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
|
//******************************************************************************
//*
//* 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
//*/
//**
///**
//******************************************************************************
//*
//*
//* @brief
//* This file contains definitions of routines that compute distortion
//* between two macro/sub blocks of identical dimensions
//*
//* @author
//* Ittiam
//*
//* @par List of Functions:
//* - ime_compute_sad_16x16()
//* - ime_compute_sad_8x8()
//* - ime_compute_sad_4x4()
//* - ime_compute_sad_16x8()
//* - ime_compute_satqd_16x16_lumainter_av8()
//*
//* @remarks
//* None
//*
//*******************************************************************************
//
///**
//******************************************************************************
//*
//* @brief computes distortion (SAD) between 2 16x16 blocks (fast mode)
//*
//* @par Description
//* This functions computes SAD between 2 16x16 blocks. There is a provision
//* for early exit if the up-to computed SAD exceeds maximum allowed SAD. To
//* compute the distortion of the entire block set u4_max_sad to USHRT_MAX.
//*
//* @param[in] pu1_src
//* UWORD8 pointer to the source
//*
//* @param[out] pu1_dst
//* UWORD8 pointer to the destination
//*
//* @param[in] src_strd
//* integer source stride
//*
//* @param[in] dst_strd
//* integer destination stride
//*
//* @param[in] i4_max_sad
//* integer maximum allowed distortion
//*
//* @param[in] pi4_mb_distortion
//* integer evaluated sad
//*
//* @remarks
//*
//******************************************************************************
//*/
.text
.p2align 2
.macro push_v_regs
stp d8, d9, [sp, #-16]!
stp d10, d11, [sp, #-16]!
stp d12, d13, [sp, #-16]!
stp d14, d15, [sp, #-16]!
.endm
.macro pop_v_regs
ldp d14, d15, [sp], #16
ldp d12, d13, [sp], #16
ldp d10, d11, [sp], #16
ldp d8, d9, [sp], #16
.endm
.global ime_compute_sad_16x16_fast_av8
ime_compute_sad_16x16_fast_av8:
push_v_regs
lsl x2, x2, #1
lsl x3, x3, #1
mov x6, #2
movi v30.8h, #0
core_loop_ime_compute_sad_16x16_fast_av8:
ld1 {v0.16b}, [x0], x2
ld1 {v1.16b}, [x1], x3
ld1 {v2.16b}, [x0], x2
ld1 {v3.16b}, [x1], x3
uabal v30.8h, v0.8b, v1.8b
uabal2 v30.8h, v0.16b, v1.16b
uabal v30.8h, v2.8b, v3.8b
uabal2 v30.8h, v2.16b, v3.16b
ld1 {v4.16b}, [x0], x2
ld1 {v5.16b}, [x1], x3
ld1 {v6.16b}, [x0], x2
ld1 {v7.16b}, [x1], x3
uabal v30.8h, v4.8b, v5.8b
uabal2 v30.8h, v4.16b, v5.16b
uabal v30.8h, v6.8b, v7.8b
uabal2 v30.8h, v6.16b, v7.16b
subs x6, x6, #1
bne core_loop_ime_compute_sad_16x16_fast_av8
addp v30.8h, v30.8h, v30.8h
uaddlp v30.4s, v30.8h
addp v30.2s, v30.2s, v30.2s
shl v30.2s, v30.2s, #1
st1 {v30.s}[0], [x5]
pop_v_regs
ret
///**
//******************************************************************************
//*
//* @brief computes distortion (SAD) between 2 16x8 blocks
//*
//*
//* @par Description
//* This functions computes SAD between 2 16x8 blocks. There is a provision
//* for early exit if the up-to computed SAD exceeds maximum allowed SAD. To
//* compute the distortion of the entire block set u4_max_sad to USHRT_MAX.
