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
1001
|
/*
* Copyright (c) 2019 ARM Limited.
*
* 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.
*/
#include "activation_float_helpers.h"
#include "helpers.h"
/** Loads the rows from 0 to n-1 in the given variables (BASENAME0 to BASENAMEn-1).
* @name LOAD_ROW_n
*
* @param[in] N0 The number of rows to load
* @param[in] DATA_TYPE The data type of variables
* @param[in] BASENAME The basename of the destination variables for the loaded rows
* @param[in] PTR The base pointer
* @param[in] OFFSET The offset within a row
* @param[in] STRIDE_Y The stride value in y-axis direction
* @param[in] Z The z-axis offset vector
* @{
*/
#define LOAD_ROW_1(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##0 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 0 * STRIDE_Y + Z##0));
#define LOAD_ROW_2(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_1(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##1 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 1 * STRIDE_Y + Z##1));
#define LOAD_ROW_3(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_2(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##2 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 2 * STRIDE_Y + Z##2));
#define LOAD_ROW_4(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_3(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##3 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 3 * STRIDE_Y + Z##3));
#define LOAD_ROW_5(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_4(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##4 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 4 * STRIDE_Y + Z##4));
#define LOAD_ROW_6(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_5(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##5 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 5 * STRIDE_Y + Z##5));
#define LOAD_ROW_7(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_6(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##6 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 6 * STRIDE_Y + Z##6));
#define LOAD_ROW_8(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_7(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##7 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 7 * STRIDE_Y + Z##7));
#define LOAD_ROW_9(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_8(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##8 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 8 * STRIDE_Y + Z##8));
#define LOAD_ROW_10(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_9(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##9 = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 9 * STRIDE_Y + Z##9));
#define LOAD_ROW_11(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_10(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##A = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 10 * STRIDE_Y + Z##A));
#define LOAD_ROW_12(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_11(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##B = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 11 * STRIDE_Y + Z##B));
#define LOAD_ROW_13(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_12(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##C = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 12 * STRIDE_Y + Z##C));
#define LOAD_ROW_14(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_13(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##D = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 13 * STRIDE_Y + Z##D));
#define LOAD_ROW_15(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_14(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##E = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 14 * STRIDE_Y + Z##E));
#define LOAD_ROW_16(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
LOAD_ROW_15(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) \
VEC_DATA_TYPE(DATA_TYPE, N0) \
BASENAME##F = VLOAD(N0)(0, (__global DATA_TYPE *)(PTR + OFFSET + 15 * STRIDE_Y + Z##F));
/** @}*/ // end of group LOAD_ROW_n
/** Load Blocks (consecutive rows and columns) with Z offset.
* @name LOAD_BLOCK
*
* Supported cases are M0=1,2,3,...,16 and N0=1,2,3,4,8,16
* The data to load is expected to have consecutive names for each row.
* E.g., for M0=3, and BASENAME=c, the expected data is c0, c1 and c2.
