forked from JishinMaster/simd_utils
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsimd_utils_constants.h
2188 lines (1863 loc) · 76.4 KB
/
simd_utils_constants.h
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
/*
* Project : SIMD_Utils
* Version : 0.2.5
* Author : JishinMaster
* Licence : BSD-2
*/
#if defined(SSE) || defined(AVX) || defined(AVX512)
#ifndef ARM
#include <immintrin.h>
#else
#if !defined(__aarch64__)
#define SSE2NEON_PRECISE_SQRT 1
#define SSE2NEON_PRECISE_DIV 1
#endif
// Also includes arm_neon.h
#include "sse2neon_wrapper.h"
#endif
#endif
#ifdef RISCV
#include <riscv_vector.h>
/* ELEN : element length, 8,16,32,64bits
VLEN : Vector Length, at least 128bits
32 registers in the 0.10 standard, plus vstart, vxsat, vxrm, vcsr, vtype, vl, vlenb
VSEW : Vector Standard Element Width (dynamic), with of the base element : 8,16,32,64,...,1024bits
(up to 64bit in the current intrinsics
LMUL : Vector register grouping => may group multiple VLEN registers, so that 1 instruction can be applied to multiple registers. If LMUL is < 1, the operation applies only to a part of the register
LMUL = 1,2,4,8, 1, 1/2, 1/4, 1/8
VLMAX = LMUL*VLEN/SEW
Vector Tail Agnostic and Vector Mask Agnostic vta and vma allow to mask operations on vector such as only part of a vector is modified
Vector Fixed-Point Rounding Mode Register vxrm for rounding mode : round-to-nearest-up rnu, round-to-nearest-even rne, round-down rdn, round-to-odd rod
Need a real CPU with CPI/latency to have better choice of instructions..
fmadd vs fmacc, load stride vs segment, etc
*/
// 0 to nearest, 1 to zero (trunc), 2 round down, 3 round up, 4 round to nearest
#define _MM_ROUND_NEAREST 0
#define _MM_ROUND_TOWARD_ZERO 1
#define _MM_ROUND_DOWN 2
#define _MM_ROUND_UP 3
// load vector float32, 8
// "1" in name means either vector scalar instructions, or load/store scalar to vector
/*
# FP multiply-accumulate, overwrites addend
vfmacc.vv vd, vs1, vs2, vm
# vd[i] = +(vs1[i] * vs2[i]) + vd[i]
vfmacc.vf vd, rs1, vs2, vm
# vd[i] = +(f[rs1] * vs2[i]) + vd[i]
# FP negate-(multiply-accumulate), overwrites subtrahend
vfnmacc.vv vd, vs1, vs2, vm
# vd[i] = -(vs1[i] * vs2[i]) - vd[i]
vfnmacc.vf vd, rs1, vs2, vm
# vd[i] = -(f[rs1] * vs2[i]) - vd[i]
# FP multiply-subtract-accumulator, overwrites subtrahend
vfmsac.vv vd, vs1, vs2, vm
# vd[i] = +(vs1[i] * vs2[i]) - vd[i]
vfmsac.vf vd, rs1, vs2, vm
# vd[i] = +(f[rs1] * vs2[i]) - vd[i]
# FP negate-(multiply-subtract-accumulator), overwrites minuend
vfnmsac.vv vd, vs1, vs2, vm
# vd[i] = -(vs1[i] * vs2[i]) + vd[i]
vfnmsac.vf vd, rs1, vs2, vm
# vd[i] = -(f[rs1] * vs2[i]) + vd[i]
# FP multiply-add, overwrites multiplicand
vfmadd.vv vd, vs1, vs2, vm
# vd[i] = +(vs1[i] * vd[i]) + vs2[i]
vfmadd.vf vd, rs1, vs2, vm
# vd[i] = +(f[rs1] * vd[i]) + vs2[i]
# FP negate-(multiply-add), overwrites multiplicand
vfnmadd.vv vd, vs1, vs2, vm
# vd[i] = -(vs1[i] * vd[i]) - vs2[i]
vfnmadd.vf vd, rs1, vs2, vm
# vd[i] = -(f[rs1] * vd[i]) - vs2[i]
# FP multiply-sub, overwrites multiplicand
vfmsub.vv vd, vs1, vs2, vm
# vd[i] = +(vs1[i] * vd[i]) - vs2[i]
vfmsub.vf vd, rs1, vs2, vm
# vd[i] = +(f[rs1] * vd[i]) - vs2[i]
# FP negate-(multiply-sub), overwrites multiplicand
vfnmsub.vv vd, vs1, vs2, vm
# vd[i] = -(vs1[i] * vd[i]) + vs2[i]
vfnmsub.vf vd, rs1, vs2, vm
# vd[i] = -(f[rs1] * vd[i]) + vs2[i]
*/
#ifndef ELEN
#define ELEN 64 // vector support elements up to 64 bits
#endif
#ifndef VECTOR_LENGTH
#define MAX_ELTS8 1024 // 1024bits*4 registers(m4) => 512 int8
#define MAX_ELTS32 256 // 1024bits*4 registers(m4) => 128 float/int32
#define MAX_ELTS64 128 // 1024bits*4 registers(m4) => 64 double/int64
#define VECTOR_LENGTH 1024
#else
#define MAX_ELTS8 VECTOR_LENGTH
#define MAX_ELTS32 VECTOR_LENGTH / 4
#define MAX_ELTS64 VECTOR_LENGTH / 8
#endif
#ifndef vfcvt_rtz_x_f_v_i32m4
#define NO_RTZ
#define vfcvt_rtz_x_f_v_i32m4 vfcvt_x_f_v_i32m4
#endif
#ifndef vfcvt_rtz_x_f_v_i32m2
#define NO_RTZ
#define vfcvt_rtz_x_f_v_i32m2 vfcvt_x_f_v_i32m2
#endif
#if ELEN >= 64
#ifndef vfcvt_rtz_x_f_v_i64m4
#define NO_RTZ
#define vfcvt_rtz_x_f_v_i64m4 vfcvt_x_f_v_i64m4
#endif
#ifndef vfcvt_rtz_x_f_v_i64m2
#define NO_RTZ
