Repository navigation
Expand file tree
/
Copy pathsimde-common.h
More file actions
970 lines (897 loc) · 36.6 KB
/
Copy pathsimde-common.h
File metadata and controls
970 lines (897 loc) · 36.6 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
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
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
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
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
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
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
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
/* SPDX-License-Identifier: MIT
*
* Permission is hereby granted, free of charge, to any person
* obtaining a copy of this software and associated documentation
* files (the "Software"), to deal in the Software without
* restriction, including without limitation the rights to use, copy,
* modify, merge, publish, distribute, sublicense, and/or sell copies
* of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be
* included in all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* Copyright:
* 2017-2020 Evan Nemerson <evan@nemerson.com>
*/
#if !defined(SIMDE_COMMON_H)
#define SIMDE_COMMON_H
#include "hedley.h"
#define SIMDE_VERSION_MAJOR 0
#define SIMDE_VERSION_MINOR 7
#define SIMDE_VERSION_MICRO 3
#define SIMDE_VERSION HEDLEY_VERSION_ENCODE(SIMDE_VERSION_MAJOR, SIMDE_VERSION_MINOR, SIMDE_VERSION_MICRO)
// Also update meson.build in the root directory of the repository
#include <stddef.h>
#include <stdint.h>
#include "simde-detect-clang.h"
#include "simde-arch.h"
#include "simde-features.h"
#include "simde-diagnostic.h"
#include "simde-math.h"
#include "simde-constify.h"
#include "simde-align.h"
/* In some situations, SIMDe has to make large performance sacrifices
* for small increases in how faithfully it reproduces an API, but
* only a relatively small number of users will actually need the API
* to be completely accurate. The SIMDE_FAST_* options can be used to
* disable these trade-offs.
*
* They can be enabled by passing -DSIMDE_FAST_MATH to the compiler, or
* the individual defines (e.g., -DSIMDE_FAST_NANS) if you only want to
* enable some optimizations. Using -ffast-math and/or
* -ffinite-math-only will also enable the relevant options. If you
* don't want that you can pass -DSIMDE_NO_FAST_* to disable them. */
/* Most programs avoid NaNs by never passing values which can result in
* a NaN; for example, if you only pass non-negative values to the sqrt
* functions, it won't generate a NaN. On some platforms, similar
* functions handle NaNs differently; for example, the _mm_min_ps SSE
* function will return 0.0 if you pass it (0.0, NaN), but the NEON
* vminq_f32 function will return NaN. Making them behave like one
* another is expensive; it requires generating a mask of all lanes
* with NaNs, then performing the operation (e.g., vminq_f32), then
* blending together the result with another vector using the mask.
*
* If you don't want SIMDe to worry about the differences between how
* NaNs are handled on the two platforms, define this (or pass
* -ffinite-math-only) */
#if !defined(SIMDE_FAST_MATH) && !defined(SIMDE_NO_FAST_MATH) && defined(__FAST_MATH__)
#define SIMDE_FAST_MATH
#endif
#if !defined(SIMDE_FAST_NANS) && !defined(SIMDE_NO_FAST_NANS)
#if defined(SIMDE_FAST_MATH)
#define SIMDE_FAST_NANS
#elif defined(__FINITE_MATH_ONLY__)
#if __FINITE_MATH_ONLY__
#define SIMDE_FAST_NANS
#endif
#endif
#endif
/* Many functions are defined as using the current rounding mode
* (i.e., the SIMD version of fegetround()) when converting to
* an integer. For example, _mm_cvtpd_epi32. Unfortunately,
* on some platforms (such as ARMv8+ where round-to-nearest is
* always used, regardless of the FPSCR register) this means we
* have to first query the current rounding mode, then choose
* the proper function (rounnd
, ceil, floor, etc.) */
#if !defined(SIMDE_FAST_ROUND_MODE) && !defined(SIMDE_NO_FAST_ROUND_MODE) && defined(SIMDE_FAST_MATH)
#define SIMDE_FAST_ROUND_MODE
#endif
/* This controls how ties are rounded. For example, does 10.5 round to
* 10 or 11? IEEE 754 specifies round-towards-even, but ARMv7 (for
* example) doesn't support it and it must be emulated (which is rather
* slow). If you're okay with just using the default for whatever arch
* you're on, you should definitely define this.
*
