-
Notifications
You must be signed in to change notification settings - Fork 40
/
blaslapack.c
1937 lines (1556 loc) · 59.5 KB
/
blaslapack.c
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
/*******************************************************************************
* Copyright (c) 2018, College of William & Mary
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* * Neither the name of the College of William & Mary nor the
* names of its contributors may be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE COLLEGE OF WILLIAM & MARY BE LIABLE FOR ANY
* DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
* ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* PRIMME: https://github.com/primme/primme
* Contact: Andreas Stathopoulos, a n d r e a s _at_ c s . w m . e d u
*******************************************************************************
* File: blaslapack.c
*
* Purpose - This file contains mostly C wrapper routines for
* calling various BLAS and LAPACK FORTRAN routines.
*
******************************************************************************/
#define THIS_FILE "../linalg/blaslapack.c"
#include <string.h> /* memmove */
#include <assert.h>
#include <math.h>
#include "numerical.h"
#ifdef SUPPORTED_TYPE
#ifdef USE_HOST
#include "blaslapack_private.h"
#ifdef USE_DOUBLE
static int free_fn_dummy (void *p, primme_context ctx) {
(void)ctx;
free(p);
return 0;
}
static int to_blas_int(PRIMME_INT i, PRIMME_BLASINT *out) {
// If the matrix size is too large for the current BLAS, report an error
if (i > PRIMME_BLASINT_MAX) return PRIMME_UNEXPECTED_FAILURE;
*out = (PRIMME_BLASINT)i;
return 0;
}
#endif
/******************************************************************************
* Function Num_check_pointer - Return no error code if the pointer is on host.
*
* NOTE: Any check is performed for now.
*
* PARAMETERS
* ---------------------------
* x pointer to check
*
******************************************************************************/
TEMPLATE_PLEASE
int Num_check_pointer_Sprimme(void *x) {
(void)x;
return 0;
}
/******************************************************************************
* Function Num_malloc_Sprimme - Allocate a vector of scalars
*
* PARAMETERS
* ---------------------------
* n The number of elements
* v returned pointer
*
******************************************************************************/
TEMPLATE_PLEASE
int Num_malloc_Sprimme(PRIMME_INT n, SCALAR **x, primme_context ctx) {
(void)ctx;
/* Quick exit */
if (n <= 0) {
*x = NULL;
return 0;
}
/* Allocate memory */
*x = (SCALAR *)malloc(sizeof(SCALAR) * n);
if (*x == NULL) return PRIMME_MALLOC_FAILURE;
/* Register the allocation */
Mem_keep_frame(ctx);
Mem_register_alloc(*x, free_fn_dummy, ctx);
return 0;
}
/******************************************************************************
* Function Num_free_Sprimme - Free allocated a vector of scalars
*
* PARAMETERS
* ---------------------------
* v allocated pointer
*
******************************************************************************/
TEMPLATE_PLEASE
int Num_free_Sprimme(SCALAR *x, primme_context ctx) {
/* Quick exit */
if (!x) return 0;
/* Deregister the allocation */
Mem_deregister_alloc(x, ctx);
/* Free pointer */
free(x);
return 0;
}
#ifdef USE_DOUBLE
/******************************************************************************
* Function Num_malloc_iprimme - Allocate a vector of integers
*
* PARAMETERS
* ---------------------------
* n The number of elements
* v returned pointer
*
******************************************************************************/
TEMPLATE_PLEASE
