-
Notifications
You must be signed in to change notification settings - Fork 550
Expand file tree
/
Copy pathbinary.cpp
More file actions
697 lines (607 loc) · 25.1 KB
/
binary.cpp
File metadata and controls
697 lines (607 loc) · 25.1 KB
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
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
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
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
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
/*******************************************************
* Copyright (c) 2014, ArrayFire
* All rights reserved.
*
* This file is distributed under 3-clause BSD license.
* The complete license agreement can be obtained at:
* http://arrayfire.com/licenses/BSD-3-Clause
********************************************************/
#include <backend.hpp>
#include <common/ArrayInfo.hpp>
#include <common/err_common.hpp>
#include <common/moddims.hpp>
#include <common/tile.hpp>
#include <handle.hpp>
#include <implicit.hpp>
#include <optypes.hpp>
#include <sparse.hpp>
#include <sparse_handle.hpp>
#include <af/arith.h>
#include <af/array.h>
#include <af/data.h>
#include <af/defines.h>
#include <arith.hpp>
#include <logic.hpp>
#include <sparse_arith.hpp>
#include <common/half.hpp>
using af::dim4;
using af::dtype;
using arrayfire::castSparse;
using arrayfire::getSparseArray;
using arrayfire::getSparseArrayBase;
using arrayfire::common::half;
using arrayfire::common::modDims;
using arrayfire::common::SparseArrayBase;
using arrayfire::common::tile;
using detail::arithOp;
using detail::arithOpD;
using detail::Array;
using detail::cdouble;
using detail::cfloat;
using detail::intl;
using detail::schar;
using detail::uchar;
using detail::uint;
using detail::uintl;
using detail::ushort;
template<typename T, af_op_t op>
static inline af_array arithOp(const af_array lhs, const af_array rhs,
const dim4 &odims) {
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
dtype type = static_cast<af::dtype>(af::dtype_traits<T>::af_type);
const detail::Array<T> &l =
linfo.getType() == type ? getArray<T>(lhs) : castArray<T>(lhs);
const detail::Array<T> &r =
rinfo.getType() == type ? getArray<T>(rhs) : castArray<T>(rhs);
return getHandle(arithOp<T, op>(l, r, odims));
}
template<typename T, af_op_t op>
static inline af_array arithOpBroadcast(const af_array lhs,
const af_array rhs) {
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
dim4 odims(1), ltile(1), rtile(1);
dim4 lshape = linfo.dims();
dim4 rshape = rinfo.dims();
for (int d = 0; d < AF_MAX_DIMS; ++d) {
DIM_ASSERT(
1, ((lshape[d] == rshape[d]) || (lshape[d] == 1 && rshape[d] > 1) ||
(lshape[d] > 1 && rshape[d] == 1)));
odims[d] = std::max(lshape[d], rshape[d]);
if (lshape[d] == rshape[d]) {
ltile[d] = rtile[d] = 1;
} else if (lshape[d] == 1 && rshape[d] > 1) {
ltile[d] = odims[d];
} else if (lshape[d] > 1 && rshape[d] == 1) {
rtile[d] = odims[d];
}
}
Array<T> lhst =
arrayfire::common::tile<T>(modDims(getArray<T>(lhs), lshape), ltile);
Array<T> rhst =
arrayfire::common::tile<T>(modDims(getArray<T>(rhs), rshape), rtile);
return getHandle(arithOp<T, op>(lhst, rhst, odims));
}
template<typename T, af_op_t op>
static inline af_array sparseArithOp(const af_array lhs, const af_array rhs) {
auto res = arithOp<T, op>(getSparseArray<T>(lhs), getSparseArray<T>(rhs));
