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
|
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_filter_block2d_first_pass_armv6|
EXPORT |vp8_filter_block2d_first_pass_16x16_armv6|
EXPORT |vp8_filter_block2d_first_pass_8x8_armv6|
EXPORT |vp8_filter_block2d_second_pass_armv6|
EXPORT |vp8_filter4_block2d_second_pass_armv6|
EXPORT |vp8_filter_block2d_first_pass_only_armv6|
EXPORT |vp8_filter_block2d_second_pass_only_armv6|
AREA |.text|, CODE, READONLY ; name this block of code
;-------------------------------------
; r0 unsigned char *src_ptr
; r1 short *output_ptr
; r2 unsigned int src_pixels_per_line
; r3 unsigned int output_width
; stack unsigned int output_height
; stack const short *vp8_filter
;-------------------------------------
; vp8_filter the input and put in the output array. Apply the 6 tap FIR filter with
; the output being a 2 byte value and the intput being a 1 byte value.
|vp8_filter_block2d_first_pass_armv6| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #40] ; vp8_filter address
ldr r7, [sp, #36] ; output height
sub r2, r2, r3 ; inside loop increments input array,
; so the height loop only needs to add
; r2 - width to the input pointer
mov r3, r3, lsl #1 ; multiply width by 2 because using shorts
add r12, r3, #16 ; square off the output
sub sp, sp, #4
ldr r4, [r11] ; load up packed filter coefficients
ldr r5, [r11, #4]
ldr r6, [r11, #8]
str r1, [sp] ; push destination to stack
mov r7, r7, lsl #16 ; height is top part of counter
; six tap filter
|height_loop_1st_6|
ldrb r8, [r0, #-2] ; load source data
ldrb r9, [r0, #-1]
ldrb r10, [r0], #2
orr r7, r7, r3, lsr #2 ; construct loop counter
|width_loop_1st_6|
ldrb r11, [r0, #-1]
pkhbt lr, r8, r9, lsl #16 ; r9 | r8
pkhbt r8, r9, r10, lsl #16 ; r10 | r9
ldrb r9, [r0]
smuad lr, lr, r4 ; apply the filter
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smuad r8, r8, r4
pkhbt r11, r11, r9, lsl #16 ; r9 | r11
smlad lr, r10, r5, lr
ldrb r10, [r0, #1]
smlad r8, r11, r5, r8
ldrb r11, [r0, #2]
sub r7, r7, #1
pkhbt r9, r9, r10, lsl #16 ; r10 | r9
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smlad lr, r9, r6, lr
smlad r11, r10, r6, r8
ands r10, r7, #0xff ; test loop counter
add lr, lr, #0x40 ; round_shift_and_clamp
ldrneb r8, [r0, #-2] ; load data for next loop
usat lr, #8, lr, asr #7
add r11, r11, #0x40
ldrneb r9, [r0, #-1]
usat r11, #8, r11, asr #7
strh lr, [r1], r12 ; result is transposed and stored, which
; will make second pass filtering easier.
ldrneb r10, [r0], #2
strh r11, [r1], r12
bne width_loop_1st_6
ldr r1, [sp] ; load and update dst address
subs r7, r7, #0x10000
add r0, r0, r2 ; move to next input line
add r1, r1, #2 ; move over to next column
str r1, [sp]
bne height_loop_1st_6
add sp, sp, #4
ldmia sp!, {r4 - r11, pc}
ENDP
; --------------------------
; 16x16 version
; -----------------------------
|vp8_filter_block2d_first_pass_16x16_armv6| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #40] ; vp8_filter address
ldr r7, [sp, #36] ; output height
add r4, r2, #18 ; preload next low
pld [r0, r4]
sub r2, r2, r3 ; inside loop increments input array,
; so the height loop only needs to add
; r2 - width to the input pointer
mov r3, r3, lsl #1 ; multiply width by 2 because using shorts
add r12, r3, #16 ; square off the output
sub sp, sp, #4
ldr r4, [r11] ; load up packed filter coefficients
ldr r5, [r11, #4]
ldr r6, [r11, #8]
str r1, [sp] ; push destination to stack
mov r7, r7, lsl #16 ; height is top part of counter
; six tap filter
|height_loop_1st_16_6|
ldrb r8, [r0, #-2] ; load source data
ldrb r9, [r0, #-1]
ldrb r10, [r0], #2
orr r7, r7, r3, lsr #2 ; construct loop counter
|width_loop_1st_16_6|
ldrb r11, [r0, #-1]
pkhbt lr, r8, r9, lsl #16 ; r9 | r8
pkhbt r8, r9, r10, lsl #16 ; r10 | r9
ldrb r9, [r0]
smuad lr, lr, r4 ; apply the filter
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smuad r8, r8, r4
pkhbt r11, r11, r9, lsl #16 ; r9 | r11
smlad lr, r10, r5, lr
ldrb r10, [r0, #1]
smlad r8, r11, r5, r8
ldrb r11, [r0, #2]
sub r7, r7, #1
pkhbt r9, r9, r10, lsl #16 ; r10 | r9
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smlad lr, r9, r6, lr
smlad r11, r10, r6, r8
ands r10, r7, #0xff ; test loop counter
add lr, lr, #0x40 ; round_shift_and_clamp
ldrneb r8, [r0, #-2] ; load data for next loop
usat lr, #8, lr, asr #7
add r11, r11, #0x40
ldrneb r9, [r0, #-1]
usat r11, #8, r11, asr #7
strh lr, [r1], r12 ; result is transposed and stored, which
; will make second pass filtering easier.