int Num_malloc_iprimme(PRIMME_INT n, int **x, primme_context ctx) {
(void)ctx;
/* Quick exit */
if (n <= 0) {
*x = NULL;
return 0;
}
/* Allocate memory */
*x = (int *)malloc(sizeof(int) * n);
if (*x == NULL) return PRIMME_MALLOC_FAILURE;
/* Register the allocation */
Mem_keep_frame(ctx);
Mem_register_alloc(*x, free_fn_dummy, ctx);
return 0;
}
/******************************************************************************
* Function Num_free_iprimme - Free allocated a vector of integers
*
* PARAMETERS
* ---------------------------
* v allocated pointer
*
******************************************************************************/
TEMPLATE_PLEASE
int Num_free_iprimme(int *x, primme_context ctx) {
/* Quick exit */
if (!x) return 0;
/* Deregister the allocation */
Mem_deregister_alloc(x, ctx);
/* Free pointer */
free(x);
return 0;
}
/******************************************************************************
* Function Num_malloc_iblasprimme - Allocate a vector of integers
*
* PARAMETERS
* ---------------------------
* n The number of elements
* v returned pointer
*
******************************************************************************/
static int Num_malloc_iblasprimme(
PRIMME_INT n, PRIMME_BLASINT **x, primme_context ctx) {
(void)ctx;
/* Quick exit */
if (n <= 0) {
*x = NULL;
return 0;
}
/* Allocate memory */
*x = (PRIMME_BLASINT *)malloc(sizeof(PRIMME_BLASINT) * n);
if (*x == NULL) return PRIMME_MALLOC_FAILURE;
/* Register the allocation */
Mem_keep_frame(ctx);
Mem_register_alloc(*x, free_fn_dummy, ctx);
return 0;
}
/******************************************************************************
* Function Num_free_iblasprimme - Free allocated a vector of integers
*
* PARAMETERS
* ---------------------------
* v allocated pointer
*
******************************************************************************/
static int Num_free_iblasprimme(PRIMME_BLASINT *x, primme_context ctx) {
/* Quick exit */
if (!x) return 0;
/* Deregister the allocation */
Mem_deregister_alloc(x, ctx);
/* Free pointer */
free(x);
return 0;
}
#endif /* USE_DOUBLE */
/*******************************************************************************
* Subroutine Num_copy_Sprimme - y(0:n*incy-1:incy) = x(0:n*incx-1:incx)
******************************************************************************/
TEMPLATE_PLEASE
int Num_copy_Sprimme(PRIMME_INT n, SCALAR *x, PRIMME_INT incx, SCALAR *y,
PRIMME_INT incy, primme_context ctx) {
/* Quick exit */
if (x == y && incx == incy) return 0;
#if (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF)
(void)ctx;
PRIMME_BLASINT ln, lincx, lincy;
CHKERR(to_blas_int(n, &ln));
CHKERR(to_blas_int(incx, &lincx));
CHKERR(to_blas_int(incy, &lincy));
XCOPY(&ln, (const BLASSCALAR *)x, &lincx, (BLASSCALAR *)y, &lincy);
return 0;
#else
return Num_copy_matrix_Sprimme(x, 1, n, incx, y, incy, ctx);
#endif
}
/******************************************************************************
* Function Num_copy_matrix_Tprimme - Copy the matrix x into y. The types of
* x and y can be different.
*
* PARAMETERS
* ---------------------------
* x The source matrix
* m The number of rows of x
* n The number of columns of x
* ldx The leading dimension of x
* y On output y = x
* ldy The leading dimension of y
*
* NOTE: x and y *can* overlap
*
******************************************************************************/
#define Num_copy_matrix_Tprimme(x, xcast, m, n, ldx, y, ldy, ctx) \
{ \
PRIMME_INT i, j; \
\
assert((m) == 0 || (n) == 0 || ((ldx) >= (m) && (ldy) >= (m))); \
