pixel-a-sve.S
15 KB
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
/*****************************************************************************
* pixel-a-sve.S: aarch64 pixel metrics
*****************************************************************************
* Copyright (C) 2009-2024 x264 project
*
* Authors: David Chen <david.chen@myais.com.cn>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program 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 General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
*
* This program is also available under a commercial proprietary license.
* For more information, contact us at licensing@x264.com.
*****************************************************************************/
#include "asm.S"
#include "pixel-a-common.S"
.arch armv8-a+sve
#if BIT_DEPTH == 8
.macro SSD_START_SVE_4
ptrue p0.h, vl4
ld1b {z16.h}, p0/z, [x0]
ld1b {z17.h}, p0/z, [x2]
add x0, x0, x1
add x2, x2, x3
sub v2.4h, v16.4h, v17.4h
ld1b {z16.h}, p0/z, [x0]
ld1b {z17.h}, p0/z, [x2]
add x0, x0, x1
add x2, x2, x3
smull v0.4s, v2.4h, v2.4h
.endm
.macro SSD_SVE_4
sub v2.4h, v16.4h, v17.4h
ld1b {z16.h}, p0/z, [x0]
ld1b {z17.h}, p0/z, [x2]
add x0, x0, x1
add x2, x2, x3
smlal v0.4s, v2.4h, v2.4h
.endm
.macro SSD_END_SVE_4
sub v2.4h, v16.4h, v17.4h
smlal v0.4s, v2.4h, v2.4h
.endm
.macro SSD_START_SVE_8
ptrue p0.h, vl8
ld1b {z16.h}, p0/z, [x0]
ld1b {z17.h}, p0/z, [x2]
add x0, x0, x1
add x2, x2, x3
sub v2.8h, v16.8h, v17.8h
ld1b {z16.h}, p0/z, [x0]
smull v0.4s, v2.4h, v2.4h
ld1b {z17.h}, p0/z, [x2]
smlal2 v0.4s, v2.8h, v2.8h
add x0, x0, x1
add x2, x2, x3
.endm
.macro SSD_SVE_8
sub v2.8h, v16.8h, v17.8h
ld1b {z16.h}, p0/z, [x0]
smlal v0.4s, v2.4h, v2.4h
ld1b {z17.h}, p0/z, [x2]
smlal2 v0.4s, v2.8h, v2.8h
add x0, x0, x1
add x2, x2, x3
.endm
.macro SSD_END_SVE_8
sub v2.8h, v16.8h, v17.8h
smlal v0.4s, v2.4h, v2.4h
smlal2 v0.4s, v2.8h, v2.8h
.endm
.macro SSD_FUNC_SVE w h
function pixel_ssd_\w\()x\h\()_sve, export=1
SSD_START_SVE_\w
.rept \h-2
SSD_SVE_\w
.endr
SSD_END_SVE_\w
addv s0, v0.4s
mov w0, v0.s[0]
ret
endfunc
.endm
.macro load_diff_fly_sve_8x8
ld1b {z1.h}, p0/z, [x2]
ld1b {z0.h}, p0/z, [x0]
add x2, x2, x3