#define vfcvt_rtz_x_f_v_i64m2 vfcvt_x_f_v_i64m2
#endif
#endif // ELEN >= 64
///////////////////// FULL VECTOR m4 //////////////
#define VSETVL32 vsetvl_e32m4
#define VSETVL16 vsetvl_e16m4
#if ELEN >= 64
#define VSETVL64 vsetvl_e64m4
#endif // ELEN >= 64
//// FLOAT
#define V_ELT_FLOAT vfloat32m4_t
#define VLOAD_FLOAT vle32_v_f32m4
#define VLOAD1_FLOAT vfmv_v_f_f32m4
#define VSTORE_FLOAT vse32_v_f32m4
#define VADD_FLOAT vfadd_vv_f32m4
#define VADD1_FLOAT vfadd_vf_f32m4
#define VSUB_FLOAT vfsub_vv_f32m4
#define VSUB1_FLOAT vfsub_vf_f32m4
#define VRSUB1_FLOAT vfrsub_vf_f32m4 // v2 = f - v1
#define VMUL_FLOAT vfmul_vv_f32m4
#define VMUL1_FLOAT vfmul_vf_f32m4
#define VDIV_FLOAT vfdiv_vv_f32m4
#define VDIV1_FLOAT vfdiv_vf_f32m4
#define VRDIV1_FLOAT vfrdiv_vf_f32m4
#define VFMACC_FLOAT vfmacc_vv_f32m4 // vd[i] = +(vs1[i] * vs2[i]) + vd[i]
#define VFMACC1_FLOAT vfmacc_vf_f32m4
#define VFMADD_FLOAT vfmadd_vv_f32m4 // vd[i] = +(vs1[i] * vd[i]) + vs2[i]
#define VFMADD1_FLOAT vfmadd_vf_f32m4
#define VFMSUB_FLOAT vfmsub_vv_f32m4 // d = a*b - c
#define VREDSUM_FLOAT vfredosum_vs_f32m4_f32m1
#define VREDMAX_FLOAT vfredmax_vs_f32m4_f32m1
#define VREDMIN_FLOAT vfredmin_vs_f32m4_f32m1
#define VMIN_FLOAT vfmin_vv_f32m4
#define VMAX_FLOAT vfmax_vv_f32m4
#define VMIN1_FLOAT vfmin_vf_f32m4
#define VMAX1_FLOAT vfmax_vf_f32m4
#define VINTERP_FLOAT_INT vreinterpret_v_f32m4_i32m4
#define VINTERP_INT_FLOAT vreinterpret_v_i32m4_f32m4
#define VCVT_RTZ_FLOAT_INT vfcvt_rtz_x_f_v_i32m4
#define VCVT_FLOAT_INT vfcvt_x_f_v_i32m4
#define VCVT_INT_FLOAT vfcvt_f_x_v_f32m4
#define VMERGE_FLOAT vmerge_vvm_f32m4
#define VMUL1_FLOAT_MASK vfmul_vf_f32m4_m
#define VSQRT_FLOAT vfsqrt_v_f32m4
#define VLE_FLOAT_STRIDE vlse32_v_f32m4
#define VEQ1_FLOAT_BOOL vmfeq_vf_f32m4_b8
#define VEQ_FLOAT_BOOL vmfeq_vv_f32m4_b8
#define VGT1_FLOAT_BOOL vmfgt_vf_f32m4_b8
#define VNE1_FLOAT_BOOL vmfne_vf_f32m4_b8
#define VLT1_FLOAT_BOOL vmflt_vf_f32m4_b8
#define VLE1_FLOAT_BOOL vmfle_vf_f32m4_b8
#define VABS_FLOAT vfabs_v_f32m4
#define VMERGE1_FLOAT vfmerge_vfm_f32m4
#define VGATHER_FLOAT vrgather_vv_f32m4
#if ELEN >= 64
//// DOUBLE
#define V_ELT_DOUBLE vfloat64m4_t
#define VLOAD_DOUBLE vle64_v_f64m4
#define VLOAD1_DOUBLE vfmv_v_f_f64m4
#define VSTORE_DOUBLE vse64_v_f64m4
#define VADD_DOUBLE vfadd_vv_f64m4
#define VADD1_DOUBLE vfadd_vf_f64m4
#define VSUB_DOUBLE vfsub_vv_f64m4
#define VSUB1_DOUBLE vfsub_vf_f64m4
#define VMUL_DOUBLE vfmul_vv_f64m4
#define VMUL1_DOUBLE vfmul_vf_f64m4
#define VDIV_DOUBLE vfdiv_vv_f64m4
#define VFMA_DOUBLE vfmacc_vv_f64m4 // d = a + b*c
#define VFMA1_DOUBLE vfmacc_vf_f64m4
#define VFMSUB_DOUBLE vfmsub_vv_f64m4 // d = a*b - c
#define VREDSUM_DOUBLE vfredosum_vs_f64m4_f64m1
#define VREDMAX_DOUBLE vfredmax_vs_f64m4_f64m1
#define VREDMIN_DOUBLE vfredmin_vs_f64m4_f64m1
#define VMIN_DOUBLE vfmin_vv_f64m4
#define VMAX_DOUBLE vfmax_vv_f64m4
#define VMIN1_DOUBLE vfmin_vf_f64m4
#define VMAX1_DOUBLE vfmax_vf_f64m4
#define VINTERP_DOUBLE_INT vreinterpret_v_f64m4_i64m4
#define VINTERP_INT_DOUBLE vreinterpret_v_i64m4_f64m4
#define VCVT_RTZ_DOUBLE_INT vfcvt_rtz_x_f_v_i64m4
#define VCVT_DOUBLE_INT vfcvt_x_f_v_i64m4
#define VCVT_INT_DOUBLE vfcvt_f_x_v_f64m4
#define VABS_DOUBLE vfabs_v_f64m4
#define VSQRT_DOUBLE vfsqrt_v_f64m4
#define VCVT_DOUBLE_FLOAT vfncvt_f_f_w_f32m2
#define VCVT_FLOAT_DOUBLE vfwcvt_f_f_v_f64m4
#endif // ELEN >= 64
//// INT
#define V_ELT_INT vint32m4_t
#define VLOAD_INT vle32_v_i32m4
#define VLOAD1_INT vmv_v_x_i32m4
#define VSTORE_INT vse32_v_i32m4
#define VADD_INT vadd_vv_i32m4
#define VADD1_INT vadd_vx_i32m4
#define VMUL_INT vmul_vv_i32m4
#define VMUL1_INT vmul_vx_i32m4
#define VSUB_INT vsub_vv_i32m4
#define VSUB1_INT vsub_vx_i32m4
#define VAND1_INT vand_vx_i32m4
#define VAND_INT vand_vv_i32m4
#define VXOR_INT vxor_vv_i32m4
#define VSLL1_INT vsll_vx_i32m4
#define VEQ1_INT_BOOL vmseq_vx_i32m4_b8
#define VEQ_INT_BOOL vmseq_vv_i32m4_b8
#define VGT1_INT_BOOL vmsgt_vx_i32m4_b8
#define VNE1_INT_BOOL vmsne_vx_i32m4_b8
#define VLT1_INT_BOOL vmslt_vx_i32m4_b8
#define VLE1_INT_BOOL vmsle_vx_i32m4_b8
#define VADD1_INT_MASK vadd_vx_i32m4_m
#define VSUB1_INT_MASK vsub_vx_i32m4_m
#define VSUB1_INT vsub_vx_i32m4