* The Z offset is expected to have consecutive names.
* E.g., for M0=3, and Z=zin, the expected Z offsets are zin0, zin1 and zin2.
*
* @param[in] M0 The number of consecutive rows
* @param[in] N0 The number of consecutive columns
* @param[in] DATA_TYPE The data type of the target
* @param[in] BASENAME The basename of the result variables
* @param[in] PTR The base pointer for the data
* @param[in] OFFSET The offset within a row
* @param[in] STRIDE_Y The stride in y-axis direction
* @param[in] Z The z-axis offset vector
* @{
*/
#define LOAD_BLOCK_STR(M0, N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) LOAD_ROW_##M0(N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z)
#define LOAD_BLOCK(M0, N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z) LOAD_BLOCK_STR(M0, N0, DATA_TYPE, BASENAME, PTR, OFFSET, STRIDE_Y, Z)
/** @} */ // end of group LOAD_BLOCK
/** Basic macros to calculate Z offset values from Z0 to Zn-1
* @name CALCULATE_Z_OFFSET_n
*
* @param[in] M0 The number of offset values to calculate
* @param[in] DATA_TYPE The data type of the results
* @param[in] Z The basename of the result variables
* @param[in] Y The work-itme ID of y-axis
* @param[in] HEIGHT_GEMM3D The height of GEMM3D
* @param[in] DEPTH_GEMM3D The depth of GEMM3D
* @param[in] CROSS_PLANE_PAD The padding required for plane changes accross the z-dimension
* @param[in] STRIDE_Y The stride value in y-axis direction
*
* @{
*/
#define CALCULATE_Z_OFFSET_1(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
Z##0 = (0 + (DATA_TYPE)(Y * (DATA_TYPE)M0)) / (DATA_TYPE)HEIGHT_GEMM3D; \
Z##0 = min((DATA_TYPE)(DEPTH_GEMM3D - 1), Z##0); \
Z##0 *= (CROSS_PLANE_PAD * STRIDE_Y);
#define CALCULATE_Z_OFFSET_2(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
CALCULATE_Z_OFFSET_1(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
Z##1 = (1 + (DATA_TYPE)(Y * (DATA_TYPE)M0)) / (DATA_TYPE)HEIGHT_GEMM3D; \
Z##1 = min((DATA_TYPE)(DEPTH_GEMM3D - 1), Z##1); \
Z##1 *= (CROSS_PLANE_PAD * STRIDE_Y);
#define CALCULATE_Z_OFFSET_3(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
CALCULATE_Z_OFFSET_2(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
Z##2 = (2 + (DATA_TYPE)(Y * (DATA_TYPE)M0)) / (DATA_TYPE)HEIGHT_GEMM3D; \
Z##2 = min((DATA_TYPE)(DEPTH_GEMM3D - 1), Z##2); \
Z##2 *= (CROSS_PLANE_PAD * STRIDE_Y);
#define CALCULATE_Z_OFFSET_4(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
CALCULATE_Z_OFFSET_3(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
Z##3 = (3 + (DATA_TYPE)(Y * (DATA_TYPE)M0)) / (DATA_TYPE)HEIGHT_GEMM3D; \
Z##3 = min((DATA_TYPE)(DEPTH_GEMM3D - 1), Z##3); \
Z##3 *= (CROSS_PLANE_PAD * STRIDE_Y);
#define CALCULATE_Z_OFFSET_5(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
CALCULATE_Z_OFFSET_4(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
Z##4 = (4 + (DATA_TYPE)(Y * (DATA_TYPE)M0)) / (DATA_TYPE)HEIGHT_GEMM3D; \
Z##4 = min((DATA_TYPE)(DEPTH_GEMM3D - 1), Z##4); \
Z##4 *= (CROSS_PLANE_PAD * STRIDE_Y);
#define CALCULATE_Z_OFFSET_6(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
CALCULATE_Z_OFFSET_5(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
Z##5 = (5 + (DATA_TYPE)(Y * (DATA_TYPE)M0)) / (DATA_TYPE)HEIGHT_GEMM3D; \
Z##5 = min((DATA_TYPE)(DEPTH_GEMM3D - 1), Z##5); \
Z##5 *= (CROSS_PLANE_PAD * STRIDE_Y);
#define CALCULATE_Z_OFFSET_7(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
CALCULATE_Z_OFFSET_6(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
Z##6 = (6 + (DATA_TYPE)(Y * (DATA_TYPE)M0)) / (DATA_TYPE)HEIGHT_GEMM3D; \
Z##6 = min((DATA_TYPE)(DEPTH_GEMM3D - 1), Z##6); \
Z##6 *= (CROSS_PLANE_PAD * STRIDE_Y);
#define CALCULATE_Z_OFFSET_8(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
CALCULATE_Z_OFFSET_7(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) \
Z##7 = (7 + (DATA_TYPE)(Y * (DATA_TYPE)M0)) / (DATA_TYPE)HEIGHT_GEMM3D; \
Z##7 = min((DATA_TYPE)(DEPTH_GEMM3D - 1), Z##7); \
Z##7 *= (CROSS_PLANE_PAD * STRIDE_Y);
/** @} */ // end of group CALCULATE_Z_OFFSET_n
/** Calculate Z offset values from Z0 to Zn-1
* @name CALCULATE_Z_OFFSET
*
* The Z offsets are expected to have consecutive names.
* E.g., for M0=3 and Z=zin, the expected names of Z offsets are zin1, zin2, zin3.
* Note that, CROSS_PLANE_PAD (cross plain padding) is required to take into account
* the possible cross plane paddings in case of the plance changes across the z-dimension.
