aboutsummaryrefslogtreecommitdiffstats
path: root/libavcodec/arm/dsputil_neon.S
blob: 33969134d73564282957fbf64420d575a8fb867e (plain) (blame)
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
/*
 * ARM NEON optimised DSP functions
 * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
 *
 * This file is part of Libav.
 *
 * Libav is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * Libav is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with Libav; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "config.h"
#include "asm.S"

        preserve8

function ff_clear_block_neon, export=1
        vmov.i16        q0,  #0
        .rept           8
        vst1.16         {q0}, [r0,:128]!
        .endr
        bx              lr
endfunc

function ff_clear_blocks_neon, export=1
        vmov.i16        q0,  #0
        .rept           8*6
        vst1.16         {q0}, [r0,:128]!
        .endr
        bx              lr
endfunc

        .macro pixels16 avg=0
.if \avg
        mov             ip,  r0
.endif
1:      vld1.64         {d0, d1},  [r1], r2
        vld1.64         {d2, d3},  [r1], r2
        vld1.64         {d4, d5},  [r1], r2
        pld             [r1, r2, lsl #2]
        vld1.64         {d6, d7},  [r1], r2
        pld             [r1]
        pld             [r1, r2]
        pld             [r1, r2, lsl #1]
.if \avg
        vld1.64         {d16,d17}, [ip,:128], r2
        vrhadd.u8       q0,  q0,  q8
        vld1.64         {d18,d19}, [ip,:128], r2
        vrhadd.u8       q1,  q1,  q9
        vld1.64         {d20,d21}, [ip,:128], r2
        vrhadd.u8       q2,  q2,  q10
        vld1.64         {d22,d23}, [ip,:128], r2
        vrhadd.u8       q3,  q3,  q11
.endif
        subs            r3,  r3,  #4
        vst1.64         {d0, d1},  [r0,:128], r2
        vst1.64         {d2, d3},  [r0,:128], r2
        vst1.64         {d4, d5},  [r0,:128], r2
        vst1.64         {d6, d7},  [r0,:128], r2
        bne             1b
        bx              lr
        .endm

        .macro pixels16_x2 vhadd=vrhadd.u8
1:      vld1.64         {d0-d2},   [r1], r2
        vld1.64         {d4-d6},   [r1], r2
        pld             [r1]
        pld             [r1, r2]
        subs            r3,  r3,  #2
        vext.8          q1,  q0,  q1,  #1
        \vhadd          q0,  q0,  q1
        vext.8          q3,  q2,  q3,  #1
        \vhadd          q2,  q2,  q3
        vst1.64         {d0, d1},  [r0,:128], r2
        vst1.64         {d4, d5},  [r0,:128], r2
        bne             1b
        bx              lr
        .endm

        .macro pixels16_y2 vhadd=vrhadd.u8
        vld1.64         {d0, d1},  [r1], r2
        vld1.64         {d2, d3},  [r1], r2
1:      subs            r3,  r3,  #2
        \vhadd          q2,  q0,  q1
        vld1.64         {d0, d1},  [r1], r2
        \vhadd          q3,  q0,  q1
        vld1.64         {d2, d3},  [r1], r2
        pld             [r1]
        pld             [r1, r2]
        vst1.64         {d4, d5},  [r0,:128], r2
        vst1.64         {d6, d7},  [r0,:128], r2
        bne             1b
        bx              lr
        .endm

        .macro pixels16_xy2 vshrn=vrshrn.u16 no_rnd=0
        vld1.64         {d0-d2},   [r1], r2
        vld1.64         {d4-d6},   [r1], r2
.if \no_rnd
        vmov.i16        q13, #1
.endif
        pld             [r1]
        pld             [r1, r2]
        vext.8          q1,  q0,  q1,  #1
        vext.8          q3,  q2,  q3,  #1
        vaddl.u8        q8,  d0,  d2
        vaddl.u8        q10, d1,  d3
        vaddl.u8        q9,  d4,  d6
        vaddl.u8        q11, d5,  d7
1:      subs            r3,  r3,  #2
        vld1.64         {d0-d2},   [r1], r2
        vadd.u16        q12, q8,  q9
        pld             [r1]
.if \no_rnd
        vadd.u16        q12, q12, q13
.endif
        vext.8          q15, q0,  q1,  #1
        vadd.u16        q1 , q10, q11
        \vshrn          d28, q12, #2
.if \no_rnd
        vadd.u16        q1,  q1,  q13
.endif
        \vshrn          d29, q1,  #2
        vaddl.u8        q8,  d0,  d30
        vld1.64         {d2-d4},   [r1], r2
        vaddl.u8        q10, d1,  d31
        vst1.64         {d28,d29}, [r0,:128], r2
        vadd.u16        q12, q8,  q9
        pld             [r1, r2]
.if \no_rnd
        vadd.u16        q12, q12, q13
.endif
        vext.8          q2,  q1,  q2,  #1
        vadd.u16        q0,  q10, q11
        \vshrn          d30, q12, #2
.if \no_rnd
        vadd.u16        q0,  q0,  q13
.endif
        \vshrn          d31, q0,  #2
        vaddl.u8        q9,  d2,  d4
        vaddl.u8        q11, d3,  d5
        vst1.64         {d30,d31}, [r0,:128], r2
        bgt             1b
        bx              lr
        .endm