#define VOR1_INT vor_vx_i32m4
#define VSRA1_INT vsra_vx_i32m4
#define VMIN_INT vmin_vv_i32m4
#define VMIN1_INT vmin_vx_i32m4
#define VMAX_INT vmax_vv_i32m4
#define VMAX1_INT vmax_vx_i32m4
#define VMERGE1_INT vmerge_vxm_i32m4
#define VMERGE_INT vmerge_vvm_i32m4
#define VNEG_INT vneg_v_i32m4
#define VREDSUM_INT vredosum_vs_i32m4_i32m1
#define VREDMAX_INT vredmax_vs_i32m4_i32m1
#define VREDMIN_INT vredmin_vs_i32m4_i32m1
#define VGATHER_INT vrgather_vv_i32m4
#define VNOT_INT vnot_v_i32m4
//// UINT
#define VLOAD_UINT vle32_v_u32m4
#define VSTORE_UINT vse32_v_u32m4
#define V_ELT_UINT vuint32m4_t
#define VCVT_FLOAT_UINT vfcvt_xu_f_v_u32m4
//// SHORT
#define V_ELT_SHORT vint16m4_t
#define VLOAD_SHORT vle16_v_i16m4
#define VLOAD1_SHORT vmv_v_x_i16m4
#define VSTORE_SHORT vse16_v_i16m4
#define VADD_SHORT vadd_vv_i16m4
#define VSUB_SHORT vsub_vv_i16m4
#define VREDSUMW_SHORT vwredsum_vs_i16m4_i32m1
#define VGT_SHORT_BOOL vmsgt_vv_i16m4_b4
#define VMERGE_SHORT vmerge_vvm_i16m4
//// BOOL for 16 bits elements
#define V_ELT_BOOL16 vbool4_t
//// BOOL for 32 bits elements
#define V_ELT_BOOL32 vbool8_t
#define VNOT_BOOL vmnot_m_b8
#define VCLEAR_BOOL vmclr_m_b8
#define VXOR_BOOL vmxor_mm_b8
#define VOR_BOOL vmor_mm_b8
#define VAND_BOOL vmand_mm_b8
#define VANDNOT_BOOL vmandn_mm_b8
/////////////////////////// HALF VECTOR, m2 ///////////////
#define VSETVL32H vsetvl_e32m2
#define VSETVL16H vsetvl_e16m2
#if ELEN >= 64
#define VSETVL64H vsetvl_e64m2
#endif // ELEN >= 64
//// FLOATH
#define V_ELT_FLOATH vfloat32m2_t
#define VLOAD_FLOATH vle32_v_f32m2
#define VLOAD1_FLOATH vfmv_v_f_f32m2
#define VLOAD_FLOATH2 vlseg2e32_v_f32m2
#define VLOAD_FLOATH_STRIDE vlse32_v_f32m2
#define VSTORE_FLOATH vse32_v_f32m2
#define VSTORE_FLOATH2 vsseg2e32_v_f32m2
#define VINTERP_FLOATH_INTH vreinterpret_v_f32m2_i32m2
#define VINTERP_INTH_FLOATH vreinterpret_v_i32m2_f32m2
#define VXOR1_INTH vxor_vx_i32m2
#define VADD_FLOATH vfadd_vv_f32m2
#define VADD1_FLOATH vfadd_vf_f32m2
#define VSUB_FLOATH vfsub_vv_f32m2
#define VSUB1_FLOATH vfsub_vf_f32m2 // v2 = v1 - f
#define VRSUB1_FLOATH vfrsub_vf_f32m2 // v2 = f - v1
#define VMUL_FLOATH vfmul_vv_f32m2
#define VMUL1_FLOATH vfmul_vf_f32m2
#define VMUL1_FLOATH_MASK vfmul_vf_f32m2_m
#define VDIV_FLOATH vfdiv_vv_f32m2
#define VDIV1_FLOATH vfdiv_vf_f32m2
#define VRDIV1_FLOATH vfrdiv_vf_f32m2
#define VFMACC_FLOATH vfmacc_vv_f32m2 // d = a + b*c
#define VFMACC1_FLOATH vfmacc_vf_f32m2
#define VFMADD_FLOATH vfmadd_vv_f32m2 // vd[i] = +(vs1[i] * vd[i]) + vs2[i]
#define VFMADD1_FLOATH vfmadd_vf_f32m2
#define VFMSUB_FLOATH vfmsub_vv_f32m2 // d = a*b - c
#define VREDSUM_FLOATH vfredosum_vs_f32m2_f32m1
#define VREDMAX_FLOATH vfredmax_vs_f32m2_f32m1
#define VREDMIN_FLOATH vfredmin_vs_f32m2_f32m1
#define VMIN_FLOATH vfmin_vv_f32m2
#define VMIN1_FLOATH vfmin_vf_f32m2
#define VMAX_FLOATH vfmax_vv_f32m2
#define VMAX1_FLOATH vfmax_vf_f32m2
#define VINTHERP_FLOATH_INTH vreinterpret_v_f32m2_i32m2
#define VINTHERP_INTH_FLOATH vreinterpret_v_i32m2_f32m2
#define VCVT_RTZ_FLOATH_INTH vfcvt_rtz_x_f_v_i32m2
#define VCVT_FLOATH_INTH vfcvt_x_f_v_i32m2
#define VCVT_INTH_FLOATH vfcvt_f_x_v_f32m2
#define VMERGE_FLOATH vmerge_vvm_f32m2
#define VSQRT_FLOATH vfsqrt_v_f32m2
#define VEQ1_FLOATH_BOOLH vmfeq_vf_f32m2_b16
#define VEQ_FLOATH_BOOLH vmfeq_vv_f32m2_b16
#define VGE1_FLOATH_BOOLH vmfge_vf_f32m2_b16
#define VGT1_FLOATH_BOOLH vmfgt_vf_f32m2_b16
#define VNE1_FLOATH_BOOLH vmfne_vf_f32m2_b16
#define VLT1_FLOATH_BOOLH vmflt_vf_f32m2_b16
#define VLE1_FLOATH_BOOLH vmfle_vf_f32m2_b16
#define VABS_FLOATH vfabs_v_f32m2
#define VMERGE1_FLOATH vfmerge_vfm_f32m2
#define VGATHER_FLOATH vrgather_vv_f32m2
#if ELEN >= 64
//// DOUBLEH
#define VSETVL64H vsetvl_e64m2
#define V_ELT_DOUBLEH vfloat64m2_t
#define VLOAD_DOUBLEH vle64_v_f64m2
#define VLOAD1_DOUBLEH vfmv_v_f_f64m2
#define VSTORE_DOUBLEH vse64_v_f64m2
#define VADD_DOUBLEH vfadd_vv_f64m2
#define VADD1_DOUBLEH vfadd_vf_f64m2
#define VSUB_DOUBLEH vfsub_vv_f64m2
#define VSUB1_DOUBLEH vfsub_vf_f64m2
#define VMUL_DOUBLEH vfmul_vv_f64m2
#define VMUL1_DOUBLEH vfmul_vf_f64m2
#define VDIV_DOUBLEH vfdiv_vv_f64m2
#define VFMA_DOUBLEH vfmacc_vv_f64m2 // d = a + b*c