*
* <!--
* | |
* | plane0 |
* | |
* |__________________|
* |******************|
* | cross_plane_pad |
* |******************|
* | |
* | plane1 |
* | |
* |__________________|
* -->
*
* @param[in] M0 The number of offset values to calculate
* @param[in] DATA_TYPE The data type of the results
* @param[in] Z The basename of the result variables
* @param[in] Y The work-itme ID of y-axis
* @param[in] HEIGHT_GEMM3D The height of GEMM3D
* @param[in] DEPTH_GEMM3D The depth of GEMM3D
* @param[in] CROSS_PLANE_PAD The padding required for plane changes accross the z-dimension
* @param[in] STRIDE_Y The stride value in y-axis direction
* @{
*/
#define CALCULATE_Z_OFFSET_STR(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) CALCULATE_Z_OFFSET_##M0(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y)
#define CALCULATE_Z_OFFSET(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y) CALCULATE_Z_OFFSET_STR(M0, DATA_TYPE, Z, Y, HEIGHT_GEMM3D, DEPTH_GEMM3D, CROSS_PLANE_PAD, STRIDE_Y)
/** @} */ // end of group CALCULATE_Z_OFFSET
/** Store the 0 to (n-1)th rows of the given variables
* @name STORE_ROW_n
*
* @param[in] N0 The size of the vectors
* @param[in] DATA_TYPE The data type of the vectors
* @param[in] BASENAME The basename of the variables
* @param[in] PTR The base pointer
* @param[in] STRIDE_Y The stride value in y-axis direction
* @param[in] Z The offset in z-axis direction
* @{
*/
#define STORE_ROW_1(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##0, 0, (__global DATA_TYPE *)(PTR + 0 * STRIDE_Y + Z##0));
#define STORE_ROW_2(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_1(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##1, 0, (__global DATA_TYPE *)(PTR + 1 * STRIDE_Y + Z##1));
#define STORE_ROW_3(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_2(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##2, 0, (__global DATA_TYPE *)(PTR + 2 * STRIDE_Y + Z##2));
#define STORE_ROW_4(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_3(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##3, 0, (__global DATA_TYPE *)(PTR + 3 * STRIDE_Y + Z##3));
#define STORE_ROW_5(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_4(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##4, 0, (__global DATA_TYPE *)(PTR + 4 * STRIDE_Y + Z##4));
#define STORE_ROW_6(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_5(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##5, 0, (__global DATA_TYPE *)(PTR + 5 * STRIDE_Y + Z##5));
#define STORE_ROW_7(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_6(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##6, 0, (__global DATA_TYPE *)(PTR + 6 * STRIDE_Y + Z##6));
#define STORE_ROW_8(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_7(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##7, 0, (__global DATA_TYPE *)(PTR + 7 * STRIDE_Y + Z##7));
#define STORE_ROW_9(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_8(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##8, 0, (__global DATA_TYPE *)(PTR + 8 * STRIDE_Y + Z##8));
#define STORE_ROW_10(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_9(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##9, 0, (__global DATA_TYPE *)(PTR + 9 * STRIDE_Y + Z##9));
#define STORE_ROW_11(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_10(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##A, 0, (__global DATA_TYPE *)(PTR + 10 * STRIDE_Y + Z##A));
#define STORE_ROW_12(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_11(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##B, 0, (__global DATA_TYPE *)(PTR + 11 * STRIDE_Y + Z##B));
#define STORE_ROW_13(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_12(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##C, 0, (__global DATA_TYPE *)(PTR + 12 * STRIDE_Y + Z##C));
#define STORE_ROW_14(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_13(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##D, 0, (__global DATA_TYPE *)(PTR + 13 * STRIDE_Y + Z##D));
#define STORE_ROW_15(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_14(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##E, 0, (__global DATA_TYPE *)(PTR + 14 * STRIDE_Y + Z##E));
#define STORE_ROW_16(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
STORE_ROW_15(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(BASENAME##F, 0, (__global DATA_TYPE *)(PTR + 15 * STRIDE_Y + Z##F));
/** @} */ // end of groupd STORE_ROW_n
/** Convert and store the 0th to (n-1)th rows of the given variables
* @name CONVERT_STORE_ROW_n
*
* @param[in] N0 The size of the vectors
* @param[in] DATA_TYPE The data type of the vectors
* @param[in] BASENAME The basename of the variables
* @param[in] PTR The base pointer
* @param[in] STRIDE_Y The stride value in y-axis direction
* @param[in] Z The offset in z-axis direction
* @{
*/
#define CONVERT_STORE_ROW_1(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##0), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 0 * STRIDE_Y + Z##0));
#define CONVERT_STORE_ROW_2(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_1(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##1), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 1 * STRIDE_Y + Z##1));
#define CONVERT_STORE_ROW_3(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_2(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##2), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 2 * STRIDE_Y + Z##2));
#define CONVERT_STORE_ROW_4(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_3(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##3), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 3 * STRIDE_Y + Z##3));
#define CONVERT_STORE_ROW_5(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_4(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##4), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 4 * STRIDE_Y + Z##4));
#define CONVERT_STORE_ROW_6(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_5(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##5), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 5 * STRIDE_Y + Z##5));
#define CONVERT_STORE_ROW_7(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_6(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##6), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 6 * STRIDE_Y + Z##6));
#define CONVERT_STORE_ROW_8(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_7(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##7), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 7 * STRIDE_Y + Z##7));
#define CONVERT_STORE_ROW_9(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_8(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##8), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 8 * STRIDE_Y + Z##8));
#define CONVERT_STORE_ROW_10(N0, DATA, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_9(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##9), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 9 * STRIDE_Y + Z##9));
#define CONVERT_STORE_ROW_11(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_10(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##A), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 10 * STRIDE_Y + Z##A));
#define CONVERT_STORE_ROW_12(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_11(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##B), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 11 * STRIDE_Y + Z##B));
#define CONVERT_STORE_ROW_13(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_12(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##C), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 12 * STRIDE_Y + Z##C));
#define CONVERT_STORE_ROW_14(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_13(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##D), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 13 * STRIDE_Y + Z##D));
#define CONVERT_STORE_ROW_15(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_14(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##E), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 14 * STRIDE_Y + Z##E));
#define CONVERT_STORE_ROW_16(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
CONVERT_STORE_ROW_15(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) \
VSTORE(N0) \
(CONVERT_SAT((BASENAME##F), VEC_DATA_TYPE(DATA_TYPE, N0)), 0, (__global DATA_TYPE *)(PTR + 15 * STRIDE_Y + Z##F));
/** @} */ // end of groupd CONVERT_STORE_ROW_n
/** Store a block of the given size M0xN0
* @name STORE_BLOCK
*
* Supported cases are M0=1,2,3,...,16 and N0=2,3,4,8,16.
