aboutsummaryrefslogtreecommitdiffstats
path: root/libavcodec/x86/vp8dsp.asm
blob: ea6dcf6179e66ec8d01fde211a9d3cc65a441350 (plain) (blame)
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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
;******************************************************************************
;* VP8 MMXEXT optimizations
;* Copyright (c) 2010 Ronald S. Bultje <rsbultje@gmail.com>
;* Copyright (c) 2010 Fiona Glaser <fiona@x264.com>
;*
;* This file is part of FFmpeg.
;*
;* FFmpeg is free software; you can redistribute it and/or
;* modify it under the terms of the GNU Lesser General Public
;* License as published by the Free Software Foundation; either
;* version 2.1 of the License, or (at your option) any later version.
;*
;* FFmpeg 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
;* Lesser General Public License for more details.
;*
;* You should have received a copy of the GNU Lesser General Public
;* License along with FFmpeg; if not, write to the Free Software
;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
;******************************************************************************

%include "libavutil/x86/x86util.asm"

SECTION_RODATA

fourtap_filter_hw_m: times 4 dw  -6, 123
                     times 4 dw  12,  -1
                     times 4 dw  -9,  93
                     times 4 dw  50,  -6
                     times 4 dw  -6,  50
                     times 4 dw  93,  -9
                     times 4 dw  -1,  12
                     times 4 dw 123,  -6

sixtap_filter_hw_m:  times 4 dw   2, -11
                     times 4 dw 108,  36
                     times 4 dw  -8,   1
                     times 4 dw   3, -16
                     times 4 dw  77,  77
                     times 4 dw -16,   3
                     times 4 dw   1,  -8
                     times 4 dw  36, 108
                     times 4 dw -11,   2

fourtap_filter_hb_m: times 8 db  -6, 123
                     times 8 db  12,  -1
                     times 8 db  -9,  93
                     times 8 db  50,  -6
                     times 8 db  -6,  50
                     times 8 db  93,  -9
                     times 8 db  -1,  12
                     times 8 db 123,  -6

sixtap_filter_hb_m:  times 8 db   2,   1
                     times 8 db -11, 108
                     times 8 db  36,  -8
                     times 8 db   3,   3
                     times 8 db -16,  77
                     times 8 db  77, -16
                     times 8 db   1,   2
                     times 8 db  -8,  36
                     times 8 db 108, -11

fourtap_filter_v_m:  times 8 dw  -6
                     times 8 dw 123
                     times 8 dw  12
                     times 8 dw  -1
                     times 8 dw  -9
                     times 8 dw  93
                     times 8 dw  50
                     times 8 dw  -6
                     times 8 dw  -6
                     times 8 dw  50
                     times 8 dw  93
                     times 8 dw  -9
                     times 8 dw  -1
                     times 8 dw  12
                     times 8 dw 123
                     times 8 dw  -6

sixtap_filter_v_m:   times 8 dw   2
                     times 8 dw -11
                     times 8 dw 108
                     times 8 dw  36
                     times 8 dw  -8
                     times 8 dw   1
                     times 8 dw   3
                     times 8 dw -16
                     times 8 dw  77
                     times 8 dw  77
                     times 8 dw -16
                     times 8 dw   3
                     times 8 dw   1
                     times 8 dw  -8
                     times 8 dw  36
                     times 8 dw 108
                     times 8 dw -11
                     times 8 dw   2

bilinear_filter_vw_m: times 8 dw 1
                      times 8 dw 2
                      times 8 dw 3
                      times 8 dw 4
                      times 8 dw 5
                      times 8 dw 6
                      times 8 dw 7

bilinear_filter_vb_m: times 8 db 7, 1
                      times 8 db 6, 2
                      times 8 db 5, 3
                      times 8 db 4, 4
                      times 8 db 3, 5
                      times 8 db 2, 6
                      times 8 db 1, 7

%ifdef PIC
%define fourtap_filter_hw  picregq
%define sixtap_filter_hw   picregq
%define fourtap_filter_hb  picregq
%define sixtap_filter_hb   picregq
%define fourtap_filter_v   picregq
%define sixtap_filter_v    picregq
%define bilinear_filter_vw picregq
%define bilinear_filter_vb picregq
%define npicregs 1
%else
%define fourtap_filter_hw  fourtap_filter_hw_m
%define sixtap_filter_hw   sixtap_filter_hw_m
%define fourtap_filter_hb  fourtap_filter_hb_m
%define sixtap_filter_hb   sixtap_filter_hb_m
%define fourtap_filter_v   fourtap_filter_v_m
%define sixtap_filter_v    sixtap_filter_v_m
%define bilinear_filter_vw bilinear_filter_vw_m
%define bilinear_filter_vb bilinear_filter_vb_m
%define npicregs 0
%endif

filter_h2_shuf:  db 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5,  6, 6,  7,  7,  8
filter_h4_shuf:  db 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7,  8, 8,  9,  9, 10

filter_h6_shuf1: db 0, 5, 1, 6, 2, 7, 3, 8, 4, 9, 5, 10, 6, 11,  7, 12
filter_h6_shuf2: db 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,  7, 7,  8,  8,  9
filter_h6_shuf3: db 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8,  9, 9, 10, 10, 11

pw_20091: times 4 dw 20091
pw_17734: times 4 dw 17734

cextern pw_3
cextern pw_4
cextern pw_64
cextern pw_256

SECTION .text

;-------------------------------------------------------------------------------
; subpel MC functions:
;
; void ff_put_vp8_epel<size>_h<htap>v<vtap>_<opt>(uint8_t *dst, int deststride,
;                                                 uint8_t *src, int srcstride,
;                                                 int height,   int mx, int my);
;-------------------------------------------------------------------------------

