-
Notifications
You must be signed in to change notification settings - Fork 259
/
Copy pathitx16_avx2.asm
8599 lines (8392 loc) · 311 KB
/
itx16_avx2.asm
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
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
; Copyright © 2021, VideoLAN and dav1d authors
; Copyright © 2021, Two Orioles, LLC
; Copyright © 2021, Matthias Dressel
; All rights reserved.
;
; Redistribution and use in source and binary forms, with or without
; modification, are permitted provided that the following conditions are met:
;
; 1. Redistributions of source code must retain the above copyright notice, this
; list of conditions and the following disclaimer.
;
; 2. Redistributions in binary form must reproduce the above copyright notice,
; this list of conditions and the following disclaimer in the documentation
; and/or other materials provided with the distribution.
;
; THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
; ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
; WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
; DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
; ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
; (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
; ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
; (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
; SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
%include "config.asm"
%include "ext/x86/x86inc.asm"
%if ARCH_X86_64
SECTION_RODATA 32
itx4_shuf: dd 0x50401600, 0xd0c09284, 0x70603422, 0xf0e0b0a6
dd 0x50401701, 0xd0c09385, 0x70603523, 0xf0e0b1a7
idct4_12_shuf: dd 0, 2, 4, 6, 1, 3, 5, 7
idct4_12_shuf2: dd 2, 0, 6, 4, 3, 1, 7, 5
iadst8_12_shuf: dd 0, 4, 1, 5, 2, 6, 3, 7
idct16_12_shuf: dd 0, 4, 1, 5, 3, 7, 2, 6
iadst16_12_shuf: dd 3, 7, 0, 4, 2, 6, 1, 5
pw_2048_m2048: dw 2048, 2048, 2048, 2048, -2048, -2048, -2048, -2048
idct4_shuf: db 0, 1, 4, 5, 12, 13, 8, 9, 2, 3, 6, 7, 14, 15, 10, 11
idct32_shuf: db 0, 1, 8, 9, 4, 5, 12, 13, 2, 3, 10, 11, 6, 7, 14, 15
%macro COEF_PAIR 2-3 0
pd_%1_%2: dd %1, %1, %2, %2
%define pd_%1 (pd_%1_%2 + 4*0)
%define pd_%2 (pd_%1_%2 + 4*2)
%if %3
dd -%2, -%2
%define pd_%2_m%2 pd_%2
%endif
%endmacro
COEF_PAIR 201, 995
COEF_PAIR 401, 1931
COEF_PAIR 799, 3406
COEF_PAIR 1380, 601
COEF_PAIR 1751, 2440
COEF_PAIR 2598, 1189
COEF_PAIR 2751, 2106
COEF_PAIR 2896, 1567, 1
COEF_PAIR 2896, 3784, 1
COEF_PAIR 3035, 3513
COEF_PAIR 3166, 3920
COEF_PAIR 3703, 3290
COEF_PAIR 3857, 4052
COEF_PAIR 4017, 2276
COEF_PAIR 4076, 3612
COEF_PAIR 4091, 3973
pd_8: dd 8
pd_m601: dd -601
pd_m1189: dd -1189
pd_m1380: dd -1380
pd_m2106: dd -2106
pd_m2598: dd -2598
pd_m2751: dd -2751
pd_m3344: dd -3344
pd_1024: dd 1024
pd_1321: dd 1321
pd_1448: dd 1448
pd_1697: dd 1697
pd_2482: dd 2482
pd_3072: dd 3072 ; 1024 + 2048
pd_3803: dd 3803
pd_5119: dd 5119 ; 1024 + 4096 - 1
pd_5120: dd 5120 ; 1024 + 4096
pd_5793: dd 5793
pd_6144: dd 6144 ; 2048 + 4096
pd_17408: dd 17408 ; 1024 + 16384
pixel_10bpc_max: times 2 dw 0x03ff
pixel_12bpc_max: times 2 dw 0x0fff
dconly_10bpc: times 2 dw 0x7c00
dconly_12bpc: times 2 dw 0x7000
clip_18b_min: dd -0x20000
clip_18b_max: dd 0x1ffff
clip_20b_min: dd -0x80000
clip_20b_max: dd 0x7ffff
const idct64_mul_16bpc
dd 4095, 101, 2967, -2824, 3745, 1660, 3822, -1474, 401, 4076, 799, 4017
dd -700, 4036, 2359, 3349, -2191, 3461, 897, 3996, -2598, -3166, -4017, -799
dd 4065, 501, 3229, -2520, 3564, 2019, 3948, -1092, 1931, 3612, 3406, 2276
dd -301, 4085, 2675, 3102, -1842, 3659, 1285, 3889, -1189, -3920, -2276, -3406
cextern deint_shuf
cextern idct64_mul
cextern pw_1697x8
cextern pw_1697x16
cextern pw_1567_3784
cextern pw_m1567_m3784
cextern pw_m3784_1567
cextern pw_2896_2896
cextern pw_m2896_2896
cextern pw_5
cextern pw_2048
cextern pw_4096