//*
//* @param[in] pu1_src
//* UWORD8 pointer to the source
//*
//* @param[out] pu1_dst
//* UWORD8 pointer to the destination
//*
//* @param[in] src_strd
//* integer source stride
//*
//* @param[in] dst_strd
//* integer destination stride
//*
//* @param[in] u4_max_sad
//* integer maximum allowed distortion
//*
//* @param[in] pi4_mb_distortion
//* integer evaluated sad
//*
//* @remarks
//*
//******************************************************************************
//*/
//
.global ime_compute_sad_16x8_av8
ime_compute_sad_16x8_av8:
//chheck what stride incremtn to use
//earlier code did not have this lsl
push_v_regs
mov x6, #2
movi v30.8h, #0
core_loop_ime_compute_sad_16x8_av8:
ld1 {v0.16b}, [x0], x2
ld1 {v1.16b}, [x1], x3
ld1 {v2.16b}, [x0], x2
ld1 {v3.16b}, [x1], x3
uabal v30.8h, v0.8b, v1.8b
uabal2 v30.8h, v0.16b, v1.16b
uabal v30.8h, v2.8b, v3.8b
uabal2 v30.8h, v2.16b, v3.16b
ld1 {v4.16b}, [x0], x2
ld1 {v5.16b}, [x1], x3
ld1 {v6.16b}, [x0], x2
ld1 {v7.16b}, [x1], x3
uabal v30.8h, v4.8b, v5.8b
uabal2 v30.8h, v4.16b, v5.16b
uabal v30.8h, v6.8b, v7.8b
uabal2 v30.8h, v6.16b, v7.16b
subs x6, x6, #1
bne core_loop_ime_compute_sad_16x8_av8
addp v30.8h, v30.8h, v30.8h
uaddlp v30.4s, v30.8h
addp v30.2s, v30.2s, v30.2s
st1 {v30.s}[0], [x5]
pop_v_regs
ret
///**
//******************************************************************************
//*
//* @brief computes distortion (SAD) between 2 16x16 blocks with early exit
//*
//* @par Description
//* This functions computes SAD between 2 16x16 blocks. There is a provision
//* for early exit if the up-to computed SAD exceeds maximum allowed SAD. To
//* compute the distortion of the entire block set u4_max_sad to USHRT_MAX.
//*
//* @param[in] pu1_src
//* UWORD8 pointer to the source
//*
//* @param[out] pu1_dst
//* UWORD8 pointer to the destination
//*
//* @param[in] src_strd
//* integer source stride
//*
//* @param[in] dst_strd
//* integer destination stride
//*
//* @param[in] i4_max_sad
//* integer maximum allowed distortion
//*
//* @param[in] pi4_mb_distortion
//* integer evaluated sad
//*
//* @remarks
//*
//******************************************************************************
//*/
.global ime_compute_sad_16x16_ea8_av8
ime_compute_sad_16x16_ea8_av8:
push_v_regs
movi v30.8h, #0
add x7, x0, x2
add x8, x1, x3
lsl x2, x2, #1
lsl x3, x3, #1
ld1 {v0.16b}, [x0], x2
ld1 {v1.16b}, [x1], x3
ld1 {v2.16b}, [x0], x2
ld1 {v3.16b}, [x1], x3
ld1 {v8.16b}, [x0], x2
ld1 {v9.16b}, [x1], x3
ld1 {v10.16b}, [x0], x2
ld1 {v11.16b}, [x1], x3