%macro FILTER_SSSE3 1
cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src, srcstride, height, mx, picreg
    lea      mxd, [mxq*3]
    mova      m3, [filter_h6_shuf2]
    mova      m4, [filter_h6_shuf3]
%ifdef PIC
    lea  picregq, [sixtap_filter_hb_m]
%endif
    mova      m5, [sixtap_filter_hb+mxq*8-48] ; set up 6tap filter in bytes
    mova      m6, [sixtap_filter_hb+mxq*8-32]
    mova      m7, [sixtap_filter_hb+mxq*8-16]

.nextrow:
    movu      m0, [srcq-2]
    mova      m1, m0
    mova      m2, m0
%if mmsize == 8
; For epel4, we need 9 bytes, but only 8 get loaded; to compensate, do the
; shuffle with a memory operand
    punpcklbw m0, [srcq+3]
%else
    pshufb    m0, [filter_h6_shuf1]
%endif
    pshufb    m1, m3
    pshufb    m2, m4
    pmaddubsw m0, m5
    pmaddubsw m1, m6
    pmaddubsw m2, m7
    paddsw    m0, m1
    paddsw    m0, m2
    pmulhrsw  m0, [pw_256]
    packuswb  m0, m0
    movh  [dstq], m0        ; store

    ; go to next line
    add     dstq, dststrideq
    add     srcq, srcstrideq
    dec  heightd            ; next row
    jg .nextrow
    REP_RET

cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg
    shl      mxd, 4
    mova      m2, [pw_256]
    mova      m3, [filter_h2_shuf]
    mova      m4, [filter_h4_shuf]
%ifdef PIC
    lea  picregq, [fourtap_filter_hb_m]
%endif
    mova      m5, [fourtap_filter_hb+mxq-16] ; set up 4tap filter in bytes
    mova      m6, [fourtap_filter_hb+mxq]

.nextrow:
    movu      m0, [srcq-1]
    mova      m1, m0
    pshufb    m0, m3
    pshufb    m1, m4
    pmaddubsw m0, m5
    pmaddubsw m1, m6
    paddsw    m0, m1
    pmulhrsw  m0, m2
    packuswb  m0, m0
    movh  [dstq], m0        ; store

    ; go to next line
    add     dstq, dststrideq
    add     srcq, srcstrideq
    dec  heightd            ; next row
    jg .nextrow
    REP_RET

cglobal put_vp8_epel%1_v4, 7, 7, 8, dst, dststride, src, srcstride, height, picreg, my
    shl      myd, 4
%ifdef PIC
    lea  picregq, [fourtap_filter_hb_m]
%endif
    mova      m5, [fourtap_filter_hb+myq-16]
    mova      m6, [fourtap_filter_hb+myq]
    mova      m7, [pw_256]

    ; read 3 lines
    sub     srcq, srcstrideq
    movh      m0, [srcq]
    movh      m1, [srcq+  srcstrideq]
    movh      m2, [srcq+2*srcstrideq]
    add     srcq, srcstrideq

.nextrow:
    movh      m3, [srcq+2*srcstrideq]      ; read new row
    mova      m4, m0
    mova      m0, m1
    punpcklbw m4, m1
    mova      m1, m2
    punpcklbw m2, m3
    pmaddubsw m4, m5
    pmaddubsw m2, m6
    paddsw    m4, m2
    mova      m2, m3
    pmulhrsw  m4, m7
    packuswb  m4, m4
    movh  [dstq], m4

    ; go to next line
    add      dstq, dststrideq
    add      srcq, srcstrideq
    dec   heightd                          ; next row
    jg .nextrow
    REP_RET

cglobal put_vp8_epel%1_v6, 7, 7, 8, dst, dststride, src, srcstride, height, picreg, my
    lea      myd, [myq*3]
%ifdef PIC
    lea  picregq, [sixtap_filter_hb_m]
%endif
    lea      myq, [sixtap_filter_hb+myq*8]

    ; read 5 lines
    sub     srcq, srcstrideq
    sub     srcq, srcstrideq
    movh      m0, [srcq]
    movh      m1, [srcq+srcstrideq]
    movh      m2, [srcq+srcstrideq*2]
    lea     srcq, [srcq+srcstrideq*2]
    add     srcq, srcstrideq
    movh      m3, [srcq]
    movh      m4, [srcq+srcstrideq]

.nextrow:
    movh      m5, [srcq+2*srcstrideq]      ; read new row
    mova      m6, m0
    punpcklbw m6, m5
    mova      m0, m1
    punpcklbw m1, m2
    mova      m7, m3
    punpcklbw m7, m4
    pmaddubsw m6, [myq-48]
    pmaddubsw m1, [myq-32]
    pmaddubsw m7, [myq-16]
    paddsw    m6, m1
    paddsw    m6, m7
    mova      m1, m2
    mova      m2, m3
    pmulhrsw  m6, [pw_256]
    mova      m3, m4
    packuswb  m6, m6
    mova      m4, m5
    movh  [dstq], m6

    ; go to next line
    add      dstq, dststrideq
    add      srcq, srcstrideq
    dec   heightd                          ; next row
    jg .nextrow
    REP_RET
%endmacro

INIT_MMX ssse3
FILTER_SSSE3 4
INIT_XMM ssse3
FILTER_SSSE3 8

; 4x4 block, H-only 4-tap filter
INIT_MMX mmxext
cglobal put_vp8_epel4_h4, 6, 6 + npicregs, 0, dst, dststride, src, srcstride, height, mx, picreg
    shl       mxd, 4
%ifdef PIC
    lea   picregq, [fourtap_filter_hw_m]
%endif
    movq      mm4, [fourtap_filter_hw+mxq-16] ; set up 4tap filter in words
    movq      mm5, [fourtap_filter_hw+mxq]
    movq      mm7, [pw_64]
    pxor      mm6, mm6