cextern pw_8192
cextern pw_16384
cextern pw_2896x8
cextern pd_2048
cextern idct_4x8_internal_8bpc_avx2.main
cextern idct_4x16_internal_8bpc_avx2.main
cextern idct_8x8_internal_8bpc_avx2.main
cextern idct_8x16_internal_8bpc_avx2.main
cextern idct_16x4_internal_8bpc_avx2.main
cextern idct_16x8_internal_8bpc_avx2.main
cextern idct_16x16_internal_8bpc_avx2.main
cextern inv_txfm_add_dct_dct_8x32_8bpc_avx2.main
cextern inv_txfm_add_dct_dct_8x32_8bpc_avx2.main_fast
cextern inv_txfm_add_dct_dct_16x32_8bpc_avx2.main_oddhalf
cextern inv_txfm_add_dct_dct_16x32_8bpc_avx2.main_oddhalf_fast
cextern inv_txfm_add_dct_dct_16x64_8bpc_avx2.main_part1
cextern inv_txfm_add_dct_dct_16x64_8bpc_avx2.main_part2_internal
cextern iadst_4x4_internal_8bpc_avx2.main
cextern iadst_4x8_internal_8bpc_avx2.main_pass2
cextern iadst_4x16_internal_8bpc_avx2.main2
cextern iadst_8x4_internal_8bpc_avx2.main
cextern iadst_8x8_internal_8bpc_avx2.main_pass2
cextern iadst_8x16_internal_8bpc_avx2.main
cextern iadst_8x16_internal_8bpc_avx2.main_pass2_end
cextern iadst_16x4_internal_8bpc_avx2.main
cextern iadst_16x8_internal_8bpc_avx2.main
cextern iadst_16x8_internal_8bpc_avx2.main_pass2_end
cextern iadst_16x16_internal_8bpc_avx2.main
cextern iadst_16x16_internal_8bpc_avx2.main_pass2_end
SECTION .text
%define m(x) mangle(private_prefix %+ _ %+ x %+ SUFFIX)
%macro WRAP_XMM 1+
INIT_XMM cpuname
%1
INIT_YMM cpuname
%endmacro
%macro IWHT4_1D_PACKED 0
; m0 = in0 in2, m1 = in1 in3
psubd m2, m0, m1 ; t2
paddd xm0, xm1 ; t0
vpermq m2, m2, q3322
vpermq m0, m0, q1100
vpermq m1, m1, q3120
psubd m3, m0, m2
psrad m3, 1
psubd m3, m1 ; t1 t3
psubd m0, m3 ; ____ out0
paddd m2, m3 ; out3 ____
%endmacro
INIT_YMM avx2
cglobal inv_txfm_add_wht_wht_4x4_16bpc, 3, 7, 6, dst, stride, c, eob, bdmax
mova xm0, [cq+16*0]
vinserti128 m0, [cq+16*2], 1
mova xm1, [cq+16*1]
vinserti128 m1, [cq+16*3], 1
pxor m4, m4
mova [cq+32*0], m4
mova [cq+32*1], m4
lea r6, [dstq+strideq*2]
psrad m0, 2
psrad m1, 2
IWHT4_1D_PACKED
punpckhdq m0, m3
punpckldq m3, m2
punpckhqdq m1, m0, m3
punpcklqdq m0, m3
IWHT4_1D_PACKED
vpblendd m0, m2, 0x33
packssdw m0, m3
vextracti128 xm2, m0, 1
punpckhdq xm1, xm0, xm2 ; out2 out1
punpckldq xm0, xm2 ; out3 out0
movq xm2, [r6 +strideq*1]
movhps xm2, [dstq+strideq*0]
movq xm3, [r6 +strideq*0]
movhps xm3, [dstq+strideq*1]
%ifidn bdmaxd, bdmaxm
movd xm5, bdmaxd
vpbroadcastw xm5, xm5
%else ; win64: load from stack
vpbroadcastw xm5, bdmaxm
%endif
paddsw xm0, xm2
paddsw xm1, xm3
pmaxsw xm0, xm4
pmaxsw xm1, xm4
pminsw xm0, xm5
pminsw xm1, xm5
movhps [dstq+strideq*0], xm0
movhps [dstq+strideq*1], xm1
movq [r6 +strideq*0], xm1
movq [r6 +strideq*1], xm0
RET
; dst1 = (src1 * coef1 - src2 * coef2 + rnd) >> 12
; dst2 = (src1 * coef2 + src2 * coef1 + rnd) >> 12
; flags: 1 = packed, 2 = inv_dst2
; skip round/shift if rnd is not a number
%macro ITX_MULSUB_2D 8-9 0 ; dst/src[1-2], tmp[1-3], rnd, coef[1-2], flags
%if %8 < 32
pmulld m%4, m%1, m%8
pmulld m%3, m%2, m%8
%else
%if %9 & 1
vbroadcasti128 m%3, [pd_%8]
%else
vpbroadcastd m%3, [pd_%8]
%endif
pmulld m%4, m%1, m%3
pmulld m%3, m%2
%endif
%if %7 < 32
pmulld m%1, m%7
pmulld m%2, m%7
%else
%if %9 & 1
vbroadcasti128 m%5, [pd_%7]
%else
vpbroadcastd m%5, [pd_%7]
%endif
pmulld m%1, m%5
pmulld m%2, m%5
%endif
%if %9 & 2
psubd m%4, m%6, m%4
psubd m%2, m%4, m%2
%else
%ifnum %6
paddd m%4, m%6
%endif
paddd m%2, m%4
%endif
%ifnum %6
paddd m%1, m%6
%endif
psubd m%1, m%3
%ifnum %6
psrad m%2, 12
psrad m%1, 12
%endif
%endmacro
%macro INV_TXFM_FN 4-5 10 ; type1, type2, eob_offset, size, bitdepth
cglobal inv_txfm_add_%1_%2_%4_%5bpc, 4, 5, 0, dst, stride, c, eob, tx2
%define %%p1 m(i%1_%4_internal_%5bpc)
; Jump to the 1st txfm function if we're not taking the fast path, which
; in turn performs an indirect jump to the 2nd txfm function.