ld1 {v12.16b}, [x0], x2
ld1 {v13.16b}, [x1], x3
ld1 {v14.16b}, [x0], x2
ld1 {v15.16b}, [x1], x3
ld1 {v16.16b}, [x0], x2
ld1 {v17.16b}, [x1], x3
ld1 {v18.16b}, [x0], x2
ld1 {v19.16b}, [x1], x3
uabal v30.8h, v0.8b, v1.8b
uabal2 v30.8h, v0.16b, v1.16b
uabal v30.8h, v2.8b, v3.8b
uabal2 v30.8h, v2.16b, v3.16b
uabal v30.8h, v8.8b, v9.8b
uabal2 v30.8h, v8.16b, v9.16b
uabal v30.8h, v10.8b, v11.8b
uabal2 v30.8h, v10.16b, v11.16b
uabal v30.8h, v12.8b, v13.8b
uabal2 v30.8h, v12.16b, v13.16b
uabal v30.8h, v14.8b, v15.8b
uabal2 v30.8h, v14.16b, v15.16b
uabal v30.8h, v16.8b, v17.8b
uabal2 v30.8h, v16.16b, v17.16b
uabal v30.8h, v18.8b, v19.8b
uabal2 v30.8h, v18.16b, v19.16b
addp v31.8h, v30.8h, v30.8h
uaddlp v31.4s, v31.8h
addp v31.2s, v31.2s, v31.2s
mov w6, v31.s[0]
cmp w6, w4
bgt end_func_16x16
//do the stuff again
ld1 {v0.16b}, [x7], x2
ld1 {v1.16b}, [x8], x3
ld1 {v2.16b}, [x7], x2
ld1 {v3.16b}, [x8], x3
ld1 {v8.16b}, [x7], x2
ld1 {v9.16b}, [x8], x3
ld1 {v10.16b}, [x7], x2
ld1 {v11.16b}, [x8], x3
ld1 {v12.16b}, [x7], x2
ld1 {v13.16b}, [x8], x3
ld1 {v14.16b}, [x7], x2
ld1 {v15.16b}, [x8], x3
ld1 {v16.16b}, [x7], x2
ld1 {v17.16b}, [x8], x3
ld1 {v18.16b}, [x7], x2
ld1 {v19.16b}, [x8], x3
uabal v30.8h, v0.8b, v1.8b
uabal2 v30.8h, v0.16b, v1.16b
uabal v30.8h, v2.8b, v3.8b
uabal2 v30.8h, v2.16b, v3.16b
uabal v30.8h, v8.8b, v9.8b
uabal2 v30.8h, v8.16b, v9.16b
uabal v30.8h, v10.8b, v11.8b
uabal2 v30.8h, v10.16b, v11.16b
uabal v30.8h, v12.8b, v13.8b
uabal2 v30.8h, v12.16b, v13.16b
uabal v30.8h, v14.8b, v15.8b
uabal2 v30.8h, v14.16b, v15.16b
uabal v30.8h, v16.8b, v17.8b
uabal2 v30.8h, v16.16b, v17.16b
uabal v30.8h, v18.8b, v19.8b
uabal2 v30.8h, v18.16b, v19.16b
addp v31.8h, v30.8h, v30.8h
uaddlp v31.4s, v31.8h
addp v31.2s, v31.2s, v31.2s
end_func_16x16:
st1 {v31.s}[0], [x5]
pop_v_regs
ret
///*
////---------------------------------------------------------------------------
//// Function Name : ime_calculate_sad2_prog_av8()
////
//// Detail Description : This function find the sad values of 4 Progressive MBs
//// at one shot
////
//// Platform : CortexAv8/NEON .
////
////-----------------------------------------------------------------------------
//*/
.global ime_calculate_sad2_prog_av8
ime_calculate_sad2_prog_av8:
// x0 = ref1 <UWORD8 *>
// x1 = ref2 <UWORD8 *>
// x2 = src <UWORD8 *>
// x3 = RefBufferWidth <UWORD32>
// stack = CurBufferWidth <UWORD32>, psad <UWORD32 *>
push_v_regs
mov x6, #8
movi v30.8h, #0
movi v31.8h, #0
core_loop_ime_calculate_sad2_prog_av8:
ld1 {v0.16b}, [x0], x3
ld1 {v1.16b}, [x1], x3
