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
|
//******************************************************************************
//*
//* Copyright (C) 2015 The Android Open Source Project
//*
//* Licensed under the Apache License, Version 2.0 (the "License");
//* you may not use this file except in compliance with the License.
//* You may obtain a copy of the License at:
//*
//* http://www.apache.org/licenses/LICENSE-2.0
//*
//* Unless required by applicable law or agreed to in writing, software
//* distributed under the License is distributed on an "AS IS" BASIS,
//* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
//* See the License for the specific language governing permissions and
//* limitations under the License.
//*
//*****************************************************************************
//* Originally developed and contributed by Ittiam Systems Pvt. Ltd, Bangalore
//*/
///*****************************************************************************/
///* */
///* File Name : ih264_deblk_chroma_av8.s */
///* */
///* Description : Contains function definitions for deblocking luma */
///* edge. Functions are coded in NEON assembly and can */
///* be compiled using ARM RVDS. */
///* */
///* List of Functions : ih264_deblk_chroma_vert_bs4_av8() */
///* ih264_deblk_chroma_vert_bslt4_av8() */
///* ih264_deblk_chroma_horz_bs4_av8() */
///* ih264_deblk_chroma_horz_bslt4_av8() */
///* Issues / Problems : None */
///* */
///* Revision History : */
///* */
///* DD MM YYYY Author(s) Changes (Describe the changes made) */
///* 28 11 2013 Ittiam Draft */
///*****************************************************************************/
.text
.p2align 2
.include "ih264_neon_macros.s"
///**
//*******************************************************************************
//*
//* @brief
//* Performs filtering of a chroma block horizontal edge when the
//* boundary strength is set to 4 in high profile
//*
//* @par Description:
//* This operation is described in Sec. 8.7.2.4 under the title
//* "Filtering process for edges for bS equal to 4" in ITU T Rec H.264.
//*
//* @param[in] x0 - pu1_src
//* Pointer to the src sample q0
//*
//* @param[in] x1 - src_strd
//* Source stride
//*
//* @param[in] x2 - alpha_cb
//* Alpha Value for the boundary in U
//*
//* @param[in] x3 - beta_cb
//* Beta Value for the boundary in U
//*
//* @param[in] sp(0) - alpha_cr
//* Alpha Value for the boundary in V
//*
//* @param[in] sp(4) - beta_cr
//* Beta Value for the boundary in V
//*
//* @returns
//* None
//*
//* @remarks
//* None
//*
//*******************************************************************************
//*/
.global ih264_deblk_chroma_horz_bs4_av8
ih264_deblk_chroma_horz_bs4_av8:
// STMFD sp!,{x4-x6,x14} //
push_v_regs
stp x19, x20, [sp, #-16]!
mov x6, x5
mov x5, x4
sub x0, x0, x1, lsl #1 //x0 = uc_edgePixel pointing to p1 of chroma
ld2 {v6.8b, v7.8b}, [x0], x1 //D6 = p1u , D7 = p1v
mov x4, x0 //Keeping a backup of the pointer p0 of chroma
ld2 {v4.8b, v5.8b}, [x0], x1 //D4 = p0u , D5 = p0v
dup v20.8b, w2 //D20 contains alpha_cb
dup v21.8b, w5 //D21 contains alpha_cr
mov v20.d[1], v21.d[0]
ld2 {v0.8b, v1.8b}, [x0], x1 //D0 = q0u , D1 = q0v
uaddl v8.8h, v6.8b, v0.8b //
uaddl v10.8h, v7.8b, v1.8b //Q4,Q5 = q0 + p1
movi v31.8b, #2 //
ld2 {v2.8b, v3.8b}, [x0] //D2 = q1u , D3 = q1v
mov v0.d[1], v1.d[0]
mov v2.d[1], v3.d[0]
mov v4.d[1], v5.d[0]
mov v6.d[1], v7.d[0]