lea tx2q, [m(i%2_%4_internal_%5bpc).pass2]
%ifidn %1_%2, dct_dct
test eobd, eobd
jnz %%p1
%else
%if %3
add eobd, %3
%endif
; jump to the 1st txfm function unless it's located directly after this
times ((%%end - %%p1) >> 31) & 1 jmp %%p1
ALIGN function_align
%%end:
%endif
%endmacro
%macro INV_TXFM_4X4_FN 2-3 10 ; type1, type2, bitdepth
INV_TXFM_FN %1, %2, 0, 4x4, %3
%ifidn %1_%2, dct_dct
vpbroadcastd xm2, [dconly_%3bpc]
%if %3 = 10
.dconly:
imul r6d, [cq], 181
mov [cq], eobd ; 0
or r3d, 4
.dconly2:
add r6d, 128
sar r6d, 8
.dconly3:
imul r6d, 181
add r6d, 2176
sar r6d, 12
movd xm0, r6d
paddsw xm0, xm2
vpbroadcastw xm0, xm0
.dconly_loop:
movq xm1, [dstq+strideq*0]
movhps xm1, [dstq+strideq*1]
paddsw xm1, xm0
psubusw xm1, xm2
movq [dstq+strideq*0], xm1
movhps [dstq+strideq*1], xm1
lea dstq, [dstq+strideq*2]
sub r3d, 2
jg .dconly_loop
WRAP_XMM RET
%else
jmp m(inv_txfm_add_dct_dct_4x4_10bpc).dconly
%endif
%endif
%endmacro
%macro IDCT4_1D_PACKED 6 ; dst/src[1-2], tmp[1-3], rnd
ITX_MULSUB_2D %1, %2, %3, %4, %5, %6, 2896_1567, 2896_3784, 1
punpckhqdq m%3, m%2, m%1 ; t3 t2
punpcklqdq m%2, m%1 ; t0 t1
paddd m%1, m%2, m%3 ; out0 out1
psubd m%2, m%3 ; out3 out2
%endmacro
%macro IDCT4_1D_PACKED_WORD 6 ; dst/src[1-2], tmp[1-3], rnd
vpbroadcastd m%5, [pw_m3784_1567]
punpckhwd m%3, m%2, m%1
vpbroadcastd m%4, [pw_1567_3784]
punpcklwd m%2, m%1
vpbroadcastd m%1, [pw_m2896_2896]
pmaddwd m%5, m%3
pmaddwd m%3, m%4
vpbroadcastd m%4, [pw_2896_2896]
pmaddwd m%1, m%2
pmaddwd m%2, m%4
REPX {paddd x, m%6}, m%5, m%3, m%1, m%2
REPX {psrad x, 12 }, m%5, m%3, m%1, m%2
packssdw m%3, m%5 ; t3 t2
packssdw m%2, m%1 ; t0 t1
paddsw m%1, m%2, m%3 ; out0 out1
psubsw m%2, m%3 ; out3 out2
%endmacro
INV_TXFM_4X4_FN dct, dct
INV_TXFM_4X4_FN dct, identity
INV_TXFM_4X4_FN dct, adst
INV_TXFM_4X4_FN dct, flipadst
cglobal idct_4x4_internal_10bpc, 0, 7, 6, dst, stride, c, eob, tx2
call .main
vbroadcasti128 m2, [idct4_shuf]
packssdw m0, m1
pshufb m0, m2
jmp tx2q
.pass2:
vextracti128 xm1, m0, 1
WRAP_XMM IDCT4_1D_PACKED_WORD 0, 1, 2, 3, 4, 5
packssdw xm5, xm5 ; pw_2048
pmulhrsw xm0, xm5
pmulhrsw xm1, xm5
movq xm2, [dstq+strideq*0]
movhps xm2, [dstq+strideq*1]
lea r6, [dstq+strideq*2]
movq xm3, [r6 +strideq*1]
movhps xm3, [r6 +strideq*0]
vpbroadcastd xm5, [pixel_10bpc_max]
pxor m4, m4
mova [cq+32*0], m4
mova [cq+32*1], m4
paddw xm0, xm2
paddw xm1, xm3
pmaxsw xm0, xm4
pmaxsw xm1, xm4
pminsw xm0, xm5
pminsw xm1, xm5
movq [dstq+strideq*0], xm0
movhps [dstq+strideq*1], xm0
movhps [r6 +strideq*0], xm1
movq [r6 +strideq*1], xm1
RET
ALIGN function_align
.main:
vpermq m0, [cq+32*0], q3120
vpermq m1, [cq+32*1], q3120
vpbroadcastd m5, [pd_2048]
.main2:
IDCT4_1D_PACKED 0, 1, 2, 3, 4, 5
ret
INV_TXFM_4X4_FN adst, dct
INV_TXFM_4X4_FN adst, adst
INV_TXFM_4X4_FN adst, flipadst
INV_TXFM_4X4_FN adst, identity
%macro IADST4_1D 0