        .macro pixels8 avg=0
1:      vld1.64         {d0}, [r1], r2
        vld1.64         {d1}, [r1], r2
        vld1.64         {d2}, [r1], r2
        pld             [r1, r2, lsl #2]
        vld1.64         {d3}, [r1], r2
        pld             [r1]
        pld             [r1, r2]
        pld             [r1, r2, lsl #1]
.if \avg
        vld1.64         {d4}, [r0,:64], r2
        vrhadd.u8       d0,  d0,  d4
        vld1.64         {d5}, [r0,:64], r2
        vrhadd.u8       d1,  d1,  d5
        vld1.64         {d6}, [r0,:64], r2
        vrhadd.u8       d2,  d2,  d6
        vld1.64         {d7}, [r0,:64], r2
        vrhadd.u8       d3,  d3,  d7
        sub             r0,  r0,  r2,  lsl #2
.endif
        subs            r3,  r3,  #4
        vst1.64         {d0}, [r0,:64], r2
        vst1.64         {d1}, [r0,:64], r2
        vst1.64         {d2}, [r0,:64], r2
        vst1.64         {d3}, [r0,:64], r2
        bne             1b
        bx              lr
        .endm

        .macro pixels8_x2 vhadd=vrhadd.u8
1:      vld1.64         {d0, d1},  [r1], r2
        vext.8          d1,  d0,  d1,  #1
        vld1.64         {d2, d3},  [r1], r2
        vext.8          d3,  d2,  d3,  #1
        pld             [r1]
        pld             [r1, r2]
        subs            r3,  r3,  #2
        vswp            d1,  d2
        \vhadd          q0,  q0,  q1
        vst1.64         {d0},      [r0,:64], r2
        vst1.64         {d1},      [r0,:64], r2
        bne             1b
        bx              lr
        .endm

        .macro pixels8_y2 vhadd=vrhadd.u8
        vld1.64         {d0},      [r1], r2
        vld1.64         {d1},      [r1], r2
1:      subs            r3,  r3,  #2
        \vhadd          d4,  d0,  d1
        vld1.64         {d0},      [r1], r2
        \vhadd          d5,  d0,  d1
        vld1.64         {d1},      [r1], r2
        pld             [r1]
        pld             [r1, r2]
        vst1.64         {d4},      [r0,:64], r2
        vst1.64         {d5},      [r0,:64], r2
        bne             1b
        bx              lr
        .endm

        .macro pixels8_xy2 vshrn=vrshrn.u16 no_rnd=0
        vld1.64         {d0, d1},  [r1], r2
        vld1.64         {d2, d3},  [r1], r2
.if \no_rnd
        vmov.i16        q11, #1
.endif
        pld             [r1]
        pld             [r1, r2]
        vext.8          d4,  d0,  d1,  #1
        vext.8          d6,  d2,  d3,  #1
        vaddl.u8        q8,  d0,  d4
        vaddl.u8        q9,  d2,  d6
1:      subs            r3,  r3,  #2
        vld1.64         {d0, d1},  [r1], r2
        pld             [r1]
        vadd.u16        q10, q8,  q9
        vext.8          d4,  d0,  d1,  #1
.if \no_rnd
        vadd.u16        q10, q10, q11
.endif
        vaddl.u8        q8,  d0,  d4
        \vshrn          d5,  q10, #2
        vld1.64         {d2, d3},  [r1], r2
        vadd.u16        q10, q8,  q9
        pld             [r1, r2]
.if \no_rnd
        vadd.u16        q10, q10, q11
.endif
        vst1.64         {d5},      [r0,:64], r2
        \vshrn          d7,  q10, #2
        vext.8          d6,  d2,  d3,  #1
        vaddl.u8        q9,  d2,  d6
        vst1.64         {d7},      [r0,:64], r2
        bgt             1b
        bx              lr
        .endm