.nextrow:
    movq      mm1, [srcq-1]                ; (ABCDEFGH) load 8 horizontal pixels

    ; first set of 2 pixels
    movq      mm2, mm1                     ; byte ABCD..
    punpcklbw mm1, mm6                     ; byte->word ABCD
    pshufw    mm0, mm2, 9                  ; byte CDEF..
    punpcklbw mm0, mm6                     ; byte->word CDEF
    pshufw    mm3, mm1, 0x94               ; word ABBC
    pshufw    mm1, mm0, 0x94               ; word CDDE
    pmaddwd   mm3, mm4                     ; multiply 2px with F0/F1
    movq      mm0, mm1                     ; backup for second set of pixels
    pmaddwd   mm1, mm5                     ; multiply 2px with F2/F3
    paddd     mm3, mm1                     ; finish 1st 2px

    ; second set of 2 pixels, use backup of above
    punpckhbw mm2, mm6                     ; byte->word EFGH
    pmaddwd   mm0, mm4                     ; multiply backed up 2px with F0/F1
    pshufw    mm1, mm2, 0x94               ; word EFFG
    pmaddwd   mm1, mm5                     ; multiply 2px with F2/F3
    paddd     mm0, mm1                     ; finish 2nd 2px

    ; merge two sets of 2 pixels into one set of 4, round/clip/store
    packssdw  mm3, mm0                     ; merge dword->word (4px)
    paddsw    mm3, mm7                     ; rounding
    psraw     mm3, 7
    packuswb  mm3, mm6                     ; clip and word->bytes
    movd   [dstq], mm3                     ; store

    ; go to next line
    add      dstq, dststrideq
    add      srcq, srcstrideq
    dec   heightd                          ; next row
    jg .nextrow
    REP_RET

; 4x4 block, H-only 6-tap filter
INIT_MMX mmxext
cglobal put_vp8_epel4_h6, 6, 6 + npicregs, 0, dst, dststride, src, srcstride, height, mx, picreg
    lea       mxd, [mxq*3]
%ifdef PIC
    lea   picregq, [sixtap_filter_hw_m]
%endif
    movq      mm4, [sixtap_filter_hw+mxq*8-48] ; set up 4tap filter in words
    movq      mm5, [sixtap_filter_hw+mxq*8-32]
    movq      mm6, [sixtap_filter_hw+mxq*8-16]
    movq      mm7, [pw_64]
    pxor      mm3, mm3

.nextrow:
    movq      mm1, [srcq-2]                ; (ABCDEFGH) load 8 horizontal pixels

    ; first set of 2 pixels
    movq      mm2, mm1                     ; byte ABCD..
    punpcklbw mm1, mm3                     ; byte->word ABCD
    pshufw    mm0, mm2, 0x9                ; byte CDEF..
    punpckhbw mm2, mm3                     ; byte->word EFGH
    punpcklbw mm0, mm3                     ; byte->word CDEF
    pshufw    mm1, mm1, 0x94               ; word ABBC
    pshufw    mm2, mm2, 0x94               ; word EFFG
    pmaddwd   mm1, mm4                     ; multiply 2px with F0/F1
    pshufw    mm3, mm0, 0x94               ; word CDDE
    movq      mm0, mm3                     ; backup for second set of pixels
    pmaddwd   mm3, mm5                     ; multiply 2px with F2/F3
    paddd     mm1, mm3                     ; add to 1st 2px cache
    movq      mm3, mm2                     ; backup for second set of pixels
    pmaddwd   mm2, mm6                     ; multiply 2px with F4/F5
    paddd     mm1, mm2                     ; finish 1st 2px

    ; second set of 2 pixels, use backup of above
    movd      mm2, [srcq+3]                ; byte FGHI (prevent overreads)
    pmaddwd   mm0, mm4                     ; multiply 1st backed up 2px with F0/F1
    pmaddwd   mm3, mm5                     ; multiply 2nd backed up 2px with F2/F3
    paddd     mm0, mm3                     ; add to 2nd 2px cache
    pxor      mm3, mm3
    punpcklbw mm2, mm3                     ; byte->word FGHI
    pshufw    mm2, mm2, 0xE9               ; word GHHI
    pmaddwd   mm2, mm6                     ; multiply 2px with F4/F5
    paddd     mm0, mm2                     ; finish 2nd 2px

    ; merge two sets of 2 pixels into one set of 4, round/clip/store
    packssdw  mm1, mm0                     ; merge dword->word (4px)
    paddsw    mm1, mm7                     ; rounding
    psraw     mm1, 7
    packuswb  mm1, mm3                     ; clip and word->bytes
    movd   [dstq], mm1                     ; store

    ; go to next line
    add      dstq, dststrideq
    add      srcq, srcstrideq
    dec   heightd                          ; next row
    jg .nextrow
    REP_RET