vpbroadcastd m5, [pd_1321]
vpbroadcastd m7, [pd_2482]
pmulld m4, m0, m5 ; 1321*in0
pmulld m6, m3, m7 ; 2482*in3
paddd m4, m6 ; 1321*in0 + 2482*in3
pmulld m6, m0, m7 ; 2482*in0
paddd m0, m3 ; in0 + in3
paddd m7, m5 ; pd_3803
pmulld m5, m2 ; 1321*in2
pmulld m3, m7 ; 3803*in3
pmulld m7, m2 ; 3803*in2
psubd m2, m0 ; in2 - in0 - in3
vpbroadcastd m0, [pd_m3344]
pmulld m1, m0 ; -t3
pmulld m2, m0 ; out2 (unrounded)
psubd m6, m5 ; 2482*in0 - 1321*in2
paddd m4, m7 ; t0
psubd m6, m3 ; t1
paddd m3, m4, m6
psubd m4, m1 ; out0 (unrounded)
psubd m6, m1 ; out1 (unrounded)
paddd m3, m1 ; out3 (unrounded)
%endmacro
cglobal iadst_4x4_internal_10bpc, 0, 7, 6, dst, stride, c, eob, tx2
call .main
vinserti128 m0, m4, xm6, 1
vinserti128 m1, m2, xm3, 1
.pass1_end:
vpbroadcastd m5, [pd_2048]
mova m2, [itx4_shuf]
paddd m0, m5
paddd m1, m5
psrad m0, 12
psrad m1, 12
packssdw m0, m1
vpermd m0, m2, m0
psrld m2, 4
pshufb m0, m2
%if WIN64
movaps xmm6, [rsp+ 8]
movaps xmm7, [rsp+24]
%endif
jmp tx2q
.pass2:
lea r6, [deint_shuf+128]
vextracti128 xm1, m0, 1
call m(iadst_4x4_internal_8bpc).main
.end:
vpbroadcastd xm4, [pw_2048]
movq xm2, [dstq+strideq*0]
movhps xm2, [dstq+strideq*1]
lea r6, [dstq+strideq*2]
movq xm3, [r6 +strideq*0]
movhps xm3, [r6 +strideq*1]
vpbroadcastd xm5, [pixel_10bpc_max]
pmulhrsw xm0, xm4
pmulhrsw xm1, xm4
pxor m4, m4
mova [cq+32*0], m4
mova [cq+32*1], m4
paddw xm0, xm2
paddw xm1, xm3
pmaxsw xm0, xm4
pmaxsw xm1, xm4
pminsw xm0, xm5
pminsw xm1, xm5
movq [dstq+strideq*0], xm0
movhps [dstq+strideq*1], xm0
movq [r6 +strideq*0], xm1
movhps [r6 +strideq*1], xm1
RET
ALIGN function_align
.main:
mova xm0, [cq+16*0]
mova xm1, [cq+16*1]
mova xm2, [cq+16*2]
mova xm3, [cq+16*3]
%if WIN64
movaps [rsp+16], xmm6
movaps [rsp+32], xmm7
%endif
.main2:
WRAP_XMM IADST4_1D
ret
INV_TXFM_4X4_FN flipadst, dct
INV_TXFM_4X4_FN flipadst, adst
INV_TXFM_4X4_FN flipadst, flipadst
INV_TXFM_4X4_FN flipadst, identity
cglobal iflipadst_4x4_internal_10bpc, 0, 7, 6, dst, stride, c, eob, tx2
call m(iadst_4x4_internal_10bpc).main
vinserti128 m0, m3, xm2, 1
vinserti128 m1, m6, xm4, 1
jmp m(iadst_4x4_internal_10bpc).pass1_end
.pass2:
lea r6, [deint_shuf+128]
vextracti128 xm1, m0, 1
call m(iadst_4x4_internal_8bpc).main
vpbroadcastd xm4, [pw_2048]
movq xm3, [dstq+strideq*1]
movhps xm3, [dstq+strideq*0]
lea r6, [dstq+strideq*2]
movq xm2, [r6 +strideq*1]
movhps xm2, [r6 +strideq*0]
vpbroadcastd xm5, [pixel_10bpc_max]
pmulhrsw xm0, xm4
pmulhrsw xm1, xm4
pxor m4, m4
mova [cq+32*0], m4
mova [cq+32*1], m4
paddw xm0, xm2
paddw xm1, xm3
pmaxsw xm0, xm4
pmaxsw xm1, xm4
pminsw xm0, xm5
pminsw xm1, xm5
movhps [dstq+strideq*0], xm1
movq [dstq+strideq*1], xm1
movhps [r6 +strideq*0], xm0
movq [r6 +strideq*1], xm0
RET
INV_TXFM_4X4_FN identity, dct
INV_TXFM_4X4_FN identity, adst
INV_TXFM_4X4_FN identity, flipadst
