1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
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
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
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
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
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
563
564
565
566
|
//@HEADER
// ************************************************************************
//
// Kokkos v. 4.0
// Copyright (2022) National Technology & Engineering
// Solutions of Sandia, LLC (NTESS).
//
// Under the terms of Contract DE-NA0003525 with NTESS,
// the U.S. Government retains certain rights in this software.
//
// Part of Kokkos, under the Apache License v2.0 with LLVM Exceptions.
// See https://kokkos.org/LICENSE for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//@HEADER
#ifndef KOKKOS_TEST_SIMD_MATH_OPS_HPP
#define KOKKOS_TEST_SIMD_MATH_OPS_HPP
#include <Kokkos_SIMD.hpp>
#include <SIMDTesting_Utilities.hpp>
template <class Abi, class Loader, class TernaryOp, class T>
void host_check_math_op_one_loader(TernaryOp ternary_op, std::size_t n,
T const* first_args, T const* second_args,
T const* third_args) {
Loader loader;
using simd_type = Kokkos::Experimental::basic_simd<T, Abi>;
constexpr std::size_t width = simd_type::size();
for (std::size_t i = 0; i < n; i += width) {
std::size_t const nremaining = n - i;
std::size_t const nlanes = Kokkos::min(nremaining, width);
simd_type first_arg;
bool const loaded_first_arg =
loader.host_load(first_args + i, nlanes, first_arg);
simd_type second_arg;
bool const loaded_second_arg =
loader.host_load(second_args + i, nlanes, second_arg);
simd_type third_arg;
bool const loaded_third_arg =
loader.host_load(third_args + i, nlanes, third_arg);
if (!(loaded_first_arg && loaded_second_arg && loaded_third_arg)) continue;
T expected_val[width];
for (std::size_t lane = 0; lane < width; ++lane) {
expected_val[lane] = ternary_op.on_host(
T(first_arg[lane]), T(second_arg[lane]), T(third_arg[lane]));
}
simd_type expected_result;
expected_result.copy_from(expected_val,
Kokkos::Experimental::simd_flag_default);
simd_type const computed_result =
ternary_op.on_host(first_arg, second_arg, third_arg);
host_check_equality(expected_result, computed_result, nlanes);
}
}
template <class Abi, class Loader, class BinaryOp, class T>
void host_check_math_op_one_loader(BinaryOp binary_op, std::size_t n,
T const* first_args, T const* second_args) {
Loader loader;
using simd_type = Kokkos::Experimental::basic_simd<T, Abi>;
constexpr std::size_t width = simd_type::size();
for (std::size_t i = 0; i < n; i += width) {
std::size_t const nremaining = n - i;
std::size_t const nlanes = Kokkos::min(nremaining, width);
if ((std::is_same_v<BinaryOp, divides> ||
std::is_same_v<BinaryOp, divides_eq>)&&nremaining < width)
continue;
simd_type first_arg;
bool const loaded_first_arg =
loader.host_load(first_args + i, nlanes, first_arg);
simd_type second_arg;
bool const loaded_second_arg =
loader.host_load(second_args + i, nlanes, second_arg);
if (!(loaded_first_arg && loaded_second_arg)) continue;
// The second argument of pow being negative and/or non-integer may provoke
// a domain error
if constexpr (std::is_same_v<BinaryOp, pow_op>) {
second_arg = Kokkos::round(Kokkos::abs(second_arg));
}
T expected_val[width];
for (std::size_t lane = 0; lane < width; ++lane) {
expected_val[lane] =
binary_op.on_host(T(first_arg[lane]), T(second_arg[lane]));
}
simd_type expected_result;
expected_result.copy_from(expected_val,
Kokkos::Experimental::simd_flag_default);
simd_type const computed_result = binary_op.on_host(first_arg, second_arg);
host_check_equality(expected_result, computed_result, nlanes);