INIT_XMM sse2
cglobal put_vp8_epel8_h4, 6, 6 + npicregs, 10, dst, dststride, src, srcstride, height, mx, picreg
    shl      mxd, 5
%ifdef PIC
    lea  picregq, [fourtap_filter_v_m]
%endif
    lea      mxq, [fourtap_filter_v+mxq-32]
    pxor      m7, m7
    mova      m4, [pw_64]
    mova      m5, [mxq+ 0]
    mova      m6, [mxq+16]
%ifdef m8
    mova      m8, [mxq+32]
    mova      m9, [mxq+48]
%endif
.nextrow:
    movq      m0, [srcq-1]
    movq      m1, [srcq-0]
    movq      m2, [srcq+1]
    movq      m3, [srcq+2]
    punpcklbw m0, m7
    punpcklbw m1, m7
    punpcklbw m2, m7
    punpcklbw m3, m7
    pmullw    m0, m5
    pmullw    m1, m6
%ifdef m8
    pmullw    m2, m8
    pmullw    m3, m9
%else
    pmullw    m2, [mxq+32]
    pmullw    m3, [mxq+48]
%endif
    paddsw    m0, m1
    paddsw    m2, m3
    paddsw    m0, m2
    paddsw    m0, m4
    psraw     m0, 7
    packuswb  m0, m7
    movh  [dstq], m0        ; store

    ; go to next line
    add     dstq, dststrideq
    add     srcq, srcstrideq
    dec  heightd            ; next row
    jg .nextrow
    REP_RET

INIT_XMM sse2
cglobal put_vp8_epel8_h6, 6, 6 + npicregs, 14, dst, dststride, src, srcstride, height, mx, picreg
    lea      mxd, [mxq*3]
    shl      mxd, 4
%ifdef PIC
    lea  picregq, [sixtap_filter_v_m]
%endif
    lea      mxq, [sixtap_filter_v+mxq-96]
    pxor      m7, m7
    mova      m6, [pw_64]
%ifdef m8
    mova      m8, [mxq+ 0]
    mova      m9, [mxq+16]
    mova     m10, [mxq+32]
    mova     m11, [mxq+48]
    mova     m12, [mxq+64]
    mova     m13, [mxq+80]
%endif
.nextrow:
    movq      m0, [srcq-2]
    movq      m1, [srcq-1]
    movq      m2, [srcq-0]
    movq      m3, [srcq+1]
    movq      m4, [srcq+2]
    movq      m5, [srcq+3]
    punpcklbw m0, m7
    punpcklbw m1, m7
    punpcklbw m2, m7
    punpcklbw m3, m7
    punpcklbw m4, m7
    punpcklbw m5, m7
%ifdef m8
    pmullw    m0, m8
    pmullw    m1, m9
    pmullw    m2, m10
    pmullw    m3, m11
    pmullw    m4, m12
    pmullw    m5, m13
%else
    pmullw    m0, [mxq+ 0]
    pmullw    m1, [mxq+16]
    pmullw    m2, [mxq+32]
    pmullw    m3, [mxq+48]
    pmullw    m4, [mxq+64]
    pmullw    m5, [mxq+80]
%endif
    paddsw    m1, m4
    paddsw    m0, m5
    paddsw    m1, m2
    paddsw    m0, m3
    paddsw    m0, m1
    paddsw    m0, m6
    psraw     m0, 7
    packuswb  m0, m7
    movh  [dstq], m0        ; store

    ; go to next line
    add     dstq, dststrideq
    add     srcq, srcstrideq
    dec  heightd            ; next row
    jg .nextrow
    REP_RET

%macro FILTER_V 1
; 4x4 block, V-only 4-tap filter
cglobal put_vp8_epel%1_v4, 7, 7, 8, dst, dststride, src, srcstride, height, picreg, my
    shl      myd, 5
%ifdef PIC
    lea  picregq, [fourtap_filter_v_m]
%endif
    lea      myq, [fourtap_filter_v+myq-32]
    mova      m6, [pw_64]
    pxor      m7, m7
    mova      m5, [myq+48]

    ; read 3 lines
    sub     srcq, srcstrideq
    movh      m0, [srcq]
    movh      m1, [srcq+  srcstrideq]
    movh      m2, [srcq+2*srcstrideq]
    add     srcq, srcstrideq
    punpcklbw m0, m7
    punpcklbw m1, m7
    punpcklbw m2, m7

.nextrow:
    ; first calculate negative taps (to prevent losing positive overflows)
    movh      m4, [srcq+2*srcstrideq]      ; read new row
    punpcklbw m4, m7
    mova      m3, m4
    pmullw    m0, [myq+0]
    pmullw    m4, m5
    paddsw    m4, m0

    ; then calculate positive taps
    mova      m0, m1
    pmullw    m1, [myq+16]
    paddsw    m4, m1
    mova      m1, m2
    pmullw    m2, [myq+32]
    paddsw    m4, m2
    mova      m2, m3

    ; round/clip/store
    paddsw    m4, m6
    psraw     m4, 7
    packuswb  m4, m7
    movh  [dstq], m4

    ; go to next line
    add     dstq, dststrideq
    add     srcq, srcstrideq
    dec  heightd                           ; next row
    jg .nextrow
    REP_RET


; 4x4 block, V-only 6-tap filter
cglobal put_vp8_epel%1_v6, 7, 7, 8, dst, dststride, src, srcstride, height, picreg, my
    shl      myd, 4
    lea      myq, [myq*3]
%ifdef PIC
    lea  picregq, [sixtap_filter_v_m]
%endif
    lea      myq, [sixtap_filter_v+myq-96]
    pxor      m7, m7

    ; read 5 lines
    sub     srcq, srcstrideq
    sub     srcq, srcstrideq
    movh      m0, [srcq]
    movh      m1, [srcq+srcstrideq]
    movh      m2, [srcq+srcstrideq*2]
    lea     srcq, [srcq+srcstrideq*2]
    add     srcq, srcstrideq
    movh      m3, [srcq]
    movh      m4, [srcq+srcstrideq]
    punpcklbw m0, m7
    punpcklbw m1, m7
    punpcklbw m2, m7
    punpcklbw m3, m7
    punpcklbw m4, m7