return getHandle(res);
}
template<typename T, af_op_t op>
static inline af_array arithSparseDenseOp(const af_array lhs,
const af_array rhs,
const bool reverse) {
if (op == af_add_t || op == af_sub_t) {
return getHandle(
arithOpD<T, op>(castSparse<T>(lhs), castArray<T>(rhs), reverse));
}
if (op == af_mul_t || op == af_div_t) {
return getHandle(
arithOp<T, op>(castSparse<T>(lhs), castArray<T>(rhs), reverse));
}
}
template<af_op_t op>
static af_err af_arith(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
const af_dtype otype = implicit(linfo.getType(), rinfo.getType());
af_array res;
if (batchMode || linfo.dims() == rinfo.dims()) {
dim4 odims = getOutDims(linfo.dims(), rinfo.dims(), batchMode);
if (odims.ndims() == 0) {
return af_create_handle(out, 0, nullptr, otype);
}
switch (otype) {
case f32: res = arithOp<float, op>(lhs, rhs, odims); break;
case f64: res = arithOp<double, op>(lhs, rhs, odims); break;
case c32: res = arithOp<cfloat, op>(lhs, rhs, odims); break;
case c64: res = arithOp<cdouble, op>(lhs, rhs, odims); break;
case s32: res = arithOp<int, op>(lhs, rhs, odims); break;
case u32: res = arithOp<uint, op>(lhs, rhs, odims); break;
case s8: res = arithOp<schar, op>(lhs, rhs, odims); break;
case u8: res = arithOp<uchar, op>(lhs, rhs, odims); break;
case b8: res = arithOp<char, op>(lhs, rhs, odims); break;
case s64: res = arithOp<intl, op>(lhs, rhs, odims); break;
case u64: res = arithOp<uintl, op>(lhs, rhs, odims); break;
case s16: res = arithOp<short, op>(lhs, rhs, odims); break;
case u16: res = arithOp<ushort, op>(lhs, rhs, odims); break;
case f16: res = arithOp<half, op>(lhs, rhs, odims); break;
default: TYPE_ERROR(0, otype);
}
} else {
if (linfo.ndims() == 0 && rinfo.ndims() == 0) {
return af_create_handle(out, 0, nullptr, otype);
}
switch (otype) {
case f32: res = arithOpBroadcast<float, op>(lhs, rhs); break;
case f64: res = arithOpBroadcast<double, op>(lhs, rhs); break;
case c32: res = arithOpBroadcast<cfloat, op>(lhs, rhs); break;
case c64: res = arithOpBroadcast<cdouble, op>(lhs, rhs); break;
case s32: res = arithOpBroadcast<int, op>(lhs, rhs); break;
case u32: res = arithOpBroadcast<uint, op>(lhs, rhs); break;
case s8: res = arithOpBroadcast<schar, op>(lhs, rhs); break;
case u8: res = arithOpBroadcast<uchar, op>(lhs, rhs); break;
case b8: res = arithOpBroadcast<char, op>(lhs, rhs); break;
case s64: res = arithOpBroadcast<intl, op>(lhs, rhs); break;
case u64: res = arithOpBroadcast<uintl, op>(lhs, rhs); break;
case s16: res = arithOpBroadcast<short, op>(lhs, rhs); break;
case u16: res = arithOpBroadcast<ushort, op>(lhs, rhs); break;
case f16: res = arithOpBroadcast<half, op>(lhs, rhs); break;
default: TYPE_ERROR(0, otype);
}
}
std::swap(*out, res);
}
CATCHALL;
return AF_SUCCESS;
}
template<af_op_t op>
static af_err af_arith_real(af_array *out, const af_array lhs,
const af_array rhs, const bool batchMode) {
try {
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
dim4 odims = getOutDims(linfo.dims(), rinfo.dims(), batchMode);
const af_dtype otype = implicit(linfo.getType(), rinfo.getType());
if (odims.ndims() == 0) {
return af_create_handle(out, 0, nullptr, otype);
}
af_array res;
switch (otype) {
case f32: res = arithOp<float, op>(lhs, rhs, odims); break;
case f64: res = arithOp<double, op>(lhs, rhs, odims); break;