}
}
template <class Abi, class Loader, class UnaryOp, class T>
void host_check_math_op_one_loader(UnaryOp unary_op, std::size_t n,
T const* args) {
Loader loader;
using simd_type = Kokkos::Experimental::basic_simd<T, Abi>;
constexpr std::size_t width = simd_type::size();
for (std::size_t i = 0; i < n; i += width) {
std::size_t const nremaining = n - i;
std::size_t const nlanes = Kokkos::min(nremaining, width);
simd_type arg;
bool const loaded_arg = loader.host_load(args + i, nlanes, arg);
if (!loaded_arg) continue;
if constexpr (std::is_same_v<UnaryOp, sqrt_op>) {
arg = Kokkos::abs(arg);
}
if constexpr (std::is_same_v<UnaryOp, log_op> ||
std::is_same_v<UnaryOp, log10_op> ||
std::is_same_v<UnaryOp, log2_op> ||
std::is_same_v<UnaryOp, tgamma_op> ||
std::is_same_v<UnaryOp, lgamma_op>) {
arg = Kokkos::abs(arg) + simd_type(0.1);
}
// These functions are defined for -1 < x < 1
if constexpr (std::is_same_v<UnaryOp, asin_op> ||
std::is_same_v<UnaryOp, acos_op> ||
std::is_same_v<UnaryOp, atanh_op>) {
arg /= simd_type(10.1);
}
// acosh is defined for x >= 1
if constexpr (std::is_same_v<UnaryOp, acosh_op>) {
arg = Kokkos::abs(arg) + simd_type(1.0);
}
typename decltype(unary_op.on_host(arg))::value_type expected_val[width];
for (std::size_t lane = 0; lane < width; ++lane) {
expected_val[lane] = unary_op.on_host_serial(T(arg[lane]));
}
decltype(unary_op.on_host(arg)) expected_result;
expected_result.copy_from(expected_val,
Kokkos::Experimental::simd_flag_default);
auto computed_result = unary_op.on_host(arg);
host_check_equality(expected_result, computed_result, nlanes);
}
}
template <class Abi, class Op, class... T>
inline void host_check_math_op_all_loaders(Op op, std::size_t n,
T const*... args) {
host_check_math_op_one_loader<Abi, load_element_aligned>(op, n, args...);
host_check_math_op_one_loader<Abi, load_masked>(op, n, args...);
host_check_math_op_one_loader<Abi, load_as_scalars>(op, n, args...);
host_check_math_op_one_loader<Abi, load_vector_aligned>(op, n, args...);
}
template <typename Abi, typename DataType, size_t n>
inline void host_check_all_math_ops(const DataType (&first_args)[n],
const DataType (&second_args)[n],
const DataType (&third_args)[n]) {
host_check_math_op_all_loaders<Abi>(plus(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(plus_eq(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(minus(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(minus_eq(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(multiplies(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(multiplies_eq(), n, first_args,
second_args);
host_check_math_op_all_loaders<Abi>(divides(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(divides_eq(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(absolutes(), n, first_args);
host_check_math_op_all_loaders<Abi>(floors(), n, first_args);
host_check_math_op_all_loaders<Abi>(ceils(), n, first_args);
host_check_math_op_all_loaders<Abi>(rounds(), n, first_args);
host_check_math_op_all_loaders<Abi>(truncates(), n, first_args);
host_check_math_op_all_loaders<Abi>(minimum(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(maximum(), n, first_args, second_args);
// TODO: Place fallback implementations for all simd integer types
if constexpr (std::is_floating_point_v<DataType>) {