* The data to store is expected to have consecutive names for each row.
* E.g., for M0=3 and basename=c, the expected names are c0, c1 and c2.
* The Z offset is expected to have consecutive names.
* E.g., for M0=3 and Z=zin, the expected z offset names are zin0, zin1 and zin2.
*
* @param[in] M0 The number of rows to store
* @param[in] N0 The size of each vector
* @param[in] DATA_TYPE The data type of the vectors
* @param[in] BASENAME The basename of the variables
* @param[in] PTR The base pointer
* @param[in] STRIDE_Y The stride value in y-axis direction
* @param[in] Z The offset in z-axis direction
* @{
*/
#define STORE_BLOCK_STR(M0, N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) STORE_ROW_##M0(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z)
#define STORE_BLOCK(M0, N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) STORE_BLOCK_STR(M0, N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z)
/** @} */ // end of group STORE_BLOCK
/** Convert and store a block of the given size M0xN0
* @name CONVERT_STORE_BLOCK
*
* Supported cases are M0=1,2,3,...,16 and N0=2,3,4,8,16.
* The data to store is expected to have consecutive names for each row.
* E.g., for M0=3 and basename=c, the expected names are c0, c1 and c2.
* The Z offset is expected to have consecutive names.
* E.g., for M0=3 and Z=zin, the expected z offset names are zin0, zin1 and zin2.
*
* @param[in] M0 The number of rows to store
* @param[in] N0 The size of each vector
* @param[in] DATA_TYPE The data type of the vectors
* @param[in] BASENAME The basename of the variables
* @param[in] PTR The base pointer
* @param[in] STRIDE_Y The stride value in y-axis direction
* @param[in] Z The offset in z-axis direction
* @{
*/
#define CONVERT_STORE_BLOCK_STR(M0, N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) CONVERT_STORE_ROW_##M0(N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z)
#define CONVERT_STORE_BLOCK(M0, N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z) CONVERT_STORE_BLOCK_STR(M0, N0, DATA_TYPE, BASENAME, PTR, STRIDE_Y, Z)
/** @} */ // end of group CONVERT_STORE_BLOCK
/** Scale the rows in the given variables (BASENAME0 to BASENAMEn-1)
* @name SCALE_ROW_n
*
* @param[in] DATA_TYPE The data type of the variables
* @param[in] BASENAME The basename of the variables
* @param[in] SCALE The scale factor
* @{
*/
#define SCALE_ROW_1(DATA_TYPE, BASENAME, SCALE) \
BASENAME##0 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_2(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_1(DATA_TYPE, BASENAME, SCALE) \
BASENAME##1 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_3(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_2(DATA_TYPE, BASENAME, SCALE) \
BASENAME##2 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_4(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_3(DATA_TYPE, BASENAME, SCALE) \
BASENAME##3 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_5(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_4(DATA_TYPE, BASENAME, SCALE) \
BASENAME##4 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_6(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_5(DATA_TYPE, BASENAME, SCALE) \
BASENAME##5 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_7(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_6(DATA_TYPE, BASENAME, SCALE) \
BASENAME##6 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_8(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_7(DATA_TYPE, BASENAME, SCALE) \
BASENAME##7 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_9(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_8(DATA_TYPE, BASENAME, SCALE) \
BASENAME##8 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_10(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_9(DATA_TYPE, BASENAME, SCALE) \
BASENAME##9 *= (DATA_TYPE)SCALE;
#define SCALE_ROW_11(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_10(DATA_TYPE, BASENAME, SCALE) \
BASENAME##A *= (DATA_TYPE)SCALE;
#define SCALE_ROW_12(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_11(DATA_TYPE, BASENAME, SCALE) \
BASENAME##B *= (DATA_TYPE)SCALE;
#define SCALE_ROW_13(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_12(DATA_TYPE, BASENAME, SCALE) \
BASENAME##C *= (DATA_TYPE)SCALE;
#define SCALE_ROW_14(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_13(DATA_TYPE, BASENAME, SCALE) \