add x0, x0, x1
ld1b {z3.h}, p0/z, [x2]
ld1b {z2.h}, p0/z, [x0]
add x2, x2, x3
add x0, x0, x1
sub v16.8h, v0.8h, v1.8h
sub v17.8h, v2.8h, v3.8h
ld1b {z5.h}, p0/z, [x2]
ld1b {z4.h}, p0/z, [x0]
add x2, x2, x3
add x0, x0, x1
ld1b {z7.h}, p0/z, [x2]
ld1b {z6.h}, p0/z, [x0]
add x2, x2, x3
add x0, x0, x1
sub v18.8h, v4.8h, v5.8h
sub v19.8h, v6.8h, v7.8h
ld1b {z1.h}, p0/z, [x2]
ld1b {z0.h}, p0/z, [x0]
add x2, x2, x3
add x0, x0, x1
ld1b {z3.h}, p0/z, [x2]
ld1b {z2.h}, p0/z, [x0]
add x2, x2, x3
add x0, x0, x1
sub v20.8h, v0.8h, v1.8h
sub v21.8h, v2.8h, v3.8h
ld1b {z5.h}, p0/z, [x2]
ld1b {z4.h}, p0/z, [x0]
add x2, x2, x3
add x0, x0, x1
ld1b {z7.h}, p0/z, [x2]
ld1b {z6.h}, p0/z, [x0]
add x2, x2, x3
add x0, x0, x1
SUMSUB_AB v0.8h, v1.8h, v16.8h, v17.8h
SUMSUB_AB v2.8h, v3.8h, v18.8h, v19.8h
sub v22.8h, v4.8h, v5.8h
sub v23.8h, v6.8h, v7.8h
.endm
.macro pixel_var_sve_8 h
function pixel_var_8x\h\()_sve, export=1
ptrue p0.h, vl8
ld1b {z16.h}, p0/z, [x0]
add x0, x0, x1
ld1b {z17.h}, p0/z, [x0]
add x0, x0, x1
mov x2, \h - 4
mul v1.8h, v16.8h, v16.8h
mul v2.8h, v17.8h, v17.8h
add v0.8h, v16.8h, v17.8h
ld1b {z18.h}, p0/z, [x0]
add x0, x0, x1
uaddlp v1.4s, v1.8h
uaddlp v2.4s, v2.8h
ld1b {z19.h}, p0/z, [x0]
add x0, x0, x1
1: subs x2, x2, #4
add v0.8h, v0.8h, v18.8h
mul v24.8h, v18.8h, v18.8h
ld1b {z20.h}, p0/z, [x0]
add x0, x0, x1
add v0.8h, v0.8h, v19.8h
mul v25.8h, v19.8h, v19.8h
uadalp v1.4s, v24.8h
ld1b {z21.h}, p0/z, [x0]
add x0, x0, x1
add v0.8h, v0.8h, v20.8h
mul v26.8h, v20.8h, v20.8h
uadalp v2.4s, v25.8h
ld1b {z18.h}, p0/z, [x0]
add x0, x0, x1
add v0.8h, v0.8h, v21.8h
mul v27.8h, v21.8h, v21.8h
uadalp v1.4s, v26.8h
ld1b {z19.h}, p0/z, [x0]
add x0, x0, x1
uadalp v2.4s, v27.8h
b.gt 1b
add v0.8h, v0.8h, v18.8h
mul v28.8h, v18.8h, v18.8h
add v0.8h, v0.8h, v19.8h
mul v29.8h, v19.8h, v19.8h
uadalp v1.4s, v28.8h
uadalp v2.4s, v29.8h
b var_end
endfunc
.endm
function var_end
add v1.4s, v1.4s, v2.4s
uaddlv s0, v0.8h
uaddlv d1, v1.4s
mov w0, v0.s[0]
mov x1, v1.d[0]
orr x0, x0, x1, lsl #32
ret
endfunc
.macro SUMSUBL_AB_SVE sum, sub, a, b
add \sum, \a, \b
sub \sub, \a, \b
.endm
function pixel_sa8d_8x8_sve, export=1
ptrue p0.h, vl8
mov x4, x30