host_check_math_op_all_loaders<Abi>(abs_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(exp_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(exp2_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(log_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(log10_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(log2_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(sqrt_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(cbrt_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(sin_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(cos_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(tan_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(asin_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(acos_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(atan_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(sinh_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(cosh_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(tanh_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(asinh_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(acosh_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(atanh_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(erf_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(erfc_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(tgamma_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(lgamma_op(), n, first_args);
host_check_math_op_all_loaders<Abi>(pow_op(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(hypot_op(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(atan2_op(), n, first_args, second_args);
host_check_math_op_all_loaders<Abi>(copysign_op(), n, first_args,
second_args);
host_check_math_op_all_loaders<Abi>(fma_op(), n, first_args, second_args,
third_args);
host_check_math_op_all_loaders<Abi>(ternary_hypot_op(), n, first_args,
second_args, third_args);
}
}
template <typename Abi, typename DataType>
inline void host_check_abi_size() {
using simd_type = Kokkos::Experimental::basic_simd<DataType, Abi>;
using mask_type = typename simd_type::mask_type;
static_assert(simd_type::size() == mask_type::size());
}
template <typename Abi, typename DataType>
inline void host_check_math_ops() {
if constexpr (is_simd_avail_v<DataType, Abi>) {
constexpr size_t alignment =
Kokkos::Experimental::basic_simd<DataType, Abi>::size() *
sizeof(DataType);
host_check_abi_size<Abi, DataType>();
if constexpr (!std::is_integral_v<DataType>) {
alignas(alignment) DataType const first_args[] = {
0.1, 0.4, 0.5, 0.7, 1.0, 1.5, -2.0, 10.0,
0.0, 1.2, -2.8, 3.0, 4.0, -0.1, 5.0, -0.2};
alignas(alignment) DataType const second_args[] = {
1.0, 0.2, 1.1, 1.8, -0.1, -3.0, -2.4, 1.0,
13.0, -3.2, -2.1, 3.0, -15.0, -0.5, -0.2, -0.2};
alignas(alignment) DataType const third_args[] = {
3.7, 2.6, 9.8, 9.3, 9.9, -5.3, 8.5, 1.6,
-3.8, -0.4, 6.1, -5.3, 6.1, -8.9, -2.5, -5.2};
host_check_all_math_ops<Abi>(first_args, second_args, third_args);
} else {
if constexpr (std::is_signed_v<DataType>) {
alignas(alignment) DataType const first_args[] = {
1, 2, -1, 10, 0, 1, -2, 10, 0, 1, -2, -3, 7, 4, -9, -15};
alignas(alignment) DataType const second_args[] = {
1, 2, 1, 1, 1, -3, -2, 1, 13, -3, -2, 10, -15, 7, 2, -10};
alignas(alignment) DataType const third_args[] = {
20, 3, -18, 3, 19, 11, 4, 20, 8, -8, 13, -18, -2, -5, -1, 11};
host_check_all_math_ops<Abi>(first_args, second_args, third_args);
} else {
alignas(alignment) DataType const first_args[] = {
1, 2, 1, 10, 0, 1, 2, 10, 0, 1, 2, 11, 5, 8, 2, 14};
alignas(alignment) DataType const second_args[] = {
1, 2, 1, 1, 1, 3, 2, 1, 13, 3, 2, 3, 6, 20, 5, 14};
alignas(alignment) DataType const third_args[] = {
10, 10, 6, 6, 16, 14, 18, 9, 19, 7, 0, 6, 2, 15, 10, 16};
host_check_all_math_ops<Abi>(first_args, second_args, third_args);
}
}
}
}
template <typename Abi, typename... DataTypes>
inline void host_check_math_ops_all_types(
Kokkos::Experimental::Impl::data_types<DataTypes...>) {
(host_check_math_ops<Abi, DataTypes>(), ...);
}
template <typename... Abis>