uabd v26.16b, v6.16b , v4.16b //Q13 = ABS(p1 - p0)
umlal v8.8h, v2.8b, v31.8b //
umlal v10.8h, v3.8b, v31.8b //Q5,Q4 = (X2(q1U) + q0U + p1U)
uabd v22.16b, v4.16b , v0.16b //Q11 = ABS(p0 - q0)
uabd v24.16b, v2.16b , v0.16b //Q12 = ABS(q1 - q0)
uaddl v14.8h, v4.8b, v2.8b //
uaddl v28.8h, v5.8b, v3.8b //Q14,Q7 = P0 + Q1
dup v16.8b, w3 //D16 contains beta_cb
dup v17.8b, w6 //D17 contains beta_cr
mov v16.d[1], v17.d[0]
umlal v14.8h, v6.8b, v31.8b //
umlal v28.8h, v7.8b, v31.8b //Q14,Q7 = (X2(p1U) + p0U + q1U)
cmhs v18.16b, v22.16b, v20.16b
cmhs v24.16b, v24.16b, v16.16b
cmhs v26.16b, v26.16b, v16.16b
rshrn v8.8b, v8.8h, #2 //
rshrn v9.8b, v10.8h, #2 //Q4 = (X2(q1U) + q0U + p1U + 2) >> 2
mov v8.d[1], v9.d[0]
orr v18.16b, v18.16b , v24.16b //Q9 = ( ABS(p0 - q0) >= Alpha ) | ( ABS(q1 - q0) >= Beta )
rshrn v10.8b, v14.8h, #2 //
rshrn v11.8b, v28.8h, #2 //Q5 = (X2(p1U) + p0U + q1U + 2) >> 2
mov v10.d[1], v11.d[0]
orr v18.16b, v18.16b , v26.16b //Q9 = ( ABS(p0 - q0) >= Alpha ) | ( ABS(q1 - q0) >= Beta ) | ( ABS(p1 - p0) >= Beta )
bit v10.16b, v4.16b , v18.16b //
bit v8.16b, v0.16b , v18.16b //
mov v11.d[0], v10.d[1]
mov v9.d[0], v8.d[1]
st2 {v10.8b, v11.8b}, [x4], x1 //
st2 {v8.8b, v9.8b}, [x4] //
// LDMFD sp!,{x4-x6,pc} //
ldp x19, x20, [sp], #16
pop_v_regs
ret
///**
//*******************************************************************************
//*
//* @brief
//* Performs filtering of a chroma block vertical edge when the
//* boundary strength is set to 4 in high profile
//*
//* @par Description:
//* This operation is described in Sec. 8.7.2.4 under the title
//* "Filtering process for edges for bS equal to 4" in ITU T Rec H.264.
//*
//* @param[in] x0 - pu1_src
//* Pointer to the src sample q0
//*
//* @param[in] x1 - src_strd
//* Source stride
//*
//* @param[in] x2 - alpha_cb
//* Alpha Value for the boundary in U
//*
//* @param[in] x3 - beta_cb
//* Beta Value for the boundary in U
//*
//* @param[in] sp(0) - alpha_cr
//* Alpha Value for the boundary in V
//*
//* @param[in] sp(4) - beta_cr
//* Beta Value for the boundary in V
//*
//* @returns
//* None
//*
//* @remarks
//* None
//*
//*******************************************************************************
//*/
.global ih264_deblk_chroma_vert_bs4_av8
ih264_deblk_chroma_vert_bs4_av8:
// STMFD sp!,{x4,x5,x12,x14}
push_v_regs
stp x19, x20, [sp, #-16]!
sub x0, x0, #4 //point x0 to p1u of row0.
mov x12, x0 //keep a back up of x0 for buffer write
add x2, x2, x4, lsl #8 //x2 = (alpha_cr,alpha_cb)
add x3, x3, x5, lsl #8 //x3 = (beta_cr,beta_cb)
ld4 {v0.h, v1.h, v2.h, v3.h}[0], [x0], x1
ld4 {v0.h, v1.h, v2.h, v3.h}[1], [x0], x1
ld4 {v0.h, v1.h, v2.h, v3.h}[2], [x0], x1
ld4 {v0.h, v1.h, v2.h, v3.h}[3], [x0], x1
ld4 {v4.h, v5.h, v6.h, v7.h}[0], [x0], x1
ld4 {v4.h, v5.h, v6.h, v7.h}[1], [x0], x1
ld4 {v4.h, v5.h, v6.h, v7.h}[2], [x0], x1
ld4 {v4.h, v5.h, v6.h, v7.h}[3], [x0], x1
mov v10.16b, v2.16b
mov v2.16b, v1.16b
mov v1.16b, v4.16b
mov v4.16b, v10.16b
mov v10.16b, v6.16b
mov v6.16b, v3.16b
mov v3.16b, v5.16b
mov v5.16b, v10.16b
dup v22.8h, w2 //Q11 = alpha
dup v24.8h, w3 //Q12 = beta
movi v31.8b, #2
mov v0.d[1], v1.d[0]
mov v2.d[1], v3.d[0]
mov v4.d[1], v5.d[0]
mov v6.d[1], v7.d[0]
uabd v8.16b, v2.16b , v4.16b //|p0-q0|
uabd v10.16b, v6.16b , v4.16b //|q1-q0|
uabd v12.16b, v0.16b , v2.16b //|p1-p0|
uaddl v14.8h, v2.8b, v6.8b
uaddl v16.8h, v3.8b, v7.8b //(p0 + q1)
cmhi v8.16b, v22.16b , v8.16b //|p0-q0| < alpha ?
