summaryrefslogtreecommitdiffstats
path: root/third_party/dav1d/src/x86/mc16_avx2.asm
blob: 42e2a5525ecff1420ff28dd7c353ec441420c644 (plain)
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
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
; Copyright © 2021, VideoLAN and dav1d authors
; Copyright © 2021, Two Orioles, LLC
; 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 64

; dav1d_obmc_masks[] * -512
const obmc_masks_avx2
            dw      0,      0,  -9728,      0, -12800,  -7168,  -2560,      0
            dw -14336, -11264,  -8192,  -5632,  -3584,  -1536,      0,      0
            dw -15360, -13824, -12288, -10752,  -9216,  -7680,  -6144,  -5120
            dw  -4096,  -3072,  -2048,  -1536,      0,      0,      0,      0
            dw -15872, -14848, -14336, -13312, -12288, -11776, -10752, -10240
            dw  -9728,  -8704,  -8192,  -7168,  -6656,  -6144,  -5632,  -4608
            dw  -4096,  -3584,  -3072,  -2560,  -2048,  -2048,  -1536,  -1024
            dw      0,      0,      0,      0,      0,      0,      0,      0

deint_shuf:     dd 0,  4,  1,  5,  2,  6,  3,  7
subpel_h_shufA: db 0,  1,  2,  3,  2,  3,  4,  5,  4,  5,  6,  7,  6,  7,  8,  9
subpel_h_shufB: db 4,  5,  6,  7,  6,  7,  8,  9,  8,  9, 10, 11, 10, 11, 12, 13
subpel_h_shuf2: db 0,  1,  2,  3,  4,  5,  6,  7,  2,  3,  4,  5,  6,  7,  8,  9
subpel_s_shuf2: db 0,  1,  2,  3,  4,  5,  6,  7,  0,  1,  2,  3,  4,  5,  6,  7
subpel_s_shuf8: db 0,  1,  8,  9,  2,  3, 10, 11,  4,  5, 12, 13,  6,  7, 14, 15
rescale_mul:    dd 0,  1,  2,  3,  4,  5,  6,  7
rescale_mul2:   dd 0,  1,  4,  5,  2,  3,  6,  7
resize_shuf:    db 0,  1,  0,  1,  0,  1,  0,  1,  0,  1,  2,  3,  4,  5,  6,  7
                db 8,  9, 10, 11, 12, 13, 14, 15, 14, 15, 14, 15, 14, 15, 14, 15
blend_shuf:     db 0,  1,  0,  1,  0,  1,  0,  1,  2,  3,  2,  3,  2,  3,  2,  3
wswap:          db 2,  3,  0,  1,  6,  7,  4,  5, 10, 11,  8,  9, 14, 15, 12, 13
bdct_lb_q: times 8 db 0
           times 8 db 4
           times 8 db 8
           times 8 db 12

prep_mul:         dw 16, 16, 4, 4
put_bilin_h_rnd:  dw 8, 8, 10, 10
put_8tap_h_rnd:   dd 34, 40
s_8tap_h_rnd:     dd 2, 8
s_8tap_h_sh:      dd 2, 4
put_s_8tap_v_rnd: dd 512, 128
put_s_8tap_v_sh:  dd 10, 8
prep_8tap_1d_rnd: dd     8 - (8192 <<  4)
prep_8tap_2d_rnd: dd    32 - (8192 <<  5)
warp8x8t_rnd:     dd 16384 - (8192 << 15)
warp8x8_shift:    dd  5,  3
warp8x8_rnd:      dw   4096,   4096,  16384,  16384
bidir_rnd:        dw -16400, -16400, -16388, -16388
bidir_mul:        dw   2048,   2048,   8192,   8192

%define pw_16 prep_mul
%define pd_512 put_s_8tap_v_rnd

pw_2:          times 2 dw 2
pw_64:         times 2 dw 64
pw_2048:       times 2 dw 2048
pw_8192:       times 2 dw 8192
pw_27615:      times 2 dw 27615
pw_32766:      times 2 dw 32766
pw_m512:       times 2 dw -512
pd_32:         dd 32
pd_63:         dd 63
pd_64:         dd 64
pd_32768:      dd 32768
pd_65538:      dd 65538
pd_m524256:    dd -524256 ; -8192 << 6 + 32
pd_0x3ff:      dd 0x3ff
pq_0x40000000: dq 0x40000000
               dd 0