ld1 {v2.16b}, [x3], x4
ld1 {v3.16b}, [x0], x3
ld1 {v4.16b}, [x1], x3
ld1 {v5.16b}, [x3], x4
uabal v30.8h, v0.8b, v2.8b
uabal2 v30.8h, v0.16b, v2.16b
uabal v31.8h, v1.8b, v2.8b
uabal2 v31.8h, v1.16b, v2.16b
uabal v30.8h, v3.8b, v5.8b
uabal2 v30.8h, v3.16b, v5.16b
uabal v31.8h, v4.8b, v5.8b
uabal2 v31.8h, v4.16b, v5.16b
ld1 {v6.16b}, [x0], x3
ld1 {v7.16b}, [x1], x3
ld1 {v8.16b}, [x3], x4
ld1 {v9.16b}, [x0], x3
ld1 {v10.16b}, [x1], x3
ld1 {v11.16b}, [x3], x4
uabal v30.8h, v6.8b, v8.8b
uabal2 v30.8h, v6.16b, v8.16b
uabal v31.8h, v7.8b, v8.8b
uabal2 v31.8h, v7.16b, v8.16b
uabal v30.8h, v9.8b, v11.8b
uabal2 v30.8h, v9.16b, v11.16b
uabal v31.8h, v10.8b, v11.8b
uabal2 v31.8h, v0.16b, v11.16b
subs x6, x6, #1
bne core_loop_ime_calculate_sad2_prog_av8
addp v30.8h, v30.8h, v31.8h
uaddlp v30.4s, v30.8h
addp v30.2s, v30.2s, v30.2s
shl v30.2s, v30.2s, #1
st1 {v30.2s}, [x5]
pop_v_regs
ret
///*
////---------------------------------------------------------------------------
//// Function Name : Calculate_Mad3_prog()
////
//// Detail Description : This function find the sad values of 4 Progressive MBs
//// at one shot
////
//// Platform : CortexA8/NEON .
////
////-----------------------------------------------------------------------------
//*/
.global ime_calculate_sad3_prog_av8
ime_calculate_sad3_prog_av8:
// x0 = ref1 <UWORD8 *>
// x1 = ref2 <UWORD8 *>
// x2 = ref3 <UWORD8 *>
// x3 = src <UWORD8 *>
// stack = RefBufferWidth <UWORD32>, CurBufferWidth <UWORD32>, psad <UWORD32 *>
// x0 = ref1 <UWORD8 *>
// x1 = ref2 <UWORD8 *>
// x2 = src <UWORD8 *>
// x3 = RefBufferWidth <UWORD32>
// stack = CurBufferWidth <UWORD32>, psad <UWORD32 *>
push_v_regs
mov x6, #16
movi v29.8h, #0
movi v30.8h, #0
movi v31.8h, #0
core_loop_ime_calculate_sad3_prog_av8:
ld1 {v0.16b}, [x0], x4
ld1 {v1.16b}, [x1], x4
ld1 {v2.16b}, [x2], x4
ld1 {v3.16b}, [x3], x5
uabal v29.8h, v0.8b, v3.8b
uabal2 v29.8h, v0.16b, v3.16b
uabal v30.8h, v1.8b, v3.8b
uabal2 v30.8h, v1.16b, v3.16b
uabal v31.8h, v2.8b, v3.8b
uabal2 v31.8h, v2.16b, v3.16b
ld1 {v4.16b}, [x0], x4
ld1 {v5.16b}, [x1], x4
ld1 {v6.16b}, [x2], x4
ld1 {v7.16b}, [x3], x5
uabal v29.8h, v4.8b, v7.8b
uabal2 v29.8h, v4.16b, v7.16b
uabal v30.8h, v5.8b, v7.8b
uabal2 v30.8h, v5.16b, v7.16b
uabal v31.8h, v6.8b, v7.8b
uabal2 v31.8h, v6.16b, v7.16b
subs x6, x6, #1
bne core_loop_ime_calculate_sad2_prog_av8
addp v30.8h, v30.8h, v31.8h
uaddlp v30.4s, v30.8h
addp v30.2s, v30.2s, v30.2s
shl v30.2s, v30.2s, #1
st1 {v30.2s}, [x5]
pop_v_regs
ret
///**