bl pixel_sa8d_8x8_sve
add v0.8h, v0.8h, v1.8h
uaddlv s0, v0.8h
mov w0, v0.s[0]
add w0, w0, #1
lsr w0, w0, #1
ret x4
endfunc
.macro sa8d_satd_sve_8x8 satd=
function pixel_sa8d_\satd\()8x8_sve
load_diff_fly_sve_8x8
SUMSUB_AB v16.8h, v18.8h, v0.8h, v2.8h
SUMSUB_AB v17.8h, v19.8h, v1.8h, v3.8h
HADAMARD4_V v20.8h, v21.8h, v22.8h, v23.8h, v0.8h, v1.8h, v2.8h, v3.8h
.ifc \satd, satd_
transpose v0.8h, v1.8h, v16.8h, v17.8h
transpose v2.8h, v3.8h, v18.8h, v19.8h
transpose v4.8h, v5.8h, v20.8h, v21.8h
transpose v6.8h, v7.8h, v22.8h, v23.8h
SUMSUB_AB v24.8h, v25.8h, v0.8h, v1.8h
SUMSUB_AB v26.8h, v27.8h, v2.8h, v3.8h
SUMSUB_AB v0.8h, v1.8h, v4.8h, v5.8h
SUMSUB_AB v2.8h, v3.8h, v6.8h, v7.8h
transpose v4.4s, v6.4s, v24.4s, v26.4s
transpose v5.4s, v7.4s, v25.4s, v27.4s
transpose v24.4s, v26.4s, v0.4s, v2.4s
transpose v25.4s, v27.4s, v1.4s, v3.4s
abs v0.8h, v4.8h
abs v1.8h, v5.8h
abs v2.8h, v6.8h
abs v3.8h, v7.8h
abs v4.8h, v24.8h
abs v5.8h, v25.8h
abs v6.8h, v26.8h
abs v7.8h, v27.8h
umax v0.8h, v0.8h, v2.8h
umax v1.8h, v1.8h, v3.8h
umax v2.8h, v4.8h, v6.8h
umax v3.8h, v5.8h, v7.8h
add v26.8h, v0.8h, v1.8h
add v27.8h, v2.8h, v3.8h
.endif
SUMSUB_AB v0.8h, v16.8h, v16.8h, v20.8h
SUMSUB_AB v1.8h, v17.8h, v17.8h, v21.8h
SUMSUB_AB v2.8h, v18.8h, v18.8h, v22.8h
SUMSUB_AB v3.8h, v19.8h, v19.8h, v23.8h
transpose v20.8h, v21.8h, v16.8h, v17.8h
transpose v4.8h, v5.8h, v0.8h, v1.8h
transpose v22.8h, v23.8h, v18.8h, v19.8h
transpose v6.8h, v7.8h, v2.8h, v3.8h
SUMSUB_AB v2.8h, v3.8h, v20.8h, v21.8h
SUMSUB_AB v24.8h, v25.8h, v4.8h, v5.8h
SUMSUB_AB v0.8h, v1.8h, v22.8h, v23.8h
SUMSUB_AB v4.8h, v5.8h, v6.8h, v7.8h
transpose v20.4s, v22.4s, v2.4s, v0.4s
transpose v21.4s, v23.4s, v3.4s, v1.4s
transpose v16.4s, v18.4s, v24.4s, v4.4s
transpose v17.4s, v19.4s, v25.4s, v5.4s
SUMSUB_AB v0.8h, v2.8h, v20.8h, v22.8h
SUMSUB_AB v1.8h, v3.8h, v21.8h, v23.8h
SUMSUB_AB v4.8h, v6.8h, v16.8h, v18.8h
SUMSUB_AB v5.8h, v7.8h, v17.8h, v19.8h
transpose v16.2d, v20.2d, v0.2d, v4.2d
transpose v17.2d, v21.2d, v1.2d, v5.2d
transpose v18.2d, v22.2d, v2.2d, v6.2d
transpose v19.2d, v23.2d, v3.2d, v7.2d
abs v16.8h, v16.8h
abs v20.8h, v20.8h
abs v17.8h, v17.8h
abs v21.8h, v21.8h