#define VFMA1_DOUBLEH vfmacc_vf_f64m2
#define VFMSUB_DOUBLEH vfmsub_vv_f64m2 // d = a*b - c
#define VREDSUM_DOUBLEH vfredosum_vs_f64m2_f64m1
#define VREDMAX_DOUBLEH vfredmax_vs_f64m2_f64m1
#define VREDMIN_DOUBLEH vfredmin_vs_f64m2_f64m1
#define VMIN_DOUBLEH vfmin_vv_f64m2
#define VMAX_DOUBLEH vfmax_vv_f64m2
#define VMIN1_DOUBLEH vfmin_vf_f64m2
#define VMAX1_DOUBLEH vfmax_vf_f64m2
#define VINTHERP_DOUBLEH_INTH vreinterpret_v_f64m2_i64m2
#define VINTHERP_INTH_DOUBLEH vreinterpret_v_i64m2_f64m2
#define VCVT_RTZ_DOUBLEH_INTH vfcvt_rtz_x_f_v_i64m2
#define VCVT_DOUBLEH_INTH vfcvt_x_f_v_i64m2
#define VCVT_INTH_DOUBLEH vfcvt_f_x_v_f64m2
#define VABS_DOUBLEH vfabs_v_f64m2
#define VSQRT_DOUBLEH vfsqrt_v_f64m2
#define VCVT_DOUBLEH_FLOATH vfncvt_f_f_w_f32m2
#define VCVT_FLOATH_DOUBLEH vfwcvt_f_f_v_f64m2
#endif // ELEN >= 64
//// INTH
#define V_ELT_INTH vint32m2_t
#define VLOAD_INTH vle32_v_i32m2
#define VLOAD1_INTH vmv_v_x_i32m2
#define VSTORE_INTH vse32_v_i32m2
#define VADD_INTH vadd_vv_i32m2
#define VADD1_INTH vadd_vx_i32m2
#define VADD1_INTH_MASK vadd_vx_i32m2_m
#define VMUL_INTH vmul_vv_i32m2
#define VMUL1_INTH vmul_vx_i32m2
#define VSUB_INTH vsub_vv_i32m2
#define VSUB1_INTH vsub_vx_i32m2
#define VSUB1_INTH_MASK vsub_vx_i32m2_m
#define VAND1_INTH vand_vx_i32m2
#define VAND_INTH vand_vv_i32m2
#define VXOR_INTH vxor_vv_i32m2
#define VSLL1_INTH vsll_vx_i32m2
#define VEQ1_INTH_BOOLH vmseq_vx_i32m2_b16
#define VGT1_INTH_BOOLH vmsgt_vx_i32m2_b16
#define VNE1_INTH_BOOLH vmsne_vx_i32m2_b16
#define VLT1_INTH_BOOLH vmflt_vf_f32m2_b16
#define VLE1_INTH_BOOLH vmsle_vx_i32m2_b16
#define VEQ_INTH_BOOLH vmseq_vv_i32m2_b16
#define VOR1_INTH vor_vx_i32m2
#define VSRA1_INTH vsra_vx_i32m2
#define VMIN_INTH vmin_vv_i32m2
#define VMIN1_INTH vmin_vx_i32m2
#define VMAX_INTH vmax_vv_i32m2
#define VMAX1_INTH vmax_vx_i32m2
#define VNOT_INTH vnot_v_i32m2
#define VMERGE_INTH vmerge_vvm_i32m2
//// UINTH
#define VLOAD_UINTH vle32_v_u32m2
#define V_ELT_UINTH vuint32m2_t
#define VCVT_FLOATH_UINTH vfcvt_xu_f_v_u32m2
//// SHORTH
#define V_ELT_SHORTH vint16m2_t
#define VLOAD_SHORTH vle16_v_i16m2
#define VLOAD1_SHORTH vmv_v_x_i16m2
#define VSTORE_SHORTH vse16_v_i16m2
#define VADD_SHORTH vadd_vv_i16m2
#define VREDSUMW_SHORTH vwredsum_vs_i16m4_i32m1
#define VCVT_INT_SHORTH vnclip_wx_i16m2
#if __riscv_v != 7000
#define VCVT_SHORTH_INT vsext_vf2_i32m4
#else
#define VCVT_SHORTH_INT(a, b) vwmul_vx_i32m4((a), 1, (b))
#endif
//// USHORTH
#define V_ELT_USHORTH vuint16m2_t
#define VLOAD_USHORTH vle16_v_u16m2
#define VSTORE_USHORTH vse16_v_u16m2
#define VCVT_UINT_USHORTH vnclipu_wx_u16m2
//// UBYTEHH
#define V_ELT_UBYTEHH vuint8m1_t
#define VLOAD_UBYTEHH vle8_v_u8m1
#define VSTORE_UBYTEHH vse8_v_u8m1
#define VCVT_USHORTH_UBYTEHH vnclipu_wx_u8m1
//// BOOL for Half length vector 32 bits elements
#define V_ELT_BOOL32H vbool16_t
#define VNOT_BOOLH vmnot_m_b16
#define VCLEAR_BOOLH vmclr_m_b16
#define VXOR_BOOLH vmxor_mm_b16
#define VOR_BOOLH vmor_mm_b16
#define VAND_BOOLH vmand_mm_b16
#define VANDNOT_BOOLH vmandn_mm_b16
//#define VANDNOT_BOOLH vmnand_mm_b16
#endif // RISCV
#ifdef ALTIVEC
#include <altivec.h>
#endif
#ifdef _MSC_VER /* visual c++ */
#define ALIGN16_BEG __declspec(align(16))
#define ALIGN16_END
#define ALIGN32_BEG
#define ALIGN32_END __declspec(align(32))
#define ALIGN64_BEG
#define ALIGN64_END __declspec(align(64))
#else /* gcc,icc, clang */
#define ALIGN16_BEG
#define ALIGN16_END __attribute__((aligned(16)))
#define ALIGN32_BEG
#define ALIGN32_END __attribute__((aligned(32)))
#define ALIGN64_BEG
#define ALIGN64_END __attribute__((aligned(64)))
#endif
static const float FOPI = 1.27323954473516f;
static const float PIO4F = 0.7853981633974483096f;
#define PIF 3.14159265358979323846f // PI
#define mPIF -3.14159265358979323846f // -PI
#define PIO2F 1.57079632679489661923f // PI/2 1.570796326794896619
#define mPIO2F -1.57079632679489661923f // -PI/2 1.570796326794896619
/* Note, these constants are for a 32-bit significand: */
/*
static const float DP1 = 0.7853851318359375f;
static const float DP2 = 1.30315311253070831298828125e-5f;
static const float DP3 = 3.03855025325309630e-11f;
static const float lossth = 65536.f;