\
if ((void *)x != (void *)y) \
for (i = 0; i < (n); i++) \
for (j = 0; j < (m); j++) \
(y)[i * (ldy) + j] = xcast(x)[i * (ldx) + j]; \
}
/******************************************************************************
* Function Num_copy_Tmatrix - Copy the matrix x with type given as a parameter
* into y
*
* PARAMETERS
* ---------------------------
* x The source matrix
* xt The type of x
* m The number of rows of x
* n The number of columns of x
* ldx The leading dimension of x
* y On output y = x
* ldy The leading dimension of y
*
* NOTE: x and y *can* overlap
*
******************************************************************************/
TEMPLATE_PLEASE
int Num_copy_Tmatrix_Sprimme(void *x, primme_op_datatype xt, PRIMME_INT m,
PRIMME_INT n, PRIMME_INT ldx, SCALAR *y, PRIMME_INT ldy,
primme_context ctx) {
(void)ctx;
/* Quick exit */
if (xt == primme_op_default || xt == PRIMME_OP_SCALAR) {
CHKERR(Num_copy_matrix_Sprimme((SCALAR*)x, m, n, ldx, y, ldy, ctx));
return 0;
}
if (m == 0 || n == 0) return 0;
/* In-place casting is not supported */
if (x == y) return PRIMME_FUNCTION_UNAVAILABLE;
#ifdef USE_COMPLEX
return Num_copy_Tmatrix_Rprimme(
x, xt, m * 2, n, ldx * 2, (REAL *)y, ldy * 2, ctx);
#else
#if defined(USE_HALF) || defined(USE_HALFCOMPLEX)
# define CAST (float)
#else
# define CAST
#endif
switch (xt) {
# ifdef SUPPORTED_HALF_TYPE
case primme_op_half: Num_copy_matrix_Tprimme((PRIMME_HALF*)x, CAST, m, n, ldx, (XSCALAR*)y, ldy, ctx); break;
# endif
case primme_op_float: Num_copy_matrix_Tprimme((float*) x, CAST, m, n, ldx, (XSCALAR*)y, ldy, ctx); break;
case primme_op_double: Num_copy_matrix_Tprimme((double*) x, CAST, m, n, ldx, (XSCALAR*)y, ldy, ctx); break;
case primme_op_quad: Num_copy_matrix_Tprimme((PRIMME_QUAD*)x, CAST, m, n, ldx, (XSCALAR*)y, ldy, ctx); break;
case primme_op_int: Num_copy_matrix_Tprimme((int*) x, CAST, m, n, ldx, (XSCALAR*)y, ldy, ctx); break;
default: CHKERR(PRIMME_FUNCTION_UNAVAILABLE);
}
#undef CAST
return 0;
#endif /* USE_COMPLEX */
}
#ifdef USE_DOUBLE
TEMPLATE_PLEASE
int Num_copy_Tmatrix_iprimme(void *x, primme_op_datatype xt, PRIMME_INT m,
PRIMME_INT n, PRIMME_INT ldx, int *y, PRIMME_INT ldy,
primme_context ctx) {
(void)ctx;
#if defined(USE_HALF) || defined(USE_HALFCOMPLEX)
# define CAST (float)
#else
# define CAST
#endif
switch (xt) {
# ifdef SUPPORTED_HALF_TYPE
case primme_op_half: Num_copy_matrix_Tprimme((PRIMME_HALF*)x, CAST, m, n, ldx, y, ldy, ctx); break;
# endif
case primme_op_float: Num_copy_matrix_Tprimme((float*) x, CAST, m, n, ldx, y, ldy, ctx); break;
case primme_op_double: Num_copy_matrix_Tprimme((double*) x, CAST, m, n, ldx, y, ldy, ctx); break;
case primme_op_quad: Num_copy_matrix_Tprimme((PRIMME_QUAD*)x, CAST, m, n, ldx, y, ldy, ctx); break;
case primme_op_int: Num_copy_matrix_Tprimme((int*) x, CAST, m, n, ldx, y, ldy, ctx); break;
default: CHKERR(PRIMME_FUNCTION_UNAVAILABLE);
}
#undef CAST
return 0;
}
#endif /* USE_DOUBLE */
/******************************************************************************
* Function Num_copy_matrix - Copy the matrix x into y
*
* PARAMETERS
* ---------------------------
* x The source matrix
* m The number of rows of x
* n The number of columns of x
* ldx The leading dimension of x
* y On output y = x
* ldy The leading dimension of y
*
* NOTE: x and y *can* overlap
*
******************************************************************************/
TEMPLATE_PLEASE
int Num_copy_matrix_Sprimme(SCALAR *x, PRIMME_INT m, PRIMME_INT n,
PRIMME_INT ldx, SCALAR *y, PRIMME_INT ldy,
primme_context ctx) {
(void)ctx;
PRIMME_INT i, j;