%macro BIDIR_JMP_TABLE 2-*
    %xdefine %1_%2_table (%%table - 2*%3)
    %xdefine %%base %1_%2_table
    %xdefine %%prefix mangle(private_prefix %+ _%1_16bpc_%2)
    %%table:
    %rep %0 - 2
        dd %%prefix %+ .w%3 - %%base
        %rotate 1
    %endrep
%endmacro

BIDIR_JMP_TABLE avg,        avx2,    4, 8, 16, 32, 64, 128
BIDIR_JMP_TABLE w_avg,      avx2,    4, 8, 16, 32, 64, 128
BIDIR_JMP_TABLE mask,       avx2,    4, 8, 16, 32, 64, 128
BIDIR_JMP_TABLE w_mask_420, avx2,    4, 8, 16, 32, 64, 128
BIDIR_JMP_TABLE w_mask_422, avx2,    4, 8, 16, 32, 64, 128
BIDIR_JMP_TABLE w_mask_444, avx2,    4, 8, 16, 32, 64, 128
BIDIR_JMP_TABLE blend,      avx2,    4, 8, 16, 32
BIDIR_JMP_TABLE blend_v,    avx2, 2, 4, 8, 16, 32
BIDIR_JMP_TABLE blend_h,    avx2, 2, 4, 8, 16, 32, 64, 128

%macro BASE_JMP_TABLE 3-*
    %xdefine %1_%2_table (%%table - %3)
    %xdefine %%base %1_%2
    %%table:
    %rep %0 - 2
        dw %%base %+ _w%3 - %%base
        %rotate 1
    %endrep
%endmacro

%xdefine put_avx2 mangle(private_prefix %+ _put_bilin_16bpc_avx2.put)
%xdefine prep_avx2 mangle(private_prefix %+ _prep_bilin_16bpc_avx2.prep)

BASE_JMP_TABLE put,  avx2, 2, 4, 8, 16, 32, 64, 128
BASE_JMP_TABLE prep, avx2,    4, 8, 16, 32, 64, 128

%macro HV_JMP_TABLE 5-*
    %xdefine %%prefix mangle(private_prefix %+ _%1_%2_16bpc_%3)
    %xdefine %%base %1_%3
    %assign %%types %4
    %if %%types & 1
        %xdefine %1_%2_h_%3_table  (%%h  - %5)
        %%h:
        %rep %0 - 4
            dw %%prefix %+ .h_w%5 - %%base
            %rotate 1
        %endrep
        %rotate 4
    %endif
    %if %%types & 2
        %xdefine %1_%2_v_%3_table  (%%v  - %5)
        %%v:
        %rep %0 - 4
            dw %%prefix %+ .v_w%5 - %%base
            %rotate 1
        %endrep
        %rotate 4
    %endif
    %if %%types & 4
        %xdefine %1_%2_hv_%3_table (%%hv - %5)
        %%hv:
        %rep %0 - 4
            dw %%prefix %+ .hv_w%5 - %%base
            %rotate 1
        %endrep
    %endif
%endmacro

HV_JMP_TABLE put,  bilin, avx2, 7, 2, 4, 8, 16, 32, 64, 128
HV_JMP_TABLE prep, bilin, avx2, 7,    4, 8, 16, 32, 64, 128

%macro SCALED_JMP_TABLE 2-*
    %xdefine %1_%2_table (%%table - %3)
    %xdefine %%base mangle(private_prefix %+ _%1_16bpc_%2)
%%table:
    %rep %0 - 2
        dw %%base %+ .w%3 - %%base
        %rotate 1
    %endrep
    %rotate 2
 %%dy_1024:
    %xdefine %1_%2_dy1_table (%%dy_1024 - %3)
    %rep %0 - 2
        dw %%base %+ .dy1_w%3 - %%base
        %rotate 1
    %endrep
    %rotate 2
 %%dy_2048:
    %xdefine %1_%2_dy2_table (%%dy_2048 - %3)
    %rep %0 - 2
        dw %%base %+ .dy2_w%3 - %%base
        %rotate 1
    %endrep
%endmacro

SCALED_JMP_TABLE put_8tap_scaled, avx2, 2, 4, 8, 16, 32, 64, 128
SCALED_JMP_TABLE prep_8tap_scaled, avx2,   4, 8, 16, 32, 64, 128