*/
/* These are for a 24-bit significand: */
static const float minus_cephes_DP1 = -0.78515625f;
static const float minus_cephes_DP2 = -2.4187564849853515625e-4f;
static const float minus_cephes_DP3 = -3.77489497744594108e-8f;
static float lossth = 8192.;
static const float T24M1 = 16777215.f;
static const float sincof[] = {-1.9515295891E-4f, 8.3321608736E-3f, -1.6666654611E-1f};
static const float coscof[] = {2.443315711809948E-5f, -1.388731625493765E-3f,
4.166664568298827E-2f};
#define SIGN_MASK 0x80000000
static const int32_t sign_mask = SIGN_MASK;
static const int32_t inv_sign_mask = ~SIGN_MASK;
#define neg_sign_mask ~0x7FFFFFFF
#define min_norm_pos 0x00800000
#define INVLN10 0.4342944819032518f // 0.4342944819f
#define INVLN2 1.4426950408889634f // 1.44269504089f
#define LN2 0.6931471805599453094172321214581765680755001343602552541206800094f
#define LN2_DIV_LN10 0.3010299956639811952137388947244930267681898814621085413104274611f
#define IMM8_FLIP_VEC 0x1B // change m128 from abcd to dcba
#define IMM8_LO_HI_VEC 0x1E // change m128 from abcd to cdab
#define IMM8_PERMUTE_128BITS_LANES 0x1 // reverse abcd efgh to efgh abcd
#ifndef M_PI
#define M_PI 3.14159265358979323846264338327950288
#endif
typedef union {
struct {
int16_t re;
int16_t im;
};
int16_t c[2];
} complex16s_t;
typedef union {
struct {
int32_t re;
int32_t im;
};
int32_t c[2];
} complex32s_t;
typedef union {
struct {
float re;
float im;
};
float c[2];
} complex32_t;
typedef union {
struct {
double re;
double im;
};
double c[2];
} complex64_t;
typedef enum {
RndZero,
RndNear,
RndFinancial,
} FloatRoundingMode;
#define c_inv_mant_mask ~0x7f800000u
#define c_cephes_SQRTHF 0.707106781186547524f
#define c_cephes_log_p0 7.0376836292E-2
#define c_cephes_log_p1 -1.1514610310E-1
#define c_cephes_log_p2 1.1676998740E-1
#define c_cephes_log_p3 -1.2420140846E-1
#define c_cephes_log_p4 +1.4249322787E-1
#define c_cephes_log_p5 -1.6668057665E-1
#define c_cephes_log_p6 +2.0000714765E-1
#define c_cephes_log_p7 -2.4999993993E-1
#define c_cephes_log_p8 +3.3333331174E-1
#define c_cephes_log_q1 -2.12194440e-4
#define c_cephes_log_q2 0.693359375
#define c_cephes_L102A 3.0078125E-1f
#define c_cephes_L102B 2.48745663981195213739E-4f
#define c_cephes_L10EA 4.3359375E-1f
#define c_cephes_L10EB 7.00731903251827651129E-4f
#define c_exp_hi 88.3762626647949f
#define c_exp_lo -88.3762626647949f
#define c_cephes_LOG2EF 1.44269504088896341f
#define c_cephes_LOG2EA 0.44269504088896340735992f
#define c_cephes_exp_C1 0.693359375f
#define c_cephes_exp_C2 -2.12194440e-4f
#define c_cephes_exp_p0 1.9875691500E-4f
#define c_cephes_exp_p1 1.3981999507E-3f
#define c_cephes_exp_p2 8.3334519073E-3f
#define c_cephes_exp_p3 4.1665795894E-2f
#define c_cephes_exp_p4 1.6666665459E-1f
#define c_cephes_exp_p5 5.0000001201E-1f
// TODO redundant with previous static float definitions, one should be removed
#define c_minus_cephes_DP1 -0.78515625f
#define c_minus_cephes_DP2 -2.4187564849853515625e-4f
#define c_minus_cephes_DP3 -3.77489497744594108e-8f
#define c_sincof_p0 -1.9515295891E-4f
#define c_sincof_p1 8.3321608736E-3f
#define c_sincof_p2 -1.6666654611E-1f
#define c_coscof_p0 2.443315711809948E-005f
#define c_coscof_p1 -1.388731625493765E-003f
#define c_coscof_p2 4.166664568298827E-002f
#define c_cephes_FOPI 1.27323954473516f // 4 / M_PI
#define ATAN_P0 8.05374449538e-2f
#define ATAN_P1 -1.38776856032E-1f
#define ATAN_P2 1.99777106478E-1f
#define ATAN_P3 -3.33329491539E-1f
#define TAN_P0 9.38540185543E-3f
#define TAN_P1 3.11992232697E-3f
#define TAN_P2 2.44301354525E-2f
#define TAN_P3 5.34112807005E-2f
#define TAN_P4 1.33387994085E-1f
#define TAN_P5 3.33331568548E-1f
#define ASIN_P0 4.2163199048E-2f
#define ASIN_P1 2.4181311049E-2f
#define ASIN_P2 4.5470025998E-2f
#define ASIN_P3 7.4953002686E-2f
#define ASIN_P4 1.6666752422E-1f
#define TANH_P0 -5.70498872745E-3f
#define TANH_P1 2.06390887954E-2f
#define TANH_P2 -5.37397155531E-2f
#define TANH_P3 1.33314422036E-1f
#define TANH_P4 -3.33332819422E-1f
#define SINH_P0 2.03721912945E-4f
#define SINH_P1 8.33028376239E-3f
#define SINH_P2 1.66667160211E-1f
#define ATANH_P0 1.81740078349E-1f