.nextrow:
    ; first calculate negative taps (to prevent losing positive overflows)
    mova      m5, m1
    pmullw    m5, [myq+16]
    mova      m6, m4
    pmullw    m6, [myq+64]
    paddsw    m6, m5

    ; then calculate positive taps
    movh      m5, [srcq+2*srcstrideq]      ; read new row
    punpcklbw m5, m7
    pmullw    m0, [myq+0]
    paddsw    m6, m0
    mova      m0, m1
    mova      m1, m2
    pmullw    m2, [myq+32]
    paddsw    m6, m2
    mova      m2, m3
    pmullw    m3, [myq+48]
    paddsw    m6, m3
    mova      m3, m4
    mova      m4, m5
    pmullw    m5, [myq+80]
    paddsw    m6, m5

    ; round/clip/store
    paddsw    m6, [pw_64]
    psraw     m6, 7
    packuswb  m6, m7
    movh  [dstq], m6

    ; go to next line
    add     dstq, dststrideq
    add     srcq, srcstrideq
    dec  heightd                           ; next row
    jg .nextrow
    REP_RET
%endmacro

INIT_MMX mmxext
FILTER_V 4
INIT_XMM sse2
FILTER_V 8

%macro FILTER_BILINEAR 1
cglobal put_vp8_bilinear%1_v, 7, 7, 7, dst, dststride, src, srcstride, height, picreg, my
    shl      myd, 4
%ifdef PIC
    lea  picregq, [bilinear_filter_vw_m]
%endif
    pxor      m6, m6
    mova      m5, [bilinear_filter_vw+myq-1*16]
    neg      myq
    mova      m4, [bilinear_filter_vw+myq+7*16]
.nextrow:
    movh      m0, [srcq+srcstrideq*0]
    movh      m1, [srcq+srcstrideq*1]
    movh      m3, [srcq+srcstrideq*2]
    punpcklbw m0, m6
    punpcklbw m1, m6
    punpcklbw m3, m6
    mova      m2, m1
    pmullw    m0, m4
    pmullw    m1, m5
    pmullw    m2, m4
    pmullw    m3, m5
    paddsw    m0, m1
    paddsw    m2, m3
    psraw     m0, 2
    psraw     m2, 2
    pavgw     m0, m6
    pavgw     m2, m6
%if mmsize == 8
    packuswb  m0, m0
    packuswb  m2, m2
    movh   [dstq+dststrideq*0], m0
    movh   [dstq+dststrideq*1], m2
%else
    packuswb  m0, m2
    movh   [dstq+dststrideq*0], m0
    movhps [dstq+dststrideq*1], m0
%endif

    lea     dstq, [dstq+dststrideq*2]
    lea     srcq, [srcq+srcstrideq*2]
    sub  heightd, 2
    jg .nextrow
    REP_RET

cglobal put_vp8_bilinear%1_h, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg
    shl      mxd, 4
%ifdef PIC
    lea  picregq, [bilinear_filter_vw_m]
%endif
    pxor      m6, m6
    mova      m5, [bilinear_filter_vw+mxq-1*16]
    neg      mxq
    mova      m4, [bilinear_filter_vw+mxq+7*16]
.nextrow:
    movh      m0, [srcq+srcstrideq*0+0]
    movh      m1, [srcq+srcstrideq*0+1]
    movh      m2, [srcq+srcstrideq*1+0]
    movh      m3, [srcq+srcstrideq*1+1]
    punpcklbw m0, m6
    punpcklbw m1, m6
    punpcklbw m2, m6
    punpcklbw m3, m6
    pmullw    m0, m4
    pmullw    m1, m5
    pmullw    m2, m4
    pmullw    m3, m5
    paddsw    m0, m1
    paddsw    m2, m3
    psraw     m0, 2
    psraw     m2, 2
    pavgw     m0, m6
    pavgw     m2, m6
%if mmsize == 8
    packuswb  m0, m0
    packuswb  m2, m2
    movh   [dstq+dststrideq*0], m0
    movh   [dstq+dststrideq*1], m2
%else
    packuswb  m0, m2
    movh   [dstq+dststrideq*0], m0
    movhps [dstq+dststrideq*1], m0
%endif

    lea     dstq, [dstq+dststrideq*2]
    lea     srcq, [srcq+srcstrideq*2]
    sub  heightd, 2
    jg .nextrow
    REP_RET
%endmacro

INIT_MMX mmxext
FILTER_BILINEAR 4
INIT_XMM sse2
FILTER_BILINEAR 8

%macro FILTER_BILINEAR_SSSE3 1
cglobal put_vp8_bilinear%1_v, 7, 7, 5, dst, dststride, src, srcstride, height, picreg, my
    shl      myd, 4
%ifdef PIC
    lea  picregq, [bilinear_filter_vb_m]
%endif
    pxor      m4, m4
    mova      m3, [bilinear_filter_vb+myq-16]
.nextrow:
    movh      m0, [srcq+srcstrideq*0]
    movh      m1, [srcq+srcstrideq*1]
    movh      m2, [srcq+srcstrideq*2]
    punpcklbw m0, m1
    punpcklbw m1, m2
    pmaddubsw m0, m3
    pmaddubsw m1, m3
    psraw     m0, 2
    psraw     m1, 2
    pavgw     m0, m4
    pavgw     m1, m4
%if mmsize==8
    packuswb  m0, m0
    packuswb  m1, m1
    movh   [dstq+dststrideq*0], m0
    movh   [dstq+dststrideq*1], m1
%else
    packuswb  m0, m1
    movh   [dstq+dststrideq*0], m0
    movhps [dstq+dststrideq*1], m0
%endif