%define table_offset(type, fn) type %+ fn %+ SUFFIX %+ _table - type %+ SUFFIX

cextern mc_subpel_filters
%define subpel_filters (mangle(private_prefix %+ _mc_subpel_filters)-8)

cextern mc_warp_filter
cextern resize_filter

SECTION .text

INIT_XMM avx2
cglobal put_bilin_16bpc, 4, 8, 0, dst, ds, src, ss, w, h, mxy
    mov                mxyd, r6m ; mx
    lea                  r7, [put_avx2]
%if UNIX64
    DECLARE_REG_TMP 8
    %define org_w r8d
    mov                 r8d, wd
%else
    DECLARE_REG_TMP 7
    %define org_w wm
%endif
    tzcnt                wd, wm
    movifnidn            hd, hm
    test               mxyd, mxyd
    jnz .h
    mov                mxyd, r7m ; my
    test               mxyd, mxyd
    jnz .v
.put:
    movzx                wd, word [r7+wq*2+table_offset(put,)]
    add                  wq, r7
    jmp                  wq
.put_w2:
    mov                 r6d, [srcq+ssq*0]
    mov                 r7d, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    mov        [dstq+dsq*0], r6d
    mov        [dstq+dsq*1], r7d
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .put_w2
    RET
.put_w4:
    mov                  r6, [srcq+ssq*0]
    mov                  r7, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    mov        [dstq+dsq*0], r6
    mov        [dstq+dsq*1], r7
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .put_w4
    RET
.put_w8:
    movu                 m0, [srcq+ssq*0]
    movu                 m1, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    mova       [dstq+dsq*0], m0
    mova       [dstq+dsq*1], m1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .put_w8
    RET
INIT_YMM avx2
.put_w16:
    movu                 m0, [srcq+ssq*0]
    movu                 m1, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    mova       [dstq+dsq*0], m0
    mova       [dstq+dsq*1], m1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .put_w16
    RET
.put_w32:
    movu                 m0, [srcq+ssq*0+32*0]
    movu                 m1, [srcq+ssq*0+32*1]
    movu                 m2, [srcq+ssq*1+32*0]
    movu                 m3, [srcq+ssq*1+32*1]
    lea                srcq, [srcq+ssq*2]
    mova  [dstq+dsq*0+32*0], m0
    mova  [dstq+dsq*0+32*1], m1
    mova  [dstq+dsq*1+32*0], m2
    mova  [dstq+dsq*1+32*1], m3
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .put_w32
    RET
.put_w64:
    movu                 m0, [srcq+32*0]
    movu                 m1, [srcq+32*1]
    movu                 m2, [srcq+32*2]
    movu                 m3, [srcq+32*3]
    add                srcq, ssq
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    mova        [dstq+32*2], m2
    mova        [dstq+32*3], m3
    add                dstq, dsq
    dec                  hd
    jg .put_w64
    RET
.put_w128:
    movu                 m0, [srcq+32*0]
    movu                 m1, [srcq+32*1]
    movu                 m2, [srcq+32*2]
    movu                 m3, [srcq+32*3]
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    mova        [dstq+32*2], m2
    mova        [dstq+32*3], m3
    movu                 m0, [srcq+32*4]
    movu                 m1, [srcq+32*5]
    movu                 m2, [srcq+32*6]
    movu                 m3, [srcq+32*7]
    add                srcq, ssq
    mova        [dstq+32*4], m0
    mova        [dstq+32*5], m1
    mova        [dstq+32*6], m2
    mova        [dstq+32*7], m3
    add                dstq, dsq
    dec                  hd
    jg .put_w128
    RET
.h:
    movd                xm5, mxyd
    mov                mxyd, r7m ; my
    vpbroadcastd         m4, [pw_16]
    vpbroadcastw         m5, xm5
    psubw                m4, m5
    test               mxyd, mxyd
    jnz .hv
    ; 12-bit is rounded twice so we can't use the same pmulhrsw approach as .v
    movzx                wd, word [r7+wq*2+table_offset(put, _bilin_h)]
    mov                 r6d, r8m ; bitdepth_max
    add                  wq, r7
    shr                 r6d, 11
    vpbroadcastd         m3, [r7-put_avx2+put_bilin_h_rnd+r6*4]