assert(m == 0 || n == 0 || (ldx >= m && ldy >= m));
/* Do nothing if x and y are the same matrix */
if (x == y && ldx == ldy)
return 0;
/* Do nothing for zero-size matrices */
if (m <= 0 || n <= 0)
return 0;
/* Copy a contiguous memory region */
if (ldx == ldy && ldx == m) {
memmove(y, x, sizeof(SCALAR) * m * n);
}
/* Copy the matrix some rows back or forward */
else if (ldx == ldy && (y > x ? y - x : x - y) < ldx) {
for (i = 0; i < n; i++)
memmove(&y[i * ldy], &x[i * ldx], sizeof(SCALAR) * m);
}
/* Copy the matrix some columns forward */
else if (ldx == ldy && y > x && y - x > ldx) {
for (i = n - 1; i >= 0; i--)
for (j = 0; j < m; j++)
y[i * ldy + j] = x[i * ldx + j];
}
/* Copy the matrix some columns backward, and other cases */
else {
/* TODO: assert x and y don't overlap */
for (i = 0; i < n; i++) {
for (j = 0; j < m; j++) {
y[i * ldy + j] = x[i * ldx + j];
}
}
}
return 0;
}
/******************************************************************************
* Function Num_zero_matrix - Zero the matrix
*
* PARAMETERS
* ---------------------------
* x The matrix
* m The number of rows of x
* n The number of columns of x
* ldx The leading dimension of x
*
******************************************************************************/
TEMPLATE_PLEASE
int Num_zero_matrix_Sprimme(SCALAR *x, PRIMME_INT m, PRIMME_INT n,
PRIMME_INT ldx, primme_context ctx) {
(void)ctx;
PRIMME_INT i,j;
for (i=0; i<n; i++)
for (j=0; j<m; j++)
SET_ZERO(x[i*ldx+j]);
return 0;
}
/*******************************************************************************
* Subroutine Num_gemm_Sprimme - C = op(A)*op(B), with C size m x n
******************************************************************************/
#if (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF)
TEMPLATE_PLEASE
int Num_gemm_Sprimme(const char *transa, const char *transb, PRIMME_INT m,
PRIMME_INT n, PRIMME_INT k, HSCALAR alpha, SCALAR *a, PRIMME_INT lda,
SCALAR *b, PRIMME_INT ldb, HSCALAR beta, SCALAR *c, PRIMME_INT ldc,
primme_context ctx) {
PRIMME_BLASINT lm, ln, lk, llda, lldb, lldc;
CHKERR(to_blas_int(m, &lm));
CHKERR(to_blas_int(n, &ln));
CHKERR(to_blas_int(k, &lk));
CHKERR(to_blas_int(lda, &llda));
CHKERR(to_blas_int(ldb, &lldb));
CHKERR(to_blas_int(ldc, &lldc));
/* Zero dimension matrix may cause problems */
if (m == 0 || n == 0) return 0;
/* Quick exit */
if (k == 0 || ABS(alpha) == 0.0) {
if (ABS(beta) == 0.0) {
Num_zero_matrix_Sprimme(c, m, n, ldc, ctx);
}
else if (beta != (HSCALAR)1.0) {
PRIMME_INT i;
for (i=0; i<n; i++) {
Num_scal_Sprimme(m, beta, &c[ldc*i], 1, ctx);
}
}
return 0;
}
if (n == 1) {
PRIMME_INT mA; PRIMME_INT nA;
if (*transa == 'n' || *transa == 'N') mA = m, nA = k;
else mA = k, nA = m;
PRIMME_INT incb = ((*transb == 'n' || *transb == 'N') ? 1 : ldb);
return Num_gemv_Sprimme(
transa, mA, nA, alpha, a, lda, b, incb, beta, c, 1, ctx);
}
XGEMM(transa, transb, &lm, &ln, &lk, (const BLASSCALAR *)&alpha,
(const BLASSCALAR *)a, &llda, (const BLASSCALAR *)b, &lldb,
(const BLASSCALAR *)&beta, (BLASSCALAR *)c, &lldc);
return 0;
}
#endif /* (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF) */
TEMPLATE_PLEASE
int Num_gemm_dhd_Sprimme(const char *transa, const char *transb, PRIMME_INT m,
PRIMME_INT n, PRIMME_INT k, HSCALAR alpha, SCALAR *a, PRIMME_INT lda,
HSCALAR *b, PRIMME_INT ldb, HSCALAR beta, SCALAR *c, PRIMME_INT ldc,
primme_context ctx) {
/* Quick exit */
if (m == 0 || n == 0 ||
((k == 0 || ABS(alpha) == 0.0) && beta == (HSCALAR)1.0))
return 0;
/* Cast the matrices a and c to HSCALAR */
HSCALAR *af = NULL, *cf = NULL;
PRIMME_INT ldaf, ldcf, ldc0 = ldc;