    lea     dstq, [dstq+dststrideq*2]
    lea     srcq, [srcq+srcstrideq*2]
    sub  heightd, 2
    jg .nextrow
    REP_RET

cglobal put_vp8_bilinear%1_h, 6, 6 + npicregs, 5, dst, dststride, src, srcstride, height, mx, picreg
    shl      mxd, 4
%ifdef PIC
    lea  picregq, [bilinear_filter_vb_m]
%endif
    pxor      m4, m4
    mova      m2, [filter_h2_shuf]
    mova      m3, [bilinear_filter_vb+mxq-16]
.nextrow:
    movu      m0, [srcq+srcstrideq*0]
    movu      m1, [srcq+srcstrideq*1]
    pshufb    m0, m2
    pshufb    m1, m2
    pmaddubsw m0, m3
    pmaddubsw m1, m3
    psraw     m0, 2
    psraw     m1, 2
    pavgw     m0, m4
    pavgw     m1, m4
%if mmsize==8
    packuswb  m0, m0
    packuswb  m1, m1
    movh   [dstq+dststrideq*0], m0
    movh   [dstq+dststrideq*1], m1
%else
    packuswb  m0, m1
    movh   [dstq+dststrideq*0], m0
    movhps [dstq+dststrideq*1], m0
%endif

    lea     dstq, [dstq+dststrideq*2]
    lea     srcq, [srcq+srcstrideq*2]
    sub  heightd, 2
    jg .nextrow
    REP_RET
%endmacro

INIT_MMX ssse3
FILTER_BILINEAR_SSSE3 4
INIT_XMM ssse3
FILTER_BILINEAR_SSSE3 8

INIT_MMX mmx
cglobal put_vp8_pixels8, 5, 5, 0, dst, dststride, src, srcstride, height
.nextrow:
    movq    mm0, [srcq+srcstrideq*0]
    movq    mm1, [srcq+srcstrideq*1]
    lea    srcq, [srcq+srcstrideq*2]
    movq [dstq+dststrideq*0], mm0
    movq [dstq+dststrideq*1], mm1
    lea    dstq, [dstq+dststrideq*2]
    sub heightd, 2
    jg .nextrow
    REP_RET

%if ARCH_X86_32
INIT_MMX mmx
cglobal put_vp8_pixels16, 5, 5, 0, dst, dststride, src, srcstride, height
.nextrow:
    movq    mm0, [srcq+srcstrideq*0+0]
    movq    mm1, [srcq+srcstrideq*0+8]
    movq    mm2, [srcq+srcstrideq*1+0]
    movq    mm3, [srcq+srcstrideq*1+8]
    lea    srcq, [srcq+srcstrideq*2]
    movq [dstq+dststrideq*0+0], mm0
    movq [dstq+dststrideq*0+8], mm1
    movq [dstq+dststrideq*1+0], mm2
    movq [dstq+dststrideq*1+8], mm3
    lea    dstq, [dstq+dststrideq*2]
    sub heightd, 2
    jg .nextrow
    REP_RET
%endif

INIT_XMM sse
cglobal put_vp8_pixels16, 5, 5, 2, dst, dststride, src, srcstride, height
.nextrow:
    movups xmm0, [srcq+srcstrideq*0]
    movups xmm1, [srcq+srcstrideq*1]
    lea    srcq, [srcq+srcstrideq*2]
    movaps [dstq+dststrideq*0], xmm0
    movaps [dstq+dststrideq*1], xmm1
    lea    dstq, [dstq+dststrideq*2]
    sub heightd, 2
    jg .nextrow
    REP_RET

;-----------------------------------------------------------------------------
; void ff_vp8_idct_dc_add_<opt>(uint8_t *dst, int16_t block[16], int stride);
;-----------------------------------------------------------------------------

%macro ADD_DC 4
    %4        m2, [dst1q+%3]
    %4        m3, [dst1q+strideq+%3]
    %4        m4, [dst2q+%3]
    %4        m5, [dst2q+strideq+%3]
    paddusb   m2, %1
    paddusb   m3, %1
    paddusb   m4, %1
    paddusb   m5, %1
    psubusb   m2, %2
    psubusb   m3, %2
    psubusb   m4, %2
    psubusb   m5, %2
    %4 [dst1q+%3], m2
    %4 [dst1q+strideq+%3], m3
    %4 [dst2q+%3], m4
    %4 [dst2q+strideq+%3], m5
%endmacro

%if ARCH_X86_32
INIT_MMX mmx
cglobal vp8_idct_dc_add, 3, 3, 0, dst, block, stride
    ; load data
    movd       m0, [blockq]

    ; calculate DC
    paddw      m0, [pw_4]
    pxor       m1, m1
    psraw      m0, 3
    movd [blockq], m1
    psubw      m1, m0
    packuswb   m0, m0
    packuswb   m1, m1
    punpcklbw  m0, m0
    punpcklbw  m1, m1
    punpcklwd  m0, m0
    punpcklwd  m1, m1

    ; add DC
    DEFINE_ARGS dst1, dst2, stride
    lea     dst2q, [dst1q+strideq*2]
    ADD_DC     m0, m1, 0, movh
    RET
%endif

%macro VP8_IDCT_DC_ADD 0
cglobal vp8_idct_dc_add, 3, 3, 6, dst, block, stride
    ; load data
    movd       m0, [blockq]
    pxor       m1, m1