cmhi v10.16b, v24.16b , v10.16b //|q1-q0| < beta ?
cmhi v12.16b, v24.16b , v12.16b //|p1-p0| < beta ?
umlal v14.8h, v0.8b, v31.8b
umlal v16.8h, v1.8b, v31.8b //2*p1 + (p0 + q1)
uaddl v18.8h, v0.8b, v4.8b
uaddl v20.8h, v1.8b, v5.8b //(p1 + q0)
and v8.16b, v8.16b , v10.16b //|p0-q0| < alpha && |q1-q0| < beta
umlal v18.8h, v6.8b, v31.8b
umlal v20.8h, v7.8b, v31.8b //2*q1 + (p1 + q0)
rshrn v14.8b, v14.8h, #2
rshrn v15.8b, v16.8h, #2 //(2*p1 + (p0 + q1) + 2) >> 2
mov v14.d[1], v15.d[0]
and v8.16b, v8.16b , v12.16b //|p0-q0| < alpha && |q1-q0| < beta && |p1-p0| < beta
rshrn v18.8b, v18.8h, #2
rshrn v19.8b, v20.8h, #2 //(2*q1 + (p1 + q0) + 2) >> 2
mov v18.d[1], v19.d[0]
bit v2.16b, v14.16b , v8.16b
bit v4.16b, v18.16b , v8.16b
mov v1.d[0], v0.d[1]
mov v3.d[0], v2.d[1]
mov v5.d[0], v4.d[1]
mov v7.d[0], v6.d[1]
mov v10.16b, v1.16b
mov v1.16b, v2.16b
mov v2.16b, v4.16b
mov v4.16b, v10.16b
mov v10.16b, v3.16b
mov v3.16b, v6.16b
mov v6.16b, v5.16b
mov v5.16b, v10.16b
st4 {v0.h, v1.h, v2.h, v3.h}[0], [x12], x1
st4 {v0.h, v1.h, v2.h, v3.h}[1], [x12], x1
st4 {v0.h, v1.h, v2.h, v3.h}[2], [x12], x1
st4 {v0.h, v1.h, v2.h, v3.h}[3], [x12], x1
st4 {v4.h, v5.h, v6.h, v7.h}[0], [x12], x1
st4 {v4.h, v5.h, v6.h, v7.h}[1], [x12], x1
st4 {v4.h, v5.h, v6.h, v7.h}[2], [x12], x1
st4 {v4.h, v5.h, v6.h, v7.h}[3], [x12], x1
// LDMFD sp!,{x4,x5,x12,pc}
ldp x19, x20, [sp], #16
pop_v_regs
ret
///**
//*******************************************************************************
//*
//* @brief
//* Performs filtering of a chroma block horizontal edge for cases where the
//* boundary strength is less than 4 in high profile
//*
//* @par Description:
//* This operation is described in Sec. 8.7.2.4 under the title
//* "Filtering process for edges for bS equal to 4" in ITU T Rec H.264.
//*
//* @param[in] x0 - pu1_src
//* Pointer to the src sample q0
//*
//* @param[in] x1 - src_strd
//* Source stride
//*
//* @param[in] x2 - alpha_cb
//* Alpha Value for the boundary in U
//*
//* @param[in] x3 - beta_cb
//* Beta Value for the boundary in U
//*
//* @param[in] sp(0) - alpha_cr
//* Alpha Value for the boundary in V
//*
//* @param[in] sp(4) - beta_cr
//* Beta Value for the boundary in V
//*
//* @param[in] sp(8) - u4_bs
//* Packed Boundary strength array
//*
//* @param[in] sp(12) - pu1_cliptab_cb
//* tc0_table for U
//*
//* @param[in] sp(16) - pu1_cliptab_cr
//* tc0_table for V
//*
//* @returns
//* None
//*
//* @remarks
//* None
//*
//*******************************************************************************
//*/
.global ih264_deblk_chroma_horz_bslt4_av8
ih264_deblk_chroma_horz_bslt4_av8:
// STMFD sp!,{x4-x9,x14} //
push_v_regs
stp x19, x20, [sp, #-16]!