inline void host_check_math_ops_all_abis(
Kokkos::Experimental::Impl::abi_set<Abis...>) {
using DataTypes = Kokkos::Experimental::Impl::data_type_set;
(host_check_math_ops_all_types<Abis>(DataTypes()), ...);
}
template <typename Abi, typename Loader, typename TernaryOp, typename T>
KOKKOS_INLINE_FUNCTION void device_check_math_op_one_loader(
TernaryOp ternary_op, std::size_t n, T const* first_args,
T const* second_args, T const* third_args) {
Loader loader;
using simd_type = Kokkos::Experimental::basic_simd<T, Abi>;
constexpr std::size_t width = simd_type::size();
for (std::size_t i = 0; i < n; i += width) {
std::size_t const nremaining = n - i;
std::size_t const nlanes = Kokkos::min(nremaining, width);
simd_type first_arg;
bool const loaded_first_arg =
loader.device_load(first_args + i, nlanes, first_arg);
simd_type second_arg;
bool const loaded_second_arg =
loader.device_load(second_args + i, nlanes, second_arg);
simd_type third_arg;
bool const loaded_third_arg =
loader.device_load(third_args + i, nlanes, third_arg);
if (!(loaded_first_arg && loaded_second_arg && loaded_third_arg)) continue;
simd_type expected_result(KOKKOS_LAMBDA(std::size_t lane) {
return ternary_op.on_device(first_arg[lane], second_arg[lane],
third_arg[lane]);
});
simd_type const computed_result =
ternary_op.on_device(first_arg, second_arg, third_arg);
device_check_equality(expected_result, computed_result, nlanes);
}
}
template <typename Abi, typename Loader, typename BinaryOp, typename T>
KOKKOS_INLINE_FUNCTION void device_check_math_op_one_loader(
BinaryOp binary_op, std::size_t n, T const* first_args,
T const* second_args) {
Loader loader;
using simd_type = Kokkos::Experimental::basic_simd<T, Abi>;
constexpr std::size_t width = simd_type::size();
for (std::size_t i = 0; i < n; i += width) {
std::size_t const nremaining = n - i;
std::size_t const nlanes = Kokkos::min(nremaining, width);
if ((std::is_same_v<BinaryOp, divides> ||
std::is_same_v<BinaryOp, divides_eq>)&&nremaining < width)
continue;
simd_type first_arg;
bool const loaded_first_arg =
loader.device_load(first_args + i, nlanes, first_arg);
simd_type second_arg;
bool const loaded_second_arg =
loader.device_load(second_args + i, nlanes, second_arg);
if (!(loaded_first_arg && loaded_second_arg)) continue;
// The second argument of pow being negative and/or non-integer may provoke
// a domain error
if constexpr (std::is_same_v<BinaryOp, pow_op>) {
second_arg = Kokkos::round(Kokkos::abs(second_arg));
}
simd_type expected_result(KOKKOS_LAMBDA(std::size_t lane) {
return binary_op.on_device(first_arg[lane], second_arg[lane]);
});
simd_type const computed_result =
binary_op.on_device(first_arg, second_arg);
device_check_equality(expected_result, computed_result, nlanes);
}
}
template <typename Abi, typename Loader, typename UnaryOp, typename T>
KOKKOS_INLINE_FUNCTION void device_check_math_op_one_loader(UnaryOp unary_op,
std::size_t n,
T const* args) {
Loader loader;
using simd_type = Kokkos::Experimental::basic_simd<T, Abi>;
constexpr std::size_t width = simd_type::size();
for (std::size_t i = 0; i < n; i += width) {
std::size_t const nremaining = n - i;
std::size_t const nlanes = Kokkos::min(nremaining, width);
simd_type arg;
bool const loaded_arg = loader.device_load(args + i, nlanes, arg);
if (!loaded_arg) continue;
if constexpr (std::is_same_v<UnaryOp, sqrt_op>) {
arg = Kokkos::abs(arg);
}
if constexpr (std::is_same_v<UnaryOp, log_op> ||
std::is_same_v<UnaryOp, log10_op> ||
std::is_same_v<UnaryOp, log2_op> ||
std::is_same_v<UnaryOp, tgamma_op> ||
std::is_same_v<UnaryOp, lgamma_op>) {
arg = Kokkos::abs(arg) + simd_type(0.1);
}
// These functions are defined for -1 < x < 1
