forked from pytorch/pytorch
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathBlasKernel.cpp
1062 lines (962 loc) · 35.2 KB
/
BlasKernel.cpp
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
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
#define TORCH_ASSERT_ONLY_METHOD_OPERATORS
#include <ATen/Context.h>
#include <ATen/Config.h>
#include <ATen/OpMathType.h>
#include <ATen/Parallel.h>
#include <ATen/cpu/vec/vec.h>
#include <ATen/native/cpu/ReducedPrecisionFloatGemvFastPathKernel.h>
#include <c10/core/ScalarType.h>
#include <c10/macros/Macros.h>
#include <c10/util/Exception.h>
#include <c10/util/Unroll.h>
#include <c10/util/complex.h>
#include <c10/util/irange.h>
#include <algorithm>
#include <climits>
#include <limits>
#if defined(__aarch64__) && !defined(C10_MOBILE)
#include <arm_neon.h>
#include <cpuinfo.h>
#endif
C10_DIAGNOSTIC_PUSH_AND_IGNORED_IF_DEFINED("-Wunused-function")
namespace {
/// Wrapper for const_cast<T*> with type-inference.
///
/// Use this to call into APIs that are not const-correct.
template <typename T>
T* remove_const(const T* x) {
return const_cast<T*>(x);
}
} // namespace
#if AT_BUILD_WITH_BLAS()
extern "C" double ddot_(int *n, double *x, int *incx, double *y, int *incy);
extern "C" void dscal_(int *n, double *a, double *x, int *incx);
extern "C" void sscal_(int *n, float *a, float *x, int *incx);
extern "C" void dgemv_(char *trans, int *m, int *n, double *alpha, double *a, int *lda, double *x, int *incx, double *beta, double *y, int *incy);
extern "C" void sgemv_(char *trans, int *m, int *n, float *alpha, float *a, int *lda, float *x, int *incx, float *beta, float *y, int *incy);
#if AT_BLAS_F2C()
# define ffloat double
#else
# define ffloat float
#endif
#if AT_BLAS_USE_CBLAS_DOT()
extern "C" float cblas_sdot(const int n, const float *x, const int incx, const float *y, const int incy);
extern "C" void cblas_cdotu_sub(const int n, const void *x, const int incx, const void *y, const int incy, void *dotu);
extern "C" void cblas_zdotu_sub(const int n, const void *x, const int incx, const void *y, const int incy, void *dotu);
extern "C" void cblas_cdotc_sub(const int n, const void *x, const int incx, const void *y, const int incy, void *dotc);
extern "C" void cblas_zdotc_sub(const int n, const void *x, const int incx, const void *y, const int incy, void *dotc);
static inline ffloat sdot_(const int *n, const float *x, const int *incx, const float *y, const int *incy)
{
return cblas_sdot(*n, x, *incx, y, *incy);
}
static inline void cdotu_(std::complex<float> *res, const int *n, const std::complex<float> *x, const int *incx,
const std::complex<float> *y, const int *incy) {
cblas_cdotu_sub(*n, x, *incx, y, *incy, res);
}
static inline void zdotu_(std::complex<double> *res, const int *n, const std::complex<double> *x, const int *incx,
const std::complex<double> *y, const int *incy) {
cblas_zdotu_sub(*n, x, *incx, y, *incy, res);
}
static inline void cdotc_(std::complex<float> *res, const int *n, const std::complex<float> *x, const int *incx,
const std::complex<float> *y, const int *incy) {
cblas_cdotc_sub(*n, x, *incx, y, *incy, res);
}
static inline void zdotc_(std::complex<double> *res, const int *n, const std::complex<double> *x, const int *incx,
const std::complex<double> *y, const int *incy) {
cblas_zdotc_sub(*n, x, *incx, y, *incy, res);
}
#else
extern "C" ffloat sdot_(int *n, float *x, int *incx, float *y, int *incy);
extern "C" void cdotu_(std::complex<float> *res, int *n, std::complex<float> *x, int *incx, std::complex<float> *y, int *incy);
extern "C" void zdotu_(std::complex<double> *res, int *n, std::complex<double> *x, int *incx, std::complex<double> *y, int *incy);
extern "C" void cdotc_(std::complex<float> *res, int *n, std::complex<float> *x, int *incx, std::complex<float> *y, int *incy);
extern "C" void zdotc_(std::complex<double> *res, int *n, std::complex<double> *x, int *incx, std::complex<double> *y, int *incy);
#endif // AT_BLAS_USE_CBLAS_DOT
#endif // AT_BUILD_WITH_BLAS
namespace at::native {
#if !defined(C10_MOBILE)
DEFINE_DISPATCH(fp16_dot_with_fp32_arith_stub);
DEFINE_DISPATCH(fp16_gemv_trans_stub);
#endif // !defined(C10_MOBILE)
namespace blas_impl {
#if !defined(C10_MOBILE)
void fp16_gemv_trans(
const int m,
const int n,
const float alpha,
const Half* a,
const int lda,
const Half* x,
const int incx,
const float beta,
Half* y,
const int incy);
float fp16_dot_with_fp32_arith(
const Half* vec1,
const Half* vec2,
int64_t len);
float fp16_dot_with_fp32_arith(
const Half* x,
const Half* a,
int64_t len) {
return fp16_dot_with_fp32_arith_stub(kCPU, x, a, len);
}
void fp16_gemv_trans(
const int m,
const int n,
const float alpha,
const Half* a,
const int lda,
const Half* x,
const int incx,
const float beta,
Half* y,
const int incy) {