INV_TXFM_4X4_FN identity, identity
cglobal iidentity_4x4_internal_10bpc, 0, 7, 6, dst, stride, c, eob, tx2
vpbroadcastd m1, [pd_5793]
pmulld m0, m1, [cq+32*0]
pmulld m1, [cq+32*1]
vpbroadcastd m5, [pd_2048]
mova m3, [itx4_shuf]
paddd m0, m5
paddd m1, m5
psrad m0, 12
psrad m1, 12
packssdw m0, m1
vpermd m0, m3, m0
psrld m3, 4
pshufb m0, m3
jmp tx2q
.pass2:
vpbroadcastd m1, [pw_1697x8]
movq xm2, [dstq+strideq*0]
movhps xm2, [dstq+strideq*1]
lea r6, [dstq+strideq*2]
pmulhrsw m1, m0
paddsw m0, m1
movq xm3, [r6 +strideq*0]
movhps xm3, [r6 +strideq*1]
vpbroadcastd xm4, [pixel_10bpc_max]
packssdw m5, m5 ; pw_2048
pmulhrsw m0, m5
pxor m5, m5
mova [cq+32*0], m5
mova [cq+32*1], m5
vextracti128 xm1, m0, 1
paddw xm0, xm2
paddw xm1, xm3
pmaxsw xm0, xm5
pmaxsw xm1, xm5
pminsw xm0, xm4
pminsw xm1, xm4
movq [dstq+strideq*0], xm0
movhps [dstq+strideq*1], xm0
movq [r6 +strideq*0], xm1
movhps [r6 +strideq*1], xm1
RET
INV_TXFM_4X4_FN dct, dct, 12
INV_TXFM_4X4_FN dct, identity, 12
INV_TXFM_4X4_FN dct, adst, 12
INV_TXFM_4X4_FN dct, flipadst, 12
cglobal idct_4x4_internal_12bpc, 0, 7, 8, dst, stride, c, eob, tx2
call m(idct_4x4_internal_10bpc).main
mova m3, [idct4_12_shuf]
mova m4, [idct4_12_shuf2]
vpermd m2, m4, m1
vpermd m1, m3, m0
jmp m(iadst_4x4_internal_12bpc).pass1_end2
.pass2:
vpbroadcastd m5, [pd_2048]
vpermq m0, m0, q3120
vpermq m1, m1, q3120
call m(idct_4x4_internal_10bpc).main2
vpermq m0, m0, q3120
vpermq m1, m1, q2031
jmp m(iadst_4x4_internal_12bpc).end
INV_TXFM_4X4_FN adst, dct, 12
INV_TXFM_4X4_FN adst, adst, 12
INV_TXFM_4X4_FN adst, flipadst, 12
INV_TXFM_4X4_FN adst, identity, 12
cglobal iadst_4x4_internal_12bpc, 0, 7, 8, dst, stride, c, eob, tx2
call m(iadst_4x4_internal_10bpc).main
vinserti128 m1, m4, xm6, 1
vinserti128 m2, xm3, 1
.pass1_end:
mova m3, [itx4_shuf]
vpbroadcastd m5, [pd_1024]
psrad m1, 1
psrad m2, 1
vpermd m1, m3, m1
vpermd m2, m3, m2
paddd m1, m5
paddd m2, m5
psrad m1, 11
psrad m2, 11
.pass1_end2:
vpbroadcastd m3, [clip_18b_min]
vpbroadcastd m4, [clip_18b_max]
punpcklqdq m0, m1, m2
punpckhqdq m1, m2
pmaxsd m0, m3
pmaxsd m1, m3
pminsd m0, m4
pminsd m1, m4
jmp tx2q
.pass2:
call .main_pass2
vinserti128 m0, m4, xm6, 1
vinserti128 m1, m2, xm3, 1
.pass2_end:
vpbroadcastd m5, [pd_2048]
paddd m0, m5
paddd m1, m5
psrad m0, 12
psrad m1, 12
.end:
%if WIN64
WIN64_RESTORE_XMM_INTERNAL
%assign xmm_regs_used 6
%endif
.end2:
vpbroadcastd m4, [pw_16384]
movq xm2, [dstq+strideq*0]
movq xm3, [dstq+strideq*1]
lea r6, [dstq+strideq*2]
movhps xm2, [r6 +strideq*0] ; dst0 dst2
movhps xm3, [r6 +strideq*1] ; dst1 dst3
vpbroadcastd m5, [pixel_12bpc_max]
vinserti128 m2, xm3, 1
psrad m0, 3
psrad m1, 3
packssdw m0, m1 ; t0 t2 t1 t3
pmulhrsw m0, m4
pxor m4, m4
mova [cq+32*0], m4
mova [cq+32*1], m4
paddw m0, m2 ; out0 out2 out1 out3
pmaxsw m0, m4
pminsw m0, m5
vextracti128 xm1, m0, 1 ; out1 out3