    jmp                  wq
.h_w2:
    movq                xm1, [srcq+ssq*0]
    movhps              xm1, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    pmullw              xm0, xm4, xm1
    psrlq               xm1, 16
    pmullw              xm1, xm5
    paddw               xm0, xm3
    paddw               xm0, xm1
    psrlw               xm0, 4
    movd       [dstq+dsq*0], xm0
    pextrd     [dstq+dsq*1], xm0, 2
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .h_w2
    RET
.h_w4:
    movq                xm0, [srcq+ssq*0]
    movhps              xm0, [srcq+ssq*1]
    movq                xm1, [srcq+ssq*0+2]
    movhps              xm1, [srcq+ssq*1+2]
    lea                srcq, [srcq+ssq*2]
    pmullw              xm0, xm4
    pmullw              xm1, xm5
    paddw               xm0, xm3
    paddw               xm0, xm1
    psrlw               xm0, 4
    movq       [dstq+dsq*0], xm0
    movhps     [dstq+dsq*1], xm0
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .h_w4
    RET
.h_w8:
    movu                xm0, [srcq+ssq*0]
    vinserti128          m0, [srcq+ssq*1], 1
    movu                xm1, [srcq+ssq*0+2]
    vinserti128          m1, [srcq+ssq*1+2], 1
    lea                srcq, [srcq+ssq*2]
    pmullw               m0, m4
    pmullw               m1, m5
    paddw                m0, m3
    paddw                m0, m1
    psrlw                m0, 4
    mova         [dstq+dsq*0], xm0
    vextracti128 [dstq+dsq*1], m0, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .h_w8
    RET
.h_w16:
    pmullw               m0, m4, [srcq+ssq*0]
    pmullw               m1, m5, [srcq+ssq*0+2]
    paddw                m0, m3
    paddw                m0, m1
    pmullw               m1, m4, [srcq+ssq*1]
    pmullw               m2, m5, [srcq+ssq*1+2]
    lea                srcq, [srcq+ssq*2]
    paddw                m1, m3
    paddw                m1, m2
    psrlw                m0, 4
    psrlw                m1, 4
    mova       [dstq+dsq*0], m0
    mova       [dstq+dsq*1], m1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .h_w16
    RET
.h_w32:
    pmullw               m0, m4, [srcq+32*0]
    pmullw               m1, m5, [srcq+32*0+2]
    paddw                m0, m3
    paddw                m0, m1
    pmullw               m1, m4, [srcq+32*1]
    pmullw               m2, m5, [srcq+32*1+2]
    add                srcq, ssq
    paddw                m1, m3
    paddw                m1, m2
    psrlw                m0, 4
    psrlw                m1, 4
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    add                dstq, dsq
    dec                  hd
    jg .h_w32
    RET
.h_w64:
.h_w128:
    movifnidn           t0d, org_w
.h_w64_loop0:
    mov                 r6d, t0d
.h_w64_loop:
    pmullw               m0, m4, [srcq+r6*2-32*1]
    pmullw               m1, m5, [srcq+r6*2-32*1+2]
    paddw                m0, m3
    paddw                m0, m1
    pmullw               m1, m4, [srcq+r6*2-32*2]
    pmullw               m2, m5, [srcq+r6*2-32*2+2]
    paddw                m1, m3
    paddw                m1, m2
    psrlw                m0, 4
    psrlw                m1, 4
    mova   [dstq+r6*2-32*1], m0
    mova   [dstq+r6*2-32*2], m1
    sub                 r6d, 32
    jg .h_w64_loop
    add                srcq, ssq
    add                dstq, dsq
    dec                  hd
    jg .h_w64_loop0
    RET
.v:
    movzx                wd, word [r7+wq*2+table_offset(put, _bilin_v)]
    shl                mxyd, 11
    movd                xm5, mxyd
    add                  wq, r7
    vpbroadcastw         m5, xm5
    jmp                  wq
.v_w2:
    movd                xm0, [srcq+ssq*0]
.v_w2_loop:
    movd                xm1, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    punpckldq           xm2, xm0, xm1
    movd                xm0, [srcq+ssq*0]
    punpckldq           xm1, xm0
    psubw               xm1, xm2
    pmulhrsw            xm1, xm5
    paddw               xm1, xm2
    movd       [dstq+dsq*0], xm1