fp16_gemv_trans_stub(kCPU, m, n, alpha, a, lda, x, incx, beta, y, incy);
}
#endif // !defined(C10_MOBILE)
#if defined(__aarch64__) && !defined(C10_MOBILE)
void fp16_gemv_notrans(
const int m,
const int n,
const float alpha,
const Half* a,
const int lda,
const Half* x,
const int incx,
const float beta,
Half* y,
const int incy);
void bf16_gemv_trans(
const int m,
const int n,
const at::BFloat16 alpha,
const at::BFloat16* a,
const int lda,
const at::BFloat16* x,
const int incx,
const at::BFloat16 beta,
at::BFloat16* y,
const int incy);
float bf16_dot_with_fp32_arith(
const at::BFloat16* vec1,
const at::BFloat16* vec2,
int64_t len);
#endif // defined(__aarch64__) && !defined(C10_MOBILE)
template <typename scalar_t>
bool scal_use_fast_path(
[[maybe_unused]] int64_t n,
[[maybe_unused]] int64_t incx) {
return false;
}
template <typename scalar_t>
bool gemv_use_fast_path(
[[maybe_unused]] char trans,
[[maybe_unused]] int64_t m,
[[maybe_unused]] int64_t n,
[[maybe_unused]] scalar_t alpha,
[[maybe_unused]] int64_t lda,
[[maybe_unused]] int64_t incx,
[[maybe_unused]] scalar_t beta,
[[maybe_unused]] int64_t incy) {
return false;
}
template <typename scalar_t>
void scal_fast_path(
[[maybe_unused]] int* n,
[[maybe_unused]] scalar_t* a,
[[maybe_unused]] scalar_t* x,
[[maybe_unused]] int* incx) {
TORCH_INTERNAL_ASSERT(
false, "scal_fast_path shouldn't be called for this configuration");
}
template <typename scalar_t>
void gemv_fast_path(
[[maybe_unused]] const char* trans,
[[maybe_unused]] const int* m,
[[maybe_unused]] const int* n,
[[maybe_unused]] const scalar_t* alpha,
[[maybe_unused]] const scalar_t* a,
[[maybe_unused]] const int* lda,
[[maybe_unused]] const scalar_t* x,
[[maybe_unused]] const int* incx,
[[maybe_unused]] const scalar_t* beta,
[[maybe_unused]] scalar_t* y,
[[maybe_unused]] const int* incy) {
TORCH_INTERNAL_ASSERT(
false, "gemv_fast_path shouldn't be called for this configuration");
}
#define INSTANTIATE(scalar_t) \
template bool scal_use_fast_path<scalar_t>(int64_t n, int64_t incx); \
template bool gemv_use_fast_path<scalar_t>(char trans, int64_t m, int64_t n, scalar_t alpha, int64_t lda, int64_t incx, scalar_t beta, int64_t incy); \
template void gemv_fast_path<scalar_t>(const char *trans, const int *m, const int *n, const scalar_t *alpha, const scalar_t *a, const int *lda, const scalar_t *x, const int *incx, const scalar_t *beta, scalar_t *y, const int *incy); \
template void scal_fast_path<scalar_t>(int *n, scalar_t *a, scalar_t *x, int *incx);
#if AT_BUILD_WITH_BLAS()
template <>
bool scal_use_fast_path<double>(int64_t n, int64_t incx) {
auto intmax = std::numeric_limits<int>::max();
return n <= intmax && incx <= intmax;
}
template <>
bool scal_use_fast_path<float>(int64_t n, int64_t incx) {
return scal_use_fast_path<double>(n, incx);
}
template <>
void scal_fast_path<double>(int *n, double *a, double *x, int *incx) {
dscal_(n, a, x, incx);
}
template <>
void scal_fast_path<float>(int *n, float *a, float *x, int *incx) {
sscal_(n, a, x, incx);
}
template <>
bool gemv_use_fast_path<float>(
[[maybe_unused]] char trans,
int64_t m,
int64_t n,
[[maybe_unused]] float alpha,
int64_t lda,
int64_t incx,
[[maybe_unused]] float beta,
int64_t incy) {
auto intmax = std::numeric_limits<int>::max();
return (m <= intmax) && (n <= intmax) && (lda <= intmax) &&
(incx > 0) && (incx <= intmax) && (incy > 0) && (incy <= intmax);
}
template <>
bool gemv_use_fast_path<double>(
[[maybe_unused]] char trans,
int64_t m,
int64_t n,
[[maybe_unused]] double alpha,
int64_t lda,
int64_t incx,
[[maybe_unused]] double beta,
int64_t incy) {
return gemv_use_fast_path<float>(
trans, m, n, (float)alpha, lda, incx, (float)beta, incy);
}
template <>
void gemv_fast_path<double>(const char *trans, const int *m, const int *n, const double *alpha, const double *a, const int *lda, const double *x, const int *incx, const double *beta, double *y, const int *incy) {
dgemv_(remove_const(trans), remove_const(m), remove_const(n), remove_const(alpha), remove_const(a), remove_const(lda), remove_const(x), remove_const(incx), remove_const(beta), y, remove_const(incy));
}
template <>
void gemv_fast_path<float>(const char *trans, const int *m, const int *n, const float *alpha, const float *a, const int *lda, const float *x, const int *incx, const float *beta, float *y, const int *incy) {
sgemv_(remove_const(trans), remove_const(m), remove_const(n), remove_const(alpha), remove_const(a), remove_const(lda), remove_const(x), remove_const(incx), remove_const(beta), y, remove_const(incy));
}
#else
INSTANTIATE(float)
INSTANTIATE(double)
#endif // AT_BUILD_WITH_BLAS
INSTANTIATE(uint8_t)
INSTANTIATE(int8_t)
INSTANTIATE(int16_t)
INSTANTIATE(int)
INSTANTIATE(int64_t)
#if !defined(C10_MOBILE)
#if !defined(__aarch64__)
// Currently, only fp16_gemv_trans is built for non-aarch64.