movq [dstq+strideq*0], xm0
movq [dstq+strideq*1], xm1
movhps [r6 +strideq*0], xm0
movhps [r6 +strideq*1], xm1
RET
.main_pass2:
vextracti128 xm3, m1, 1
mova xm2, xm1
vextracti128 xm1, m0, 1
jmp m(iadst_4x4_internal_10bpc).main2
INV_TXFM_4X4_FN flipadst, dct, 12
INV_TXFM_4X4_FN flipadst, adst, 12
INV_TXFM_4X4_FN flipadst, flipadst, 12
INV_TXFM_4X4_FN flipadst, identity, 12
cglobal iflipadst_4x4_internal_12bpc, 0, 7, 8, dst, stride, c, eob, tx2
call m(iadst_4x4_internal_10bpc).main
vinserti128 m1, m3, xm2, 1
vinserti128 m2, m6, xm4, 1
jmp m(iadst_4x4_internal_12bpc).pass1_end
.pass2:
call m(iadst_4x4_internal_12bpc).main_pass2
vinserti128 m0, m3, xm2, 1
vinserti128 m1, m6, xm4, 1
jmp m(iadst_4x4_internal_12bpc).pass2_end
INV_TXFM_4X4_FN identity, dct, 12
INV_TXFM_4X4_FN identity, adst, 12
INV_TXFM_4X4_FN identity, flipadst, 12
INV_TXFM_4X4_FN identity, identity, 12
cglobal iidentity_4x4_internal_12bpc, 0, 7, 8, dst, stride, c, eob, tx2
mova m2, [itx4_shuf]
vpbroadcastd m3, [pd_1697]
vpermd m0, m2, [cq+32*0]
vpermd m2, m2, [cq+32*1]
vpbroadcastd m5, [pd_2048]
pmulld m1, m3, m0
pmulld m3, m2
paddd m1, m5
paddd m3, m5
psrad m1, 12
psrad m3, 12
paddd m1, m0
paddd m2, m3
jmp m(iadst_4x4_internal_12bpc).pass1_end2
.pass2:
; m0 = in0 in1
; m1 = in2 in3
vpbroadcastd m3, [pd_5793]
vpbroadcastd m5, [pd_2048]
pmulld m0, m3
pmulld m1, m3
paddd m0, m5 ; 2048
paddd m1, m5
psrad m0, 12
psrad m1, 12
jmp m(iadst_4x4_internal_12bpc).end
%macro INV_TXFM_4X8_FN 2-3 10 ; type1, type2, bitdepth
INV_TXFM_FN %1, %2, 0, 4x8, %3
%ifidn %1_%2, dct_dct
vpbroadcastd xm2, [dconly_%3bpc]
%if %3 = 10
.dconly:
imul r6d, [cq], 181
mov [cq], eobd ; 0
or r3d, 8
add r6d, 128
sar r6d, 8
imul r6d, 181
jmp m(inv_txfm_add_dct_dct_4x4_10bpc).dconly2
%else
jmp m(inv_txfm_add_dct_dct_4x8_10bpc).dconly
%endif
%endif
%endmacro
%macro IDCT4_1D 8 ; src[1-4], tmp[1-3], rnd
ITX_MULSUB_2D %2, %4, %5, %6, %7, %8, 1567, 3784 ; t2, t3
vpbroadcastd m%5, [pd_2896]
pmulld m%1, m%5
pmulld m%3, m%5
paddd m%1, m%8
paddd m%5, m%1, m%3
psubd m%1, m%3
psrad m%5, 12 ; t0
psrad m%1, 12 ; t1
psubd m%3, m%1, m%2
paddd m%2, m%1
paddd m%1, m%5, m%4
psubd m%4, m%5, m%4
%endmacro
INV_TXFM_4X8_FN dct, dct
INV_TXFM_4X8_FN dct, identity
INV_TXFM_4X8_FN dct, adst
INV_TXFM_4X8_FN dct, flipadst
cglobal idct_4x8_internal_10bpc, 0, 7, 8, dst, stride, c, eob, tx2
.pass1:
vpbroadcastd m3, [pd_2896]
pmulld m0, m3, [cq+32*0]
pmulld m1, m3, [cq+32*1]
pmulld m2, m3, [cq+32*2]
pmulld m3, m3, [cq+32*3]
vpbroadcastd m7, [pd_2048]
REPX {paddd x, m7}, m0, m1, m2, m3
REPX {psrad x, 12}, m0, m1, m2, m3
IDCT4_1D 0, 1, 2, 3, 4, 5, 6, 7
jmp tx2q
.pass2:
packssdw m0, m2
packssdw m1, m3
lea r6, [deint_shuf+128]
punpckhwd m2, m0, m1
punpcklwd m0, m1
punpckhdq m1, m0, m2 ; 2 3
punpckldq m0, m2 ; 0 1
vextracti128 xm2, m0, 1 ; 4 5
vextracti128 xm3, m1, 1 ; 6 7