* Note that we don't use this macro to avoid correct implementations
* in functions which are explicitly about rounding (such as vrnd* on
* NEON, _mm_round_* on x86, etc.); it is only used for code where
* rounding is a component in another function, and even then it isn't
* usually a problem since such functions will use the current rounding
* mode. */
#if !defined(SIMDE_FAST_ROUND_TIES) && !defined(SIMDE_NO_FAST_ROUND_TIES) && defined(SIMDE_FAST_MATH)
#define SIMDE_FAST_ROUND_TIES
#endif
/* For functions which convert from one type to another (mostly from
* floating point to integer types), sometimes we need to do a range
* check and potentially return a different result if the value
* falls outside that range. Skipping this check can provide a
* performance boost, at the expense of faithfulness to the API we're
* emulating. */
#if !defined(SIMDE_FAST_CONVERSION_RANGE) && !defined(SIMDE_NO_FAST_CONVERSION_RANGE) && defined(SIMDE_FAST_MATH)
#define SIMDE_FAST_CONVERSION_RANGE
#endif
#if \
HEDLEY_HAS_BUILTIN(__builtin_constant_p) || \
HEDLEY_GCC_VERSION_CHECK(3,4,0) || \
HEDLEY_INTEL_VERSION_CHECK(13,0,0) || \
HEDLEY_TINYC_VERSION_CHECK(0,9,19) || \
HEDLEY_ARM_VERSION_CHECK(4,1,0) || \
HEDLEY_IBM_VERSION_CHECK(13,1,0) || \
HEDLEY_TI_CL6X_VERSION_CHECK(6,1,0) || \
(HEDLEY_SUNPRO_VERSION_CHECK(5,10,0) && !defined(__cplusplus)) || \
HEDLEY_CRAY_VERSION_CHECK(8,1,0) || \
HEDLEY_MCST_LCC_VERSION_CHECK(1,25,10)
#define SIMDE_CHECK_CONSTANT_(expr) (__builtin_constant_p(expr))
#elif defined(__cplusplus) && (__cplusplus > 201703L)
#include <type_traits>
#define SIMDE_CHECK_CONSTANT_(expr) (std::is_constant_evaluated())
#endif
#if !defined(SIMDE_NO_CHECK_IMMEDIATE_CONSTANT)
#if defined(SIMDE_CHECK_CONSTANT_) && \
SIMDE_DETECT_CLANG_VERSION_CHECK(9,0,0) && \
(!defined(__apple_build_version__) || ((__apple_build_version__ < 11000000) || (__apple_build_version__ >= 12000000)))
#define SIMDE_REQUIRE_CONSTANT(arg) HEDLEY_REQUIRE_MSG(SIMDE_CHECK_CONSTANT_(arg), "`" #arg "' must be constant")
#else
#define SIMDE_REQUIRE_CONSTANT(arg)
#endif
#else
#define SIMDE_REQUIRE_CONSTANT(arg)
#endif
#define SIMDE_REQUIRE_RANGE(arg, min, max) \
HEDLEY_REQUIRE_MSG((((arg) >= (min)) && ((arg) <= (max))), "'" #arg "' must be in [" #min ", " #max "]")
#define SIMDE_REQUIRE_CONSTANT_RANGE(arg, min, max) \
SIMDE_REQUIRE_CONSTANT(arg) \
SIMDE_REQUIRE_RANGE(arg, min, max)
/* A copy of HEDLEY_STATIC_ASSERT, except we don't define an empty
* fallback if we can't find an implementation; instead we have to
* check if SIMDE_STATIC_ASSERT is defined before using it. */
#if \
!defined(__cplusplus) && ( \
(defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)) || \
HEDLEY_HAS_FEATURE(c_static_assert) || \
HEDLEY_GCC_VERSION_CHECK(6,0,0) || \
HEDLEY_INTEL_VERSION_CHECK(13,0,0) || \
defined(_Static_assert) \
)
# define SIMDE_STATIC_ASSERT(expr, message) _Static_assert(expr, message)
#elif \
(defined(__cplusplus) && (__cplusplus >= 201103L)) || \
HEDLEY_MSVC_VERSION_CHECK(16,0,0)
# define SIMDE_STATIC_ASSERT(expr, message) HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_(static_assert(expr, message))
#endif
/* Statement exprs */
#if \
HEDLEY_GNUC_VERSION_CHECK(2,95,0) || \
HEDLEY_TINYC_VERSION_CHECK(0,9,26) || \
HEDLEY_INTEL_VERSION_CHECK(9,0,0) || \
HEDLEY_PGI_VERSION_CHECK(18,10,0) || \
HEDLEY_SUNPRO_VERSION_CHECK(5,12,0) || \
HEDLEY_IBM_VERSION_CHECK(11,1,0) || \
HEDLEY_MCST_LCC_VERSION_CHECK(1,25,10)
#define SIMDE_STATEMENT_EXPR_(expr) (__extension__ expr)
#endif
#if defined(SIMDE_CHECK_CONSTANT_) && defined(SIMDE_STATIC_ASSERT)
#define SIMDE_ASSERT_CONSTANT_(v) SIMDE_STATIC_ASSERT(SIMDE_CHECK_CONSTANT_(v), #v " must be constant.")
#endif
#if \
(HEDLEY_HAS_ATTRIBUTE(may_alias) && !defined(HEDLEY_SUNPRO_VERSION)) || \
HEDLEY_GCC_VERSION_CHECK(3,3,0) || \
HEDLEY_INTEL_VERSION_CHECK(13,0,0) || \
HEDLEY_IBM_VERSION_CHECK(13,1,0)
# define SIMDE_MAY_ALIAS __attribute__((__may_alias__))
#else
# define SIMDE_MAY_ALIAS
#endif
/* Lots of compilers support GCC-style vector extensions, but many
don't support all the features. Define different macros depending
on support for
* SIMDE_VECTOR - Declaring a vector.
* SIMDE_VECTOR_OPS - basic operations (binary and unary).
* SIMDE_VECTOR_NEGATE - negating a vector
* SIMDE_VECTOR_SCALAR - For binary operators, the second argument
can be a scalar, in which case the result is as if that scalar
had been broadcast to all lanes of a vector.