template <>
bool gemv_use_fast_path<at::Half>(
char trans,
[[maybe_unused]] int64_t m,
[[maybe_unused]] int64_t n,
at::Half alpha,
[[maybe_unused]] int64_t lda,
[[maybe_unused]] int64_t incx,
[[maybe_unused]] at::Half beta,
[[maybe_unused]] int64_t incy) {
// clang is capable of constant-folding fp16_ieee_from_fp32_value,
// so use it to get simple integer comparisons.
// https://godbolt.org/z/v936hroYb
using c10::detail::fp16_ieee_from_fp32_value;;
return (trans == 'T' || trans == 't') && incx == 1 &&
alpha.x == fp16_ieee_from_fp32_value(1.0f);
}
template <>
void gemv_fast_path<at::Half>(
const char* trans,
const int* m,
const int* n,
const at::Half* alpha,
const at::Half* a,
const int* lda,
const at::Half* x,
const int* incx,
const at::Half* beta,
at::Half* y,
const int* incy) {
TORCH_INTERNAL_ASSERT_DEBUG_ONLY(trans[0] == 'T' || trans[0] == 't');
fp16_gemv_trans(
*m,
*n,
*alpha,
a,
*lda,
x,
*incx,
*beta,
y,
*incy);
}
INSTANTIATE(c10::BFloat16)
#else
template <>
bool scal_use_fast_path<at::Half>(
[[maybe_unused]] int64_t n,
[[maybe_unused]] int64_t incx) {
return false;
}
template <>
bool gemv_use_fast_path<at::Half>(
char trans,
[[maybe_unused]] int64_t m,
[[maybe_unused]] int64_t n,
at::Half alpha,
[[maybe_unused]] int64_t lda,
[[maybe_unused]] int64_t incx,
at::Half beta,
[[maybe_unused]] int64_t incy) {
return incx == 1 && c10::detail::fp16_from_bits(alpha.x) == 1.0f &&
// TODO: enable nonzero beta for fp16_gemv_notrans
(c10::detail::fp16_from_bits(beta.x) == 0.0f || trans == 't' || trans == 'T');
}
template <>
bool gemv_use_fast_path<at::BFloat16>(
[[maybe_unused]] char trans,
[[maybe_unused]] int64_t m,
[[maybe_unused]] int64_t n,
at::BFloat16 alpha,
[[maybe_unused]] int64_t lda,
[[maybe_unused]] int64_t incx,
at::BFloat16 beta,
[[maybe_unused]] int64_t incy) {
return (trans == 'T' || trans == 't') && incx == 1 && alpha == 1.0 &&
beta == 0.0;
}
// We need the shift for reduce(), hence the extra constants.
static constexpr auto kF16ElementsPerIterationShift = 7;
static constexpr auto kF16ElementsPerIteration = 1 << kF16ElementsPerIterationShift;
static_assert(kF16ElementsPerIteration == 128);
static constexpr auto kF16ElementsPerRegisterShift = 3;
static constexpr auto kF16ElementsPerRegister = 1 << kF16ElementsPerRegisterShift;
static_assert(kF16ElementsPerRegister == 8);
static constexpr auto kF16RegistersPerIterationShift = kF16ElementsPerIterationShift - kF16ElementsPerRegisterShift;
static constexpr auto kF16RegistersPerIteration = 1 << kF16RegistersPerIterationShift;
static_assert(kF16RegistersPerIteration == kF16ElementsPerIteration / kF16ElementsPerRegister);
// We need the shift for reduce(), hence the extra constants.