abs v18.8h, v18.8h
abs v22.8h, v22.8h
abs v19.8h, v19.8h
abs v23.8h, v23.8h
umax v16.8h, v16.8h, v20.8h
umax v17.8h, v17.8h, v21.8h
umax v18.8h, v18.8h, v22.8h
umax v19.8h, v19.8h, v23.8h
add v0.8h, v16.8h, v17.8h
add v1.8h, v18.8h, v19.8h
ret
endfunc
.endm
.macro HADAMARD_AC_SVE w h
function pixel_hadamard_ac_\w\()x\h\()_sve, export=1
ptrue p0.h, vl8
movrel x5, mask_ac_4_8
mov x4, x30
ld1 {v30.8h,v31.8h}, [x5]
movi v28.16b, #0
movi v29.16b, #0
bl hadamard_ac_8x8_sve
.if \h > 8
bl hadamard_ac_8x8_sve
.endif
.if \w > 8
sub x0, x0, x1, lsl #3
add x0, x0, #8
bl hadamard_ac_8x8_sve
.endif
.if \w * \h == 256
sub x0, x0, x1, lsl #4
bl hadamard_ac_8x8_sve
.endif
addv s1, v29.4s
addv s0, v28.4s
mov w1, v1.s[0]
mov w0, v0.s[0]
lsr w1, w1, #2
lsr w0, w0, #1
orr x0, x0, x1, lsl #32
ret x4
endfunc
.endm
// v28: satd v29: sa8d v30: mask_ac4 v31: mask_ac8
function hadamard_ac_8x8_sve
ld1b {z16.h}, p0/z, [x0]
add x0, x0, x1
ld1b {z17.h}, p0/z, [x0]
add x0, x0, x1
ld1b {z18.h}, p0/z, [x0]
add x0, x0, x1
ld1b {z19.h}, p0/z, [x0]
add x0, x0, x1
SUMSUBL_AB_SVE v0.8h, v1.8h, v16.8h, v17.8h
ld1b {z20.h}, p0/z, [x0]
add x0, x0, x1
ld1b {z21.h}, p0/z, [x0]
add x0, x0, x1
SUMSUBL_AB_SVE v2.8h, v3.8h, v18.8h, v19.8h
ld1b {z22.h}, p0/z, [x0]
add x0, x0, x1
ld1b {z23.h}, p0/z, [x0]
add x0, x0, x1
SUMSUBL_AB_SVE v4.8h, v5.8h, v20.8h, v21.8h
SUMSUBL_AB_SVE v6.8h, v7.8h, v22.8h, v23.8h
SUMSUB_ABCD v16.8h, v18.8h, v17.8h, v19.8h, v0.8h, v2.8h, v1.8h, v3.8h
SUMSUB_ABCD v20.8h, v22.8h, v21.8h, v23.8h, v4.8h, v6.8h, v5.8h, v7.8h
transpose v0.8h, v1.8h, v16.8h, v17.8h
transpose v2.8h, v3.8h, v18.8h, v19.8h
transpose v4.8h, v5.8h, v20.8h, v21.8h
transpose v6.8h, v7.8h, v22.8h, v23.8h
SUMSUB_AB v16.8h, v17.8h, v0.8h, v1.8h
SUMSUB_AB v18.8h, v19.8h, v2.8h, v3.8h
SUMSUB_AB v20.8h, v21.8h, v4.8h, v5.8h
SUMSUB_AB v22.8h, v23.8h, v6.8h, v7.8h
transpose v0.4s, v2.4s, v16.4s, v18.4s
transpose v1.4s, v3.4s, v17.4s, v19.4s
transpose v4.4s, v6.4s, v20.4s, v22.4s
transpose v5.4s, v7.4s, v21.4s, v23.4s
SUMSUB_AB v16.8h, v18.8h, v0.8h, v2.8h
SUMSUB_AB v17.8h, v19.8h, v1.8h, v3.8h
SUMSUB_ABCD v20.8h, v22.8h, v21.8h, v23.8h, v4.8h, v6.8h, v5.8h, v7.8h
abs v0.8h, v16.8h