if constexpr (std::is_same_v<UnaryOp, asin_op> ||
std::is_same_v<UnaryOp, acos_op> ||
std::is_same_v<UnaryOp, atanh_op>) {
arg /= simd_type(10.1);
}
// acosh is defined for x >= 1
if constexpr (std::is_same_v<UnaryOp, acosh_op>) {
arg = Kokkos::abs(arg) + simd_type(1.0);
}
auto computed_result = unary_op.on_device(arg);
decltype(computed_result) expected_result(KOKKOS_LAMBDA(std::size_t lane) {
return unary_op.on_device_serial(arg[lane]);
});
device_check_equality(expected_result, computed_result, nlanes);
}
}
template <typename Abi, typename Op, typename... T>
KOKKOS_INLINE_FUNCTION void device_check_math_op_all_loaders(Op op,
std::size_t n,
T const*... args) {
device_check_math_op_one_loader<Abi, load_element_aligned>(op, n, args...);
device_check_math_op_one_loader<Abi, load_masked>(op, n, args...);
device_check_math_op_one_loader<Abi, load_as_scalars>(op, n, args...);
device_check_math_op_one_loader<Abi, load_vector_aligned>(op, n, args...);
}
template <typename Abi, typename DataType, size_t n>
KOKKOS_INLINE_FUNCTION void device_check_all_math_ops(
const DataType (&first_args)[n], const DataType (&second_args)[n],
const DataType (&third_args)[n]) {
device_check_math_op_all_loaders<Abi>(plus(), n, first_args, second_args);
device_check_math_op_all_loaders<Abi>(plus_eq(), n, first_args, second_args);
device_check_math_op_all_loaders<Abi>(minus(), n, first_args, second_args);
device_check_math_op_all_loaders<Abi>(minus_eq(), n, first_args, second_args);
device_check_math_op_all_loaders<Abi>(multiplies(), n, first_args,
second_args);
device_check_math_op_all_loaders<Abi>(multiplies_eq(), n, first_args,
second_args);
device_check_math_op_all_loaders<Abi>(divides(), n, first_args, second_args);
device_check_math_op_all_loaders<Abi>(divides_eq(), n, first_args,
second_args);
device_check_math_op_all_loaders<Abi>(absolutes(), n, first_args);
device_check_math_op_all_loaders<Abi>(floors(), n, first_args);
device_check_math_op_all_loaders<Abi>(ceils(), n, first_args);
device_check_math_op_all_loaders<Abi>(rounds(), n, first_args);
device_check_math_op_all_loaders<Abi>(truncates(), n, first_args);
device_check_math_op_all_loaders<Abi>(minimum(), n, first_args, second_args);
device_check_math_op_all_loaders<Abi>(maximum(), n, first_args, second_args);
if constexpr (std::is_floating_point_v<DataType>) {
device_check_math_op_all_loaders<Abi>(abs_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(exp_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(exp2_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(log_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(log10_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(log2_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(sqrt_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(cbrt_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(sin_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(cos_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(tan_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(asin_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(acos_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(atan_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(sinh_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(cosh_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(tanh_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(asinh_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(acosh_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(atanh_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(erf_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(erfc_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(tgamma_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(lgamma_op(), n, first_args);