PRIMME_INT ma = (*transa == 'N' || *transa == 'n') ? m : k;
PRIMME_INT na = (*transa == 'N' || *transa == 'n') ? k : m;
CHKERR(Num_matrix_astype_Sprimme(a, ma, na, lda, PRIMME_OP_SCALAR,
(void **)&af, &ldaf, PRIMME_OP_HSCALAR, 1 /* alloc */, 1 /* copy */,
ctx));
CHKERR(Num_matrix_astype_Sprimme(c, m, n, ldc, PRIMME_OP_SCALAR,
(void **)&cf, &ldcf, PRIMME_OP_HSCALAR, 1 /* alloc */, 1 /* copy */,
ctx));
/* Call the kernel */
CHKERR(Num_gemm_SHprimme(transa, transb, m, n, k, alpha, af, ldaf, b, ldb,
beta, cf, ldcf, ctx));
/* Copy back c and destroy the cast matrices */
if (a != (SCALAR*)af) CHKERR(Num_free_SHprimme(af, ctx));
CHKERR(Num_matrix_astype_Sprimme(cf, m, n, ldcf, PRIMME_OP_HSCALAR,
(void **)&c, &ldc0, PRIMME_OP_SCALAR, -1 /* destroy */, 1 /* copy */,
ctx));
return 0;
}
TEMPLATE_PLEASE
int Num_gemm_ddh_Sprimme(const char *transa, const char *transb, PRIMME_INT m,
PRIMME_INT n, PRIMME_INT k, HSCALAR alpha, SCALAR *a, PRIMME_INT lda,
SCALAR *b, PRIMME_INT ldb, HSCALAR beta, HSCALAR *c, PRIMME_INT ldc,
primme_context ctx) {
/* Quick exit */
if (m == 0 || n == 0 ||
((k == 0 || ABS(alpha) == 0.0) && beta == (HSCALAR)1.0))
return 0;
/* If input matrices are going to be cast and the operation is the inner */
/* product between, that is, C = A'*B, then stream the operation */
PRIMME_INT K = k;
if (PRIMME_OP_SCALAR != PRIMME_OP_HSCALAR &&
(*transa == 'C' || *transa == 'C') &&
(*transb == 'N' || *transb == 'n') && k > PRIMME_BLOCK_SIZE) {
K = PRIMME_BLOCK_SIZE;
}
/* Cast the matrices a and b to HSCALAR */
HSCALAR *af = NULL, *bf = NULL;
PRIMME_INT ldaf, ldbf;
PRIMME_INT na = (*transa == 'N' || *transa == 'n') ? k : m;
PRIMME_INT nb = (*transb == 'N' || *transb == 'n') ? n : k;
PRIMME_INT i;
for (i=0; i<k; i+=K, K=min(K,k-i)) {
CHKERR(Num_matrix_astype_Sprimme(&a[i], K, na, lda, PRIMME_OP_SCALAR,
(void **)&af, &ldaf, PRIMME_OP_HSCALAR,
i == 0 /* alloc the first time */, 1 /* copy */, ctx));
CHKERR(Num_matrix_astype_Sprimme(&b[i], K, nb, ldb, PRIMME_OP_SCALAR,
(void **)&bf, &ldbf, PRIMME_OP_HSCALAR,
i == 0 /* alloc the first time */, 1 /* copy */, ctx));
/* Call the kernel */
CHKERR(Num_gemm_SHprimme(transa, transb, m, n, K, alpha, af, ldaf, bf,
ldbf, beta, c, ldc, ctx));
/* Update beta */
beta = (HSCALAR)1.0;
}
if (a != (SCALAR *)af) CHKERR(Num_free_SHprimme(af, ctx));
if (b != (SCALAR *)bf) CHKERR(Num_free_SHprimme(bf, ctx));
return 0;
}
/*******************************************************************************
* Subroutine Num_hemm_Sprimme - C = A*B or B*A where A is Hermitian,
* where C size m x n.
******************************************************************************/
#if (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF)
TEMPLATE_PLEASE
int Num_hemm_Sprimme(const char *side, const char *uplo, PRIMME_INT m,
PRIMME_INT n, HSCALAR alpha, SCALAR *a, PRIMME_INT lda, SCALAR *b,
PRIMME_INT ldb, HSCALAR beta, SCALAR *c, PRIMME_INT ldc,
primme_context ctx) {
(void)ctx;
PRIMME_BLASINT lm, ln, llda, lldb, lldc;
CHKERR(to_blas_int(m, &lm));
CHKERR(to_blas_int(n, &ln));
CHKERR(to_blas_int(lda, &llda));
CHKERR(to_blas_int(ldb, &lldb));
CHKERR(to_blas_int(ldc, &lldc));
/* Zero dimension matrix may cause problems */
if (m == 0 || n == 0) return 0;
XHEMM(side, uplo, &lm, &ln, (const BLASSCALAR *)&alpha,
(const BLASSCALAR *)a, &llda, (const BLASSCALAR *)b, &lldb,
(const BLASSCALAR *)&beta, (BLASSCALAR *)c, &lldc);
return 0;
}
#endif
/*******************************************************************************
* Subroutine Num_trmm_Sprimme - B = A*B or B*A where A is triangular,
* with B size m x n.