//******************************************************************************
//*
//* @brief computes distortion (SAD) for sub-pel motion estimation
//*
//* @par Description
//* This functions computes SAD for all the 8 half pel points
//*
//* @param[out] pi4_sad
//* integer evaluated sad
//* pi4_sad[0] - half x
//* pi4_sad[1] - half x - 1
//* pi4_sad[2] - half y
//* pi4_sad[3] - half y - 1
//* pi4_sad[4] - half xy
//* pi4_sad[5] - half xy - 1
//* pi4_sad[6] - half xy - strd
//* pi4_sad[7] - half xy - 1 - strd
//*
//* @remarks
//*
//******************************************************************************
//*/
.text
.p2align 2
.global ime_sub_pel_compute_sad_16x16_av8
ime_sub_pel_compute_sad_16x16_av8:
push_v_regs
sub x7, x1, #1 //x left
sub x8, x2, x5 //y top
sub x9, x3, #1 //xy left
sub x10, x3, x5 //xy top
sub x11, x10, #1 //xy top left
movi v24.8h, #0
movi v25.8h, #0
movi v26.8h, #0
movi v27.8h, #0
movi v28.8h, #0
movi v29.8h, #0
movi v30.8h, #0
movi v31.8h, #0
mov x12, #16
core_loop_ime_sub_pel_compute_sad_16x16_av8:
ld1 {v0.16b}, [x0], x4 //src
ld1 {v1.16b}, [x1], x5 //x
ld1 {v2.16b}, [x7], x5 //x left
ld1 {v3.16b}, [x2], x5 //y
ld1 {v9.16b}, [x8], x5 //y top
ld1 {v10.16b}, [x3], x5 //xy
ld1 {v11.16b}, [x9], x5 //xy left
ld1 {v12.16b}, [x10], x5 //xy top
ld1 {v13.16b}, [x11], x5 //xy top left
uabal v24.8h, v0.8b, v1.8b
uabal2 v24.8h, v0.16b, v1.16b
uabal v25.8h, v0.8b, v2.8b
uabal2 v25.8h, v0.16b, v2.16b
uabal v26.8h, v0.8b, v3.8b
uabal2 v26.8h, v0.16b, v3.16b
uabal v27.8h, v0.8b, v9.8b
uabal2 v27.8h, v0.16b, v9.16b
uabal v28.8h, v0.8b, v10.8b
uabal2 v28.8h, v0.16b, v10.16b
uabal v29.8h, v0.8b, v11.8b
uabal2 v29.8h, v0.16b, v11.16b
uabal v30.8h, v0.8b, v12.8b
uabal2 v30.8h, v0.16b, v12.16b
uabal v31.8h, v0.8b, v13.8b
uabal2 v31.8h, v0.16b, v13.16b
subs x12, x12, #1
bne core_loop_ime_sub_pel_compute_sad_16x16_av8
addp v24.8h, v24.8h, v25.8h
addp v26.8h, v26.8h, v27.8h
addp v28.8h, v28.8h, v29.8h
addp v30.8h, v30.8h, v31.8h
uaddlp v24.4s, v24.8h
uaddlp v26.4s, v26.8h
uaddlp v28.4s, v28.8h
uaddlp v30.4s, v30.8h
addp v24.4s, v24.4s, v26.4s
addp v25.4s, v28.4s, v30.4s
st1 {v24.4s-v25.4s}, [x6]
pop_v_regs
ret
///**
//******************************************************************************
//*
//* @brief computes distortion (SAD) between 2 16x16 blocks
//*
//* @par Description
//* This functions computes SAD between 2 16x16 blocks. There is a provision
//* for early exit if the up-to computed SAD exceeds maximum allowed SAD. To
//* compute the distortion of the entire block set u4_max_sad to USHRT_MAX.