* SIMDE_VECTOR_SUBSCRIPT - Supports array subscript notation for
extracting/inserting a single element.=
SIMDE_VECTOR can be assumed if any others are defined, the
others are independent. */
#if !defined(SIMDE_NO_VECTOR)
# if \
HEDLEY_GCC_VERSION_CHECK(4,8,0)
# define SIMDE_VECTOR(size) __attribute__((__vector_size__(size)))
# define SIMDE_VECTOR_OPS
# define SIMDE_VECTOR_NEGATE
# define SIMDE_VECTOR_SCALAR
# define SIMDE_VECTOR_SUBSCRIPT
# elif HEDLEY_INTEL_VERSION_CHECK(16,0,0)
# define SIMDE_VECTOR(size) __attribute__((__vector_size__(size)))
# define SIMDE_VECTOR_OPS
# define SIMDE_VECTOR_NEGATE
/* ICC only supports SIMDE_VECTOR_SCALAR for constants */
# define SIMDE_VECTOR_SUBSCRIPT
# elif \
HEDLEY_GCC_VERSION_CHECK(4,1,0) || \
HEDLEY_INTEL_VERSION_CHECK(13,0,0) || \
HEDLEY_MCST_LCC_VERSION_CHECK(1,25,10)
# define SIMDE_VECTOR(size) __attribute__((__vector_size__(size)))
# define SIMDE_VECTOR_OPS
# elif HEDLEY_SUNPRO_VERSION_CHECK(5,12,0)
# define SIMDE_VECTOR(size) __attribute__((__vector_size__(size)))
# elif HEDLEY_HAS_ATTRIBUTE(vector_size)
# define SIMDE_VECTOR(size) __attribute__((__vector_size__(size)))
# define SIMDE_VECTOR_OPS
# define SIMDE_VECTOR_NEGATE
# define SIMDE_VECTOR_SUBSCRIPT
# if SIMDE_DETECT_CLANG_VERSION_CHECK(5,0,0)
# define SIMDE_VECTOR_SCALAR
# endif
# endif
/* GCC and clang have built-in functions to handle shuffling and
converting of vectors, but the implementations are slightly
different. This macro is just an abstraction over them. Note that
elem_size is in bits but vec_size is in bytes. */
# if !defined(SIMDE_NO_SHUFFLE_VECTOR) && defined(SIMDE_VECTOR_SUBSCRIPT)
HEDLEY_DIAGNOSTIC_PUSH
/* We don't care about -Wvariadic-macros; all compilers that support
* shufflevector/shuffle support them. */
# if HEDLEY_HAS_WARNING("-Wc++98-compat-pedantic")
# pragma clang diagnostic ignored "-Wc++98-compat-pedantic"
# endif
# if HEDLEY_HAS_WARNING("-Wvariadic-macros") || HEDLEY_GCC_VERSION_CHECK(4,0,0)
# pragma GCC diagnostic ignored "-Wvariadic-macros"
# endif
# if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
# define SIMDE_SHUFFLE_VECTOR_(elem_size, vec_size, a, b, ...) __builtin_shufflevector(a, b, __VA_ARGS__)
# elif HEDLEY_GCC_HAS_BUILTIN(__builtin_shuffle,4,7,0) && !defined(__INTEL_COMPILER)
# define SIMDE_SHUFFLE_VECTOR_(elem_size, vec_size, a, b, ...) (__extension__ ({ \
int##elem_size##_t SIMDE_VECTOR(vec_size) simde_shuffle_ = { __VA_ARGS__ }; \
__builtin_shuffle(a, b, simde_shuffle_); \
}))
# endif
HEDLEY_DIAGNOSTIC_POP
# endif
/* TODO: this actually works on XL C/C++ without SIMDE_VECTOR_SUBSCRIPT
but the code needs to be refactored a bit to take advantage. */
# if !defined(SIMDE_NO_CONVERT_VECTOR) && defined(SIMDE_VECTOR_SUBSCRIPT)
# if HEDLEY_HAS_BUILTIN(__builtin_convertvector) || HEDLEY_GCC_VERSION_CHECK(9,0,0)
# if HEDLEY_GCC_VERSION_CHECK(9,0,0) && !HEDLEY_GCC_VERSION_CHECK(9,3,0)
/* https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93557 */
# define SIMDE_CONVERT_VECTOR_(to, from) ((to) = (__extension__({ \
__typeof__(from) from_ = (from); \
((void) from_); \
__builtin_convertvector(from_, __typeof__(to)); \
})))
# else
# define SIMDE_CONVERT_VECTOR_(to, from) ((to) = __builtin_convertvector((from), __typeof__(to)))
# endif
# endif
# endif
#endif
/* Since we currently require SUBSCRIPT before using a vector in a
union, we define these as dependencies of SUBSCRIPT. They are
likely to disappear in the future, once SIMDe learns how to make
use of vectors without using the union members. Do not use them
in your code unless you're okay with it breaking when SIMDe
changes. */
#if defined(SIMDE_VECTOR_SUBSCRIPT)
# if defined(SIMDE_VECTOR_OPS)
# define SIMDE_VECTOR_SUBSCRIPT_OPS
# endif
# if defined(SIMDE_VECTOR_SCALAR)
# define SIMDE_VECTOR_SUBSCRIPT_SCALAR
# endif
#endif
#if !defined(SIMDE_DISABLE_OPENMP)
#if !defined(SIMDE_ENABLE_OPENMP) && ((defined(_OPENMP) && (_OPENMP >= 201307L)) || (defined(_OPENMP_SIMD) && (_OPENMP_SIMD >= 201307L))) || defined(HEDLEY_MCST_LCC_VERSION)
#define SIMDE_ENABLE_OPENMP
#endif
#endif
#if !defined(SIMDE_ENABLE_CILKPLUS) && (defined(__cilk) || defined(HEDLEY_INTEL_VERSION))
# define SIMDE_ENABLE_CILKPLUS
#endif
#if defined(SIMDE_ENABLE_OPENMP)
# define SIMDE_VECTORIZE HEDLEY_PRAGMA(omp simd)
# define SIMDE_VECTORIZE_SAFELEN(l) HEDLEY_PRAGMA(omp simd safelen(l))
# if defined(__clang__)
# define SIMDE_VECTORIZE_REDUCTION(r) \
HEDLEY_DIAGNOSTIC_PUSH \
