apply{&p}.apply(); }
Paul's avatar
Paul committed
933
934

} // namespace cpu
Paul's avatar
Paul committed
935
} // namespace MIGRAPHX_INLINE_NS
Paul's avatar
Paul committed
936
} // namespace migraphx