    ; calculate DC
    paddw      m0, [pw_4]
    movd [blockq], m1
    DEFINE_ARGS dst1, dst2, stride
    lea     dst2q, [dst1q+strideq*2]
    movd       m2, [dst1q]
    movd       m3, [dst1q+strideq]
    movd       m4, [dst2q]
    movd       m5, [dst2q+strideq]
    psraw      m0, 3
    pshuflw    m0, m0, 0
    punpcklqdq m0, m0
    punpckldq  m2, m3
    punpckldq  m4, m5
    punpcklbw  m2, m1
    punpcklbw  m4, m1
    paddw      m2, m0
    paddw      m4, m0
    packuswb   m2, m4
    movd   [dst1q], m2
%if cpuflag(sse4)
    pextrd [dst1q+strideq], m2, 1
    pextrd [dst2q], m2, 2
    pextrd [dst2q+strideq], m2, 3
%else
    psrldq     m2, 4
    movd [dst1q+strideq], m2
    psrldq     m2, 4
    movd [dst2q], m2
    psrldq     m2, 4
    movd [dst2q+strideq], m2
%endif
    RET
%endmacro

INIT_XMM sse2
VP8_IDCT_DC_ADD
INIT_XMM sse4
VP8_IDCT_DC_ADD

;-----------------------------------------------------------------------------
; void ff_vp8_idct_dc_add4y_<opt>(uint8_t *dst, int16_t block[4][16], int stride);
;-----------------------------------------------------------------------------

%if ARCH_X86_32
INIT_MMX mmx
cglobal vp8_idct_dc_add4y, 3, 3, 0, dst, block, stride
    ; load data
    movd      m0, [blockq+32*0] ; A
    movd      m1, [blockq+32*2] ; C
    punpcklwd m0, [blockq+32*1] ; A B
    punpcklwd m1, [blockq+32*3] ; C D
    punpckldq m0, m1        ; A B C D
    pxor      m6, m6

    ; calculate DC
    paddw     m0, [pw_4]
    movd [blockq+32*0], m6
    movd [blockq+32*1], m6
    movd [blockq+32*2], m6
    movd [blockq+32*3], m6
    psraw     m0, 3
    psubw     m6, m0
    packuswb  m0, m0
    packuswb  m6, m6
    punpcklbw m0, m0 ; AABBCCDD
    punpcklbw m6, m6 ; AABBCCDD
    movq      m1, m0
    movq      m7, m6
    punpcklbw m0, m0 ; AAAABBBB
    punpckhbw m1, m1 ; CCCCDDDD
    punpcklbw m6, m6 ; AAAABBBB
    punpckhbw m7, m7 ; CCCCDDDD

    ; add DC
    DEFINE_ARGS dst1, dst2, stride
    lea    dst2q, [dst1q+strideq*2]
    ADD_DC    m0, m6, 0, mova
    ADD_DC    m1, m7, 8, mova
    RET
%endif

INIT_XMM sse2
cglobal vp8_idct_dc_add4y, 3, 3, 6, dst, block, stride
    ; load data
    movd      m0, [blockq+32*0] ; A
    movd      m1, [blockq+32*2] ; C
    punpcklwd m0, [blockq+32*1] ; A B
    punpcklwd m1, [blockq+32*3] ; C D
    punpckldq m0, m1        ; A B C D
    pxor      m1, m1

    ; calculate DC
    paddw     m0, [pw_4]
    movd [blockq+32*0], m1
    movd [blockq+32*1], m1
    movd [blockq+32*2], m1
    movd [blockq+32*3], m1
    psraw     m0, 3
    psubw     m1, m0
    packuswb  m0, m0
    packuswb  m1, m1
    punpcklbw m0, m0
    punpcklbw m1, m1
    punpcklbw m0, m0
    punpcklbw m1, m1

    ; add DC
    DEFINE_ARGS dst1, dst2, stride
    lea    dst2q, [dst1q+strideq*2]
    ADD_DC    m0, m1, 0, mova
    RET

;-----------------------------------------------------------------------------
; void ff_vp8_idct_dc_add4uv_<opt>(uint8_t *dst, int16_t block[4][16], int stride);
;-----------------------------------------------------------------------------

INIT_MMX mmx
cglobal vp8_idct_dc_add4uv, 3, 3, 0, dst, block, stride
    ; load data
    movd      m0, [blockq+32*0] ; A
    movd      m1, [blockq+32*2] ; C
    punpcklwd m0, [blockq+32*1] ; A B
    punpcklwd m1, [blockq+32*3] ; C D
    punpckldq m0, m1        ; A B C D
    pxor      m6, m6

    ; calculate DC
    paddw     m0, [pw_4]
    movd [blockq+32*0], m6
    movd [blockq+32*1], m6
    movd [blockq+32*2], m6
    movd [blockq+32*3], m6
    psraw     m0, 3
    psubw     m6, m0
    packuswb  m0, m0
    packuswb  m6, m6
    punpcklbw m0, m0 ; AABBCCDD
    punpcklbw m6, m6 ; AABBCCDD
    movq      m1, m0
    movq      m7, m6
    punpcklbw m0, m0 ; AAAABBBB
    punpckhbw m1, m1 ; CCCCDDDD
    punpcklbw m6, m6 ; AAAABBBB
    punpckhbw m7, m7 ; CCCCDDDD

    ; add DC
    DEFINE_ARGS dst1, dst2, stride
    lea    dst2q, [dst1q+strideq*2]
    ADD_DC    m0, m6, 0, mova
    lea    dst1q, [dst1q+strideq*4]
    lea    dst2q, [dst2q+strideq*4]
    ADD_DC    m1, m7, 0, mova
    RET

;-----------------------------------------------------------------------------
; void ff_vp8_idct_add_<opt>(uint8_t *dst, int16_t block[16], int stride);
;-----------------------------------------------------------------------------