_Pragma("clang diagnostic ignored \"-Wsign-conversion\"") \
HEDLEY_PRAGMA(omp simd reduction(r)) \
HEDLEY_DIAGNOSTIC_POP
# else
# define SIMDE_VECTORIZE_REDUCTION(r) HEDLEY_PRAGMA(omp simd reduction(r))
# endif
# define SIMDE_VECTORIZE_ALIGNED(a) HEDLEY_PRAGMA(omp simd aligned(a))
#elif defined(SIMDE_ENABLE_CILKPLUS)
# define SIMDE_VECTORIZE HEDLEY_PRAGMA(simd)
# define SIMDE_VECTORIZE_SAFELEN(l) HEDLEY_PRAGMA(simd vectorlength(l))
# define SIMDE_VECTORIZE_REDUCTION(r) HEDLEY_PRAGMA(simd reduction(r))
# define SIMDE_VECTORIZE_ALIGNED(a) HEDLEY_PRAGMA(simd aligned(a))
#elif defined(__clang__) && !defined(HEDLEY_IBM_VERSION)
# define SIMDE_VECTORIZE HEDLEY_PRAGMA(clang loop vectorize(enable))
# define SIMDE_VECTORIZE_SAFELEN(l) HEDLEY_PRAGMA(clang loop vectorize_width(l))
# define SIMDE_VECTORIZE_REDUCTION(r) SIMDE_VECTORIZE
# define SIMDE_VECTORIZE_ALIGNED(a)
#elif HEDLEY_GCC_VERSION_CHECK(4,9,0)
# define SIMDE_VECTORIZE HEDLEY_PRAGMA(GCC ivdep)
# define SIMDE_VECTORIZE_SAFELEN(l) SIMDE_VECTORIZE
# define SIMDE_VECTORIZE_REDUCTION(r) SIMDE_VECTORIZE
# define SIMDE_VECTORIZE_ALIGNED(a)
#elif HEDLEY_CRAY_VERSION_CHECK(5,0,0)
# define SIMDE_VECTORIZE HEDLEY_PRAGMA(_CRI ivdep)
# define SIMDE_VECTORIZE_SAFELEN(l) SIMDE_VECTORIZE
# define SIMDE_VECTORIZE_REDUCTION(r) SIMDE_VECTORIZE
# define SIMDE_VECTORIZE_ALIGNED(a)
#else
# define SIMDE_VECTORIZE
# define SIMDE_VECTORIZE_SAFELEN(l)
# define SIMDE_VECTORIZE_REDUCTION(r)
# define SIMDE_VECTORIZE_ALIGNED(a)
#endif
#define SIMDE_MASK_NZ_(v, mask) (((v) & (mask)) | !((v) & (mask)))
/* Intended for checking coverage, you should never use this in
production. */
#if defined(SIMDE_NO_INLINE)
# define SIMDE_FUNCTION_ATTRIBUTES HEDLEY_NEVER_INLINE static
#else
# define SIMDE_FUNCTION_ATTRIBUTES HEDLEY_ALWAYS_INLINE static
#endif
#if \
HEDLEY_HAS_ATTRIBUTE(unused) || \
HEDLEY_GCC_VERSION_CHECK(2,95,0)
# define SIMDE_FUNCTION_POSSIBLY_UNUSED_ __attribute__((__unused__))
#else
# define SIMDE_FUNCTION_POSSIBLY_UNUSED_
#endif
#if HEDLEY_HAS_WARNING("-Wused-but-marked-unused")
# define SIMDE_DIAGNOSTIC_DISABLE_USED_BUT_MARKED_UNUSED _Pragma("clang diagnostic ignored \"-Wused-but-marked-unused\"")
#else
# define SIMDE_DIAGNOSTIC_DISABLE_USED_BUT_MARKED_UNUSED
#endif
#if defined(_MSC_VER)
# define SIMDE_BEGIN_DECLS_ HEDLEY_DIAGNOSTIC_PUSH __pragma(warning(disable:4996 4204)) HEDLEY_BEGIN_C_DECLS
# define SIMDE_END_DECLS_ HEDLEY_DIAGNOSTIC_POP HEDLEY_END_C_DECLS
#else
# define SIMDE_BEGIN_DECLS_ \
HEDLEY_DIAGNOSTIC_PUSH \
SIMDE_DIAGNOSTIC_DISABLE_USED_BUT_MARKED_UNUSED \
HEDLEY_BEGIN_C_DECLS
# define SIMDE_END_DECLS_ \
HEDLEY_END_C_DECLS \
HEDLEY_DIAGNOSTIC_POP
#endif
#if defined(__SIZEOF_INT128__)
# define SIMDE_HAVE_INT128_
HEDLEY_DIAGNOSTIC_PUSH
SIMDE_DIAGNOSTIC_DISABLE_PEDANTIC_
typedef __int128 simde_int128;
typedef unsigned __int128 simde_uint128;
HEDLEY_DIAGNOSTIC_POP
#endif
#if !defined(SIMDE_ENDIAN_LITTLE)
# define SIMDE_ENDIAN_LITTLE 1234
#endif
#if !defined(SIMDE_ENDIAN_BIG)
# define SIMDE_ENDIAN_BIG 4321
#endif
#if !defined(SIMDE_ENDIAN_ORDER)
/* GCC (and compilers masquerading as GCC) define __BYTE_ORDER__. */
# if defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_LITTLE
# elif defined(__BYTE_ORDER__) && defined(__ORDER_BIG_ENDIAN__) && (__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_BIG
/* TI defines _BIG_ENDIAN or _LITTLE_ENDIAN */
# elif defined(_BIG_ENDIAN)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_BIG
# elif defined(_LITTLE_ENDIAN)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_LITTLE
/* We know the endianness of some common architectures. Common
* architectures not listed (ARM, POWER, MIPS, etc.) here are
* bi-endian. */
# elif defined(__amd64) || defined(_M_X64) || defined(__i386) || defined(_M_IX86)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_LITTLE
# elif defined(__s390x__) || defined(__zarch__)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_BIG
/* Looks like we'll have to rely on the platform. If we're missing a
* platform, please let us know. */