BASENAME##D *= (DATA_TYPE)SCALE;
#define SCALE_ROW_15(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_14(DATA_TYPE, BASENAME, SCALE) \
BASENAME##E *= (DATA_TYPE)SCALE;
#define SCALE_ROW_16(DATA_TYPE, BASENAME, SCALE) \
SCALE_ROW_15(DATA_TYPE, BASENAME, SCALE) \
BASENAME##F *= (DATA_TYPE)SCALE;
/** @} */ // end of group SCALE_ROW_n
/** Scale elements stored in a block (BASENAME)
* @name SCALE_BLOCK
*
* Supported cases are N=1,2,3,...,16
*
* @param[in] N The number of rows in the block
* @param[in] DATA_TYPE The data type of the block
* @param[in] BASENAME The basename of the block
* @param[in] SCALE The scale factor
* @{
*/
#define SCALE_BLOCK_STR(N, DATA_TYPE, BASENAME, SCALE) SCALE_ROW_##N(DATA_TYPE, BASENAME, SCALE)
#define SCALE_BLOCK(N, DATA_TYPE, BASENAME, SCALE) SCALE_BLOCK_STR(N, DATA_TYPE, BASENAME, SCALE)
/** @} */ // end of group SCALE_BLOCK
/** Create a new vector containing the values at the given index for a set of given vectors
* @name COLUMN_VECTORn
*
* @param[in] IDX_COL The index value
* @param[in] BASENAME The basename of the destination vectors
* @param[in] X The basename of the source vectors
* @{
*/
#define COLUMN_VECTOR1(IDX_COL, BASENAME, X) \
uchar BASENAME##IDX_COL = (uchar)((X##0).s##IDX_COL);
#define COLUMN_VECTOR2(IDX_COL, BASENAME, X) \
uchar2 BASENAME##IDX_COL = (uchar2)((X##0).s##IDX_COL, (X##1).s##IDX_COL);
#define COLUMN_VECTOR3(IDX_COL, BASENAME, X) \
uchar3 BASENAME##IDX_COL = (uchar3)((X##0).s##IDX_COL, (X##1).s##IDX_COL, (X##2).s##IDX_COL);
#define COLUMN_VECTOR4(IDX_COL, BASENAME, X) \
uchar4 BASENAME##IDX_COL = (uchar4)((X##0).s##IDX_COL, (X##1).s##IDX_COL, (X##2).s##IDX_COL, (X##3).s##IDX_COL);
#define COLUMN_VECTOR8(IDX_COL, BASENAME, X) \
uchar8 BASENAME##IDX_COL = (uchar8)((X##0).s##IDX_COL, (X##1).s##IDX_COL, (X##2).s##IDX_COL, (X##3).s##IDX_COL, (X##4).s##IDX_COL, (X##5).s##IDX_COL, (X##6).s##IDX_COL, (X##7).s##IDX_COL);
#define COLUMN_VECTOR16(IDX_COL, BASENAME, X) \
uchar16 BASENAME##IDX_COL = (uchar16)((X##0).s##IDX_COL, (X##1).s##IDX_COL, (X##2).s##IDX_COL, (X##3).s##IDX_COL, (X##4).s##IDX_COL, (X##5).s##IDX_COL, (X##6).s##IDX_COL, (X##7).s##IDX_COL, (X##8).s##IDX_COL, (X##9).s##IDX_COL, (X##A).s##IDX_COL, (X##B).s##IDX_COL, (X##C).s##IDX_COL, (X##D).s##IDX_COL, (X##E).s##IDX_COL, (X##F).s##IDX_COL);
/** @} */ // end of group COLUMN_VECTORn
/** Create transposed vectors of the given vectors
* @name TRANSPOSE_K0Xn
*
* @param[in] K0 The size of the source vectors
* @param[in] BASENAME The basename of transposed vectors
* @param[in] B The basename of source vectors for transposition
* @{
*/
#define TRANSPOSE_K0X1(K0, BASENAME, B) \
COLUMN_VECTOR(K0, 0, BASENAME, B);
#define TRANSPOSE_K0X2(K0, BASENAME, B) \
TRANSPOSE_K0X1(K0, BASENAME, B); \
COLUMN_VECTOR(K0, 1, BASENAME, B);
#define TRANSPOSE_K0X3(K0, BASENAME, B) \
TRANSPOSE_K0X2(K0, BASENAME, B); \
COLUMN_VECTOR(K0, 2, BASENAME, B);
#define TRANSPOSE_K0X4(K0, BASENAME, B) \
TRANSPOSE_K0X3(K0, BASENAME, B); \
COLUMN_VECTOR(K0, 3, BASENAME, B);
#define TRANSPOSE_K0X8(K0, BASENAME, B) \
TRANSPOSE_K0X4(K0, BASENAME, B); \
COLUMN_VECTOR(K0, 4, BASENAME, B); \
COLUMN_VECTOR(K0, 5, BASENAME, B); \
COLUMN_VECTOR(K0, 6, BASENAME, B); \
COLUMN_VECTOR(K0, 7, BASENAME, B);
#define TRANSPOSE_K0X16(K0, BASENAME, B) \
TRANSPOSE_K0X8(K0, BASENAME, B); \
COLUMN_VECTOR(K0, 8, BASENAME, B); \
COLUMN_VECTOR(K0, 9, BASENAME, B); \
COLUMN_VECTOR(K0, A, BASENAME, B); \
COLUMN_VECTOR(K0, B, BASENAME, B); \
COLUMN_VECTOR(K0, C, BASENAME, B); \
COLUMN_VECTOR(K0, D, BASENAME, B); \
COLUMN_VECTOR(K0, E, BASENAME, B); \
COLUMN_VECTOR(K0, F, BASENAME, B);
/** @} */ // end of group TRANSPOSE_K0Xn
/** Create column vectors to contain the values at the given index for a set of given vectors
*
* @param[in] K0 The number of source vectors
* @param[in] IDX_COL The index value
* @param[in] BASENAME The basename of the destination vectors
* @param[in] B The basename of the source vectors
*/
#define COLUMN_VECTOR(K0, IDX_COL, BASENAME, B) \
CONCAT(COLUMN_VECTOR, K0) \
(IDX_COL, BASENAME, B);
/** Create transposed vectors form the given source vectors
*
* @param[in] K0 The size of source vectors
* @param[in] N0 The number of source vectors
* @param[in] BASENAME The basename of transposed vectors