case s32: res = arithOp<int, op>(lhs, rhs, odims); break;
case u32: res = arithOp<uint, op>(lhs, rhs, odims); break;
case s8: res = arithOp<schar, op>(lhs, rhs, odims); break;
case u8: res = arithOp<uchar, op>(lhs, rhs, odims); break;
case b8: res = arithOp<char, op>(lhs, rhs, odims); break;
case s64: res = arithOp<intl, op>(lhs, rhs, odims); break;
case u64: res = arithOp<uintl, op>(lhs, rhs, odims); break;
case s16: res = arithOp<short, op>(lhs, rhs, odims); break;
case u16: res = arithOp<ushort, op>(lhs, rhs, odims); break;
case f16: res = arithOp<half, op>(lhs, rhs, odims); break;
default: TYPE_ERROR(0, otype);
}
std::swap(*out, res);
}
CATCHALL;
return AF_SUCCESS;
}
template<af_op_t op>
static af_err af_arith_sparse(af_array *out, const af_array lhs,
const af_array rhs) {
try {
const SparseArrayBase linfo = getSparseArrayBase(lhs);
const SparseArrayBase rinfo = getSparseArrayBase(rhs);
ARG_ASSERT(1, (linfo.getStorage() == rinfo.getStorage()));
ARG_ASSERT(1, (linfo.dims() == rinfo.dims()));
ARG_ASSERT(1, (linfo.getStorage() == AF_STORAGE_CSR));
const af_dtype otype = implicit(linfo.getType(), rinfo.getType());
af_array res;
switch (otype) {
case f32: res = sparseArithOp<float, op>(lhs, rhs); break;
case f64: res = sparseArithOp<double, op>(lhs, rhs); break;
case c32: res = sparseArithOp<cfloat, op>(lhs, rhs); break;
case c64: res = sparseArithOp<cdouble, op>(lhs, rhs); break;
default: TYPE_ERROR(0, otype);
}
std::swap(*out, res);
}
CATCHALL;
return AF_SUCCESS;
}
template<af_op_t op>
static af_err af_arith_sparse_dense(af_array *out, const af_array lhs,
const af_array rhs,
const bool reverse = false) {
try {
const SparseArrayBase linfo = getSparseArrayBase(lhs);
if (linfo.ndims() > 2) {
AF_ERROR(
"Sparse-Dense arithmetic operations cannot be used in batch "
"mode",
AF_ERR_BATCH);
}
const ArrayInfo &rinfo = getInfo(rhs);
const af_dtype otype = implicit(linfo.getType(), rinfo.getType());
af_array res;
switch (otype) {
case f32:
res = arithSparseDenseOp<float, op>(lhs, rhs, reverse);
break;
case f64:
res = arithSparseDenseOp<double, op>(lhs, rhs, reverse);
break;
case c32:
res = arithSparseDenseOp<cfloat, op>(lhs, rhs, reverse);
break;
case c64:
res = arithSparseDenseOp<cdouble, op>(lhs, rhs, reverse);
break;
default: TYPE_ERROR(0, otype);
}
std::swap(*out, res);
}
CATCHALL;
return AF_SUCCESS;
}
af_err af_add(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
// Check if inputs are sparse
const ArrayInfo &linfo = getInfo(lhs, false);
const ArrayInfo &rinfo = getInfo(rhs, false);
if (linfo.isSparse() && rinfo.isSparse()) {
return af_arith_sparse<af_add_t>(out, lhs, rhs);
}
if (linfo.isSparse() && !rinfo.isSparse()) {
return af_arith_sparse_dense<af_add_t>(out, lhs, rhs);
}
if (!linfo.isSparse() && rinfo.isSparse()) {
// second operand(Array) of af_arith call should be dense
return af_arith_sparse_dense<af_add_t>(out, rhs, lhs, true);
}
return af_arith<af_add_t>(out, lhs, rhs, batchMode);
}
CATCHALL;
}
af_err af_mul(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
// Check if inputs are sparse
const ArrayInfo &linfo = getInfo(lhs, false);
const ArrayInfo &rinfo = getInfo(rhs, false);
if (linfo.isSparse() && rinfo.isSparse()) {
// return af_arith_sparse<af_mul_t>(out, lhs, rhs);
// MKL doesn't have mul or div support yet, hence