# elif defined(_WIN32)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_LITTLE
# elif defined(sun) || defined(__sun) /* Solaris */
# include <sys/byteorder.h>
# if defined(_LITTLE_ENDIAN)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_LITTLE
# elif defined(_BIG_ENDIAN)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_BIG
# endif
# elif defined(__APPLE__)
# include <libkern/OSByteOrder.h>
# if defined(__LITTLE_ENDIAN__)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_LITTLE
# elif defined(__BIG_ENDIAN__)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_BIG
# endif
# elif defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || defined(__bsdi__) || defined(__DragonFly__) || defined(BSD)
# include <machine/endian.h>
# if defined(__BYTE_ORDER) && (__BYTE_ORDER == __LITTLE_ENDIAN)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_LITTLE
# elif defined(__BYTE_ORDER) && (__BYTE_ORDER == __BIG_ENDIAN)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_BIG
# endif
# elif defined(__linux__) || defined(__linux) || defined(__gnu_linux__)
# include <endian.h>
# if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && (__BYTE_ORDER == __LITTLE_ENDIAN)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_LITTLE
# elif defined(__BYTE_ORDER) && defined(__BIG_ENDIAN) && (__BYTE_ORDER == __BIG_ENDIAN)
# define SIMDE_ENDIAN_ORDER SIMDE_ENDIAN_BIG
# endif
# endif
#endif
#if \
HEDLEY_HAS_BUILTIN(__builtin_bswap64) || \
HEDLEY_GCC_VERSION_CHECK(4,3,0) || \
HEDLEY_IBM_VERSION_CHECK(13,1,0) || \
HEDLEY_INTEL_VERSION_CHECK(13,0,0)
#define simde_bswap64(v) __builtin_bswap64(v)
#elif HEDLEY_MSVC_VERSION_CHECK(13,10,0)
#define simde_bswap64(v) _byteswap_uint64(v)
#else
SIMDE_FUNCTION_ATTRIBUTES
uint64_t
simde_bswap64(uint64_t v) {
return
((v & (((uint64_t) 0xff) << 56)) >> 56) |
((v & (((uint64_t) 0xff) << 48)) >> 40) |
((v & (((uint64_t) 0xff) << 40)) >> 24) |
((v & (((uint64_t) 0xff) << 32)) >> 8) |
((v & (((uint64_t) 0xff) << 24)) << 8) |
((v & (((uint64_t) 0xff) << 16)) << 24) |
((v & (((uint64_t) 0xff) << 8)) << 40) |
((v & (((uint64_t) 0xff) )) << 56);
}
#endif
#if !defined(SIMDE_ENDIAN_ORDER)
# error Unknown byte order; please file a bug
#else
# if SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE
# define simde_endian_bswap64_be(value) simde_bswap64(value)
# define simde_endian_bswap64_le(value) (value)
# elif SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_BIG
# define simde_endian_bswap64_be(value) (value)
# define simde_endian_bswap64_le(value) simde_bswap64(value)
# endif
#endif
/* TODO: we should at least make an attempt to detect the correct
types for simde_float32/float64 instead of just assuming float and
double. */
#if !defined(SIMDE_FLOAT32_TYPE)
# define SIMDE_FLOAT32_TYPE float
# define SIMDE_FLOAT32_C(value) value##f
#else
# define SIMDE_FLOAT32_C(value) ((SIMDE_FLOAT32_TYPE) value)
#endif
typedef SIMDE_FLOAT32_TYPE simde_float32;
#if !defined(SIMDE_FLOAT64_TYPE)
# define SIMDE_FLOAT64_TYPE double
# define SIMDE_FLOAT64_C(value) value
#else
# define SIMDE_FLOAT64_C(value) ((SIMDE_FLOAT64_TYPE) value)
#endif
typedef SIMDE_FLOAT64_TYPE simde_float64;
#if defined(__cplusplus)
typedef bool simde_bool;
#elif defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L)
typedef _Bool simde_bool;
#elif defined(bool)
typedef bool simde_bool;
#else
#include <stdbool.h>
typedef bool simde_bool;
#endif
#if HEDLEY_HAS_WARNING("-Wbad-function-cast")
# define SIMDE_CONVERT_FTOI(T,v) \
HEDLEY_DIAGNOSTIC_PUSH \
_Pragma("clang diagnostic ignored \"-Wbad-function-cast\"") \
HEDLEY_STATIC_CAST(T, (v)) \
HEDLEY_DIAGNOSTIC_POP
#else
# define SIMDE_CONVERT_FTOI(T,v) ((T) (v))
#endif
/* TODO: detect compilers which support this outside of C11 mode */
#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)
#define SIMDE_CHECKED_REINTERPRET_CAST(to, from, value) _Generic((value), to: (value), default: (_Generic((value), from: ((to) (value)))))
#define SIMDE_CHECKED_STATIC_CAST(to, from, value) _Generic((value), to: (value), default: (_Generic((value), from: ((to) (value)))))
#else
#define SIMDE_CHECKED_REINTERPRET_CAST(to, from, value) HEDLEY_REINTERPRET_CAST(to, value)
#define SIMDE_CHECKED_STATIC_CAST(to, from, value) HEDLEY_STATIC_CAST(to, value)
#endif
#if HEDLEY_HAS_WARNING("-Wfloat-equal")
# define SIMDE_DIAGNOSTIC_DISABLE_FLOAT_EQUAL _Pragma("clang diagnostic ignored \"-Wfloat-equal\"")
#elif HEDLEY_GCC_VERSION_CHECK(3,0,0)
# define SIMDE_DIAGNOSTIC_DISABLE_FLOAT_EQUAL _Pragma("GCC diagnostic ignored \"-Wfloat-equal\"")
#else
# define SIMDE_DIAGNOSTIC_DISABLE_FLOAT_EQUAL
#endif
/* Some functions can trade accuracy for speed. For those functions
you can control the trade-off using this macro. Possible values:
0: prefer speed
1: reasonable trade-offs
2: prefer accuracy */