* @param[in] B The basename of source vectors for transposition
*
*/
#define TRANSPOSE_K0XN0(K0, N0, BASENAME, B) \
CONCAT(TRANSPOSE_K0X, N0) \
(K0, BASENAME, B);
/** Add the variables (BIAS0 to BIASn-1) to the others (BASENAME0 to BASENAMEn-1)
* @name ADD_ROW_n
*
* @param[in] BASENAME The basename of the destination variables
* @param[in] BIAS The basename of the added variables
* @{
*/
#define ADD_ROW_1(BASENAME, BIAS) \
BASENAME##0 += BIAS##0;
#define ADD_ROW_2(BASENAME, BIAS) \
ADD_ROW_1(BASENAME, BIAS) \
BASENAME##1 += BIAS##1;
#define ADD_ROW_3(BASENAME, BIAS) \
ADD_ROW_2(BASENAME, BIAS) \
BASENAME##2 += BIAS##2;
#define ADD_ROW_4(BASENAME, BIAS) \
ADD_ROW_3(BASENAME, BIAS) \
BASENAME##3 += BIAS##3;
#define ADD_ROW_5(BASENAME, BIAS) \
ADD_ROW_4(BASENAME, BIAS) \
BASENAME##4 += BIAS##4;
#define ADD_ROW_6(BASENAME, BIAS) \
ADD_ROW_5(BASENAME, BIAS) \
BASENAME##5 += BIAS##5;
#define ADD_ROW_7(BASENAME, BIAS) \
ADD_ROW_6(BASENAME, BIAS) \
BASENAME##6 += BIAS##6;
#define ADD_ROW_8(BASENAME, BIAS) \
ADD_ROW_7(BASENAME, BIAS) \
BASENAME##7 += BIAS##7;
#define ADD_ROW_9(BASENAME, BIAS) \
ADD_ROW_8(BASENAME, BIAS) \
BASENAME##8 += BIAS##8;
#define ADD_ROW_10(BASENAME, BIAS) \
ADD_ROW_9(BASENAME, BIAS) \
BASENAME##9 += BIAS##9;
#define ADD_ROW_11(BASENAME, BIAS) \
ADD_ROW_10(BASENAME, BIAS) \
BASENAME##A += BIAS##A;
#define ADD_ROW_12(BASENAME, BIAS) \
ADD_ROW_11(BASENAME, BIAS) \
BASENAME##B += BIAS##B;
#define ADD_ROW_13(BASENAME, BIAS) \
ADD_ROW_12(BASENAME, BIAS) \
BASENAME##C += BIAS##C;
#define ADD_ROW_14(BASENAME, BIAS) \
ADD_ROW_13(BASENAME, BIAS) \
BASENAME##D += BIAS##D;
#define ADD_ROW_15(BASENAME, BIAS) \
ADD_ROW_14(BASENAME, BIAS) \
BASENAME##E += BIAS##E;
#define ADD_ROW_16(BASENAME, BIAS) \
ADD_ROW_15(BASENAME, BIAS) \
BASENAME##F += BIAS##F;
/** @} */ // end of group ADD_ROW_n
/** Add the block (BIAS) to another block (BASENAME)
* @name ADD_BLOCK
*
* Supported cases are N=1,2,3,...,16
*
* @param[in] N The number of vectors in the block
* @param[in] BASENAME The basename of the destination variables
* @param[in] BIAS The basename of the added variables
* @{
*/
#define ADD_BLOCK_STR(N, BASENAME, BIAS) ADD_ROW_##N(BASENAME, BIAS)
#define ADD_BLOCK(N, BASENAME, BIAS) ADD_BLOCK_STR(N, BASENAME, BIAS)
/** @} */ // end of group ADD_BLOCK
/** Broadcast (add single value) to the each element of the destination variables
* @name ADD_ROW_BROADCAST_n
*
* @param[in] BASENAME The basename of the destination variables
* @param[in] BIAS The variable containing the value to add
* @{
*/
#define ADD_ROW_BROADCAST_1(BASENAME, BIAS) \
BASENAME##0 += BIAS;
#define ADD_ROW_BROADCAST_2(BASENAME, BIAS) \
ADD_ROW_BROADCAST_1(BASENAME, BIAS) \
BASENAME##1 += BIAS;
#define ADD_ROW_BROADCAST_3(BASENAME, BIAS) \
ADD_ROW_BROADCAST_2(BASENAME, BIAS) \
BASENAME##2 += BIAS;
#define ADD_ROW_BROADCAST_4(BASENAME, BIAS) \
ADD_ROW_BROADCAST_3(BASENAME, BIAS) \
BASENAME##3 += BIAS;
#define ADD_ROW_BROADCAST_5(BASENAME, BIAS) \
ADD_ROW_BROADCAST_4(BASENAME, BIAS) \
BASENAME##4 += BIAS;
#define ADD_ROW_BROADCAST_6(BASENAME, BIAS) \
ADD_ROW_BROADCAST_5(BASENAME, BIAS) \
BASENAME##5 += BIAS;
#define ADD_ROW_BROADCAST_7(BASENAME, BIAS) \
ADD_ROW_BROADCAST_6(BASENAME, BIAS) \
BASENAME##6 += BIAS;
#define ADD_ROW_BROADCAST_8(BASENAME, BIAS) \
ADD_ROW_BROADCAST_7(BASENAME, BIAS) \
BASENAME##7 += BIAS;
#define ADD_ROW_BROADCAST_9(BASENAME, BIAS) \
ADD_ROW_BROADCAST_8(BASENAME, BIAS) \
BASENAME##8 += BIAS;
#define ADD_ROW_BROADCAST_10(BASENAME, BIAS) \
ADD_ROW_BROADCAST_9(BASENAME, BIAS) \
BASENAME##9 += BIAS;
#define ADD_ROW_BROADCAST_11(BASENAME, BIAS) \
ADD_ROW_BROADCAST_10(BASENAME, BIAS) \
BASENAME##A += BIAS;
#define ADD_ROW_BROADCAST_12(BASENAME, BIAS) \
ADD_ROW_BROADCAST_11(BASENAME, BIAS) \
BASENAME##B += BIAS;
#define ADD_ROW_BROADCAST_13(BASENAME, BIAS) \
ADD_ROW_BROADCAST_12(BASENAME, BIAS) \
BASENAME##C += BIAS;
#define ADD_ROW_BROADCAST_14(BASENAME, BIAS) \
ADD_ROW_BROADCAST_13(BASENAME, BIAS) \
BASENAME##D += BIAS;
#define ADD_ROW_BROADCAST_15(BASENAME, BIAS) \
ADD_ROW_BROADCAST_14(BASENAME, BIAS) \
BASENAME##E += BIAS;
#define ADD_ROW_BROADCAST_16(BASENAME, BIAS) \
ADD_ROW_BROADCAST_15(BASENAME, BIAS) \
BASENAME##F += BIAS;
/** Broadcast (add a value) to the each element of the destination block (BASENAME)
* @name ADD_BLOCK_BROADCAST
*
* Supported cases are N=1,2,3,...,16.