        .macro pixfunc pfx name suf rnd_op args:vararg
function ff_\pfx\name\suf\()_neon, export=1
        \name \rnd_op \args
endfunc
        .endm

        .macro pixfunc2 pfx name args:vararg
        pixfunc \pfx \name
        pixfunc \pfx \name \args
        .endm

function ff_put_h264_qpel16_mc00_neon, export=1
        mov             r3,  #16
endfunc

        pixfunc  put_ pixels16
        pixfunc2 put_ pixels16_x2,  _no_rnd, vhadd.u8
        pixfunc2 put_ pixels16_y2,  _no_rnd, vhadd.u8
        pixfunc2 put_ pixels16_xy2, _no_rnd, vshrn.u16, 1

function ff_avg_h264_qpel16_mc00_neon, export=1
        mov             r3,  #16
endfunc

        pixfunc  avg_ pixels16,, 1

function ff_put_h264_qpel8_mc00_neon, export=1
        mov             r3,  #8
endfunc

        pixfunc  put_ pixels8
        pixfunc2 put_ pixels8_x2,   _no_rnd, vhadd.u8
        pixfunc2 put_ pixels8_y2,   _no_rnd, vhadd.u8
        pixfunc2 put_ pixels8_xy2,  _no_rnd, vshrn.u16, 1

function ff_avg_h264_qpel8_mc00_neon, export=1
        mov             r3,  #8
endfunc

        pixfunc  avg_ pixels8,, 1

function ff_put_pixels_clamped_neon, export=1
        vld1.64         {d16-d19}, [r0,:128]!
        vqmovun.s16     d0, q8
        vld1.64         {d20-d23}, [r0,:128]!
        vqmovun.s16     d1, q9
        vld1.64         {d24-d27}, [r0,:128]!
        vqmovun.s16     d2, q10
        vld1.64         {d28-d31}, [r0,:128]!
        vqmovun.s16     d3, q11
        vst1.64         {d0},      [r1,:64], r2
        vqmovun.s16     d4, q12
        vst1.64         {d1},      [r1,:64], r2
        vqmovun.s16     d5, q13
        vst1.64         {d2},      [r1,:64], r2
        vqmovun.s16     d6, q14
        vst1.64         {d3},      [r1,:64], r2
        vqmovun.s16     d7, q15
        vst1.64         {d4},      [r1,:64], r2
        vst1.64         {d5},      [r1,:64], r2
        vst1.64         {d6},      [r1,:64], r2
        vst1.64         {d7},      [r1,:64], r2
        bx              lr
endfunc

function ff_put_signed_pixels_clamped_neon, export=1
        vmov.u8         d31, #128
        vld1.64         {d16-d17}, [r0,:128]!
        vqmovn.s16      d0, q8
        vld1.64         {d18-d19}, [r0,:128]!
        vqmovn.s16      d1, q9
        vld1.64         {d16-d17}, [r0,:128]!
        vqmovn.s16      d2, q8
        vld1.64         {d18-d19}, [r0,:128]!
        vadd.u8         d0, d0, d31
        vld1.64         {d20-d21}, [r0,:128]!
        vadd.u8         d1, d1, d31
        vld1.64         {d22-d23}, [r0,:128]!
        vadd.u8         d2, d2, d31
        vst1.64         {d0},      [r1,:64], r2
        vqmovn.s16      d3, q9
        vst1.64         {d1},      [r1,:64], r2
        vqmovn.s16      d4, q10
        vst1.64         {d2},      [r1,:64], r2
        vqmovn.s16      d5, q11
        vld1.64         {d24-d25}, [r0,:128]!
        vadd.u8         d3, d3, d31
        vld1.64         {d26-d27}, [r0,:128]!
        vadd.u8         d4, d4, d31
        vadd.u8         d5, d5, d31
        vst1.64         {d3},      [r1,:64], r2
        vqmovn.s16      d6, q12
        vst1.64         {d4},      [r1,:64], r2
        vqmovn.s16      d7, q13
        vst1.64         {d5},      [r1,:64], r2
        vadd.u8         d6, d6, d31
        vadd.u8         d7, d7, d31
        vst1.64         {d6},      [r1,:64], r2
        vst1.64         {d7},      [r1,:64], r2
        bx              lr
endfunc