#if !defined(SIMDE_ACCURACY_PREFERENCE)
# define SIMDE_ACCURACY_PREFERENCE 1
#endif
#if defined(__STDC_HOSTED__)
# define SIMDE_STDC_HOSTED __STDC_HOSTED__
#else
# if \
defined(HEDLEY_PGI_VERSION) || \
defined(HEDLEY_MSVC_VERSION)
# define SIMDE_STDC_HOSTED 1
# else
# define SIMDE_STDC_HOSTED 0
# endif
#endif
/* Try to deal with environments without a standard library. */
#if !defined(simde_memcpy)
#if HEDLEY_HAS_BUILTIN(__builtin_memcpy)
#define simde_memcpy(dest, src, n) __builtin_memcpy(dest, src, n)
#endif
#endif
#if !defined(simde_memset)
#if HEDLEY_HAS_BUILTIN(__builtin_memset)
#define simde_memset(s, c, n) __builtin_memset(s, c, n)
#endif
#endif
#if !defined(simde_memcmp)
#if HEDLEY_HAS_BUILTIN(__builtin_memcmp)
#define simde_memcmp(s1, s2, n) __builtin_memcmp(s1, s2, n)
#endif
#endif
#if !defined(simde_memcpy) || !defined(simde_memset) || !defined(simde_memcmp)
#if !defined(SIMDE_NO_STRING_H)
#if defined(__has_include)
#if !__has_include(<string.h>)
#define SIMDE_NO_STRING_H
#endif
#elif (SIMDE_STDC_HOSTED == 0)
#define SIMDE_NO_STRING_H
#endif
#endif
#if !defined(SIMDE_NO_STRING_H)
#include <string.h>
#if !defined(simde_memcpy)
#define simde_memcpy(dest, src, n) memcpy(dest, src, n)
#endif
#if !defined(simde_memset)
#define simde_memset(s, c, n) memset(s, c, n)
#endif
#if !defined(simde_memcmp)
#define simde_memcmp(s1, s2, n) memcmp(s1, s2, n)
#endif
#else
/* These are meant to be portable, not fast. If you're hitting them you
* should think about providing your own (by defining the simde_memcpy
* macro prior to including any SIMDe files) or submitting a patch to
* SIMDe so we can detect your system-provided memcpy/memset, like by
* adding your compiler to the checks for __builtin_memcpy and/or
* __builtin_memset. */
#if !defined(simde_memcpy)
SIMDE_FUNCTION_ATTRIBUTES
void
simde_memcpy_(void* dest, const void* src, size_t len) {
char* dest_ = HEDLEY_STATIC_CAST(char*, dest);
char* src_ = HEDLEY_STATIC_CAST(const char*, src);
for (size_t i = 0 ; i < len ; i++) {
dest_[i] = src_[i];
}
}
#define simde_memcpy(dest, src, n) simde_memcpy_(dest, src, n)
#endif
#if !defined(simde_memset)
SIMDE_FUNCTION_ATTRIBUTES
void
simde_memset_(void* s, int c, size_t len) {
char* s_ = HEDLEY_STATIC_CAST(char*, s);
char c_ = HEDLEY_STATIC_CAST(char, c);
for (size_t i = 0 ; i < len ; i++) {
s_[i] = c_[i];
}
}
#define simde_memset(s, c, n) simde_memset_(s, c, n)
#endif
#if !defined(simde_memcmp)
SIMDE_FUCTION_ATTRIBUTES
int
simde_memcmp_(const void *s1, const void *s2, size_t n) {
unsigned char* s1_ = HEDLEY_STATIC_CAST(unsigned char*, s1);
unsigned char* s2_ = HEDLEY_STATIC_CAST(unsigned char*, s2);
for (size_t i = 0 ; i < len ; i++) {
if (s1_[i] != s2_[i]) {
return (int) (s1_[i] - s2_[i]);
}
}
return 0;
}
#define simde_memcmp(s1, s2, n) simde_memcmp_(s1, s2, n)
#endif
#endif
#endif
#if defined(FE_ALL_EXCEPT)
#define SIMDE_HAVE_FENV_H
#elif defined(__has_include)
#if __has_include(<fenv.h>)
#include <fenv.h>
#define SIMDE_HAVE_FENV_H
#endif
#elif SIMDE_STDC_HOSTED == 1
#include <fenv.h>
#define SIMDE_HAVE_FENV_H
#endif
#if defined(EXIT_FAILURE)
#define SIMDE_HAVE_STDLIB_H
#elif defined(__has_include)
#if __has_include(<stdlib.h>)
#include <stdlib.h>
#define SIMDE_HAVE_STDLIB_H
#endif
#elif SIMDE_STDC_HOSTED == 1
#include <stdlib.h>
#define SIMDE_HAVE_STDLIB_H
#endif
#if defined(__has_include)
# if defined(__cplusplus) && (__cplusplus >= 201103L) && __has_include(<cfenv>)
# include <cfenv>
# elif __has_include(<fenv.h>)
# include <fenv.h>
# endif
# if __has_include(<stdlib.h>)
# include <stdlib.h>
# endif
#elif SIMDE_STDC_HOSTED == 1
# include <stdlib.h>
# include <fenv.h>
#endif
#define SIMDE_DEFINE_CONVERSION_FUNCTION_(Name, T_To, T_From) \
static HEDLEY_ALWAYS_INLINE HEDLEY_CONST \
T_To \
Name (T_From value) { \
T_To r; \
simde_memcpy(&r, &value, sizeof(r)); \
return r; \
}
#include "check.h"
/* GCC/clang have a bunch of functionality in builtins which we would
* like to access, but the suffixes indicate whether the operate on
* int, long, or long long, not fixed width types (e.g., int32_t).
* we use these macros to attempt to map from fixed-width to the
* names GCC uses. Note that you should still cast the input(s) and
* return values (to/from SIMDE_BUILTIN_TYPE_*_) since often even if
* types are the same size they may not be compatible according to the
* compiler. For example, on x86 long and long lonsg are generally