ldrneb r10, [r0], #2
strh r11, [r1], r12
bne width_loop_1st_16_6
ldr r1, [sp] ; load and update dst address
subs r7, r7, #0x10000
add r0, r0, r2 ; move to next input line
add r11, r2, #34 ; adding back block width(=16)
pld [r0, r11] ; preload next low
add r1, r1, #2 ; move over to next column
str r1, [sp]
bne height_loop_1st_16_6
add sp, sp, #4
ldmia sp!, {r4 - r11, pc}
ENDP
; --------------------------
; 8x8 version
; -----------------------------
|vp8_filter_block2d_first_pass_8x8_armv6| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #40] ; vp8_filter address
ldr r7, [sp, #36] ; output height
add r4, r2, #10 ; preload next low
pld [r0, r4]
sub r2, r2, r3 ; inside loop increments input array,
; so the height loop only needs to add
; r2 - width to the input pointer
mov r3, r3, lsl #1 ; multiply width by 2 because using shorts
add r12, r3, #16 ; square off the output
sub sp, sp, #4
ldr r4, [r11] ; load up packed filter coefficients
ldr r5, [r11, #4]
ldr r6, [r11, #8]
str r1, [sp] ; push destination to stack
mov r7, r7, lsl #16 ; height is top part of counter
; six tap filter
|height_loop_1st_8_6|
ldrb r8, [r0, #-2] ; load source data
ldrb r9, [r0, #-1]
ldrb r10, [r0], #2
orr r7, r7, r3, lsr #2 ; construct loop counter
|width_loop_1st_8_6|
ldrb r11, [r0, #-1]
pkhbt lr, r8, r9, lsl #16 ; r9 | r8
pkhbt r8, r9, r10, lsl #16 ; r10 | r9
ldrb r9, [r0]
smuad lr, lr, r4 ; apply the filter
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smuad r8, r8, r4
pkhbt r11, r11, r9, lsl #16 ; r9 | r11
smlad lr, r10, r5, lr
ldrb r10, [r0, #1]
smlad r8, r11, r5, r8
ldrb r11, [r0, #2]
sub r7, r7, #1
pkhbt r9, r9, r10, lsl #16 ; r10 | r9
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smlad lr, r9, r6, lr
smlad r11, r10, r6, r8
ands r10, r7, #0xff ; test loop counter
add lr, lr, #0x40 ; round_shift_and_clamp
ldrneb r8, [r0, #-2] ; load data for next loop
usat lr, #8, lr, asr #7
add r11, r11, #0x40
ldrneb r9, [r0, #-1]
usat r11, #8, r11, asr #7
strh lr, [r1], r12 ; result is transposed and stored, which
; will make second pass filtering easier.