******************************************************************************/
#if (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF)
TEMPLATE_PLEASE
int Num_trmm_Sprimme(const char *side, const char *uplo, const char *transa,
const char *diag, PRIMME_INT m, PRIMME_INT n, HSCALAR alpha, SCALAR *a,
PRIMME_INT lda, SCALAR *b, PRIMME_INT ldb, primme_context ctx) {
(void)ctx;
PRIMME_BLASINT lm, ln, llda, lldb;
CHKERR(to_blas_int(m, &lm));
CHKERR(to_blas_int(n, &ln));
CHKERR(to_blas_int(lda, &llda));
CHKERR(to_blas_int(ldb, &lldb));
/* Zero dimension matrix may cause problems */
if (m == 0 || n == 0) return 0;
XTRMM(side, uplo, transa, diag, &lm, &ln, (const BLASSCALAR *)&alpha,
(const BLASSCALAR *)a, &llda, (BLASSCALAR *)b, &lldb);
return 0;
}
#endif /* (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF) */
/*******************************************************************************
* Subroutine Num_gemv_Sprimme - y = alpha*A*x + beta*y, with A size m x n
******************************************************************************/
#if (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF)
TEMPLATE_PLEASE
int Num_gemv_Sprimme(const char *transa, PRIMME_INT m, PRIMME_INT n,
HSCALAR alpha, SCALAR *a, PRIMME_INT lda, SCALAR *x, PRIMME_INT incx,
HSCALAR beta, SCALAR *y, PRIMME_INT incy, primme_context ctx) {
PRIMME_BLASINT lm, ln, llda, lincx, lincy;
CHKERR(to_blas_int(n, &ln));
CHKERR(to_blas_int(lda, &llda));
CHKERR(to_blas_int(incx, &lincx));
CHKERR(to_blas_int(incy, &lincy));
/* Zero dimension matrix may cause problems */
int tA = (*transa != 'n' && *transa != 'N' ? 1 : 0);
PRIMME_INT mA = tA ? n : m, nA = tA ? m : n;
if (mA == 0) return 0;
/* Quick exit */
if (nA == 0 || ABS(alpha) == 0.0) {
if (ABS(beta) == 0.0) {
Num_zero_matrix_Sprimme(y, 1, mA, incy, ctx);
}
else {
Num_scal_Sprimme(mA, beta, y, incy, ctx);
}
return 0;
}
if (mA == 1) {
if (ABS(beta) != 0.0)
y[0] *= beta;
else
y[0] = 0.0;
if (tA) {
y[0] += Num_dot_Sprimme(nA, a, 1, x, incx, ctx) * alpha;
} else {
y[0] += Num_dot_Sprimme(nA, a, lda, x, incx, ctx) * alpha;
}
return 0;
}
while(m > 0) {
lm = (PRIMME_BLASINT)min(m, PRIMME_BLASINT_MAX-1);
XGEMV(transa, &lm, &ln, (const BLASSCALAR *)&alpha, (const BLASSCALAR *)a,
&llda, (const BLASSCALAR *)x, &lincx, (const BLASSCALAR *)&beta,
(BLASSCALAR *)y, &lincy);
m -= (PRIMME_INT)lm;
a += lm;
if (!tA) {
y += lm;
}
else {
x += lm;
beta = 1.0;
}
}
return 0;
}
#endif /* (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF) */
/*******************************************************************************
* Subroutine Num_gemv_ddh_Sprimme - y = alpha*A*x + beta*y, with A size m x n
* NOTE: A and x are in device and y is in host
******************************************************************************/
TEMPLATE_PLEASE
int Num_gemv_ddh_Sprimme(const char *transa, PRIMME_INT m, PRIMME_INT n,
HSCALAR alpha, SCALAR *a, PRIMME_INT lda, SCALAR *x, PRIMME_INT incx,
HSCALAR beta, HSCALAR *y, PRIMME_INT incy, primme_context ctx) {
/* Cast the matrices a and x to HSCALAR */
HSCALAR *af = NULL, *xf = NULL;
PRIMME_INT ldaf, incxf;
PRIMME_INT mx = (*transa == 'N' || *transa == 'n') ? n : m;