//*
//* @param[in] pu1_src
//* UWORD8 pointer to the source
//*
//* @param[out] pu1_dst
//* UWORD8 pointer to the destination
//*
//* @param[in] src_strd
//* integer source stride
//*
//* @param[in] dst_strd
//* integer destination stride
//*
//* @param[in] i4_max_sad
//* integer maximum allowed distortion
//*
//* @param[in] pi4_mb_distortion
//* integer evaluated sad
//*
//* @remarks
//*
//******************************************************************************
//*/
.global ime_compute_sad_16x16_av8
ime_compute_sad_16x16_av8:
push_v_regs
mov x6, #4
movi v30.8h, #0
core_loop_ime_compute_sad_16x16_av8:
ld1 {v0.16b}, [x0], x2
ld1 {v1.16b}, [x1], x3
ld1 {v2.16b}, [x0], x2
ld1 {v3.16b}, [x1], x3
uabal v30.8h, v0.8b, v1.8b
uabal2 v30.8h, v0.16b, v1.16b
uabal v30.8h, v2.8b, v3.8b
uabal2 v30.8h, v2.16b, v3.16b
ld1 {v4.16b}, [x0], x2
ld1 {v5.16b}, [x1], x3
ld1 {v6.16b}, [x0], x2
ld1 {v7.16b}, [x1], x3
uabal v30.8h, v4.8b, v5.8b
uabal2 v30.8h, v4.16b, v5.16b
uabal v30.8h, v6.8b, v7.8b
uabal2 v30.8h, v6.16b, v7.16b
subs x6, x6, #1
bne core_loop_ime_compute_sad_16x16_av8
addp v30.8h, v30.8h, v30.8h
uaddlp v30.4s, v30.8h
addp v30.2s, v30.2s, v30.2s
st1 {v30.s}[0], [x5]
pop_v_regs
ret
///*
////---------------------------------------------------------------------------
//// Function Name : Calculate_Mad4_prog()
////
//// Detail Description : This function find the sad values of 4 Progressive MBs
//// at one shot
////
//// Platform : CortexA8/NEON .
////
////-----------------------------------------------------------------------------
//*/
.global ime_calculate_sad4_prog_av8
ime_calculate_sad4_prog_av8:
push_v_regs
sub x5, x0, #1 //left
add x6, x0, #1 //right
sub x7, x0, x2 //top
add x8, x0, x2 //bottom
movi v28.8h, #0
movi v29.8h, #0
movi v30.8h, #0
movi v31.8h, #0
mov x9, #16
core_loop_ime_calculate_sad4_prog_av8:
ld1 {v0.16b}, [x1], x3
ld1 {v1.16b}, [x5], x2
ld1 {v2.16b}, [x6], x2
ld1 {v3.16b}, [x7], x2
ld1 {v9.16b}, [x8], x2
uabal v28.8h, v0.8b, v1.8b
uabal2 v28.8h, v0.16b, v1.16b
uabal v29.8h, v0.8b, v2.8b
uabal2 v29.8h, v0.16b, v2.16b
uabal v30.8h, v0.8b, v3.8b
uabal2 v30.8h, v0.16b, v3.16b
uabal v31.8h, v0.8b, v9.8b
uabal2 v31.8h, v0.16b, v9.16b
subs x9, x9, #1
bne core_loop_ime_calculate_sad4_prog_av8
addp v28.8h, v28.8h, v29.8h
addp v30.8h, v30.8h, v31.8h
uaddlp v28.4s, v28.8h
uaddlp v30.4s, v30.8h
addp v28.4s, v28.4s, v30.4s
st1 {v28.4s}, [x4]
pop_v_regs
ret
//*****************************************************************************
//*
//* Function Name : ime_compute_satqd_16x16_lumainter_av8
//* Description : This fucntion computes SAD for a 16x16 block.
// : It also computes if any 4x4 block will have a nonzero coefficent after transform and quant
//
// Arguments : x0 :pointer to src buffer
// x1 :pointer to est buffer
// x2 :source stride
// x3 :est stride
// STACk :Threshold,distotion,is_nonzero
//*
//* Values Returned : NONE
//*
//* Register Usage : x0-x11
//* Stack Usage :
//* Cycles : Around
//* Interruptiaility : Interruptable
//*
//* Known Limitations
//* \Assumptions :
//*
//* Revision History :
//* DD MM YYYY Author(s) Changes
//* 14 04 2014 Harinarayanan K K First version
//*
//*****************************************************************************
.global ime_compute_satqd_16x16_lumainter_av8
ime_compute_satqd_16x16_lumainter_av8:
//x0 :pointer to src buffer
//x1 :pointer to est buffer
//x2 :Source stride
//x3 :Pred stride
//x4 :Threshold pointer
//x5 :Distortion,ie SAD
//x6 :is nonzero
//x7 :loop counter
push_v_regs
stp d8, d9, [sp, #-16]!