call m(idct_4x8_internal_8bpc).main
vpbroadcastd xm4, [pw_2048]
REPX {pmulhrsw x, xm4}, xm0, xm1, xm2, xm3
lea r3, [strideq*3]
lea r6, [dstq+strideq*4]
movq xm4, [dstq+strideq*0]
movhps xm4, [dstq+strideq*1]
movq xm5, [dstq+r3 ]
movhps xm5, [dstq+strideq*2]
movq xm6, [r6 +strideq*0]
movhps xm6, [r6 +strideq*1]
movq xm7, [r6 +r3 ]
movhps xm7, [r6 +strideq*2]
paddw xm0, xm4 ; 0 1
paddw xm1, xm5 ; 3 2
paddw xm2, xm6 ; 4 5
paddw xm3, xm7 ; 7 6
vpbroadcastd xm5, [pixel_10bpc_max]
pxor m4, m4
REPX {mova [cq+32*x], m4}, 0, 1, 2, 3
REPX {pmaxsw x, xm4}, xm0, xm1, xm2, xm3
REPX {pminsw x, xm5}, xm0, xm1, xm2, xm3
movq [dstq+strideq*0], xm0
movhps [dstq+strideq*1], xm0
movhps [dstq+strideq*2], xm1
movq [dstq+r3 ], xm1
movq [r6 +strideq*0], xm2
movhps [r6 +strideq*1], xm2
movhps [r6 +strideq*2], xm3
movq [r6 +r3 ], xm3
RET
INV_TXFM_4X8_FN adst, dct
INV_TXFM_4X8_FN adst, adst
INV_TXFM_4X8_FN adst, flipadst
INV_TXFM_4X8_FN adst, identity
cglobal iadst_4x8_internal_10bpc, 0, 7, 8, dst, stride, c, eob, tx2
call m(iadst_8x4_internal_10bpc).main
vpbroadcastd m5, [pd_2048]
paddd m0, m5, m4
paddd m1, m5, m6
paddd m2, m5
paddd m3, m5
.pass1_end:
REPX {psrad x, 12}, m0, m1, m2, m3
jmp tx2q
.pass2:
call .pass2_main
mova xm4, [pw_2048_m2048]
REPX {pmulhrsw x, xm4}, xm0, xm1, xm2, xm3
.end:
lea r3, [strideq*3]
lea r6, [dstq+strideq*4]
movq xm4, [dstq+strideq*0]
movhps xm4, [dstq+strideq*1]
movq xm5, [dstq+strideq*2]
movhps xm5, [dstq+r3 ]
movq xm6, [r6 +strideq*0]
movhps xm6, [r6 +strideq*1]
movq xm7, [r6 +strideq*2]
movhps xm7, [r6 +r3 ]
paddw xm0, xm4 ; 0 1
paddw xm1, xm5 ; 2 3
paddw xm2, xm6 ; 4 5
paddw xm3, xm7 ; 6 7
vpbroadcastd xm5, [pixel_10bpc_max]
pxor m4, m4
REPX {mova [cq+32*x], m4}, 0, 1, 2, 3
REPX {pmaxsw x, xm4}, xm0, xm1, xm2, xm3
REPX {pminsw x, xm5}, xm0, xm1, xm2, xm3
movq [dstq+strideq*0], xm0
movhps [dstq+strideq*1], xm0
movq [dstq+strideq*2], xm1
movhps [dstq+r3 ], xm1
movq [r6 +strideq*0], xm2
movhps [r6 +strideq*1], xm2
movq [r6 +strideq*2], xm3
movhps [r6 +r3 ], xm3
RET
ALIGN function_align
.pass2_main:
packssdw m0, m2
packssdw m1, m3
lea r6, [deint_shuf+128]
punpcklwd m4, m0, m1
punpckhwd m0, m1
punpckhdq m5, m4, m0
punpckldq m4, m0
vextracti128 xm2, m4, 1 ; 4 5
vextracti128 xm3, m5, 1 ; 6 7
pshufd xm4, xm4, q1032 ; 1 0
pshufd xm5, xm5, q1032 ; 3 2
jmp m(iadst_4x8_internal_8bpc).main_pass2
ALIGN function_align
.main:
vpbroadcastd m8, [clip_18b_min]
vpbroadcastd m9, [clip_18b_max]
.main2:
vbroadcasti128 m0, [cq+16*0]
vbroadcasti128 m2, [cq+16*2]
vbroadcasti128 m3, [cq+16*5]
vbroadcasti128 m1, [cq+16*7]
vpbroadcastd m6, [pd_2896]
shufpd m0, m2, 0x0c ; 0 2
shufpd m1, m3, 0x0c ; 7 5
vbroadcasti128 m2, [cq+16*4]
vbroadcasti128 m4, [cq+16*6]
vbroadcasti128 m5, [cq+16*1]
vbroadcasti128 m3, [cq+16*3]
vpbroadcastd m7, [pd_2048]
shufpd m2, m4, 0x0c ; 4 6