mov x8, x7
mov x7, x6
ldr x9, [sp, #80]
sub x0, x0, x1, lsl #1 //x0 = uc_edgePixelU pointing to p1 of chroma U
rev w7, w7 //
mov v12.2s[0], w7 //D12[0] = ui_Bs
ld1 {v16.s}[0], [x8] //D16[0] contains cliptab_cb
ld1 {v17.s}[0], [x9] //D17[0] contains cliptab_cr
ld2 {v6.8b, v7.8b}, [x0], x1 //Q3=p1
tbl v14.8b, {v16.16b}, v12.8b //Retreiving cliptab values for U
tbl v28.8b, {v17.16b}, v12.8b //Retrieving cliptab values for V
uxtl v12.8h, v12.8b //Q6 = uc_Bs in each 16 bit scalar
mov x6, x0 //Keeping a backup of the pointer to chroma U P0
ld2 {v4.8b, v5.8b}, [x0], x1 //Q2=p0
movi v30.8b, #1 //
dup v20.8b, w2 //D20 contains alpha_cb
dup v21.8b, w4 //D21 contains alpha_cr
mov v20.d[1], v21.d[0]
ld2 {v0.8b, v1.8b}, [x0], x1 //Q0=q0
uxtl v14.8h, v14.8b //
uxtl v28.8h, v28.8b //
mov v15.d[0], v28.d[0] //D14 has cliptab values for U, D15 for V
mov v14.d[1], v28.d[0]
ld2 {v2.8b, v3.8b}, [x0] //Q1=q1
usubl v10.8h, v1.8b, v5.8b //
usubl v8.8h, v0.8b, v4.8b //Q5,Q4 = (q0 - p0)
mov v6.d[1], v7.d[0]
mov v4.d[1], v5.d[0]
uabd v26.16b, v6.16b , v4.16b //Q13 = ABS(p1 - p0)
shl v10.8h, v10.8h, #2 //Q5 = (q0 - p0)<<2
mov v0.d[1], v1.d[0]
uabd v22.16b, v4.16b , v0.16b //Q11 = ABS(p0 - q0)
shl v8.8h, v8.8h, #2 //Q4 = (q0 - p0)<<2
mov v14.d[1], v15.d[0]
sli v14.8h, v14.8h, #8
mov v15.d[0], v14.d[1]
mov v2.d[1], v3.d[0]
uabd v24.16b, v2.16b , v0.16b //Q12 = ABS(q1 - q0)
cmhs v18.16b, v22.16b, v20.16b
usubl v20.8h, v6.8b, v2.8b //Q10 = (p1 - q1)L
usubl v6.8h, v7.8b, v3.8b //Q3 = (p1 - q1)H
dup v16.8b, w3 //Q8 contains beta_cb
dup v17.8b, w5 //Q8 contains beta_cr
mov v16.d[1], v17.d[0]
add v8.8h, v8.8h , v20.8h //
add v10.8h, v10.8h , v6.8h //Q5,Q4 = [ (q0 - p0)<<2 ] + (p1 - q1)
cmhs v24.16b, v24.16b, v16.16b
cmgt v12.4h, v12.4h, #0
sqrshrn v8.8b, v8.8h, #3 //
sqrshrn v9.8b, v10.8h, #3 //Q4 = i_macro = (((q0 - p0)<<2) + (p1 - q1) + 4)>>3
mov v8.d[1], v9.d[0]
add v14.8b, v14.8b , v30.8b //D14 = C = C0+1 for U
cmhs v26.16b, v26.16b, v16.16b
orr v18.16b, v18.16b , v24.16b //Q9 = ( ABS(p0 - q0) >= Alpha ) | ( ABS(q1 - q0) >= Beta )
abs v6.16b, v8.16b //Q4 = ABS (i_macro)
add v15.8b, v15.8b , v30.8b //D15 = C = C0+1 for V
mov v14.d[1], v15.d[0]
mov v13.8b, v12.8b
mov v12.d[1], v13.d[0] //
orr v18.16b, v18.16b , v26.16b //Q9 = ( ABS(p0 - q0) >= Alpha ) | ( ABS(q1 - q0) >= Beta ) | ( ABS(p1 - p0) >= Beta )
umin v14.16b, v6.16b , v14.16b //Q7 = delta = (ABS(i_macro) > C) ? C : ABS(i_macro)
bic v12.16b, v12.16b , v18.16b //final condition
cmge v8.16b, v8.16b, #0
and v14.16b, v14.16b , v12.16b //Making delta zero in places where values shouldn be filterd
uqadd v16.16b, v4.16b , v14.16b //Q8 = p0 + delta
uqsub v4.16b, v4.16b , v14.16b //Q2 = p0 - delta