static constexpr auto kF32ElementsPerIterationShift = 5;
static constexpr auto kF32ElementsPerIteration = 1 << kF32ElementsPerIterationShift;
static_assert(kF32ElementsPerIteration == 32);
static constexpr auto kF32ElementsPerRegisterShift = 2;
static constexpr auto kF32ElementsPerRegister = 1 << kF32ElementsPerRegisterShift;
static_assert(kF32ElementsPerRegister == 4);
static constexpr auto kF32RegisterPairsPerIteration = 4;
static constexpr auto kF32RegistersPerIteration = kF32RegisterPairsPerIteration * 2;
static constexpr auto kF32RegistersPerIterationShift = 3;
static_assert(kF32RegistersPerIteration == kF32ElementsPerIteration / kF32ElementsPerRegister);
static_assert(kF32RegistersPerIteration == 1 << kF32RegistersPerIterationShift);
static float32x4_t to_bfloat16(uint16x4_t u16) {
int32x4_t shift = vdupq_n_s32(16);
return vreinterpretq_f32_u32(vshlq_u32(vmovl_u16(u16), shift));
}
static inline float32x4_t f32_fma(float32x4_t a, float32x4_t b, float32x4_t c) {
#ifdef __ARM_FEATURE_FMA
return vfmaq_f32(a, b, c);
#else
return vaddq_f32(a, vmulq_f32(b, c));
#endif
}
static float32x4_t f32_fma_bf16(float32x4_t a, uint16x4_t b, uint16x4_t c) {
return f32_fma(a, to_bfloat16(b), to_bfloat16(c));
}
#if defined(__clang__) && __clang_major__ > 15
// https://godbolt.org/z/z8P4Yncra
#define COMPILER_SUPPORTS_BF16_TARGET 1
#elif !defined(__clang__) && defined(__GNUC__) && __GNUC__ >= 10
// https://gcc.gnu.org/gcc-10/changes.html
// https://godbolt.org/z/cdGG7vn8o
#define COMPILER_SUPPORTS_BF16_TARGET 1
#else
#define COMPILER_SUPPORTS_BF16_TARGET 0
#endif
#if COMPILER_SUPPORTS_BF16_TARGET
#define TARGET_ARM_BF16_ATTRIBUTE __attribute__((target("arch=armv8.2-a+bf16")))
TARGET_ARM_BF16_ATTRIBUTE static C10_ALWAYS_INLINE float32x4_t
f32_dot_bf16(float32x4_t a, bfloat16x8_t b, bfloat16x8_t c) {
return vbfdotq_f32(a, b, c);
}
TARGET_ARM_BF16_ATTRIBUTE static C10_ALWAYS_INLINE void
dot_with_fp32_arith_main_inner_loop_bfdot(
const BFloat16* vec1,
const BFloat16* vec2,
vec::VectorizedN<float, kF32RegistersPerIteration>& sum,
int registerPairIndex) {
const bfloat16x8_t temp_vec1 = vld1q_bf16(reinterpret_cast<const __bf16*>(
&vec1[registerPairIndex * 2 * vec::Vectorized<float>::size()]));
const bfloat16x8_t temp_vec2 = vld1q_bf16(reinterpret_cast<const __bf16*>(
&vec2[registerPairIndex * 2 * vec::Vectorized<float>::size()]));
sum[registerPairIndex] =
f32_dot_bf16(sum[registerPairIndex], temp_vec1, temp_vec2);
}
// See NOTE [GCC code duplication] below for why we have _bfdot and
// _no_bfdot versions of
// dot_with_fp32_arith_vectorized_tail_inner_loop.
TARGET_ARM_BF16_ATTRIBUTE C10_ALWAYS_INLINE
static void dot_with_fp32_arith_vectorized_tail_inner_loop_bfdot(
const at::BFloat16* vec1,
const at::BFloat16* vec2,
vec::Vectorized<float>* tail_sum,
int idx) {
const auto temp_vec1 = vld1_u16(reinterpret_cast<const uint16_t*>(&vec1[idx]));
const auto temp_vec2 = vld1_u16(reinterpret_cast<const uint16_t*>(&vec2[idx]));
*tail_sum = f32_fma_bf16(*tail_sum, temp_vec1, temp_vec2);
}
#else
#define TARGET_ARM_BF16_ATTRIBUTE
#endif // COMPILER_SUPPORTS_BF16_TARGET
static C10_ALWAYS_INLINE void dot_with_fp32_arith_main_inner_loop_no_bfdot(
const BFloat16* vec1,
const BFloat16* vec2,
vec::VectorizedN<float, kF32RegistersPerIteration>& sum,
int registerPairIndex) {
const uint16x8_t temp_vec1 = vld1q_u16(reinterpret_cast<const uint16_t*>(
&vec1[registerPairIndex * 2 * vec::Vectorized<float>::size()]));