#define ATANH_P1 8.24370301058E-2f
#define ATANH_P2 1.46691431730E-1f
#define ATANH_P3 1.99782164500E-1f
#define ATANH_P4 3.33337300303E-1f
#define LOGE2F 0.693147180559945309f
#define ASINH_P0 2.0122003309E-2f
#define ASINH_P1 -4.2699340972E-2f
#define ASINH_P2 7.4847586088E-2f
#define ASINH_P3 -1.6666288134E-1f
#define ACOSH_P0 1.7596881071E-3f
#define ACOSH_P1 -7.5272886713E-3f
#define ACOSH_P2 2.6454905019E-2f
#define ACOSH_P3 -1.1784741703E-1f
#define ACOSH_P4 1.4142135263E0f
#define cephes_CBRT2 1.25992104989487316477f
#define cephes_CBRT4 1.58740105196819947475f
#define cephes_invCBRT2 0.7937005259840997373740956123328f
#define cephes_invCBRT4 0.6299605249474365823842821870329f
#define CBRTF_P0 -0.13466110473359520655053f
#define CBRTF_P1 0.54664601366395524503440f
#define CBRTF_P2 -0.95438224771509446525043f
#define CBRTF_P3 1.1399983354717293273738f
#define CBRTF_P4 0.40238979564544752126924f
#define TANPI8F 0.414213562373095048802f // tan(pi/8) => 0.4142135623730950
#define TAN3PI8F 2.414213562373095048802f // tan(3*pi/8) => 2.414213562373095
static const float MAXNUMF = 3.4028234663852885981170418348451692544e38f;
#define MAXLOGF 88.72283905206835f
#define MINLOGF -103.278929903431851103f
#define MAXLOGFDIV2 44.361419526034176f
#ifdef ALTIVEC
#define ALTIVEC_LEN_FLOAT 4
#define ALTIVEC_LEN_INT32 4
#define ALTIVEC_LEN_INT16 8
#define ALTIVEC_LEN_BYTES 16
#define ALIGN16_BEG
#define ALIGN16_END __attribute__((aligned(16)))
#define _PI8_CONST(Name, Val) \
static const ALIGN16_BEG unsigned char _pi8_##Name[16] ALIGN16_END = {Val, Val, Val, Val, Val, Val, \
Val, Val, Val, Val, Val, Val, Val, Val, Val, Val}
_PI8_CONST(0, 0x00);
_PI8_CONST(ff, 0xFF);
typedef __vector float v4sf;
typedef __vector int v4si;
typedef __vector unsigned int v4ui;
typedef __vector short v8ss;
typedef __vector unsigned short v8us;
typedef __vector unsigned char v16u8;
typedef __vector char v16s8;
typedef __vector bool int v4bi;
typedef struct {
v4sf val[2];
} v4sfx2;
// extract real and imaginary part with
// v4sf re = vec_perm(vec1, vec2, re_mask);
// v4sf im = vec_perm(vec1, vec2, im_mask);
static const v16u8 re_mask = {0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27};
static const v16u8 im_mask = {4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31};
static const v16u8 reim_mask_hi = {0, 1, 2, 3, 16, 17, 18, 19, 4, 5, 6, 7, 20, 21, 22, 23};
static const v16u8 reim_mask_lo = {8, 9, 10, 11, 24, 25, 26, 27, 12, 13, 14, 15, 28, 29, 30, 31};
static const v16u8 flip_vector = {12, 13, 14, 15, 8, 9, 10, 11, 4, 5, 6, 7, 0, 1, 2, 3};
#endif
#ifdef SSE
#define SSE_LEN_BYTES 16 // Size of SSE lane
#define SSE_LEN_INT16 8 // number of int16 with an SSE lane
#define SSE_LEN_INT32 4 // number of int32 with an SSE lane
#define SSE_LEN_FLOAT 4 // number of float with an SSE lane
#define SSE_LEN_DOUBLE 2 // number of double with an SSE lane
typedef __m128d v2sd; // vector of 2 double (sse)
typedef __m128i v2sid; // vector of 2 int64 (sse2)
#ifdef ARM
typedef float32x4_t v4sf; // vector of 4 float
typedef float32x4x2_t v4sfx2; // vector of 4 float
typedef uint32x4_t v4su; // vector of 4 uint32
typedef int32x4_t v4si; // vector of 4 uint32
typedef float32x4x2_t v4sfx2;
#if defined(__aarch64__)
typedef float64x2x2_t v2sdx2;
#else
typedef float32x4x2_t v2sdx2;
#endif
typedef int8x16_t v8ss;
typedef uint8x16_t v8us;
typedef uint16x8_t v16u8;
typedef uint16x8_t v16s8;
#else
typedef __m128 v4sf; // vector of 4 float (sse1)
typedef __m128i v4si; // vector of 4 int (sse2)
typedef __m128i v8ss;
typedef __m128i v8us;
typedef __m128i v16u8;
typedef __m128i v16s8;
typedef struct {
v4sf val[2];
} v4sfx2;
typedef struct {
v2sd val[2];
} v2sdx2;
#endif // ARM
#define ROUNDTONEAREST (_MM_FROUND_TO_NEAREST_INT | _MM_FROUND_NO_EXC)
#define ROUNDTOFLOOR (_MM_FROUND_TO_NEG_INF | _MM_FROUND_NO_EXC)
#define ROUNDTOCEIL (_MM_FROUND_TO_POS_INF | _MM_FROUND_NO_EXC)
#define ROUNDTOZERO (_MM_FROUND_TO_ZERO | _MM_FROUND_NO_EXC)
#define _PD_CONST(Name, Val) \
static const ALIGN16_BEG double _pd_##Name[2] ALIGN16_END = {Val, Val}
#define _PI64_CONST(Name, Val) \
static const ALIGN16_BEG int64_t _pi64_##Name[2] ALIGN16_END = {Val, Val}