function ff_add_pixels_clamped_neon, export=1
        mov             r3, r1
        vld1.64         {d16},   [r1,:64], r2
        vld1.64         {d0-d1}, [r0,:128]!
        vaddw.u8        q0, q0, d16
        vld1.64         {d17},   [r1,:64], r2
        vld1.64         {d2-d3}, [r0,:128]!
        vqmovun.s16     d0, q0
        vld1.64         {d18},   [r1,:64], r2
        vaddw.u8        q1, q1, d17
        vld1.64         {d4-d5}, [r0,:128]!
        vaddw.u8        q2, q2, d18
        vst1.64         {d0},    [r3,:64], r2
        vqmovun.s16     d2, q1
        vld1.64         {d19},   [r1,:64], r2
        vld1.64         {d6-d7}, [r0,:128]!
        vaddw.u8        q3, q3, d19
        vqmovun.s16     d4, q2
        vst1.64         {d2},    [r3,:64], r2
        vld1.64         {d16},   [r1,:64], r2
        vqmovun.s16     d6, q3
        vld1.64         {d0-d1}, [r0,:128]!
        vaddw.u8        q0, q0, d16
        vst1.64         {d4},    [r3,:64], r2
        vld1.64         {d17},   [r1,:64], r2
        vld1.64         {d2-d3}, [r0,:128]!
        vaddw.u8        q1, q1, d17
        vst1.64         {d6},    [r3,:64], r2
        vqmovun.s16     d0, q0
        vld1.64         {d18},   [r1,:64], r2
        vld1.64         {d4-d5}, [r0,:128]!
        vaddw.u8        q2, q2, d18
        vst1.64         {d0},    [r3,:64], r2
        vqmovun.s16     d2, q1
        vld1.64         {d19},   [r1,:64], r2
        vqmovun.s16     d4, q2
        vld1.64         {d6-d7}, [r0,:128]!
        vaddw.u8        q3, q3, d19
        vst1.64         {d2},    [r3,:64], r2
        vqmovun.s16     d6, q3
        vst1.64         {d4},    [r3,:64], r2
        vst1.64         {d6},    [r3,:64], r2
        bx              lr
endfunc

function ff_vector_fmul_neon, export=1
        subs            r3,  r3,  #8
        vld1.64         {d0-d3},  [r1,:128]!
        vld1.64         {d4-d7},  [r2,:128]!
        vmul.f32        q8,  q0,  q2
        vmul.f32        q9,  q1,  q3
        beq             3f
        bics            ip,  r3,  #15
        beq             2f
1:      subs            ip,  ip,  #16
        vld1.64         {d0-d1},  [r1,:128]!
        vld1.64         {d4-d5},  [r2,:128]!
        vmul.f32        q10, q0,  q2
        vld1.64         {d2-d3},  [r1,:128]!
        vld1.64         {d6-d7},  [r2,:128]!
        vmul.f32        q11, q1,  q3
        vst1.64         {d16-d19},[r0,:128]!
        vld1.64         {d0-d1},  [r1,:128]!
        vld1.64         {d4-d5},  [r2,:128]!
        vmul.f32        q8,  q0,  q2
        vld1.64         {d2-d3},  [r1,:128]!
        vld1.64         {d6-d7},  [r2,:128]!
        vmul.f32        q9,  q1,  q3
        vst1.64         {d20-d23},[r0,:128]!
        bne             1b
        ands            r3,  r3,  #15
        beq             3f
2:      vld1.64         {d0-d1},  [r1,:128]!
        vld1.64         {d4-d5},  [r2,:128]!
        vst1.64         {d16-d17},[r0,:128]!
        vmul.f32        q8,  q0,  q2
        vld1.64         {d2-d3},  [r1,:128]!
        vld1.64         {d6-d7},  [r2,:128]!
        vst1.64         {d18-d19},[r0,:128]!
        vmul.f32        q9,  q1,  q3
3:      vst1.64         {d16-d19},[r0,:128]!
        bx              lr
endfunc