    pextrd     [dstq+dsq*1], xm1, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .v_w2_loop
    RET
.v_w4:
    movq                xm0, [srcq+ssq*0]
.v_w4_loop:
    movq                xm1, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    punpcklqdq          xm2, xm0, xm1
    movq                xm0, [srcq+ssq*0]
    punpcklqdq          xm1, xm0
    psubw               xm1, xm2
    pmulhrsw            xm1, xm5
    paddw               xm1, xm2
    movq       [dstq+dsq*0], xm1
    movhps     [dstq+dsq*1], xm1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .v_w4_loop
    RET
.v_w8:
    movu                xm0, [srcq+ssq*0]
.v_w8_loop:
    vbroadcasti128       m1, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    vpblendd             m2, m0, m1, 0xf0
    vbroadcasti128       m0, [srcq+ssq*0]
    vpblendd             m1, m0, 0xf0
    psubw                m1, m2
    pmulhrsw             m1, m5
    paddw                m1, m2
    mova         [dstq+dsq*0], xm1
    vextracti128 [dstq+dsq*1], m1, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .v_w8_loop
    RET
.v_w32:
    movu                 m0, [srcq+ssq*0+32*0]
    movu                 m1, [srcq+ssq*0+32*1]
.v_w32_loop:
    movu                 m2, [srcq+ssq*1+32*0]
    movu                 m3, [srcq+ssq*1+32*1]
    lea                srcq, [srcq+ssq*2]
    psubw                m4, m2, m0
    pmulhrsw             m4, m5
    paddw                m4, m0
    movu                 m0, [srcq+ssq*0+32*0]
    mova  [dstq+dsq*0+32*0], m4
    psubw                m4, m3, m1
    pmulhrsw             m4, m5
    paddw                m4, m1
    movu                 m1, [srcq+ssq*0+32*1]
    mova  [dstq+dsq*0+32*1], m4
    psubw                m4, m0, m2
    pmulhrsw             m4, m5
    paddw                m4, m2
    mova  [dstq+dsq*1+32*0], m4
    psubw                m4, m1, m3
    pmulhrsw             m4, m5
    paddw                m4, m3
    mova  [dstq+dsq*1+32*1], m4
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .v_w32_loop
    RET
.v_w16:
.v_w64:
.v_w128:
    movifnidn           t0d, org_w
    add                 t0d, t0d
    mov                  r4, srcq
    lea                 r6d, [hq+t0*8-256]
    mov                  r7, dstq
.v_w16_loop0:
    movu                 m0, [srcq+ssq*0]
.v_w16_loop:
    movu                 m3, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    psubw                m1, m3, m0
    pmulhrsw             m1, m5
    paddw                m1, m0
    movu                 m0, [srcq+ssq*0]
    psubw                m2, m0, m3
    pmulhrsw             m2, m5
    paddw                m2, m3
    mova       [dstq+dsq*0], m1
    mova       [dstq+dsq*1], m2
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .v_w16_loop
    add                  r4, 32
    add                  r7, 32
    movzx                hd, r6b
    mov                srcq, r4
    mov                dstq, r7
    sub                 r6d, 1<<8
    jg .v_w16_loop0
    RET
.hv:
    movzx                wd, word [r7+wq*2+table_offset(put, _bilin_hv)]
    WIN64_SPILL_XMM       8
    shl                mxyd, 11
    vpbroadcastd         m3, [pw_2]
    movd                xm6, mxyd
    vpbroadcastd         m7, [pw_8192]
    add                  wq, r7
    vpbroadcastw         m6, xm6
    test          dword r8m, 0x800
    jnz .hv_12bpc
    psllw                m4, 2
    psllw                m5, 2
    vpbroadcastd         m7, [pw_2048]
.hv_12bpc:
    jmp                  wq
.hv_w2:
    vpbroadcastq        xm1, [srcq+ssq*0]
    pmullw              xm0, xm4, xm1
    psrlq               xm1, 16
    pmullw              xm1, xm5
    paddw               xm0, xm3
    paddw               xm0, xm1
    psrlw               xm0, 2
.hv_w2_loop:
    movq                xm2, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    movhps              xm2, [srcq+ssq*0]
    pmullw              xm1, xm4, xm2
    psrlq               xm2, 16
    pmullw              xm2, xm5
    paddw               xm1, xm3
    paddw               xm1, xm2
    psrlw               xm1, 2              ; 1 _ 2 _