// this is commented out although alternative cpu code exists
return AF_ERR_NOT_SUPPORTED;
}
if (linfo.isSparse() && !rinfo.isSparse()) {
return af_arith_sparse_dense<af_mul_t>(out, lhs, rhs);
}
if (!linfo.isSparse() && rinfo.isSparse()) {
return af_arith_sparse_dense<af_mul_t>(
out, rhs, lhs,
true); // dense should be rhs
}
return af_arith<af_mul_t>(out, lhs, rhs, batchMode);
}
CATCHALL;
}
af_err af_sub(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
// Check if inputs are sparse
const ArrayInfo &linfo = getInfo(lhs, false);
const ArrayInfo &rinfo = getInfo(rhs, false);
if (linfo.isSparse() && rinfo.isSparse()) {
return af_arith_sparse<af_sub_t>(out, lhs, rhs);
}
if (linfo.isSparse() && !rinfo.isSparse()) {
return af_arith_sparse_dense<af_sub_t>(out, lhs, rhs);
}
if (!linfo.isSparse() && rinfo.isSparse()) {
return af_arith_sparse_dense<af_sub_t>(
out, rhs, lhs,
true); // dense should be rhs
}
return af_arith<af_sub_t>(out, lhs, rhs, batchMode);
}
CATCHALL;
}
af_err af_div(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
// Check if inputs are sparse
const ArrayInfo &linfo = getInfo(lhs, false);
const ArrayInfo &rinfo = getInfo(rhs, false);
if (linfo.isSparse() && rinfo.isSparse()) {
// return af_arith_sparse<af_div_t>(out, lhs, rhs);
// MKL doesn't have mul or div support yet, hence
// this is commented out although alternative cpu code exists
return AF_ERR_NOT_SUPPORTED;
}
if (linfo.isSparse() && !rinfo.isSparse()) {
return af_arith_sparse_dense<af_div_t>(out, lhs, rhs);
}
if (!linfo.isSparse() && rinfo.isSparse()) {
// Division by sparse is currently not allowed - for convinence of
// dealing with division by 0
// return af_arith_sparse_dense<af_div_t>(out, rhs, lhs, true); //
// dense should be rhs
return AF_ERR_NOT_SUPPORTED;
}
return af_arith<af_div_t>(out, lhs, rhs, batchMode);
}
CATCHALL;
}
af_err af_maxof(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_arith<af_max_t>(out, lhs, rhs, batchMode);
}
af_err af_minof(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_arith<af_min_t>(out, lhs, rhs, batchMode);
}
af_err af_rem(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_arith_real<af_rem_t>(out, lhs, rhs, batchMode);
}
af_err af_mod(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_arith_real<af_mod_t>(out, lhs, rhs, batchMode);
}
af_err af_pow(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
if (rinfo.isComplex()) {
af_array log_lhs, log_res;
af_array res;
AF_CHECK(af_log(&log_lhs, lhs));
AF_CHECK(af_mul(&log_res, log_lhs, rhs, batchMode));
AF_CHECK(af_exp(&res, log_res));
AF_CHECK(af_release_array(log_lhs));
AF_CHECK(af_release_array(log_res));
std::swap(*out, res);
return AF_SUCCESS;
}
if (linfo.isComplex()) {
af_array mag, angle;
af_array mag_res, angle_res;
af_array real_res, imag_res, cplx_res;
af_array res;
AF_CHECK(af_abs(&mag, lhs));
AF_CHECK(af_arg(&angle, lhs));
AF_CHECK(af_pow(&mag_res, mag, rhs, batchMode));
AF_CHECK(af_mul(&angle_res, angle, rhs, batchMode));
AF_CHECK(af_cos(&real_res, angle_res));
AF_CHECK(af_sin(&imag_res, angle_res));
AF_CHECK(af_cplx2(&cplx_res, real_res, imag_res, batchMode));
AF_CHECK(af_mul(&res, mag_res, cplx_res, batchMode));
AF_CHECK(af_release_array(mag));
AF_CHECK(af_release_array(angle));
AF_CHECK(af_release_array(mag_res));
AF_CHECK(af_release_array(angle_res));
AF_CHECK(af_release_array(real_res));