stp d10, d11, [sp, #-16]!
stp d12, d13, [sp, #-16]!
stp d14, d15, [sp, #-16]!
ld1 {v30.8h}, [x4]
dup v20.4h, v30.h[1] //ls1
dup v24.4h, v30.h[0] //ls2
dup v21.4h, v30.h[5] //ls3
dup v25.4h, v30.h[7] //ls4
dup v22.4h, v30.h[3] //ls5
dup v26.4h, v30.h[4] //ls6
dup v23.4h, v30.h[6] //ls7
dup v27.4h, v30.h[2] //ls8
mov v20.d[1], v24.d[0]
mov v21.d[1], v25.d[0]
mov v22.d[1], v26.d[0]
mov v23.d[1], v27.d[0]
add x4, x4, #16
ld1 {v29.h}[0], [x4]
dup v29.4h, v29.h[0]
movi v31.8h, #0
mov x7, #4
core_loop_satqd_ime_compute_satqd_16x16_lumainter:
ld1 {v0.16b}, [x0], x2
ld1 {v1.16b}, [x1], x3
ld1 {v2.16b}, [x0], x2
ld1 {v3.16b}, [x1], x3
ld1 {v4.16b}, [x0], x2
ld1 {v5.16b}, [x1], x3
ld1 {v6.16b}, [x0], x2
ld1 {v7.16b}, [x1], x3
uabdl v10.8h, v0.8b, v1.8b
uabdl2 v15.8h, v0.16b, v1.16b
uabdl v11.8h, v2.8b, v3.8b
uabdl2 v16.8h, v2.16b, v3.16b
uabdl v12.8h, v4.8b, v5.8b
uabdl2 v17.8h, v4.16b, v5.16b
uabdl v13.8h, v6.8b, v7.8b
uabdl2 v18.8h, v6.16b, v7.16b
add v0.8h, v10.8h, v13.8h
add v1.8h, v11.8h, v12.8h
add v2.8h, v15.8h, v18.8h
add v3.8h, v16.8h, v17.8h
//v0 : S1 S4 S4 S1 A1 A4 A4 A1
//v1 : S2 S3 S3 S2 A2 A3 A3 A2
//v2 : B1 B4 B4 B1 X1 X4 X4 X1
//v3 : B3 B2 B2 B3 X3 X2 X2 X3
trn1 v4.8h, v0.8h, v1.8h
trn2 v5.8h, v0.8h, v1.8h
trn1 v6.8h, v2.8h, v3.8h
trn2 v7.8h, v2.8h, v3.8h
trn1 v0.4s, v4.4s, v6.4s
trn2 v2.4s, v4.4s, v6.4s
trn1 v1.4s, v5.4s, v7.4s
trn2 v3.4s, v5.4s, v7.4s
add v4.8h, v0.8h, v3.8h
add v5.8h, v1.8h, v2.8h
//v4 : S1 S2 B1 B2 A1 A2 X1 X2
//v5 : S4 S3 B4 B3 A4 A3 X4 X3
//compute sad for each 4x4 block
add v6.8h, v4.8h, v5.8h
addp v19.8h, v6.8h, v6.8h
//duplicate the sad into 128 bit so that we can compare using 128bit
add v31.4h, v31.4h, v19.4h
//sad_2 = sad_1<<1;
shl v28.8h, v19.8h, #1
//sad_2 - pu2_thrsh
sub v24.8h, v28.8h, v20.8h
sub v25.8h, v28.8h, v21.8h
sub v26.8h, v28.8h, v22.8h
sub v27.8h, v28.8h, v23.8h
trn1 v0.4s, v4.4s, v5.4s
trn2 v1.4s, v4.4s, v5.4s
//v0 : S1 S2 S4 S3 A1 A2 A4 A3
//v1 : B1 B2 B4 B3 X1 X2 X4 X3
trn1 v4.8h, v0.8h, v1.8h
trn2 v5.8h, v0.8h, v1.8h
//v4 : S1 B1 S4 B4 A1 X1 A4 X4
//v5 : S2 B2 S3 B3 A2 X2 A3 X3