* both 64 bits, but platforms vary on whether an int64_t is mapped
* to a long or long long. */
#include <limits.h>
HEDLEY_DIAGNOSTIC_PUSH
SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC_
#if (INT8_MAX == INT_MAX) && (INT8_MIN == INT_MIN)
#define SIMDE_BUILTIN_SUFFIX_8_
#define SIMDE_BUILTIN_TYPE_8_ int
#elif (INT8_MAX == LONG_MAX) && (INT8_MIN == LONG_MIN)
#define SIMDE_BUILTIN_SUFFIX_8_ l
#define SIMDE_BUILTIN_TYPE_8_ long
#elif (INT8_MAX == LLONG_MAX) && (INT8_MIN == LLONG_MIN)
#define SIMDE_BUILTIN_SUFFIX_8_ ll
#define SIMDE_BUILTIN_TYPE_8_ long long
#endif
#if (INT16_MAX == INT_MAX) && (INT16_MIN == INT_MIN)
#define SIMDE_BUILTIN_SUFFIX_16_
#define SIMDE_BUILTIN_TYPE_16_ int
#elif (INT16_MAX == LONG_MAX) && (INT16_MIN == LONG_MIN)
#define SIMDE_BUILTIN_SUFFIX_16_ l
#define SIMDE_BUILTIN_TYPE_16_ long
#elif (INT16_MAX == LLONG_MAX) && (INT16_MIN == LLONG_MIN)
#define SIMDE_BUILTIN_SUFFIX_16_ ll
#define SIMDE_BUILTIN_TYPE_16_ long long
#endif
#if (INT32_MAX == INT_MAX) && (INT32_MIN == INT_MIN)
#define SIMDE_BUILTIN_SUFFIX_32_
#define SIMDE_BUILTIN_TYPE_32_ int
#elif (INT32_MAX == LONG_MAX) && (INT32_MIN == LONG_MIN)
#define SIMDE_BUILTIN_SUFFIX_32_ l
#define SIMDE_BUILTIN_TYPE_32_ long
#elif (INT32_MAX == LLONG_MAX) && (INT32_MIN == LLONG_MIN)
#define SIMDE_BUILTIN_SUFFIX_32_ ll
#define SIMDE_BUILTIN_TYPE_32_ long long
#endif
#if (INT64_MAX == INT_MAX) && (INT64_MIN == INT_MIN)
#define SIMDE_BUILTIN_SUFFIX_64_
#define SIMDE_BUILTIN_TYPE_64_ int
#elif (INT64_MAX == LONG_MAX) && (INT64_MIN == LONG_MIN)
#define SIMDE_BUILTIN_SUFFIX_64_ l
#define SIMDE_BUILTIN_TYPE_64_ long
#elif (INT64_MAX == LLONG_MAX) && (INT64_MIN == LLONG_MIN)
#define SIMDE_BUILTIN_SUFFIX_64_ ll
#define SIMDE_BUILTIN_TYPE_64_ long long
#endif
#if defined(SIMDE_BUILTIN_SUFFIX_8_)
#define SIMDE_BUILTIN_8_(name) HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_8_)
#define SIMDE_BUILTIN_HAS_8_(name) HEDLEY_HAS_BUILTIN(HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_8_))
#else
#define SIMDE_BUILTIN_HAS_8_(name) 0
#endif
#if defined(SIMDE_BUILTIN_SUFFIX_16_)
#define SIMDE_BUILTIN_16_(name) HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_16_)
#define SIMDE_BUILTIN_HAS_16_(name) HEDLEY_HAS_BUILTIN(HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_16_))
#else
#define SIMDE_BUILTIN_HAS_16_(name) 0
#endif
#if defined(SIMDE_BUILTIN_SUFFIX_32_)
#define SIMDE_BUILTIN_32_(name) HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_32_)
#define SIMDE_BUILTIN_HAS_32_(name) HEDLEY_HAS_BUILTIN(HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_32_))
#else
#define SIMDE_BUILTIN_HAS_32_(name) 0
#endif
#if defined(SIMDE_BUILTIN_SUFFIX_64_)
#define SIMDE_BUILTIN_64_(name) HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_64_)
#define SIMDE_BUILTIN_HAS_64_(name) HEDLEY_HAS_BUILTIN(HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_64_))
#else
#define SIMDE_BUILTIN_HAS_64_(name) 0
#endif
HEDLEY_DIAGNOSTIC_POP
/* Sometimes we run into problems with specific versions of compilers
which make the native versions unusable for us. Often this is due
to missing functions, sometimes buggy implementations, etc. These
macros are how we check for specific bugs. As they are fixed we'll
start only defining them for problematic compiler versions. */
#if !defined(SIMDE_IGNORE_COMPILER_BUGS)
# if defined(HEDLEY_GCC_VERSION)
# if !HEDLEY_GCC_VERSION_CHECK(4,9,0)
# define SIMDE_BUG_GCC_REV_208793
# endif
# if !HEDLEY_GCC_VERSION_CHECK(5,0,0)
# define SIMDE_BUG_GCC_BAD_MM_SRA_EPI32 /* TODO: find relevant bug or commit */
# endif
# if !HEDLEY_GCC_VERSION_CHECK(6,0,0)
# define SIMDE_BUG_GCC_SIZEOF_IMMEDIATE
# endif
# if !HEDLEY_GCC_VERSION_CHECK(4,6,0)
# define SIMDE_BUG_GCC_BAD_MM_EXTRACT_EPI8 /* TODO: find relevant bug or commit */
# endif
# if !HEDLEY_GCC_VERSION_CHECK(8,0,0)
# define SIMDE_BUG_GCC_REV_247851
# endif
# if !HEDLEY_GCC_VERSION_CHECK(10,0,0)
# define SIMDE_BUG_GCC_REV_274313
# define SIMDE_BUG_GCC_91341
# endif
# if !HEDLEY_GCC_VERSION_CHECK(9,0,0) && defined(SIMDE_ARCH_AARCH64)
# define SIMDE_BUG_GCC_ARM_SHIFT_SCALAR
# endif
# if !HEDLEY_GCC_VERSION_CHECK(9,0,0) && defined(SIMDE_ARCH_AARCH64)
# define SIMDE_BUG_GCC_BAD_VEXT_REV32
# endif
# if defined(SIMDE_ARCH_X86) && !defined(SIMDE_ARCH_AMD64)
# define SIMDE_BUG_GCC_94482
# endif
# if (defined(SIMDE_ARCH_X86) && !defined(SIMDE_ARCH_AMD64)) || defined(SIMDE_ARCH_ZARCH)
# define SIMDE_BUG_GCC_53784
# endif
# if