AF_CHECK(af_release_array(imag_res));
AF_CHECK(af_release_array(cplx_res));
std::swap(*out, res);
return AF_SUCCESS;
}
}
CATCHALL;
return af_arith_real<af_pow_t>(out, lhs, rhs, batchMode);
}
af_err af_root(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
if (linfo.isComplex() || rinfo.isComplex()) {
af_array log_lhs, log_res;
af_array res;
AF_CHECK(af_log(&log_lhs, lhs));
AF_CHECK(af_div(&log_res, log_lhs, rhs, batchMode));
AF_CHECK(af_exp(&res, log_res));
std::swap(*out, res);
return AF_SUCCESS;
}
af_array one;
AF_CHECK(af_constant(&one, 1, linfo.ndims(), linfo.dims().get(),
linfo.getType()));
af_array inv_lhs;
AF_CHECK(af_div(&inv_lhs, one, lhs, batchMode));
AF_CHECK(af_arith_real<af_pow_t>(out, rhs, inv_lhs, batchMode));
AF_CHECK(af_release_array(one));
AF_CHECK(af_release_array(inv_lhs));
}
CATCHALL;
return AF_SUCCESS;
}
af_err af_atan2(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
const af_dtype type = implicit(lhs, rhs);
if (type != f16 && type != f32 && type != f64) {
AF_ERROR("Only floating point arrays are supported for atan2 ",
AF_ERR_NOT_SUPPORTED);
}
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
dim4 odims = getOutDims(linfo.dims(), rinfo.dims(), batchMode);
if (odims.ndims() == 0) {
return af_create_handle(out, 0, nullptr, type);
}
af_array res;
switch (type) {
case f16: res = arithOp<half, af_atan2_t>(lhs, rhs, odims); break;
case f32: res = arithOp<float, af_atan2_t>(lhs, rhs, odims); break;
case f64: res = arithOp<double, af_atan2_t>(lhs, rhs, odims); break;
default: TYPE_ERROR(0, type);
}
std::swap(*out, res);
}
CATCHALL;
return AF_SUCCESS;
}
af_err af_hypot(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
const af_dtype type = implicit(lhs, rhs);
if (type != f16 && type != f32 && type != f64) {
AF_ERROR("Only floating point arrays are supported for hypot ",
AF_ERR_NOT_SUPPORTED);
}
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
dim4 odims = getOutDims(linfo.dims(), rinfo.dims(), batchMode);
if (odims.ndims() == 0) {
return af_create_handle(out, 0, nullptr, type);
}
af_array res;
switch (type) {
case f16: res = arithOp<half, af_hypot_t>(lhs, rhs, odims); break;
case f32: res = arithOp<float, af_hypot_t>(lhs, rhs, odims); break;
case f64: res = arithOp<double, af_hypot_t>(lhs, rhs, odims); break;
default: TYPE_ERROR(0, type);
}
std::swap(*out, res);
}
CATCHALL;
return AF_SUCCESS;
}
template<typename T, af_op_t op>
static inline af_array logicOp(const af_array lhs, const af_array rhs,
const dim4 &odims) {
af_array res =
getHandle(logicOp<T, op>(castArray<T>(lhs), castArray<T>(rhs), odims));
return res;
}
template<af_op_t op>
static af_err af_logic(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
const af_dtype type = implicit(lhs, rhs);
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
dim4 odims = getOutDims(linfo.dims(), rinfo.dims(), batchMode);
if (odims.ndims() == 0) {
return af_create_handle(out, 0, nullptr, type);
}
af_array res;
switch (type) {
case f32: res = logicOp<float, op>(lhs, rhs, odims); break;
case f64: res = logicOp<double, op>(lhs, rhs, odims); break;
case c32: res = logicOp<cfloat, op>(lhs, rhs, odims); break;
case c64: res = logicOp<cdouble, op>(lhs, rhs, odims); break;
case s32: res = logicOp<int, op>(lhs, rhs, odims); break;
case u32: res = logicOp<uint, op>(lhs, rhs, odims); break;