function ff_vector_fmul_window_neon, export=1
        push            {r4,r5,lr}
        ldr             lr,  [sp, #12]
        sub             r2,  r2,  #8
        sub             r5,  lr,  #2
        add             r2,  r2,  r5, lsl #2
        add             r4,  r3,  r5, lsl #3
        add             ip,  r0,  r5, lsl #3
        mov             r5,  #-16
        vld1.64         {d0,d1},  [r1,:128]!
        vld1.64         {d2,d3},  [r2,:128], r5
        vld1.64         {d4,d5},  [r3,:128]!
        vld1.64         {d6,d7},  [r4,:128], r5
1:      subs            lr,  lr,  #4
        vmul.f32        d22, d0,  d4
        vrev64.32       q3,  q3
        vmul.f32        d23, d1,  d5
        vrev64.32       q1,  q1
        vmul.f32        d20, d0,  d7
        vmul.f32        d21, d1,  d6
        beq             2f
        vmla.f32        d22, d3,  d7
        vld1.64         {d0,d1},  [r1,:128]!
        vmla.f32        d23, d2,  d6
        vld1.64         {d18,d19},[r2,:128], r5
        vmls.f32        d20, d3,  d4
        vld1.64         {d24,d25},[r3,:128]!
        vmls.f32        d21, d2,  d5
        vld1.64         {d6,d7},  [r4,:128], r5
        vmov            q1,  q9
        vrev64.32       q11, q11
        vmov            q2,  q12
        vswp            d22, d23
        vst1.64         {d20,d21},[r0,:128]!
        vst1.64         {d22,d23},[ip,:128], r5
        b               1b
2:      vmla.f32        d22, d3,  d7
        vmla.f32        d23, d2,  d6
        vmls.f32        d20, d3,  d4
        vmls.f32        d21, d2,  d5
        vrev64.32       q11, q11
        vswp            d22, d23
        vst1.64         {d20,d21},[r0,:128]!
        vst1.64         {d22,d23},[ip,:128], r5
        pop             {r4,r5,pc}
endfunc

#if CONFIG_VORBIS_DECODER
function ff_vorbis_inverse_coupling_neon, export=1
        vmov.i32        q10, #1<<31
        subs            r2,  r2,  #4
        mov             r3,  r0
        mov             r12, r1
        beq             3f

        vld1.32         {d24-d25},[r1,:128]!
        vld1.32         {d22-d23},[r0,:128]!
        vcle.s32        q8,  q12, #0
        vand            q9,  q11, q10
        veor            q12, q12, q9
        vand            q2,  q12, q8
        vbic            q3,  q12, q8
        vadd.f32        q12, q11, q2
        vsub.f32        q11, q11, q3
1:      vld1.32         {d2-d3},  [r1,:128]!
        vld1.32         {d0-d1},  [r0,:128]!
        vcle.s32        q8,  q1,  #0
        vand            q9,  q0,  q10
        veor            q1,  q1,  q9
        vst1.32         {d24-d25},[r3, :128]!
        vst1.32         {d22-d23},[r12,:128]!
        vand            q2,  q1,  q8
        vbic            q3,  q1,  q8
        vadd.f32        q1,  q0,  q2
        vsub.f32        q0,  q0,  q3
        subs            r2,  r2,  #8
        ble             2f
        vld1.32         {d24-d25},[r1,:128]!
        vld1.32         {d22-d23},[r0,:128]!
        vcle.s32        q8,  q12, #0
        vand            q9,  q11, q10
        veor            q12, q12, q9
        vst1.32         {d2-d3},  [r3, :128]!
        vst1.32         {d0-d1},  [r12,:128]!
        vand            q2,  q12, q8
        vbic            q3,  q12, q8
        vadd.f32        q12, q11, q2
        vsub.f32        q11, q11, q3
        b               1b

2:      vst1.32         {d2-d3},  [r3, :128]!
        vst1.32         {d0-d1},  [r12,:128]!
        it              lt
        bxlt            lr

3:      vld1.32         {d2-d3},  [r1,:128]
        vld1.32         {d0-d1},  [r0,:128]
        vcle.s32        q8,  q1,  #0
        vand            q9,  q0,  q10
        veor            q1,  q1,  q9
        vand            q2,  q1,  q8
        vbic            q3,  q1,  q8
        vadd.f32        q1,  q0,  q2
        vsub.f32        q0,  q0,  q3
        vst1.32         {d2-d3},  [r0,:128]!
        vst1.32         {d0-d1},  [r1,:128]!
        bx              lr
endfunc
#endif