CHKERR(Num_matrix_astype_Sprimme(a, m, n, lda, PRIMME_OP_SCALAR,
(void **)&af, &ldaf, PRIMME_OP_HSCALAR, 1 /* alloc */, 1 /* copy */,
ctx));
CHKERR(Num_matrix_astype_Sprimme(x, 1, mx, incx, PRIMME_OP_SCALAR,
(void **)&xf, &incxf, PRIMME_OP_HSCALAR, 1 /* alloc */, 1 /* copy */,
ctx));
/* Call the kernel */
CHKERR(Num_gemv_SHprimme(
transa, m, n, alpha, af, m, xf, 1, beta, y, incy, ctx));
if (a != (SCALAR*)af) CHKERR(Num_free_SHprimme(af, ctx));
if (x != (SCALAR*)xf) CHKERR(Num_free_SHprimme(xf, ctx));
return 0;
}
/*******************************************************************************
* Subroutine Num_gemv_dhd_Sprimme - y = alpha*A*x + beta*y, with A size m x n
* NOTE: A and y are in device and x is in host
******************************************************************************/
TEMPLATE_PLEASE
int Num_gemv_dhd_Sprimme(const char *transa, PRIMME_INT m, PRIMME_INT n,
HSCALAR alpha, SCALAR *a, PRIMME_INT lda, HSCALAR *x, PRIMME_INT incx,
HSCALAR beta, SCALAR *y, PRIMME_INT incy, primme_context ctx) {
/* Cast the matrices a and y to HSCALAR */
HSCALAR *af = NULL, *yf = NULL;
PRIMME_INT ldaf, incyf, incy0 = incy;
PRIMME_INT my = (*transa == 'N' || *transa == 'n') ? m : n;
CHKERR(Num_matrix_astype_Sprimme(a, m, n, lda, PRIMME_OP_SCALAR,
(void **)&af, &ldaf, PRIMME_OP_HSCALAR, 1 /* alloc */, 1 /* copy */,
ctx));
CHKERR(Num_matrix_astype_Sprimme(y, 1, my, incy, PRIMME_OP_SCALAR,
(void **)&yf, &incyf, PRIMME_OP_HSCALAR, 1 /* alloc */, 1 /* copy */,
ctx));
CHKERR(Num_gemv_SHprimme(
transa, m, n, alpha, af, m, x, incx, beta, yf, incyf, ctx));
/* Copy back y and destroy cast matrices */
if (a != (SCALAR*)af) CHKERR(Num_free_SHprimme(af, ctx));
CHKERR(Num_matrix_astype_Sprimme(yf, 1, my, incyf, PRIMME_OP_HSCALAR,
(void **)&y, &incy0, PRIMME_OP_SCALAR, -1 /* destroy */, 1 /* copy */,
ctx));
return 0;
}
/*******************************************************************************
* Subroutine Num_axpy_Sprimme - y += alpha*x
******************************************************************************/
TEMPLATE_PLEASE
int Num_axpy_Sprimme(PRIMME_INT n, HSCALAR alpha, SCALAR *x, PRIMME_INT incx,
SCALAR *y, PRIMME_INT incy, primme_context ctx) {
(void)ctx;
#if (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF)
PRIMME_BLASINT ln, lincx, lincy;
CHKERR(to_blas_int(incx, &lincx));
CHKERR(to_blas_int(incy, &lincy));
while(n > 0) {
ln = (PRIMME_BLASINT)min(n, PRIMME_BLASINT_MAX-1);
XAXPY(&ln, (const BLASSCALAR *)&alpha, (const BLASSCALAR *)x, &lincx,
(BLASSCALAR *)y, &lincy);
n -= (PRIMME_INT)ln;
x += ln;
y += ln;
}
#else
PRIMME_INT i;
for (i = 0; i < n; i++)
PLUS_EQUAL(y[incy * i], alpha * TO_COMPLEX(x[incx * i]));
#endif
return 0;
}
/*******************************************************************************
* Subroutine Num_dot_Sprimme - y'*x
******************************************************************************/
TEMPLATE_PLEASE
HSCALAR Num_dot_Sprimme(PRIMME_INT n, SCALAR *x, PRIMME_INT incx, SCALAR *y,
PRIMME_INT incy, primme_context ctx) {
(void)ctx;
/* NOTE: vecLib doesn't follow BLAS reference for sdot */
#if defined(USE_COMPLEX) || (defined(USE_FLOAT) && (defined(__APPLE__) || defined(__MACH__))) || (defined(USE_HALF) && !defined(BLASLAPACK_WITH_HALF))