    shufpd              xm2, xm0, xm1, 0x01 ; 0 _ 1 _
    mova                xm0, xm1
    psubw               xm1, xm2
    paddw               xm1, xm1
    pmulhw              xm1, xm6
    paddw               xm1, xm2
    pmulhrsw            xm1, xm7
    movd       [dstq+dsq*0], xm1
    pextrd     [dstq+dsq*1], xm1, 2
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .hv_w2_loop
    RET
.hv_w4:
    pmullw              xm0, xm4, [srcq+ssq*0-8]
    pmullw              xm1, xm5, [srcq+ssq*0-6]
    paddw               xm0, xm3
    paddw               xm0, xm1
    psrlw               xm0, 2
.hv_w4_loop:
    movq                xm1, [srcq+ssq*1]
    movq                xm2, [srcq+ssq*1+2]
    lea                srcq, [srcq+ssq*2]
    movhps              xm1, [srcq+ssq*0]
    movhps              xm2, [srcq+ssq*0+2]
    pmullw              xm1, xm4
    pmullw              xm2, xm5
    paddw               xm1, xm3
    paddw               xm1, xm2
    psrlw               xm1, 2              ; 1 2
    shufpd              xm2, xm0, xm1, 0x01 ; 0 1
    mova                xm0, xm1
    psubw               xm1, xm2
    paddw               xm1, xm1
    pmulhw              xm1, xm6
    paddw               xm1, xm2
    pmulhrsw            xm1, xm7
    movq       [dstq+dsq*0], xm1
    movhps     [dstq+dsq*1], xm1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .hv_w4_loop
    RET
.hv_w8:
    pmullw              xm0, xm4, [srcq+ssq*0]
    pmullw              xm1, xm5, [srcq+ssq*0+2]
    paddw               xm0, xm3
    paddw               xm0, xm1
    psrlw               xm0, 2
    vinserti128          m0, xm0, 1
.hv_w8_loop:
    movu                xm1, [srcq+ssq*1]
    movu                xm2, [srcq+ssq*1+2]
    lea                srcq, [srcq+ssq*2]
    vinserti128          m1, [srcq+ssq*0], 1
    vinserti128          m2, [srcq+ssq*0+2], 1
    pmullw               m1, m4
    pmullw               m2, m5
    paddw                m1, m3
    paddw                m1, m2
    psrlw                m1, 2            ; 1 2
    vperm2i128           m2, m0, m1, 0x21 ; 0 1
    mova                 m0, m1
    psubw                m1, m2
    paddw                m1, m1
    pmulhw               m1, m6
    paddw                m1, m2
    pmulhrsw             m1, m7
    mova         [dstq+dsq*0], xm1
    vextracti128 [dstq+dsq*1], m1, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .hv_w8_loop
    RET
.hv_w16:
.hv_w32:
.hv_w64:
.hv_w128:
%if UNIX64
    lea                 r6d, [r8*2-32]
%else
    mov                 r6d, wm
    lea                 r6d, [r6*2-32]
%endif
    mov                  r4, srcq
    lea                 r6d, [hq+r6*8]
    mov                  r7, dstq
.hv_w16_loop0:
    pmullw               m0, m4, [srcq+ssq*0]
    pmullw               m1, m5, [srcq+ssq*0+2]
    paddw                m0, m3
    paddw                m0, m1
    psrlw                m0, 2
.hv_w16_loop:
    pmullw               m1, m4, [srcq+ssq*1]
    pmullw               m2, m5, [srcq+ssq*1+2]
    lea                srcq, [srcq+ssq*2]
    paddw                m1, m3
    paddw                m1, m2
    psrlw                m1, 2
    psubw                m2, m1, m0
    paddw                m2, m2
    pmulhw               m2, m6
    paddw                m2, m0
    pmulhrsw             m2, m7
    mova       [dstq+dsq*0], m2
    pmullw               m0, m4, [srcq+ssq*0]
    pmullw               m2, m5, [srcq+ssq*0+2]
    paddw                m0, m3
    paddw                m0, m2
    psrlw                m0, 2
    psubw                m2, m0, m1
    paddw                m2, m2
    pmulhw               m2, m6
    paddw                m2, m1
    pmulhrsw             m2, m7
    mova       [dstq+dsq*1], m2
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .hv_w16_loop
    add                  r4, 32
    add                  r7, 32
    movzx                hd, r6b
    mov                srcq, r4
    mov                dstq, r7
    sub                 r6d, 1<<8
    jg .hv_w16_loop0
    RET