function ff_vector_fmul_scalar_neon, export=1
VFP     len .req r2
NOVFP   len .req r3
VFP     vdup.32         q8,  d0[0]
NOVFP   vdup.32         q8,  r2
        bics            r12, len, #15
        beq             3f
        vld1.32         {q0},[r1,:128]!
        vld1.32         {q1},[r1,:128]!
1:      vmul.f32        q0,  q0,  q8
        vld1.32         {q2},[r1,:128]!
        vmul.f32        q1,  q1,  q8
        vld1.32         {q3},[r1,:128]!
        vmul.f32        q2,  q2,  q8
        vst1.32         {q0},[r0,:128]!
        vmul.f32        q3,  q3,  q8
        vst1.32         {q1},[r0,:128]!
        subs            r12, r12, #16
        beq             2f
        vld1.32         {q0},[r1,:128]!
        vst1.32         {q2},[r0,:128]!
        vld1.32         {q1},[r1,:128]!
        vst1.32         {q3},[r0,:128]!
        b               1b
2:      vst1.32         {q2},[r0,:128]!
        vst1.32         {q3},[r0,:128]!
        ands            len, len, #15
        it              eq
        bxeq            lr
3:      vld1.32         {q0},[r1,:128]!
        vmul.f32        q0,  q0,  q8
        vst1.32         {q0},[r0,:128]!
        subs            len, len, #4
        bgt             3b
        bx              lr
        .unreq          len
endfunc

function ff_vector_fmac_scalar_neon, export=1
VFP     len .req r2
VFP     acc .req r3
NOVFP   len .req r3
NOVFP   acc .req r2
VFP     vdup.32         q15, d0[0]
NOVFP   vdup.32         q15, r2
        bics            r12, len, #15
        mov             acc, r0
        beq             3f
        vld1.32         {q0},     [r1,:128]!
        vld1.32         {q8},     [acc,:128]!
        vld1.32         {q1},     [r1,:128]!
        vld1.32         {q9},     [acc,:128]!
1:      vmla.f32        q8,  q0,  q15
        vld1.32         {q2},     [r1,:128]!
        vld1.32         {q10},    [acc,:128]!
        vmla.f32        q9,  q1,  q15
        vld1.32         {q3},     [r1,:128]!
        vld1.32         {q11},    [acc,:128]!
        vmla.f32        q10, q2,  q15
        vst1.32         {q8},     [r0,:128]!
        vmla.f32        q11, q3,  q15
        vst1.32         {q9},     [r0,:128]!
        subs            r12, r12, #16
        beq             2f
        vld1.32         {q0},     [r1,:128]!
        vld1.32         {q8},     [acc,:128]!
        vst1.32         {q10},    [r0,:128]!
        vld1.32         {q1},     [r1,:128]!
        vld1.32         {q9},     [acc,:128]!
        vst1.32         {q11},    [r0,:128]!
        b               1b
2:      vst1.32         {q10},    [r0,:128]!
        vst1.32         {q11},    [r0,:128]!
        ands            len, len, #15
        it              eq
        bxeq            lr
3:      vld1.32         {q0},     [r1,:128]!
        vld1.32         {q8},     [acc,:128]!
        vmla.f32        q8,  q0,  q15
        vst1.32         {q8},     [r0,:128]!
        subs            len, len, #4
        bgt             3b
        bx              lr
        .unreq          len
endfunc

function ff_butterflies_float_neon, export=1
1:      vld1.32         {q0},[r0,:128]
        vld1.32         {q1},[r1,:128]
        vsub.f32        q2,  q0,  q1
        vadd.f32        q1,  q0,  q1
        vst1.32         {q2},[r1,:128]!
        vst1.32         {q1},[r0,:128]!
        subs            r2,  r2,  #4
        bgt             1b
        bx              lr
endfunc

function ff_scalarproduct_float_neon, export=1
        vmov.f32        q2,  #0.0
1:      vld1.32         {q0},[r0,:128]!
        vld1.32         {q1},[r1,:128]!
        vmla.f32        q2,  q0,  q1
        subs            r2,  r2,  #4
        bgt             1b
        vadd.f32        d0,  d4,  d5
        vpadd.f32       d0,  d0,  d0
NOVFP   vmov.32         r0,  d0[0]
        bx              lr
endfunc