defined(SIMDE_ARCH_X86) || defined(SIMDE_ARCH_AMD64)
# if HEDLEY_GCC_VERSION_CHECK(4,3,0) /* -Wsign-conversion */
# define SIMDE_BUG_GCC_95144
# endif
# if !HEDLEY_GCC_VERSION_CHECK(11,0,0)
# define SIMDE_BUG_GCC_95483
# endif
# define SIMDE_BUG_GCC_98521
# endif
# if !HEDLEY_GCC_VERSION_CHECK(9,4,0) && defined(SIMDE_ARCH_AARCH64)
# define SIMDE_BUG_GCC_94488
# endif
# if !HEDLEY_GCC_VERSION_CHECK(9,1,0) && defined(SIMDE_ARCH_AARCH64)
# define SIMDE_BUG_GCC_REV_264019
# endif
# if defined(SIMDE_ARCH_ARM)
# define SIMDE_BUG_GCC_95399
# define SIMDE_BUG_GCC_95471
# elif defined(SIMDE_ARCH_POWER)
# define SIMDE_BUG_GCC_95227
# define SIMDE_BUG_GCC_95782
# elif defined(SIMDE_ARCH_X86) || defined(SIMDE_ARCH_AMD64)
# if !HEDLEY_GCC_VERSION_CHECK(10,2,0) && !defined(__OPTIMIZE__)
# define SIMDE_BUG_GCC_96174
# endif
# elif defined(SIMDE_ARCH_ZARCH)
# if !HEDLEY_GCC_VERSION_CHECK(9,0,0)
# define SIMDE_BUG_GCC_95782
# endif
# endif
# define SIMDE_BUG_GCC_95399
# elif defined(__clang__)
# if defined(SIMDE_ARCH_AARCH64)
# define SIMDE_BUG_CLANG_45541
# define SIMDE_BUG_CLANG_46844
# define SIMDE_BUG_CLANG_48257
# if SIMDE_DETECT_CLANG_VERSION_CHECK(10,0,0) && SIMDE_DETECT_CLANG_VERSION_NOT(11,0,0)
# define SIMDE_BUG_CLANG_BAD_VI64_OPS
# endif
# if SIMDE_DETECT_CLANG_VERSION_NOT(9,0,0)
# define SIMDE_BUG_CLANG_GIT_4EC445B8
# define SIMDE_BUG_CLANG_REV_365298 /* 0464e07c8f6e3310c28eb210a4513bc2243c2a7e */
# endif
# endif
# if defined(SIMDE_ARCH_ARM)
# if !SIMDE_DETECT_CLANG_VERSION_CHECK(11,0,0)
# define SIMDE_BUG_CLANG_BAD_VGET_SET_LANE_TYPES
# endif
# endif
# if defined(SIMDE_ARCH_POWER)
# define SIMDE_BUG_CLANG_46770
# endif
# if defined(_ARCH_PWR9) && !SIMDE_DETECT_CLANG_VERSION_CHECK(12,0,0) && !defined(__OPTIMIZE__)
# define SIMDE_BUG_CLANG_POWER9_16x4_BAD_SHIFT
# endif
# if defined(SIMDE_ARCH_X86) || defined(SIMDE_ARCH_AMD64)
# if SIMDE_DETECT_CLANG_VERSION_NOT(5,0,0)
# define SIMDE_BUG_CLANG_REV_298042 /* 6afc436a7817a52e78ae7bcdc3faafd460124cac */
# endif
# if SIMDE_DETECT_CLANG_VERSION_NOT(3,7,0)
# define SIMDE_BUG_CLANG_REV_234560 /* b929ad7b1726a32650a8051f69a747fb6836c540 */
# endif
# if SIMDE_DETECT_CLANG_VERSION_CHECK(3,8,0) && SIMDE_DETECT_CLANG_VERSION_NOT(5,0,0)
# define SIMDE_BUG_CLANG_BAD_MADD
# endif
# if SIMDE_DETECT_CLANG_VERSION_CHECK(4,0,0) && SIMDE_DETECT_CLANG_VERSION_NOT(5,0,0)
# define SIMDE_BUG_CLANG_REV_299346 /* ac9959eb533a58482ea4da6c4db1e635a98de384 */
# endif
# if SIMDE_DETECT_CLANG_VERSION_NOT(8,0,0)
# define SIMDE_BUG_CLANG_REV_344862 /* eae26bf73715994c2bd145f9b6dc3836aa4ffd4f */
# endif
# if HEDLEY_HAS_WARNING("-Wsign-conversion") && SIMDE_DETECT_CLANG_VERSION_NOT(11,0,0)
# define SIMDE_BUG_CLANG_45931
# endif
# if HEDLEY_HAS_WARNING("-Wvector-conversion") && SIMDE_DETECT_CLANG_VERSION_NOT(11,0,0)
# define SIMDE_BUG_CLANG_44589
# endif
# define SIMDE_BUG_CLANG_48673
# endif
# define SIMDE_BUG_CLANG_45959
# elif defined(HEDLEY_MSVC_VERSION)
# if defined(SIMDE_ARCH_X86)
# define SIMDE_BUG_MSVC_ROUND_EXTRACT
# endif
# elif defined(HEDLEY_INTEL_VERSION)
# define SIMDE_BUG_INTEL_857088
# elif defined(HEDLEY_MCST_LCC_VERSION)
# define SIMDE_BUG_MCST_LCC_MISSING_AVX_LOAD_STORE_M128_FUNCS
# endif
#endif
/* GCC and Clang both have the same issue:
* https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95144
* https://bugs.llvm.org/show_bug.cgi?id=45931
* This is just an easy way to work around it.
*/
#if \
(HEDLEY_HAS_WARNING("-Wsign-conversion") && SIMDE_DETECT_CLANG_VERSION_NOT(11,0,0)) || \
HEDLEY_GCC_VERSION_CHECK(4,3,0)
# define SIMDE_BUG_IGNORE_SIGN_CONVERSION(expr) (__extension__ ({ \
HEDLEY_DIAGNOSTIC_PUSH \
HEDLEY_DIAGNOSTIC_POP \
_Pragma("GCC diagnostic ignored \"-Wsign-conversion\"") \
__typeof__(expr) simde_bug_ignore_sign_conversion_v_= (expr); \
HEDLEY_DIAGNOSTIC_PUSH \
simde_bug_ignore_sign_conversion_v_; \
}))
#else
# define SIMDE_BUG_IGNORE_SIGN_CONVERSION(expr) (expr)
#endif
/* Usually the shift count is signed (for example, NEON or SSE).
* OTOH, unsigned is good for PPC (vec_srl uses unsigned), and the only option for E2K.
* Further info: https://github.com/simd-everywhere/simde/pull/700
*/
#if defined(SIMDE_ARCH_E2K) || defined(SIMDE_ARCH_POWER)
#define SIMDE_CAST_VECTOR_SHIFT_COUNT(width, value) HEDLEY_STATIC_CAST(uint##width##_t, (value))
#else
#define SIMDE_CAST_VECTOR_SHIFT_COUNT(width, value) HEDLEY_STATIC_CAST(int##width##_t, (value))
#endif
#endif /* !defined(SIMDE_COMMON_H) */