uqadd v18.16b, v0.16b , v14.16b //Q9 = q0 + delta
uqsub v0.16b, v0.16b , v14.16b //Q0 = q0 - delta
bif v16.16b, v4.16b , v8.16b //Q8 = (i_macro >= 0 ) ? (p0+delta) : (p0-delta)
bif v0.16b, v18.16b , v8.16b //Q0 = (i_macro >= 0 ) ? (q0-delta) : (q0+delta)
mov v17.d[0], v16.d[1]
mov v1.d[0], v0.d[1]
st2 {v16.8b, v17.8b}, [x6], x1 //
st2 {v0.8b, v1.8b}, [x6] //
ldp x19, x20, [sp], #16
pop_v_regs
ret
///**
//*******************************************************************************
//*
//* @brief
//* Performs filtering of a chroma block vertical edge for cases where the
//* boundary strength is less than 4 in high profile
//*
//* @par Description:
//* This operation is described in Sec. 8.7.2.4 under the title
//* "Filtering process for edges for bS equal to 4" in ITU T Rec H.264.
//*
//* @param[in] x0 - pu1_src
//* Pointer to the src sample q0
//*
//* @param[in] x1 - src_strd
//* Source stride
//*
//* @param[in] x2 - alpha_cb
//* Alpha Value for the boundary in U
//*
//* @param[in] x3 - beta_cb
//* Beta Value for the boundary in U
//*
//* @param[in] sp(0) - alpha_cr
//* Alpha Value for the boundary in V
//*
//* @param[in] sp(4) - beta_cr
//* Beta Value for the boundary in V
//*
//* @param[in] sp(8) - u4_bs
//* Packed Boundary strength array
//*
//* @param[in] sp(12) - pu1_cliptab_cb
//* tc0_table for U
//*
//* @param[in] sp(16) - pu1_cliptab_cr
//* tc0_table for V
//*
//* @returns
//* None
//*
//* @remarks
//* None
//*
//*******************************************************************************
//*/
.global ih264_deblk_chroma_vert_bslt4_av8
ih264_deblk_chroma_vert_bslt4_av8:
// STMFD sp!,{x4-x7,x10-x12,x14}
push_v_regs
stp x19, x20, [sp, #-16]!
mov x10, x7
ldr x11, [sp, #80] //x6 = u4_bs
sub x0, x0, #4 //point x0 to p1u of row0.
add x2, x2, x4, lsl #8
add x3, x3, x5, lsl #8
mov x12, x0 //keep a back up of x0 for buffer write
ld4 {v0.h, v1.h, v2.h, v3.h}[0], [x0], x1
ld4 {v0.h, v1.h, v2.h, v3.h}[1], [x0], x1
ld4 {v0.h, v1.h, v2.h, v3.h}[2], [x0], x1
ld4 {v0.h, v1.h, v2.h, v3.h}[3], [x0], x1
ld4 {v4.h, v5.h, v6.h, v7.h}[0], [x0], x1
ld4 {v4.h, v5.h, v6.h, v7.h}[1], [x0], x1
ld4 {v4.h, v5.h, v6.h, v7.h}[2], [x0], x1
ld4 {v4.h, v5.h, v6.h, v7.h}[3], [x0], x1
mov v10.16b, v2.16b
mov v2.16b, v1.16b
mov v1.16b, v4.16b
mov v4.16b, v10.16b
mov v10.16b, v6.16b
mov v6.16b, v3.16b
mov v3.16b, v5.16b
mov v5.16b, v10.16b
dup v22.8h, w2 //Q11 = alpha
mov v2.d[1], v3.d[0]
mov v4.d[1], v5.d[0]
uabd v8.16b, v2.16b , v4.16b //|p0-q0|
dup v24.8h, w3 //Q12 = beta
mov v25.d[0], v24.d[1]
mov v6.d[1], v7.d[0]
mov v0.d[1], v1.d[0]
uabd v10.16b, v6.16b , v4.16b //|q1-q0|
uabd v12.16b, v0.16b , v2.16b //|p1-p0|
cmhi v8.16b, v22.16b , v8.16b //|p0-q0| < alpha ?