abs v4.8h, v20.8h
abs v1.8h, v17.8h
abs v5.8h, v21.8h
abs v2.8h, v18.8h
abs v6.8h, v22.8h
abs v3.8h, v19.8h
abs v7.8h, v23.8h
add v0.8h, v0.8h, v4.8h
add v1.8h, v1.8h, v5.8h
and v0.16b, v0.16b, v30.16b
add v2.8h, v2.8h, v6.8h
add v3.8h, v3.8h, v7.8h
add v0.8h, v0.8h, v2.8h
add v1.8h, v1.8h, v3.8h
uadalp v28.4s, v0.8h
uadalp v28.4s, v1.8h
SUMSUB_AB v6.8h, v7.8h, v23.8h, v19.8h
SUMSUB_AB v4.8h, v5.8h, v22.8h, v18.8h
SUMSUB_AB v2.8h, v3.8h, v21.8h, v17.8h
SUMSUB_AB v1.8h, v0.8h, v16.8h, v20.8h
transpose v16.2d, v17.2d, v6.2d, v7.2d
transpose v18.2d, v19.2d, v4.2d, v5.2d
transpose v20.2d, v21.2d, v2.2d, v3.2d
abs v16.8h, v16.8h
abs v17.8h, v17.8h
abs v18.8h, v18.8h
abs v19.8h, v19.8h
abs v20.8h, v20.8h
abs v21.8h, v21.8h
transpose v7.2d, v6.2d, v1.2d, v0.2d
umax v3.8h, v16.8h, v17.8h
umax v2.8h, v18.8h, v19.8h
umax v1.8h, v20.8h, v21.8h
SUMSUB_AB v4.8h, v5.8h, v7.8h, v6.8h
add v2.8h, v2.8h, v3.8h
add v2.8h, v2.8h, v1.8h
and v4.16b, v4.16b, v31.16b
add v2.8h, v2.8h, v2.8h
abs v5.8h, v5.8h
abs v4.8h, v4.8h
add v2.8h, v2.8h, v5.8h
add v2.8h, v2.8h, v4.8h
uadalp v29.4s, v2.8h
ret
endfunc
SSD_FUNC_SVE 4, 4
SSD_FUNC_SVE 4, 8
SSD_FUNC_SVE 4, 16
SSD_FUNC_SVE 8, 4
SSD_FUNC_SVE 8, 8
pixel_var_sve_8 8
pixel_var_sve_8 16
sa8d_satd_sve_8x8
HADAMARD_AC_SVE 8, 8
HADAMARD_AC_SVE 8, 16
HADAMARD_AC_SVE 16, 8
HADAMARD_AC_SVE 16, 16
#else /* BIT_DEPTH == 10 */
.macro SSD_START_SVE_4
ptrue p0.s, vl4
ld1h {z16.s}, p0/z, [x0]
ld1h {z17.s}, p0/z, [x2]
add x0, x0, x1, lsl #1
add x2, x2, x3, lsl #1
sub v2.4s, v16.4s, v17.4s
ld1h {z16.s}, p0/z, [x0]
ld1h {z17.s}, p0/z, [x2]
add x0, x0, x1, lsl #1
add x2, x2, x3, lsl #1
mul v0.4s, v2.4s, v2.4s
.endm
.macro SSD_SVE_4
sub v2.4s, v16.4s, v17.4s
ld1h {z16.s}, p0/z, [x0]
ld1h {z17.s}, p0/z, [x2]
add x0, x0, x1, lsl #1
add x2, x2, x3, lsl #1
mla v0.4s, v2.4s, v2.4s
.endm
.macro SSD_END_SVE_4
sub v2.4s, v16.4s, v17.4s
mla v0.4s, v2.4s, v2.4s
.endm
.macro SSD_FUNC_SVE w h
function pixel_ssd_\w\()x\h\()_sve, export=1
SSD_START_SVE_\w
.rept \h-2
SSD_SVE_\w
.endr
SSD_END_SVE_\w
addv s0, v0.4s
fmov w0, s0
ret
endfunc
.endm
SSD_FUNC_SVE 4, 4
SSD_FUNC_SVE 4, 8
SSD_FUNC_SVE 4, 16
#endif /* BIT_DEPTH == 8 */