ldrneb r10, [r0], #2
strh r11, [r1], r12
bne width_loop_1st_8_6
ldr r1, [sp] ; load and update dst address
subs r7, r7, #0x10000
add r0, r0, r2 ; move to next input line
add r11, r2, #18 ; adding back block width(=8)
pld [r0, r11] ; preload next low
add r1, r1, #2 ; move over to next column
str r1, [sp]
bne height_loop_1st_8_6
add sp, sp, #4
ldmia sp!, {r4 - r11, pc}
ENDP
;---------------------------------
; r0 short *src_ptr,
; r1 unsigned char *output_ptr,
; r2 unsigned int output_pitch,
; r3 unsigned int cnt,
; stack const short *vp8_filter
;---------------------------------
|vp8_filter_block2d_second_pass_armv6| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #36] ; vp8_filter address
sub sp, sp, #4
mov r7, r3, lsl #16 ; height is top part of counter
str r1, [sp] ; push destination to stack
ldr r4, [r11] ; load up packed filter coefficients
ldr r5, [r11, #4]
ldr r6, [r11, #8]
pkhbt r12, r5, r4 ; pack the filter differently
pkhbt r11, r6, r5
sub r0, r0, #4 ; offset input buffer
|height_loop_2nd|
ldr r8, [r0] ; load the data
ldr r9, [r0, #4]
orr r7, r7, r3, lsr #1 ; loop counter
|width_loop_2nd|
smuad lr, r4, r8 ; apply filter
sub r7, r7, #1
smulbt r8, r4, r8
ldr r10, [r0, #8]
smlad lr, r5, r9, lr
smladx r8, r12, r9, r8
ldrh r9, [r0, #12]
smlad lr, r6, r10, lr
smladx r8, r11, r10, r8
add r0, r0, #4
smlatb r10, r6, r9, r8
add lr, lr, #0x40 ; round_shift_and_clamp
ands r8, r7, #0xff
usat lr, #8, lr, asr #7
add r10, r10, #0x40
strb lr, [r1], r2 ; the result is transposed back and stored
usat r10, #8, r10, asr #7
ldrne r8, [r0] ; load data for next loop
ldrne r9, [r0, #4]
strb r10, [r1], r2
bne width_loop_2nd
ldr r1, [sp] ; update dst for next loop
subs r7, r7, #0x10000
add r0, r0, #16 ; updata src for next loop
add r1, r1, #1
str r1, [sp]
bne height_loop_2nd
add sp, sp, #4
ldmia sp!, {r4 - r11, pc}
ENDP
;---------------------------------
; r0 short *src_ptr,
; r1 unsigned char *output_ptr,
; r2 unsigned int output_pitch,
; r3 unsigned int cnt,
; stack const short *vp8_filter
;---------------------------------
|vp8_filter4_block2d_second_pass_armv6| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #36] ; vp8_filter address
mov r7, r3, lsl #16 ; height is top part of counter
ldr r4, [r11] ; load up packed filter coefficients
add lr, r1, r3 ; save final destination pointer
ldr r5, [r11, #4]
ldr r6, [r11, #8]
pkhbt r12, r5, r4 ; pack the filter differently
pkhbt r11, r6, r5
mov r4, #0x40 ; rounding factor (for smlad{x})
|height_loop_2nd_4|
ldrd r8, r9, [r0, #-4] ; load the data
orr r7, r7, r3, lsr #1 ; loop counter
|width_loop_2nd_4|
ldr r10, [r0, #4]!
smladx r6, r9, r12, r4 ; apply filter
pkhbt r8, r9, r8
smlad r5, r8, r12, r4
pkhbt r8, r10, r9
smladx r6, r10, r11, r6
sub r7, r7, #1
smlad r5, r8, r11, r5
mov r8, r9 ; shift the data for the next loop
mov r9, r10
usat r6, #8, r6, asr #7 ; shift and clamp
usat r5, #8, r5, asr #7
strb r5, [r1], r2 ; the result is transposed back and stored
tst r7, #0xff
strb r6, [r1], r2
bne width_loop_2nd_4
subs r7, r7, #0x10000
add r0, r0, #16 ; update src for next loop
sub r1, lr, r7, lsr #16 ; update dst for next loop
bne height_loop_2nd_4
ldmia sp!, {r4 - r11, pc}
ENDP
;------------------------------------
; r0 unsigned char *src_ptr
; r1 unsigned char *output_ptr,
; r2 unsigned int src_pixels_per_line
; r3 unsigned int cnt,
; stack unsigned int output_pitch,
; stack const short *vp8_filter
;------------------------------------
|vp8_filter_block2d_first_pass_only_armv6| PROC
stmdb sp!, {r4 - r11, lr}
add r7, r2, r3 ; preload next low
add r7, r7, #2
pld [r0, r7]
ldr r4, [sp, #36] ; output pitch
ldr r11, [sp, #40] ; HFilter address
sub sp, sp, #8
mov r7, r3