const uint16x8_t temp_vec2 = vld1q_u16(reinterpret_cast<const uint16_t*>(
&vec2[registerPairIndex * 2 * vec::Vectorized<float>::size()]));
sum[2 * registerPairIndex] = f32_fma_bf16(
sum[2 * registerPairIndex],
vget_low_u16(temp_vec1),
vget_low_u16(temp_vec2));
sum[2 * registerPairIndex + 1] = f32_fma_bf16(
sum[2 * registerPairIndex + 1],
vget_high_u16(temp_vec1),
vget_high_u16(temp_vec2));
}
static C10_ALWAYS_INLINE void dot_with_fp32_arith_vectorized_tail_inner_loop_no_bfdot(
const at::BFloat16* vec1,
const at::BFloat16* vec2,
vec::Vectorized<float>* tail_sum,
int idx) {
const auto temp_vec1 = vld1_u16(reinterpret_cast<const uint16_t*>(&vec1[idx]));
const auto temp_vec2 = vld1_u16(reinterpret_cast<const uint16_t*>(&vec2[idx]));
*tail_sum = f32_fma_bf16(*tail_sum, temp_vec1, temp_vec2);
}
namespace {
static inline float reduce(vec::VectorizedN<float, kF32RegistersPerIteration>& x) {
int offset = kF32RegistersPerIteration;
c10::ForcedUnroll<kF32RegistersPerIterationShift>{}([&offset, &x](auto idx) {
offset /= 2;
for (int i = 0; i < offset; ++i) {
x[i] = vaddq_f32(x[i], x[offset + i]);
}
});
return vaddvq_f32(x[0]);
}
#if COMPILER_SUPPORTS_BF16_TARGET
template <int n>
struct ForcedUnrollTargetBFloat16 {
template <typename Func>
TARGET_ARM_BF16_ATTRIBUTE C10_ALWAYS_INLINE void operator()(const Func& f) const {
ForcedUnrollTargetBFloat16<n - 1>{}(f);
f(n - 1);
}
};
template <>
struct ForcedUnrollTargetBFloat16<1> {
template <typename Func>
TARGET_ARM_BF16_ATTRIBUTE C10_ALWAYS_INLINE void operator()(const Func& f) const {
f(0);
}
};
C10_ALWAYS_INLINE TARGET_ARM_BF16_ATTRIBUTE auto
dot_with_fp32_arith_main_loop_bfdot(
const BFloat16* vec1,
const BFloat16* vec2,
int64_t len) {
vec::VectorizedN<float, kF32RegistersPerIteration> sum(0);
const auto len_aligned = len & ~(kF32ElementsPerIteration - 1);
for (int j = 0; j < len_aligned ; j += kF32ElementsPerIteration) {
const auto* vec1_ = vec1 + j;
const auto* vec2_ = vec2 + j;
ForcedUnrollTargetBFloat16<kF32RegisterPairsPerIteration>{}([vec1_, vec2_, &sum](auto k)
C10_ALWAYS_INLINE_ATTRIBUTE TARGET_ARM_BF16_ATTRIBUTE {
dot_with_fp32_arith_main_inner_loop_bfdot(vec1_, vec2_, sum, k);
});
}
return reduce(sum);
}
#endif // COMPILER_SUPPORTS_BF16_TARGET
template <typename T>
C10_ALWAYS_INLINE auto
dot_with_fp32_arith_main_loop_no_bfdot(
const T* vec1,
const T* vec2,
int64_t len) {
vec::VectorizedN<float, kF32RegistersPerIteration> sum(0);
const auto len_aligned = len & ~(kF32ElementsPerIteration - 1);
for (int j = 0; j < len_aligned ; j += kF32ElementsPerIteration) {
const auto* vec1_ = vec1 + j;
const auto* vec2_ = vec2 + j;
c10::ForcedUnroll<kF32RegisterPairsPerIteration>{}([vec1_, vec2_, &sum](auto k) C10_ALWAYS_INLINE_ATTRIBUTE {
dot_with_fp32_arith_main_inner_loop_no_bfdot(vec1_, vec2_, sum, k);
});
}
return reduce(sum);
}
template <typename T>
struct half_to_float16 {
using type = T;
};
#ifdef __aarch64__
template <>
struct half_to_float16<Half> {
using type = float16_t;
};
#endif
template <typename T>
using half_to_float16_t = typename half_to_float16<T>::type;
// NOTE [GCC code duplication]: The first attempt at landing BFDOT support with
// TARGET_ARM_BF16_ATTRIBUTE failed because unlike clang, GCC will not
// allow inlining a non-bf16-specific function into a bf16-specific
// function. We can work around this by duplicating the code into the
// bfdot and non-bfdot callsites. The code is in this macro to avoid
// actual copy/paste.