#define _PD_CONST_TYPE(Name, Type, Val) \
static const ALIGN16_BEG Type _pd_##Name[2] ALIGN16_END = {Val, Val}
#endif // SSE
#if defined(SSE) || defined(ALTIVEC)
#define _PS_CONST(Name, Val) \
static const ALIGN16_BEG float _ps_##Name[4] ALIGN16_END = {Val, Val, Val, Val}
#define _PI32_CONST(Name, Val) \
static const ALIGN16_BEG int _pi32_##Name[4] ALIGN16_END = {Val, Val, Val, Val}
#define _PS_CONST_TYPE(Name, Type, Val) \
static const ALIGN16_BEG Type _ps_##Name[4] ALIGN16_END = {Val, Val, Val, Val}
// Warning, declared in reverse order since it's little endian :
// const v4sf conj_mask = _mm_set_ps(-1.0f, 1.0f, -1.0f, 1.0f);
#if defined(SSE)
static const float _ps_conj_mask[4] __attribute__((aligned(16))) = {1.0f, -1.0f, 1.0f, -1.0f};
#else // ALTIVEC, big endian
static const float _ps_conj_mask[4] __attribute__((aligned(16))) = {-1.0f, 1.0f, -1.0f, 1.0f};
#endif
/////////////// INT //////////////////
_PI32_CONST(0, 0);
_PI32_CONST(1, 1);
_PI32_CONST(inv1, ~1);
_PI32_CONST(2, 2);
_PI32_CONST(4, 4);
_PI32_CONST(0x7e, 0x7e);
_PI32_CONST(0x7f, 0x7f);
/////////////// SINGLE //////////////////
_PS_CONST(0, 0.0f);
_PS_CONST(min0, -0.0f);
_PS_CONST(1, 1.0f);
_PS_CONST(0p3, 0.333333333333f);
_PS_CONST(min0p3, -0.333333333333f);
_PS_CONST(0p5, 0.5f);
_PS_CONST(min1, -1.0f);
_PS_CONST(min2, -2.0f);
_PS_CONST(min0p5, -0.5f);
_PS_CONST(3, 3.0f);
/* the smallest non denormalized float number */
_PS_CONST_TYPE(min_norm_pos, int, 0x00800000);
_PS_CONST_TYPE(mant_mask, int, 0x7f800000);
_PS_CONST_TYPE(inv_mant_mask, int, ~0x7f800000); // 1000 0000 0111 1111 1111 1111 1111 1111
_PS_CONST_TYPE(sign_mask, int, (int) 0x80000000);
_PS_CONST_TYPE(inv_sign_mask, int, ~0x80000000);
_PS_CONST(cephes_SQRTHF, 0.707106781186547524f);
_PS_CONST(cephes_log_p0, 7.0376836292E-2f);
_PS_CONST(cephes_log_p1, -1.1514610310E-1f);
_PS_CONST(cephes_log_p2, 1.1676998740E-1f);
_PS_CONST(cephes_log_p3, -1.2420140846E-1f);
_PS_CONST(cephes_log_p4, +1.4249322787E-1f);
_PS_CONST(cephes_log_p5, -1.6668057665E-1f);
_PS_CONST(cephes_log_p6, +2.0000714765E-1f);
_PS_CONST(cephes_log_p7, -2.4999993993E-1f);
_PS_CONST(cephes_log_p8, +3.3333331174E-1f);
_PS_CONST(cephes_log_q1, -2.12194440e-4f);
_PS_CONST(cephes_log_q2, 0.693359375f);
_PS_CONST(exp_hi, 88.3762626647949f);
_PS_CONST(exp_lo, -88.3762626647949f);
_PS_CONST(cephes_LOG2EF, 1.44269504088896341f);
_PS_CONST(cephes_exp_C1, 0.693359375f);
_PS_CONST(cephes_exp_C2, -2.12194440e-4f);
_PS_CONST(cephes_exp_p0, 1.9875691500E-4f);
_PS_CONST(cephes_exp_p1, 1.3981999507E-3f);
_PS_CONST(cephes_exp_p2, 8.3334519073E-3f);
_PS_CONST(cephes_exp_p3, 4.1665795894E-2f);
_PS_CONST(cephes_exp_p4, 1.6666665459E-1f);
_PS_CONST(cephes_exp_p5, 5.0000001201E-1f);
_PS_CONST(minus_cephes_DP1, -0.78515625f);
_PS_CONST(minus_cephes_DP2, -2.4187564849853515625e-4f);
_PS_CONST(minus_cephes_DP3, -3.77489497744594108e-8f);
_PS_CONST(sincof_p0, -1.9515295891E-4f);
_PS_CONST(sincof_p1, 8.3321608736E-3f);
_PS_CONST(sincof_p2, -1.6666654611E-1f);
_PS_CONST(coscof_p0, 2.443315711809948E-005f);
_PS_CONST(coscof_p1, -1.388731625493765E-003f);
_PS_CONST(coscof_p2, 4.166664568298827E-002f);
_PS_CONST(cephes_FOPI, 1.27323954473516f); // 4 / M_PI
// For tanf
_PS_CONST(DP123, 0.78515625 + 2.4187564849853515625e-4 + 3.77489497744594108e-8);
// Neg values to better migrate to FMA
_PS_CONST(DP1, -0.78515625f);
_PS_CONST(DP2, -2.4187564849853515625E-4f);
_PS_CONST(DP3, -3.77489497744594108E-8f);
_PS_CONST(FOPI, 1.27323954473516f); /* 4/pi */
_PS_CONST(TAN_P0, 9.38540185543E-3f);
_PS_CONST(TAN_P1, 3.11992232697E-3f);
_PS_CONST(TAN_P2, 2.44301354525E-2f);
_PS_CONST(TAN_P3, 5.34112807005E-2f);
_PS_CONST(TAN_P4, 1.33387994085E-1f);
_PS_CONST(TAN_P5, 3.33331568548E-1f);
_PS_CONST(ASIN_P0, 4.2163199048E-2f);
_PS_CONST(ASIN_P1, 2.4181311049E-2f);
_PS_CONST(ASIN_P2, 4.5470025998E-2f);
_PS_CONST(ASIN_P3, 7.4953002686E-2f);
_PS_CONST(ASIN_P4, 1.6666752422E-1f);
_PS_CONST(PIF, 3.14159265358979323846f); // PI
_PS_CONST(mPIF, -3.14159265358979323846f); // -PI
_PS_CONST(PIO2F, 1.57079632679489661923f); // PI/2 1.570796326794896619
_PS_CONST(mPIO2F, -1.57079632679489661923f); // -PI/2 1.570796326794896619