shufpd m3, m5, 0x0c ; 3 1
REPX {pmulld x, m6}, m0, m1, m2, m3
REPX {paddd x, m7}, m0, m1, m2, m3
REPX {psrad x, 12}, m0, m1, m2, m3
.main3:
ITX_MULSUB_2D 1, 0, 4, 5, 6, 7, 401_1931, 4076_3612, 1
ITX_MULSUB_2D 3, 2, 4, 5, 6, 7, 3166_3920, 2598_1189, 1
psubd m4, m0, m2 ; t4 t6
paddd m0, m2 ; t0 t2
psubd m2, m1, m3 ; t5 t7
paddd m1, m3 ; t1 t3
REPX {pmaxsd x, m8}, m4, m2, m0, m1
REPX {pminsd x, m9}, m4, m2, m0, m1
pxor m5, m5
psubd m5, m4
vpblendd m4, m2, 0xcc ; t4 t7
vpblendd m2, m5, 0xcc ; t5 -t6
ITX_MULSUB_2D 4, 2, 3, 5, 6, 7, 1567, 3784
vpbroadcastd m5, [pd_2896]
vbroadcasti128 m6, [pw_2048_m2048] ; + + - -
punpckhqdq m3, m0, m1
punpcklqdq m0, m1
psubd m1, m0, m3 ; t2 t3
paddd m0, m3 ; out0 -out7
punpckhqdq m3, m4, m2 ; t7a t6a
punpcklqdq m4, m2 ; t5a t4a
psubd m2, m4, m3 ; t7 t6
paddd m4, m3 ; out6 -out1
REPX {pmaxsd x, m8}, m1, m2
REPX {pminsd x, m9}, m1, m2
vpblendd m3, m1, m2, 0xcc
shufpd m1, m2, 0x05
pmulld m3, m5
pmulld m5, m1
psignd m0, m6 ; out0 out7
psignd m4, m6 ; out6 out1
paddd m3, m7
psubd m2, m3, m5
paddd m5, m3
psrad m2, 12 ; out4 -out5
psrad m5, 12 ; -out3 out2
ret
INV_TXFM_4X8_FN flipadst, dct
INV_TXFM_4X8_FN flipadst, adst
INV_TXFM_4X8_FN flipadst, flipadst
INV_TXFM_4X8_FN flipadst, identity
cglobal iflipadst_4x8_internal_10bpc, 0, 7, 8, dst, stride, c, eob, tx2
call m(iadst_8x4_internal_10bpc).main
vpbroadcastd m5, [pd_2048]
paddd m0, m5, m3
paddd m1, m5, m2
paddd m2, m5, m6
paddd m3, m5, m4
jmp m(iadst_4x8_internal_10bpc).pass1_end
.pass2:
call m(iadst_4x8_internal_10bpc).pass2_main
mova xm4, [pw_2048_m2048]
REPX {pmulhrsw x, xm4}, xm3, xm2, xm1, xm0
lea r3, [strideq*3]
lea r6, [dstq+strideq*4]
movq xm4, [dstq+strideq*1]
movhps xm4, [dstq+strideq*0]
movq xm5, [dstq+r3 ]
movhps xm5, [dstq+strideq*2]
movq xm6, [r6 +strideq*1]
movhps xm6, [r6 +strideq*0]
movq xm7, [r6 +r3 ]
movhps xm7, [r6 +strideq*2]
paddw xm3, xm4 ; 1 0
paddw xm2, xm5 ; 3 2
paddw xm1, xm6 ; 5 4
paddw xm0, xm7 ; 7 6
vpbroadcastd xm5, [pixel_10bpc_max]
pxor m4, m4
REPX {mova [cq+32*x], m4}, 0, 1, 2, 3
REPX {pmaxsw x, xm4}, xm3, xm2, xm1, xm0
REPX {pminsw x, xm5}, xm3, xm2, xm1, xm0
movhps [dstq+strideq*0], xm3
movq [dstq+strideq*1], xm3
movhps [dstq+strideq*2], xm2
movq [dstq+r3 ], xm2
movhps [r6 +strideq*0], xm1
movq [r6 +strideq*1], xm1
movhps [r6 +strideq*2], xm0
movq [r6 +r3 ], xm0
RET
INV_TXFM_4X8_FN identity, dct
INV_TXFM_4X8_FN identity, adst
INV_TXFM_4X8_FN identity, flipadst
INV_TXFM_4X8_FN identity, identity
cglobal iidentity_4x8_internal_10bpc, 0, 7, 8, dst, stride, c, eob, tx2
.pass1:
vpbroadcastd m3, [pd_2896]
pmulld m0, m3, [cq+32*0]
pmulld m1, m3, [cq+32*1]
pmulld m2, m3, [cq+32*2]
pmulld m3, [cq+32*3]
vpbroadcastd m5, [pd_2048]
vpbroadcastd m4, [pd_5793]
REPX {paddd x, m5}, m0, m1, m2, m3
REPX {psrad x, 12}, m0, m1, m2, m3
REPX {pmulld x, m4}, m0, m1, m2, m3