#define DOT_WITH_FP32_ARITH_TAIL_AFTER_MAIN_LOOP_BODY(bfdot_suffix) \
/* First-tier tail fixup: make sure we handle workloads that can */ \
/* benefit from vectorization, but don't fit into our fully unrolled */ \
/* loop above. */ \
vec::Vectorized<float> tail_sum(0); \
const auto len_aligned = len & ~(kF32ElementsPerIteration - 1); \
const auto len_aligned_4 = len & ~3; \
for (int j = len_aligned; j < len_aligned_4; j += 4) { \
dot_with_fp32_arith_vectorized_tail_inner_loop##bfdot_suffix(vec1, vec2, &tail_sum, j); \
} \
auto reduced_tail = vpaddq_f32(tail_sum, tail_sum); \
reduced_sum += vgetq_lane_f32(vpaddq_f32(reduced_tail, reduced_tail), 0); \
\
/* Second-tier tail fixup: handle all workloads. */ \
for (int j = len_aligned_4; j < len; ++j) { \
/* We use half_to_float16_t here because changing to Half was */ \
/* causing arithmetic to at fp16 precision, but the necessary */ \
/* necessary behavior to pass python test/test_mps.py -k */ \
/* test_output_grad_match_nn_functional_linear_cpu_float16 is */ \
/* fp32. (I'm not sure exactly why this fixes it.) */ \
half_to_float16_t<std::decay_t<decltype(vec1[j])>> x1 = vec1[j]; \
half_to_float16_t<std::decay_t<decltype(vec2[j])>> x2 = vec2[j]; \
reduced_sum += x1 * x2; \
} \
return reduced_sum
#if COMPILER_SUPPORTS_BF16_TARGET
TARGET_ARM_BF16_ATTRIBUTE float
dot_with_fp32_arith_bfdot(const BFloat16* vec1, const BFloat16* vec2, int64_t len) {
auto reduced_sum = dot_with_fp32_arith_main_loop_bfdot(vec1, vec2, len);
DOT_WITH_FP32_ARITH_TAIL_AFTER_MAIN_LOOP_BODY(_bfdot);
}
#endif // COMPILER_SUPPORTS_BF16_TARGET
template <typename T>
C10_ALWAYS_INLINE float
dot_with_fp32_arith_no_bfdot(const T* vec1, const T* vec2, int64_t len) {
auto reduced_sum = dot_with_fp32_arith_main_loop_no_bfdot(vec1, vec2, len);
DOT_WITH_FP32_ARITH_TAIL_AFTER_MAIN_LOOP_BODY(_no_bfdot);
}
#undef DOT_WITH_FP32_ARITH_TAIL_AFTER_MAIN_LOOP_BODY
} // namespace
float bf16_dot_with_fp32_arith(const at::BFloat16* vec1, const at::BFloat16* vec2, int64_t len) {
#if COMPILER_SUPPORTS_BF16_TARGET
if (cpuinfo_has_arm_bf16()) {
return dot_with_fp32_arith_bfdot(vec1, vec2, len);
} else
#endif
{
return dot_with_fp32_arith_no_bfdot(vec1, vec2, len);
}
}
static void bf16_gemv_trans_fp32_arith_by_dot_products(const int m, const int n, const at::BFloat16* a, const int lda, const at::BFloat16 *x, at::BFloat16* y, int incy) {
parallel_for(0, n, 1, [&](int begin, int end) {
for (int i = begin; i < end; ++i) {
y[i * incy] = bf16_dot_with_fp32_arith(x, a + lda * i, m);
}
});
}
void bf16_gemv_trans(
const int m,
const int n,
const at::BFloat16 alpha,
const at::BFloat16* a,
const int lda,
const at::BFloat16* x,
const int incx,
const at::BFloat16 beta,
at::BFloat16* y,
const int incy) {
TORCH_INTERNAL_ASSERT_DEBUG_ONLY(incx == 1 && alpha == 1.0 && beta == 0.0);
return bf16_gemv_trans_fp32_arith_by_dot_products(m, n, a, lda, x, y, incy);
}
#ifdef __ARM_FEATURE_FP16_SCALAR_ARITHMETIC
static void fp16_gemv_notrans_fp16_arith(int m, int n, const float16_t* a, const int lda, const float16_t *x, float16_t *y) {
for (auto j = 0; j < n; j++) {
auto vecCol = vdup_n_f16(x[j]);
const auto* column = a + lda * j;
for (auto i = 0; i < m; i += 4) {
auto yf16 = y + i;
auto matRow = vld1_f16(column + i);
auto resVec = j != 0 ? vld1_f16(yf16) : vdup_n_f16(0);
resVec = vfma_lane_f16(resVec, matRow, vecCol, 0);
vst1_f16(yf16, resVec);
}
}
}
#endif
static void fp16_gemv_notrans_fp32_arith(int m, int n, const float16_t* a, const int lda, const float16_t *x, float16_t *y) {
std::vector<float> sum(m);
for (auto j = 0; j < n; j++) {
auto vecCol = vdup_n_f32(x[j]);
const auto* column = a + lda * j;
for (auto i = 0; i < m; i += 4) {
auto sf32 = sum.data() + i;
auto matRow = vcvt_f32_f16(vld1_f16(column + i));
auto resVec = j != 0 ? vld1q_f32(sf32) : vdupq_n_f32(0);
resVec = vfmaq_lane_f32(resVec, matRow, vecCol, 0);
vst1q_f32(sf32, resVec);
}
}
for (auto i = 0; i < m; i+= 4) {
vst1_f16(y + i, vcvt_f16_f32(vld1q_f32(sum.data() + i)));