_PS_CONST(PIO4F, 0.785398163397448309615f); // PI/4 0.7853981633974483096
_PS_CONST(TANPI8F, 0.414213562373095048802f); // tan(pi/8) => 0.4142135623730950
_PS_CONST(TAN3PI8F, 2.414213562373095048802f); // tan(3*pi/8) => 2.414213562373095
_PS_CONST(ATAN_P0, 8.05374449538e-2f);
_PS_CONST(ATAN_P1, -1.38776856032E-1f);
_PS_CONST(ATAN_P2, 1.99777106478E-1f);
_PS_CONST(ATAN_P3, -3.33329491539E-1f);
_PS_CONST_TYPE(pos_sign_mask, int, (int) 0x7FFFFFFF);
_PS_CONST_TYPE(neg_sign_mask, int, (int) ~0x7FFFFFFF);
_PS_CONST(MAXLOGF, 88.72283905206835f);
_PS_CONST(MAXLOGFDIV2, 44.361419526034176f);
_PS_CONST(MINLOGF, -103.278929903431851103f);
_PS_CONST(cephes_exp_minC1, -0.693359375f);
_PS_CONST(cephes_exp_minC2, 2.12194440e-4f);
_PS_CONST(0p625, 0.625f);
_PS_CONST(TANH_P0, -5.70498872745E-3f);
_PS_CONST(TANH_P1, 2.06390887954E-2f);
_PS_CONST(TANH_P2, -5.37397155531E-2f);
_PS_CONST(TANH_P3, 1.33314422036E-1f);
_PS_CONST(TANH_P4, -3.33332819422E-1f);
_PS_CONST(MAXNUMF, 3.4028234663852885981170418348451692544e38f);
_PS_CONST(minMAXNUMF, -3.4028234663852885981170418348451692544e38f);
_PS_CONST(SINH_P0, 2.03721912945E-4f);
_PS_CONST(SINH_P1, 8.33028376239E-3f);
_PS_CONST(SINH_P2, 1.66667160211E-1f);
_PS_CONST(1emin4, 1e-4f);
_PS_CONST(ATANH_P0, 1.81740078349E-1f);
_PS_CONST(ATANH_P1, 8.24370301058E-2f);
_PS_CONST(ATANH_P2, 1.46691431730E-1f);
_PS_CONST(ATANH_P3, 1.99782164500E-1f);
_PS_CONST(ATANH_P4, 3.33337300303E-1f);
_PS_CONST_TYPE(zero, int, (int) 0x00000000);
_PS_CONST(1500, 1500.0f);
_PS_CONST(LOGE2F, 0.693147180559945309f);
_PS_CONST(ASINH_P0, 2.0122003309E-2f);
_PS_CONST(ASINH_P1, -4.2699340972E-2f);
_PS_CONST(ASINH_P2, 7.4847586088E-2f);
_PS_CONST(ASINH_P3, -1.6666288134E-1f);
_PS_CONST(ACOSH_P0, 1.7596881071E-3f);
_PS_CONST(ACOSH_P1, -7.5272886713E-3f);
_PS_CONST(ACOSH_P2, 2.6454905019E-2f);
_PS_CONST(ACOSH_P3, -1.1784741703E-1f);
_PS_CONST(ACOSH_P4, 1.4142135263E0f);
/* For log10f */
_PS_CONST(cephes_L102A, 3.0078125E-1f);
_PS_CONST(cephes_L102B, 2.48745663981195213739E-4f);
_PS_CONST(cephes_L10EA, 4.3359375E-1f);
_PS_CONST(cephes_L10EB, 7.00731903251827651129E-4f);
/* For log2f */
_PS_CONST(cephes_LOG2EA, 0.44269504088896340735992f);
/* For cbrtf */
_PS_CONST(cephes_CBRT2, 1.25992104989487316477f);
_PS_CONST(cephes_CBRT4, 1.58740105196819947475f);
_PS_CONST(cephes_invCBRT2, 0.7937005259840997373740956123328f);
_PS_CONST(cephes_invCBRT4, 0.6299605249474365823842821870329f);
_PS_CONST(CBRTF_P0, -0.13466110473359520655053f);
_PS_CONST(CBRTF_P1, 0.54664601366395524503440f);
_PS_CONST(CBRTF_P2, -0.95438224771509446525043f);
_PS_CONST(CBRTF_P3, 1.1399983354717293273738f);
_PS_CONST(CBRTF_P4, 0.40238979564544752126924f);
#endif // SSE/ARM || ALTIVEC
#ifdef SSE // or ARM. no double precision for ALTIVEC
/////////////// INT64 //////////////
_PI64_CONST(1, 1);
_PI64_CONST(inv1, ~1);
_PI64_CONST(2, 2);
_PI64_CONST(4, 4);
_PI64_CONST(0x7f, 0x7f);
_PI64_CONST(0x3ff, 0x3ff);
_PI64_CONST(1023, (unsigned int) 1023);
/////////////// DOUBLE //////////////////
_PD_CONST_TYPE(zero, int, (int) 0x00000000);
_PD_CONST_TYPE(min_norm_pos, int64_t, 0x380ffff83ce549caL);
_PD_CONST_TYPE(mant_mask, int64_t, 0xFFFFFFFFFFFFFL);
_PD_CONST_TYPE(inv_mant_mask, int64_t, (int64_t) 0x800FFFFFFFFFFFFFL);
_PD_CONST_TYPE(sign_mask, int64_t, (int64_t) 0x8000000000000000L);
_PD_CONST_TYPE(inv_sign_mask, int64_t, ~0x8000000000000000L);
_PD_CONST_TYPE(pos_sign_mask, int64_t, 0xFFFFFFFFFFFFFL);
_PD_CONST_TYPE(epi64_mask, double, 0x0018000000000000L);
_PD_CONST(minus_cephes_DP1, -7.85398125648498535156E-1);
_PD_CONST(minus_cephes_DP2, -3.77489470793079817668E-8);
_PD_CONST(minus_cephes_DP3, -2.69515142907905952645E-15);
_PD_CONST(sincof_p0, 1.58962301576546568060E-10);
_PD_CONST(sincof_p1, -2.50507477628578072866E-8);
_PD_CONST(sincof_p2, 2.75573136213857245213E-6);
_PD_CONST(sincof_p3, -1.98412698295895385996E-4);
_PD_CONST(sincof_p4, 8.33333333332211858878E-3);
_PD_CONST(sincof_p5, -1.66666666666666307295E-1);
_PD_CONST(coscof_p0, -1.13585365213876817300E-11);
_PD_CONST(coscof_p1, 2.08757008419747316778E-9);
_PD_CONST(coscof_p2, -2.75573141792967388112E-7);
_PD_CONST(coscof_p3, 2.48015872888517045348E-5);