*
* @param[in] N The number of vectors in the block
* @param[in] BASENAME The basename of the destination variables
* @param[in] BIAS The variable containing the value to add
* @{
*/
#define ADD_BLOCK_BROADCAST_STR(N, BASENAME, BIAS) ADD_ROW_BROADCAST_##N(BASENAME, BIAS)
#define ADD_BLOCK_BROADCAST(N, BASENAME, BIAS) ADD_BLOCK_BROADCAST_STR(N, BASENAME, BIAS)
/** @} */ // end of group ADD_BLOCK_BROADCAST
/** Apply activation to the given variables
* @name ACTIVATION_ROW_n
*
* @param[in] ACTIVATION_TYPE The type of the activation
* @param[in] DATA_TYPE The data type of the vectors
* @param[in] BASENAME The basename of the variables
* @param[in] A_VAL Additional value required by the activation
* @param[in] B_VAL Additional value required by the activation
* @{
*/
#define ACTIVATION_ROW_1(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##0 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##0, A_VAL, B_VAL);
#define ACTIVATION_ROW_2(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_1(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##1 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##1, A_VAL, B_VAL);
#define ACTIVATION_ROW_3(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_2(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##2 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##2, A_VAL, B_VAL);
#define ACTIVATION_ROW_4(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_3(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##3 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##3, A_VAL, B_VAL);
#define ACTIVATION_ROW_5(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_4(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##4 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##4, A_VAL, B_VAL);
#define ACTIVATION_ROW_6(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_5(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##5 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##5, A_VAL, B_VAL);
#define ACTIVATION_ROW_7(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_6(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##6 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##6, A_VAL, B_VAL);
#define ACTIVATION_ROW_8(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_7(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##7 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##7, A_VAL, B_VAL);
#define ACTIVATION_ROW_9(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_8(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##8 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##8, A_VAL, B_VAL);
#define ACTIVATION_ROW_10(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_9(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##9 = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##9, A_VAL, B_VAL);
#define ACTIVATION_ROW_11(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_10(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##A = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##A, A_VAL, B_VAL);
#define ACTIVATION_ROW_12(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_11(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##B = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##B, A_VAL, B_VAL);
#define ACTIVATION_ROW_13(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_12(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##C = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##C, A_VAL, B_VAL);
#define ACTIVATION_ROW_14(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_13(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##D = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##D, A_VAL, B_VAL);
#define ACTIVATION_ROW_15(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_14(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##E = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##E, A_VAL, B_VAL);
#define ACTIVATION_ROW_16(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
ACTIVATION_ROW_15(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) \
BASENAME##F = ACTIVATION(ACTIVATION_TYPE, DATA_TYPE, BASENAME##F, A_VAL, B_VAL);
/** @} */ // end of group ACTIVATION_ROW_n
/** Apply activation to a block (BASENAME)
* @name ACTIVATION_BLOCK
*
* Supported cases are N=1,2,3,...,16.