}
}
void fp16_gemv_notrans(
const int m,
const int n,
const float alpha,
const Half* a,
const int lda,
const Half* x,
const int incx,
const float beta,
Half* y,
const int incy) {
if (incx == 1 && alpha == 1.0 && beta == 0.0 && m % 4 == 0 && incy == 1) {
#ifdef __ARM_FEATURE_FP16_SCALAR_ARITHMETIC
if (at::globalContext().allowFP16ReductionCPU()) {
return fp16_gemv_notrans_fp16_arith(m, n, reinterpret_cast<const float16_t*>(a), lda, reinterpret_cast<const float16_t*>(x), reinterpret_cast<float16_t*>(y));
}
#endif
return fp16_gemv_notrans_fp32_arith(m, n, reinterpret_cast<const float16_t*>(a), lda, reinterpret_cast<const float16_t*>(x), reinterpret_cast<float16_t*>(y));
}
std::vector<float> sum(m);
for (const auto j : c10::irange(n)) {
const auto* column_ = a + lda * j;
auto z = alpha * x[j * incx];
for (const auto i : c10::irange(m)) {
sum[i] += z * column_[i];
}
}
if (beta == 0.0) {
for (const auto i : c10::irange(m)) {
y[i * incy] = sum[i];
}
} else {
for (const auto i : c10::irange(m)) {
y[i * incy] += sum[i];
}
}
}
template <>
void gemv_fast_path<at::Half>(
const char* trans,
const int* m,
const int* n,
const at::Half* alpha,
const at::Half* a,
const int* lda,
const at::Half* x,
const int* incx,
const at::Half* beta,
at::Half* y,
const int* incy) {
using namespace c10::detail;
if ((trans[0] == 'T') || (trans[0] == 't')) {
fp16_gemv_trans(
*m,
*n,
*alpha,
a,
*lda,
x,
*incx,
*beta,
y,
*incy);
} else {
fp16_gemv_notrans(
*m,
*n,
*alpha,
a,
*lda,
x,
*incx,
*beta,
y,
*incy);
}
}
template <>
void gemv_fast_path<at::BFloat16>(
const char* trans,
const int* m,
const int* n,
const at::BFloat16* alpha,
const at::BFloat16* a,
const int* lda,
const at::BFloat16* x,
const int* incx,
const at::BFloat16* beta,
at::BFloat16* y,
const int* incy) {
TORCH_INTERNAL_ASSERT_DEBUG_ONLY(trans[0] == 'T' || trans[0] == 't');
bf16_gemv_trans(
*m,
*n,
*alpha,
a,
*lda,
x,
*incx,
*beta,
y,
*incy);
}
// Note that the above block was an else, so it's active if __aarch64__ *is* defined.
#endif // !defined(__aarch64__)
#else // !defined(C10_MOBILE))
INSTANTIATE(c10::Half)
INSTANTIATE(c10::BFloat16)
#endif // !defined(C10_MOBILE)
#undef INSTANTIATE
} // namespace blas_impl
template <typename scalar_t>
inline void scal(int64_t n, scalar_t a, scalar_t *x, int64_t incx)
{
if (n == 1) incx = 1;
#if AT_BUILD_WITH_BLAS()
if (blas_impl::scal_use_fast_path<scalar_t>(n, incx)) {
int i_n = (int)n;
int i_incx = (int)incx;
blas_impl::scal_fast_path<scalar_t>(&i_n, &a, x, &i_incx);
return;
}
#endif
for (const auto i : c10::irange(n)) {
if (a == scalar_t(0)) {
x[i * incx] = 0;
} else {
x[i * incx] *= a;
}
}
}
template<typename scalar_t>
void gemv(char trans, int64_t m, int64_t n, scalar_t alpha, const scalar_t *a, int64_t lda, const scalar_t *x, int64_t incx, scalar_t beta, scalar_t *y, int64_t incy) {
if(n == 1) lda = m;
#if AT_BUILD_WITH_BLAS()
if (blas_impl::gemv_use_fast_path<scalar_t>(trans, m, n, alpha, lda, incx, beta, incy)) {
TORCH_CHECK(lda >= std::max<int64_t>(1L, m), "lda should be at least max(1,", m, "), but have ", lda);
int i_m = (int)m;
int i_n = (int)n;
int i_lda = (int)lda;
int i_incx = (int)incx;
int i_incy = (int)incy;
blas_impl::gemv_fast_path<scalar_t>(&trans, &i_m, &i_n, &alpha, a, &i_lda, x, &i_incx, &beta, y, &i_incy);
return;
}
#endif
using opmath_t = at::opmath_type<scalar_t>;
if ((trans == 'T') || (trans == 't')) {
for (const auto i : c10::irange(n)) {
opmath_t sum = 0;
const scalar_t *row_ = a + lda * i;
for (const auto j : c10::irange(m)) {
sum += x[j * incx] * row_[j];
}
if (beta == scalar_t(0)) {
y[i * incy] = alpha * sum;
} else {
y[i * incy] = beta * y[i * incy] + alpha * sum;
}
}
} else {
if (beta != scalar_t(1) && beta != scalar_t(0)) scal<scalar_t>(m, beta, y, incy);
constexpr bool is_low_precision = !std::is_same_v<opmath_t, scalar_t>;
std::vector<opmath_t> sum;
if constexpr (is_low_precision) {
sum.resize(m);
}
for (const auto j : c10::irange(n)) {