function ff_vector_fmul_reverse_neon, export=1
        add             r2,  r2,  r3,  lsl #2
        sub             r2,  r2,  #32
        mov             r12, #-32
        vld1.32         {q0-q1},  [r1,:128]!
        vld1.32         {q2-q3},  [r2,:128], r12
1:      pld             [r1, #32]
        vrev64.32       q3,  q3
        vmul.f32        d16, d0,  d7
        vmul.f32        d17, d1,  d6
        pld             [r2, #-32]
        vrev64.32       q2,  q2
        vmul.f32        d18, d2,  d5
        vmul.f32        d19, d3,  d4
        subs            r3,  r3,  #8
        beq             2f
        vld1.32         {q0-q1},  [r1,:128]!
        vld1.32         {q2-q3},  [r2,:128], r12
        vst1.32         {q8-q9},  [r0,:128]!
        b               1b
2:      vst1.32         {q8-q9},  [r0,:128]!
        bx              lr
endfunc

function ff_vector_fmul_add_neon, export=1
        ldr             r12, [sp]
        vld1.32         {q0-q1},  [r1,:128]!
        vld1.32         {q8-q9},  [r2,:128]!
        vld1.32         {q2-q3},  [r3,:128]!
        vmul.f32        q10, q0,  q8
        vmul.f32        q11, q1,  q9
1:      vadd.f32        q12, q2,  q10
        vadd.f32        q13, q3,  q11
        pld             [r1, #16]
        pld             [r2, #16]
        pld             [r3, #16]
        subs            r12, r12, #8
        beq             2f
        vld1.32         {q0},     [r1,:128]!
        vld1.32         {q8},     [r2,:128]!
        vmul.f32        q10, q0,  q8
        vld1.32         {q1},     [r1,:128]!
        vld1.32         {q9},     [r2,:128]!
        vmul.f32        q11, q1,  q9
        vld1.32         {q2-q3},  [r3,:128]!
        vst1.32         {q12-q13},[r0,:128]!
        b               1b
2:      vst1.32         {q12-q13},[r0,:128]!
        bx              lr
endfunc

function ff_vector_clipf_neon, export=1
VFP     vdup.32         q1,  d0[1]
VFP     vdup.32         q0,  d0[0]
NOVFP   vdup.32         q0,  r2
NOVFP   vdup.32         q1,  r3
NOVFP   ldr             r2,  [sp]
        vld1.f32        {q2},[r1,:128]!
        vmin.f32        q10, q2,  q1
        vld1.f32        {q3},[r1,:128]!
        vmin.f32        q11, q3,  q1
1:      vmax.f32        q8,  q10, q0
        vmax.f32        q9,  q11, q0
        subs            r2,  r2,  #8
        beq             2f
        vld1.f32        {q2},[r1,:128]!
        vmin.f32        q10, q2,  q1
        vld1.f32        {q3},[r1,:128]!
        vmin.f32        q11, q3,  q1
        vst1.f32        {q8},[r0,:128]!
        vst1.f32        {q9},[r0,:128]!
        b               1b
2:      vst1.f32        {q8},[r0,:128]!
        vst1.f32        {q9},[r0,:128]!
        bx              lr
endfunc

function ff_apply_window_int16_neon, export=1
        push            {r4,lr}
        add             r4,  r1,  r3,  lsl #1
        add             lr,  r0,  r3,  lsl #1
        sub             r4,  r4,  #16
        sub             lr,  lr,  #16
        mov             r12, #-16
1:
        vld1.16         {q0},     [r1,:128]!
        vld1.16         {q2},     [r2,:128]!
        vld1.16         {q1},     [r4,:128], r12
        vrev64.16       q3,  q2
        vqrdmulh.s16    q0,  q0,  q2
        vqrdmulh.s16    d2,  d2,  d7
        vqrdmulh.s16    d3,  d3,  d6
        vst1.16         {q0},     [r0,:128]!
        vst1.16         {q1},     [lr,:128], r12
        subs            r3,  r3,  #16
        bgt             1b

        pop             {r4,pc}
endfunc

function ff_vector_clip_int32_neon, export=1
        vdup.32         q0,  r2
        vdup.32         q1,  r3
        ldr             r2,  [sp]
1:
        vld1.32         {q2-q3},  [r1,:128]!
        vmin.s32        q2,  q2,  q1
        vmin.s32        q3,  q3,  q1
        vmax.s32        q2,  q2,  q0
        vmax.s32        q3,  q3,  q0
        vst1.32         {q2-q3},  [r0,:128]!
        subs            r2,  r2,  #8
        bgt             1b
        bx              lr
endfunc