mov v7.s[0], v4.s[1]
mov v7.s[1], v4.s[3]
mov v6.s[0], v5.s[1] // V4 //S1 B1 A1 X1
mov v6.s[1], v5.s[3] // V5 //S2 B2 A2 X2
mov v4.s[1], v4.s[2] // V6 //S3 B3 A3 X3
mov v5.s[1], v5.s[2] // V7 //S4 B4 A4 X4
shl v0.4h, v4.4h, #1 //S1<<1
shl v1.4h, v5.4h, #1 //S2<<1
shl v2.4h, v6.4h, #1 //S3<<1
shl v3.4h, v7.4h, #1 //S4<<1
add v8.4h, v5.4h, v6.4h //(s2[j] + s3[j]))
add v9.4h, v4.4h, v7.4h //(s1[j] + s4[j]))
add v10.4h, v6.4h, v7.4h //(s3[j] + s4[j]))
sub v11.4h, v6.4h, v0.4h //(s3[j] - (s1[j]<<1))
sub v12.4h, v7.4h, v1.4h //(s4[j] - (s2[j]<<1))
add v13.4h, v4.4h, v5.4h //(s1[j] + s2[j]))
sub v14.4h, v5.4h, v3.4h //(s2[j] - (s4[j]<<1)))
sub v15.4h, v4.4h, v2.4h //(s1[j] - (s3[j]<<1)))
mov v8.d[1], v9.d[0]
mov v10.d[1], v11.d[0]
mov v12.d[1], v13.d[0]
mov v14.d[1], v15.d[0]
cmge v0.8h, v24.8h, v8.8h //ls1 ls2
cmge v1.8h, v25.8h, v10.8h //ls3 ls4
cmge v2.8h, v26.8h, v12.8h //ls5 ls6
cmge v3.8h, v27.8h, v14.8h //ls7 ls8
cmge v4.4h, v19.4h, v29.4h //sad
orr v0.16b, v0.16b, v1.16b
orr v2.16b, v2.16b, v3.16b
orr v2.16b, v0.16b, v2.16b
xtn v2.8b, v2.8h
orr v2.8b, v2.8b, v4.8b
//if the comparison is non zero, out
mov x4, v2.d[0]
cmp x4, #0
bne core_loop_compute_sad_pre
subs x7, x7, #1
bne core_loop_satqd_ime_compute_satqd_16x16_lumainter
b satdq_end_func
core_loop_compute_sad:
ld1 {v0.16b}, [x0], x2
ld1 {v1.16b}, [x1], x3
ld1 {v2.16b}, [x0], x2
ld1 {v3.16b}, [x1], x3
uabal v31.8h, v0.8b, v1.8b
uabal2 v31.8h, v0.16b, v1.16b
uabal v31.8h, v2.8b, v3.8b
uabal2 v31.8h, v2.16b, v3.16b
ld1 {v4.16b}, [x0], x2
ld1 {v5.16b}, [x1], x3
ld1 {v6.16b}, [x0], x2
ld1 {v7.16b}, [x1], x3
uabal v31.8h, v4.8b, v5.8b
uabal2 v31.8h, v4.16b, v5.16b
uabal v31.8h, v6.8b, v7.8b
uabal2 v31.8h, v6.16b, v7.16b
core_loop_compute_sad_pre:
subs x7, x7, #1
bne core_loop_compute_sad
satdq_end_func:
mov x7, #1
cmp x4, #0
csel x7, x4, x7, eq
str w7, [x6]
addp v31.8h, v31.8h, v31.8h
uaddlp v31.4s, v31.8h
addp v31.2s, v31.2s, v31.2s
st1 {v31.s}[0], [x5]
ldp d14, d15, [sp], #16
ldp d12, d13, [sp], #16
ldp d10, d11, [sp], #16
ldp d8, d9, [sp], #16
pop_v_regs
ret
|