; calculate %1=mul_35468(%1)-mul_20091(%2); %2=mul_20091(%1)+mul_35468(%2)
;           this macro assumes that m6/m7 have words for 20091/17734 loaded
%macro VP8_MULTIPLY_SUMSUB 4
    mova      %3, %1
    mova      %4, %2
    pmulhw    %3, m6 ;20091(1)
    pmulhw    %4, m6 ;20091(2)
    paddw     %3, %1
    paddw     %4, %2
    paddw     %1, %1
    paddw     %2, %2
    pmulhw    %1, m7 ;35468(1)
    pmulhw    %2, m7 ;35468(2)
    psubw     %1, %4
    paddw     %2, %3
%endmacro

; calculate x0=%1+%3; x1=%1-%3
;           x2=mul_35468(%2)-mul_20091(%4); x3=mul_20091(%2)+mul_35468(%4)
;           %1=x0+x3 (tmp0); %2=x1+x2 (tmp1); %3=x1-x2 (tmp2); %4=x0-x3 (tmp3)
;           %5/%6 are temporary registers
;           we assume m6/m7 have constant words 20091/17734 loaded in them
%macro VP8_IDCT_TRANSFORM4x4_1D 6
    SUMSUB_BA         w, %3,  %1,  %5     ;t0, t1
    VP8_MULTIPLY_SUMSUB m%2, m%4, m%5,m%6 ;t2, t3
    SUMSUB_BA         w, %4,  %3,  %5     ;tmp0, tmp3
    SUMSUB_BA         w, %2,  %1,  %5     ;tmp1, tmp2
    SWAP                 %4,  %1
    SWAP                 %4,  %3
%endmacro

%macro VP8_IDCT_ADD 0
cglobal vp8_idct_add, 3, 3, 0, dst, block, stride
    ; load block data
    movq         m0, [blockq+ 0]
    movq         m1, [blockq+ 8]
    movq         m2, [blockq+16]
    movq         m3, [blockq+24]
    movq         m6, [pw_20091]
    movq         m7, [pw_17734]
%if cpuflag(sse)
    xorps      xmm0, xmm0
    movaps [blockq+ 0], xmm0
    movaps [blockq+16], xmm0
%else
    pxor         m4, m4
    movq [blockq+ 0], m4
    movq [blockq+ 8], m4
    movq [blockq+16], m4
    movq [blockq+24], m4
%endif

    ; actual IDCT
    VP8_IDCT_TRANSFORM4x4_1D 0, 1, 2, 3, 4, 5
    TRANSPOSE4x4W            0, 1, 2, 3, 4
    paddw        m0, [pw_4]
    VP8_IDCT_TRANSFORM4x4_1D 0, 1, 2, 3, 4, 5
    TRANSPOSE4x4W            0, 1, 2, 3, 4

    ; store
    pxor         m4, m4
    DEFINE_ARGS dst1, dst2, stride
    lea       dst2q, [dst1q+2*strideq]
    STORE_DIFFx2 m0, m1, m6, m7, m4, 3, dst1q, strideq
    STORE_DIFFx2 m2, m3, m6, m7, m4, 3, dst2q, strideq

    RET
%endmacro

%if ARCH_X86_32
INIT_MMX mmx
VP8_IDCT_ADD
%endif
INIT_MMX sse
VP8_IDCT_ADD

;-----------------------------------------------------------------------------
; void ff_vp8_luma_dc_wht(int16_t block[4][4][16], int16_t dc[16])
;-----------------------------------------------------------------------------

%macro SCATTER_WHT 3
    movd dc1d, m%1
    movd dc2d, m%2
    mov [blockq+2*16*(0+%3)], dc1w
    mov [blockq+2*16*(1+%3)], dc2w
    shr  dc1d, 16
    shr  dc2d, 16
    psrlq m%1, 32
    psrlq m%2, 32
    mov [blockq+2*16*(4+%3)], dc1w
    mov [blockq+2*16*(5+%3)], dc2w
    movd dc1d, m%1
    movd dc2d, m%2
    mov [blockq+2*16*(8+%3)], dc1w
    mov [blockq+2*16*(9+%3)], dc2w
    shr  dc1d, 16
    shr  dc2d, 16
    mov [blockq+2*16*(12+%3)], dc1w
    mov [blockq+2*16*(13+%3)], dc2w
%endmacro

%macro HADAMARD4_1D 4
    SUMSUB_BADC w, %2, %1, %4, %3
    SUMSUB_BADC w, %4, %2, %3, %1
    SWAP %1, %4, %3
%endmacro

%macro VP8_DC_WHT 0
cglobal vp8_luma_dc_wht, 2, 3, 0, block, dc1, dc2
    movq          m0, [dc1q]
    movq          m1, [dc1q+8]
    movq          m2, [dc1q+16]
    movq          m3, [dc1q+24]
%if cpuflag(sse)
    xorps      xmm0, xmm0
    movaps [dc1q+ 0], xmm0
    movaps [dc1q+16], xmm0
%else
    pxor         m4, m4
    movq  [dc1q+ 0], m4
    movq  [dc1q+ 8], m4
    movq  [dc1q+16], m4
    movq  [dc1q+24], m4
%endif
    HADAMARD4_1D  0, 1, 2, 3
    TRANSPOSE4x4W 0, 1, 2, 3, 4
    paddw         m0, [pw_3]
    HADAMARD4_1D  0, 1, 2, 3
    psraw         m0, 3
    psraw         m1, 3
    psraw         m2, 3
    psraw         m3, 3
    SCATTER_WHT   0, 1, 0
    SCATTER_WHT   2, 3, 2
    RET
%endmacro

%if ARCH_X86_32
INIT_MMX mmx
VP8_DC_WHT
%endif
INIT_MMX sse
VP8_DC_WHT