sub r2, r2, r3 ; inside loop increments input array,
; so the height loop only needs to add
; r2 - width to the input pointer
sub r4, r4, r3
str r4, [sp] ; save modified output pitch
str r2, [sp, #4]
mov r2, #0x40
ldr r4, [r11] ; load up packed filter coefficients
ldr r5, [r11, #4]
ldr r6, [r11, #8]
; six tap filter
|height_loop_1st_only_6|
ldrb r8, [r0, #-2] ; load data
ldrb r9, [r0, #-1]
ldrb r10, [r0], #2
mov r12, r3, lsr #1 ; loop counter
|width_loop_1st_only_6|
ldrb r11, [r0, #-1]
pkhbt lr, r8, r9, lsl #16 ; r9 | r8
pkhbt r8, r9, r10, lsl #16 ; r10 | r9
ldrb r9, [r0]
;; smuad lr, lr, r4
smlad lr, lr, r4, r2
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
;; smuad r8, r8, r4
smlad r8, r8, r4, r2
pkhbt r11, r11, r9, lsl #16 ; r9 | r11
smlad lr, r10, r5, lr
ldrb r10, [r0, #1]
smlad r8, r11, r5, r8
ldrb r11, [r0, #2]
subs r12, r12, #1
pkhbt r9, r9, r10, lsl #16 ; r10 | r9
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smlad lr, r9, r6, lr
smlad r10, r10, r6, r8
;; add lr, lr, #0x40 ; round_shift_and_clamp
ldrneb r8, [r0, #-2] ; load data for next loop
usat lr, #8, lr, asr #7
;; add r10, r10, #0x40
strb lr, [r1], #1 ; store the result
usat r10, #8, r10, asr #7
ldrneb r9, [r0, #-1]
strb r10, [r1], #1
ldrneb r10, [r0], #2
bne width_loop_1st_only_6
ldr lr, [sp] ; load back output pitch
ldr r12, [sp, #4] ; load back output pitch
subs r7, r7, #1
add r0, r0, r12 ; updata src for next loop
add r11, r12, r3 ; preload next low
add r11, r11, #2
pld [r0, r11]
add r1, r1, lr ; update dst for next loop
bne height_loop_1st_only_6
add sp, sp, #8
ldmia sp!, {r4 - r11, pc}
ENDP ; |vp8_filter_block2d_first_pass_only_armv6|
;------------------------------------
; r0 unsigned char *src_ptr,
; r1 unsigned char *output_ptr,
; r2 unsigned int src_pixels_per_line
; r3 unsigned int cnt,
; stack unsigned int output_pitch,
; stack const short *vp8_filter
;------------------------------------
|vp8_filter_block2d_second_pass_only_armv6| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #40] ; VFilter address
ldr r12, [sp, #36] ; output pitch
mov r7, r3, lsl #16 ; height is top part of counter
sub r0, r0, r2, lsl #1 ; need 6 elements for filtering, 2 before, 3 after
sub sp, sp, #8
ldr r4, [r11] ; load up packed filter coefficients
ldr r5, [r11, #4]
ldr r6, [r11, #8]
str r0, [sp] ; save r0 to stack
str r1, [sp, #4] ; save dst to stack
; six tap filter
|width_loop_2nd_only_6|
ldrb r8, [r0], r2 ; load data
orr r7, r7, r3 ; loop counter
ldrb r9, [r0], r2
ldrb r10, [r0], r2
|height_loop_2nd_only_6|
; filter first column in this inner loop, than, move to next colum.
ldrb r11, [r0], r2
pkhbt lr, r8, r9, lsl #16 ; r9 | r8
pkhbt r8, r9, r10, lsl #16 ; r10 | r9
ldrb r9, [r0], r2
smuad lr, lr, r4
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smuad r8, r8, r4
pkhbt r11, r11, r9, lsl #16 ; r9 | r11
smlad lr, r10, r5, lr
ldrb r10, [r0], r2
smlad r8, r11, r5, r8
ldrb r11, [r0]
sub r7, r7, #2
sub r0, r0, r2, lsl #2
pkhbt r9, r9, r10, lsl #16 ; r10 | r9
pkhbt r10, r10, r11, lsl #16 ; r11 | r10
smlad lr, r9, r6, lr
smlad r10, r10, r6, r8
ands r9, r7, #0xff
add lr, lr, #0x40 ; round_shift_and_clamp
ldrneb r8, [r0], r2 ; load data for next loop
usat lr, #8, lr, asr #7
add r10, r10, #0x40
strb lr, [r1], r12 ; store the result for the column
usat r10, #8, r10, asr #7
ldrneb r9, [r0], r2
strb r10, [r1], r12
ldrneb r10, [r0], r2
bne height_loop_2nd_only_6
ldr r0, [sp]
ldr r1, [sp, #4]
subs r7, r7, #0x10000
add r0, r0, #1 ; move to filter next column
str r0, [sp]
add r1, r1, #1
str r1, [sp, #4]
bne width_loop_2nd_only_6
add sp, sp, #8
ldmia sp!, {r4 - r11, pc}
ENDP ; |vp8_filter_block2d_second_pass_only_armv6|
END
|