*
* @param[in] N The number of vectors in the block
* @param[in] ACTIVATION_TYPE The type of the activation
* @param[in] DATA_TYPE The data type of the vectors
* @param[in] BASENAME The basename of the variables
* @param[in] A_VAL Additional value required by the activation
* @param[in] B_VAL Additional value required by the activation
* @{
*/
#define ACTIVATION_BLOCK_STR(N, ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) ACTIVATION_ROW_##N(ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL)
#define ACTIVATION_BLOCK(N, ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL) ACTIVATION_BLOCK_STR(N, ACTIVATION_TYPE, DATA_TYPE, BASENAME, A_VAL, B_VAL)
/** @} */ // end of group ACTIVATION_BLOCK
/** Apply convert_<data_type> to the given variables
* @name CONVERT_ROW_n
*
* @param[in] N The size of the vectors
* @param[in] DATA_TYPE The data type of the vectors
* @param[in] BASENAME_SRC The basename of the source variables
* @param[in] BASENAME_DST The basename of the destination variables
*/
#define CONVERT_ROW_1(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##0 = CONVERT(BASENAME_SRC##0, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_2(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_1(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##1 = CONVERT(BASENAME_SRC##1, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_3(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_2(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##2 = CONVERT(BASENAME_SRC##2, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_4(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_3(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##3 = CONVERT(BASENAME_SRC##3, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_5(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_4(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##4 = CONVERT(BASENAME_SRC##4, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_6(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_5(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##5 = CONVERT(BASENAME_SRC##5, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_7(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_6(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##6 = CONVERT(BASENAME_SRC##6, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_8(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_7(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##7 = CONVERT(BASENAME_SRC##7, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_9(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_8(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##8 = CONVERT(BASENAME_SRC##8, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_10(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_9(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##9 = CONVERT(BASENAME_SRC##9, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_11(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_10(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##A = CONVERT(BASENAME_SRC##A, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_12(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_11(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##B = CONVERT(BASENAME_SRC##B, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_13(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_12(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##C = CONVERT(BASENAME_SRC##C, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_14(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_13(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##D = CONVERT(BASENAME_SRC##D, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_15(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_14(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##E = CONVERT(BASENAME_SRC##E, VEC_DATA_TYPE(DATA_TYPE, N));
#define CONVERT_ROW_16(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
CONVERT_ROW_15(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) \
VEC_DATA_TYPE(DATA_TYPE, N) \
BASENAME_DST##F = CONVERT(BASENAME_SRC##F, VEC_DATA_TYPE(DATA_TYPE, N));
/** @} */ // end of group CONVERT_ROW_n
/** Apply convert_<data_type> to a block (BASENAME_SRC) and save to another block (BASENAME_DST)
* @name CONVERT_BLOCK
*
* Supported cases N=1,2,3,...,16.
*
* @param[in] M The number of vectors to convert
* @param[in] N The size of the vectors
* @param[in] DATA_TYPE The data type of the vectors
* @param[in] BASENAME_SRC The basename of the source variables
* @param[in] BASENAME_DST The basename of the destination variables
*/
#define CONVERT_BLOCK_STR(M, N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) CONVERT_ROW_##M(N, DATA_TYPE, BASENAME_SRC, BASENAME_DST)
#define CONVERT_BLOCK(M, N, DATA_TYPE, BASENAME_SRC, BASENAME_DST) CONVERT_BLOCK_STR(M, N, DATA_TYPE, BASENAME_SRC, BASENAME_DST)
/** @} */ // end of group CONVERT_BLOCK
|