case s8: res = logicOp<schar, op>(lhs, rhs, odims); break;
case u8: res = logicOp<uchar, op>(lhs, rhs, odims); break;
case b8: res = logicOp<char, op>(lhs, rhs, odims); break;
case s64: res = logicOp<intl, op>(lhs, rhs, odims); break;
case u64: res = logicOp<uintl, op>(lhs, rhs, odims); break;
case s16: res = logicOp<short, op>(lhs, rhs, odims); break;
case u16: res = logicOp<ushort, op>(lhs, rhs, odims); break;
case f16: res = logicOp<half, op>(lhs, rhs, odims); break;
default: TYPE_ERROR(0, type);
}
std::swap(*out, res);
}
CATCHALL;
return AF_SUCCESS;
}
af_err af_eq(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_logic<af_eq_t>(out, lhs, rhs, batchMode);
}
af_err af_neq(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_logic<af_neq_t>(out, lhs, rhs, batchMode);
}
af_err af_gt(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_logic<af_gt_t>(out, lhs, rhs, batchMode);
}
af_err af_ge(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_logic<af_ge_t>(out, lhs, rhs, batchMode);
}
af_err af_lt(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_logic<af_lt_t>(out, lhs, rhs, batchMode);
}
af_err af_le(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_logic<af_le_t>(out, lhs, rhs, batchMode);
}
af_err af_and(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_logic<af_and_t>(out, lhs, rhs, batchMode);
}
af_err af_or(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_logic<af_or_t>(out, lhs, rhs, batchMode);
}
template<typename T, af_op_t op>
static inline af_array bitOp(const af_array lhs, const af_array rhs,
const dim4 &odims) {
af_array res =
getHandle(bitOp<T, op>(castArray<T>(lhs), castArray<T>(rhs), odims));
return res;
}
template<af_op_t op>
static af_err af_bitwise(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
try {
const af_dtype type = implicit(lhs, rhs);
const ArrayInfo &linfo = getInfo(lhs);
const ArrayInfo &rinfo = getInfo(rhs);
dim4 odims = getOutDims(linfo.dims(), rinfo.dims(), batchMode);
if (odims.ndims() == 0) {
return af_create_handle(out, 0, nullptr, type);
}
af_array res;
switch (type) {
case s32: res = bitOp<int, op>(lhs, rhs, odims); break;
case u32: res = bitOp<uint, op>(lhs, rhs, odims); break;
case s8: res = bitOp<schar, op>(lhs, rhs, odims); break;
case u8: res = bitOp<uchar, op>(lhs, rhs, odims); break;
case b8: res = bitOp<char, op>(lhs, rhs, odims); break;
case s64: res = bitOp<intl, op>(lhs, rhs, odims); break;
case u64: res = bitOp<uintl, op>(lhs, rhs, odims); break;
case s16: res = bitOp<short, op>(lhs, rhs, odims); break;
case u16: res = bitOp<ushort, op>(lhs, rhs, odims); break;
default: TYPE_ERROR(0, type);
}
std::swap(*out, res);
}
CATCHALL;
return AF_SUCCESS;
}
af_err af_bitand(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_bitwise<af_bitand_t>(out, lhs, rhs, batchMode);
}
af_err af_bitor(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_bitwise<af_bitor_t>(out, lhs, rhs, batchMode);
}
af_err af_bitxor(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_bitwise<af_bitxor_t>(out, lhs, rhs, batchMode);
}
af_err af_bitshiftl(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_bitwise<af_bitshiftl_t>(out, lhs, rhs, batchMode);
}
af_err af_bitshiftr(af_array *out, const af_array lhs, const af_array rhs,
const bool batchMode) {
return af_bitwise<af_bitshiftr_t>(out, lhs, rhs, batchMode);
}