const scalar_t *column_ = a + lda * j;
opmath_t z = alpha * static_cast<opmath_t>(x[j * incx]);
for (const auto i : c10::irange(m)) {
//output values are ignored if beta is 0, and set to 0, nans and infs are not propagated
if (j==0 && beta==scalar_t(0)) {
if constexpr (!is_low_precision) {
y[i * incy] = 0;
}
}
if constexpr (is_low_precision) {
sum[i] += z * column_[i];
} else {
y[i * incy] += z * column_[i];
}
}
}
if constexpr (is_low_precision) {
if (beta == scalar_t(0)) {
for (const auto i : c10::irange(m)) {
y[i * incy] = sum[i];
}
} else {
for (const auto i : c10::irange(m)) {
y[i * incy] += sum[i];
}
}
}
}
return;
}
#define INSTANTIATE(scalar_t, _) \
template void gemv<scalar_t>(char trans, int64_t m, int64_t n, scalar_t alpha, const scalar_t *a, int64_t lda, const scalar_t *x, int64_t incx, scalar_t beta, scalar_t *y, int64_t incy);
AT_FORALL_SCALAR_TYPES_AND2(BFloat16, Half, INSTANTIATE)
AT_FORALL_COMPLEX_TYPES(INSTANTIATE)
#undef INSTANTIATE
namespace blas_impl {
#if AT_BUILD_WITH_BLAS()
static float dot_fast_path(int n, float* x, int incx, float* y, int incy) {
// NOLINTNEXTLINE(bugprone-narrowing-conversions,cppcoreguidelines-narrowing-conversions)
return sdot_(&n, x, &incx, y, &incy);
}
static double dot_fast_path(int n, double* x, int incx, double* y, int incy) {
return ddot_(&n, x, &incx, y, &incy);
}
static c10::complex<float> vdot_fast_path(int n, c10::complex<float>* x, int incx, c10::complex<float>* y, int incy) {
c10::complex<float> result;
cdotc_(reinterpret_cast<std::complex<float>* >(&result), &n, reinterpret_cast<std::complex<float>*>(x), &incx, reinterpret_cast<std::complex<float>*>(y), &incy);
return result;
}
static c10::complex<double> vdot_fast_path(int n, c10::complex<double>* x, int incx, c10::complex<double>* y, int incy) {
c10::complex<double> result;
zdotc_(reinterpret_cast<std::complex<double>* >(&result), &n, reinterpret_cast<std::complex<double>*>(x), &incx, reinterpret_cast<std::complex<double>*>(y), &incy);
return result;
}
static c10::complex<double> dot_fast_path(int n, c10::complex<double>* x, int incx, c10::complex<double>* y, int incy) {
c10::complex<double> result;
zdotu_(reinterpret_cast<std::complex<double>* >(&result), &n, reinterpret_cast<std::complex<double>*>(x), &incx, reinterpret_cast<std::complex<double>*>(y), &incy);
return result;
}
static c10::complex<float> dot_fast_path(int n, c10::complex<float>* x, int incx, c10::complex<float>* y, int incy) {
c10::complex<float> result;
cdotu_(reinterpret_cast<std::complex<float>* >(&result), &n, reinterpret_cast<std::complex<float>*>(x), &incx, reinterpret_cast<std::complex<float>*>(y), &incy);
return result;
}
#endif
template <typename scalar_t, typename Functor>
scalar_t dot_naive(
int64_t n,
scalar_t* x,
int64_t incx,
scalar_t* y,
int64_t incy,
Functor op) {
// NOLINTNEXTLINE(cppcoreguidelines-init-variables)
int64_t i;
using opmath_t = at::opmath_type<scalar_t>;
opmath_t sum = 0;
for (i = 0; i < n; i++) {
sum += op(static_cast<opmath_t>(x[i * incx]), static_cast<opmath_t>(y[i * incy]));
}
return static_cast<scalar_t>(sum);
}
} // namespace blas_impl
template <typename scalar_t>
scalar_t dot_impl_floating(int64_t n, scalar_t* x, int64_t incx, scalar_t* y, int64_t incy)
{
if (n == 1) {
incx = 1;
incy = 1;
}
#if AT_BUILD_WITH_BLAS()
if ((n <= INT_MAX) && (incx <= INT_MAX) && (incy <= INT_MAX)) {
return blas_impl::dot_fast_path(n, x, incx, y, incy);
} else {
return blas_impl::dot_naive(n, x, incx, y, incy, std::multiplies<scalar_t>{});
}
#else
{ return blas_impl::dot_naive(n, x, incx, y, incy, std::multiplies<scalar_t>{}); }
#endif
}
template <typename scalar_t>
scalar_t dot_impl(int64_t n, scalar_t* x, int64_t incx, scalar_t* y, int64_t incy) {
if (n == 1) {
incx = 1;
incy = 1;
}
return blas_impl::dot_naive(n, x, incx, y, incy, std::multiplies<scalar_t>{});
}
template <>
float dot_impl(int64_t n, float* x, int64_t incx, float* y, int64_t incy) {
return dot_impl_floating(n, x, incx, y, incy);
}
template <>