/* ---- Explicit implementation of the zdotc() --- */
PRIMME_INT i;
HSCALAR zdotc = 0.0;
if (n <= 0) return(zdotc);
if (incx == 1 && incy == 1) {
for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
zdotc += CONJ(TO_COMPLEX(x[i])) * TO_COMPLEX(y[i]);
}
}
else {
for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */
zdotc += CONJ(TO_COMPLEX(x[i*incx])) * TO_COMPLEX(y[i*incy]);
}
}
return zdotc;
/* -- end of explicit implementation of the zdotc() - */
#else
PRIMME_BLASINT ln, lincx, lincy;
CHKERR(to_blas_int(incx, &lincx));
CHKERR(to_blas_int(incy, &lincy));
HSCALAR r = 0.0;
while(n > 0) {
ln = (PRIMME_BLASINT)min(n, PRIMME_BLASINT_MAX-1);
r += XDOT(&ln, (const BLASSCALAR *)x, &lincx, (BLASSCALAR *)y, &lincy);
n -= (PRIMME_INT)ln;
x += ln;
y += ln;
}
return r;
#endif
}
/*******************************************************************************
* Subroutine Num_larnv_Sprimme - x(0:n*incy-1:incy) = rand(0:n-1)
******************************************************************************/
TEMPLATE_PLEASE
int Num_larnv_Sprimme(int idist, PRIMME_INT *iseed, PRIMME_INT length,
SCALAR *x, primme_context ctx) {
(void)ctx;
#ifdef USE_COMPLEX
/* Lapack's R core library doesn't have zlarnv. The functionality is */
/* replaced by calling the REAL version doubling the length. */
assert(idist < 4); /* complex distributions are not supported */
return Num_larnv_Rprimme(idist, iseed, length*2, (REAL*)x, ctx);
#elif (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF)
PRIMME_BLASINT lidist;
CHKERR(to_blas_int(idist, &lidist));
PRIMME_BLASINT llength;
PRIMME_BLASINT temp[4];
PRIMME_BLASINT *liseed;
int i;
if (sizeof(PRIMME_INT) == sizeof(PRIMME_BLASINT)) {
liseed = (PRIMME_BLASINT*)iseed; /* cast avoid compiler warning */
} else {
liseed = temp;
for(i=0; i<4; i++)
liseed[i] = (PRIMME_BLASINT)iseed[i];
}
while(length > 0) {
llength = (PRIMME_BLASINT)min(length, PRIMME_BLASINT_MAX-1);
XLARNV(&lidist, liseed, &llength, x);
length -= (PRIMME_INT)llength;
x += llength;
}
if (sizeof(PRIMME_INT) != sizeof(PRIMME_BLASINT))
for(i=0; i<4; i++)
iseed[i] = (PRIMME_INT)liseed[i];
return 0;
#else
/* Call the HSCALAR kernel and copy back the result */
HSCALAR *xf = NULL;
CHKERR(Num_malloc_SHprimme(length, &xf, ctx));
CHKERR(Num_larnv_SHprimme(idist, iseed, length, xf, ctx));
CHKERR(Num_matrix_astype_SHprimme(xf, 1, length, 1, PRIMME_OP_HSCALAR,
(void **)&x, NULL, PRIMME_OP_SCALAR, -1 /* destroy */, 1 /* copy */,
ctx));
return 0;
#endif
}
/*******************************************************************************
* Subroutine Num_scal_Sprimme - x(0:n*incx-1:incx) *= alpha
******************************************************************************/
TEMPLATE_PLEASE
int Num_scal_Sprimme(PRIMME_INT n, HSCALAR alpha, SCALAR *x, PRIMME_INT incx,
primme_context ctx) {
(void)ctx;
#if (!defined(USE_HALF) && !defined(USE_HALFCOMPLEX)) || defined(BLASLAPACK_WITH_HALF)
PRIMME_BLASINT ln, lincx;
CHKERR(to_blas_int(incx, &lincx));
while(n > 0) {
ln = (PRIMME_BLASINT)min(n, PRIMME_BLASINT_MAX-1);
XSCAL(&ln, (const BLASSCALAR *)&alpha, (BLASSCALAR *)x, &lincx);
n -= (PRIMME_INT)ln;
x += ln;