device_check_math_op_all_loaders<Abi>(pow_op(), n, first_args, second_args);
device_check_math_op_all_loaders<Abi>(hypot_op(), n, first_args,
second_args);
device_check_math_op_all_loaders<Abi>(atan2_op(), n, first_args,
second_args);
device_check_math_op_all_loaders<Abi>(copysign_op(), n, first_args,
second_args);
device_check_math_op_all_loaders<Abi>(fma_op(), n, first_args, second_args,
third_args);
device_check_math_op_all_loaders<Abi>(ternary_hypot_op(), n, first_args,
second_args, third_args);
}
}
template <typename Abi, typename DataType>
KOKKOS_INLINE_FUNCTION void device_check_abi_size() {
using simd_type = Kokkos::Experimental::basic_simd<DataType, Abi>;
using mask_type = typename simd_type::mask_type;
static_assert(simd_type::size() == mask_type::size());
}
template <typename Abi, typename DataType>
KOKKOS_INLINE_FUNCTION void device_check_math_ops() {
if constexpr (is_type_v<Kokkos::Experimental::basic_simd<DataType, Abi>>) {
device_check_abi_size<Abi, DataType>();
if constexpr (!std::is_integral_v<DataType>) {
DataType const first_args[] = {0.1, 0.4, 0.5, 0.7, 1.0, 1.5,
-2.0, 10.0, 0.0, 1.2, -2.8, 3.0,
4.0, -0.1, 5.0, -0.2};
DataType const second_args[] = {1.0, 0.2, 1.1, 1.8, -0.1, -3.0,
-2.4, 1.0, 13.0, -3.2, -2.1, 3.0,
-15.0, -0.5, -0.2, -0.2};
DataType const third_args[] = {3.7, 2.6, 9.8, 9.3, 9.9, -5.3,
8.5, 1.6, -3.8, -0.4, 6.1, -5.3,
6.1, -8.9, -2.5, -5.2};
device_check_all_math_ops<Abi>(first_args, second_args, third_args);
} else {
if constexpr (std::is_signed_v<DataType>) {
DataType const first_args[] = {1, 2, -1, 10, 0, 1, -2, 10,
0, 1, -2, -3, 7, 4, -9, -15};
DataType const second_args[] = {1, 2, 1, 1, 1, -3, -2, 1,
13, -3, -2, 10, -15, 7, 2, -10};
DataType const third_args[] = {20, 3, -18, 3, 19, 11, 4, 20,
8, -8, 13, -18, -2, -5, -1, 11};
device_check_all_math_ops<Abi>(first_args, second_args, third_args);
} else {
DataType const first_args[] = {1, 2, 1, 10, 0, 1, 2, 10,
0, 1, 2, 11, 5, 8, 2, 14};
DataType const second_args[] = {1, 2, 1, 1, 1, 3, 2, 1,
13, 3, 2, 3, 6, 20, 5, 14};
DataType const third_args[] = {10, 10, 6, 6, 16, 14, 18, 9,
19, 7, 0, 6, 2, 15, 10, 16};
device_check_all_math_ops<Abi>(first_args, second_args, third_args);
}
}
}
}
template <typename Abi, typename... DataTypes>
KOKKOS_INLINE_FUNCTION void device_check_math_ops_all_types(
Kokkos::Experimental::Impl::data_types<DataTypes...>) {
(device_check_math_ops<Abi, DataTypes>(), ...);
}
template <typename... Abis>
KOKKOS_INLINE_FUNCTION void device_check_math_ops_all_abis(
Kokkos::Experimental::Impl::abi_set<Abis...>) {
using DataTypes = Kokkos::Experimental::Impl::data_type_set;
(device_check_math_ops_all_types<Abis>(DataTypes()), ...);
}
class simd_device_math_ops_functor {
public:
KOKKOS_INLINE_FUNCTION void operator()(int) const {
device_check_math_ops_all_abis(
Kokkos::Experimental::Impl::device_abi_set());
}
};
TEST(simd, host_math_ops) {
host_check_math_ops_all_abis(Kokkos::Experimental::Impl::host_abi_set());
}
TEST(simd, device_math_ops) {
#ifdef KOKKOS_ENABLE_OPENMPTARGET // FIXME_OPENMPTARGET
GTEST_SKIP()
<< "skipping because of a non-deterministic failure reporting: "
"Failure to synchronize stream (nil): Error in "
"cuStreamSynchronize: an illegal memory access was encountered";
#endif
#if defined(KOKKOS_ENABLE_OPENACC) && \
defined(KOKKOS_COMPILER_CLANG) // FIXME_CLACC
GTEST_SKIP()
<< "skipping because of a non-deterministic failure reporting: "
"Failure to synchronize stream (nil): Error in "
"cuStreamSynchronize: an illegal memory access was encountered";
#endif
Kokkos::parallel_for(1, simd_device_math_ops_functor());
}
#endif
|