usubl v14.8h, v0.8b, v6.8b
cmhi v10.16b, v24.16b , v10.16b //|q1-q0| < beta ?
usubl v16.8h, v1.8b, v7.8b //(p1 - q1)
cmhi v12.16b, v24.16b , v12.16b //|p1-p0| < beta ?
usubl v18.8h, v4.8b, v2.8b
and v8.16b, v8.16b , v10.16b //|p0-q0| < alpha && |q1-q0| < beta
usubl v20.8h, v5.8b, v3.8b //(q0 - p0)
movi v28.8h, #4
ld1 {v24.s}[0], [x10] //Load ClipTable for U
ld1 {v25.s}[0], [x11] //Load ClipTable for V
rev w6, w6 //Blocking strengths
and v8.16b, v8.16b , v12.16b //|p0-q0| < alpha && |q1-q0| < beta && |p1-p0| < beta
mov v10.s[0], w6
mla v14.8h, v18.8h , v28.8h
mla v16.8h, v20.8h , v28.8h //4*(q0 - p0) + (p1 - q1)
uxtl v10.8h, v10.8b
sli v10.4h, v10.4h, #8
tbl v12.8b, {v24.16b}, v10.8b //tC0 for U
tbl v13.8b, {v25.16b}, v10.8b //tC0 for V
zip1 v31.8b, v12.8b, v13.8b
zip2 v13.8b, v12.8b, v13.8b
mov v12.8b, v31.8b
mov v12.d[1], v13.d[0]
uxtl v10.4s, v10.4h
sli v10.4s, v10.4s, #16
movi v24.16b, #1
add v12.16b, v12.16b , v24.16b //tC0 + 1
cmhs v10.16b, v10.16b , v24.16b
and v8.16b, v8.16b , v10.16b //|p0-q0| < alpha && |q1-q0| < beta && |p1-p0| < beta && u4_bs != 0
// Q0 - Q3(inputs),
// Q4 (|p0-q0| < alpha && |q1-q0| < beta && |p1-p0| < beta && u4_bs != 0),
// Q6 (tC)
srshr v14.8h, v14.8h, #3
srshr v16.8h, v16.8h, #3 //(((q0 - p0) << 2) + (p1 - q1) + 4) >> 3)
cmgt v18.8h, v14.8h , #0
cmgt v20.8h, v16.8h , #0
xtn v18.8b, v18.8h
xtn v19.8b, v20.8h //Q9 = sign(delta)
mov v18.d[1], v19.d[0]
abs v14.8h, v14.8h
abs v16.8h, v16.8h
xtn v14.8b, v14.8h
xtn v15.8b, v16.8h
mov v14.d[1], v15.d[0]
umin v14.16b, v14.16b , v12.16b //Q7 = |delta|
uqadd v20.16b, v2.16b , v14.16b //p0+|delta|
uqadd v22.16b, v4.16b , v14.16b //q0+|delta|
uqsub v24.16b, v2.16b , v14.16b //p0-|delta|
uqsub v26.16b, v4.16b , v14.16b //q0-|delta|
bit v24.16b, v20.16b , v18.16b //p0 + delta
bit v22.16b, v26.16b , v18.16b //q0 - delta
bit v2.16b, v24.16b , v8.16b
bit v4.16b, v22.16b , v8.16b
mov v1.d[0], v0.d[1]
mov v3.d[0], v2.d[1]
mov v5.d[0], v4.d[1]
mov v7.d[0], v6.d[1]
mov v10.16b, v1.16b
mov v1.16b, v2.16b
mov v2.16b, v4.16b
mov v4.16b, v10.16b
mov v10.16b, v3.16b
mov v3.16b, v6.16b
mov v6.16b, v5.16b
mov v5.16b, v10.16b
st4 {v0.h, v1.h, v2.h, v3.h}[0], [x12], x1
st4 {v0.h, v1.h, v2.h, v3.h}[1], [x12], x1
st4 {v0.h, v1.h, v2.h, v3.h}[2], [x12], x1
st4 {v0.h, v1.h, v2.h, v3.h}[3], [x12], x1
st4 {v4.h, v5.h, v6.h, v7.h}[0], [x12], x1
st4 {v4.h, v5.h, v6.h, v7.h}[1], [x12], x1
st4 {v4.h, v5.h, v6.h, v7.h}[2], [x12], x1
st4 {v4.h, v5.h, v6.h, v7.h}[3], [x12], x1
ldp x19, x20, [sp], #16
pop_v_regs
ret
|