cglobal prep_bilin_16bpc, 3, 7, 0, tmp, src, stride, w, h, mxy, stride3
    movifnidn          mxyd, r5m ; mx
    lea                  r6, [prep_avx2]
%if UNIX64
    DECLARE_REG_TMP 7
    %define org_w r7d
%else
    DECLARE_REG_TMP 6
    %define org_w r5m
%endif
    mov               org_w, wd
    tzcnt                wd, wm
    movifnidn            hd, hm
    test               mxyd, mxyd
    jnz .h
    mov                mxyd, r6m ; my
    test               mxyd, mxyd
    jnz .v
.prep:
    movzx                wd, word [r6+wq*2+table_offset(prep,)]
    mov                 r5d, r7m ; bitdepth_max
    vpbroadcastd         m5, [r6-prep_avx2+pw_8192]
    add                  wq, r6
    shr                 r5d, 11
    vpbroadcastd         m4, [r6-prep_avx2+prep_mul+r5*4]
    lea            stride3q, [strideq*3]
    jmp                  wq
.prep_w4:
    movq                xm0, [srcq+strideq*0]
    movhps              xm0, [srcq+strideq*1]
    vpbroadcastq         m1, [srcq+strideq*2]
    vpbroadcastq         m2, [srcq+stride3q ]
    lea                srcq, [srcq+strideq*4]
    vpblendd             m0, m1, 0x30
    vpblendd             m0, m2, 0xc0
    pmullw               m0, m4
    psubw                m0, m5
    mova             [tmpq], m0
    add                tmpq, 32
    sub                  hd, 4
    jg .prep_w4
    RET
.prep_w8:
    movu                xm0, [srcq+strideq*0]
    vinserti128          m0, [srcq+strideq*1], 1
    movu                xm1, [srcq+strideq*2]
    vinserti128          m1, [srcq+stride3q ], 1
    lea                srcq, [srcq+strideq*4]
    pmullw               m0, m4
    pmullw               m1, m4
    psubw                m0, m5
    psubw                m1, m5
    mova        [tmpq+32*0], m0
    mova        [tmpq+32*1], m1
    add                tmpq, 32*2
    sub                  hd, 4
    jg .prep_w8
    RET
.prep_w16:
    pmullw               m0, m4, [srcq+strideq*0]
    pmullw               m1, m4, [srcq+strideq*1]
    pmullw               m2, m4, [srcq+strideq*2]
    pmullw               m3, m4, [srcq+stride3q ]
    lea                srcq, [srcq+strideq*4]
    psubw                m0, m5
    psubw                m1, m5
    psubw                m2, m5
    psubw                m3, m5
    mova        [tmpq+32*0], m0
    mova        [tmpq+32*1], m1
    mova        [tmpq+32*2], m2
    mova        [tmpq+32*3], m3
    add                tmpq, 32*4
    sub                  hd, 4
    jg .prep_w16
    RET
.prep_w32:
    pmullw               m0, m4, [srcq+strideq*0+32*0]
    pmullw               m1, m4, [srcq+strideq*0+32*1]
    pmullw               m2, m4, [srcq+strideq*1+32*0]
    pmullw               m3, m4, [srcq+strideq*1+32*1]
    lea                srcq, [srcq+strideq*2]
    psubw                m0, m5
    psubw                m1, m5
    psubw                m2, m5
    psubw                m3, m5
    mova        [tmpq+32*0], m0
    mova        [tmpq+32*1], m1
    mova        [tmpq+32*2], m2
    mova        [tmpq+32*3], m3
    add                tmpq, 32*4
    sub                  hd, 2
    jg .prep_w32
    RET
.prep_w64:
    pmullw               m0, m4, [srcq+32*0]
    pmullw               m1, m4, [srcq+32*1]
    pmullw               m2, m4, [srcq+32*2]
    pmullw               m3, m4, [srcq+32*3]
    add                srcq, strideq
    psubw                m0, m5
    psubw                m1, m5
    psubw                m2, m5
    psubw                m3, m5
    mova        [tmpq+32*0], m0
    mova        [tmpq+32*1], m1
    mova        [tmpq+32*2], m2
    mova        [tmpq+32*3], m3
    add                tmpq, 32*4
    dec                  hd
    jg .prep_w64
    RET
.prep_w128:
    pmullw               m0, m4, [srcq+32*0]
    pmullw               m1, m4, [srcq+32*1]
    pmullw               m2, m4, [srcq+32*2]
    pmullw               m3, m4, [srcq+32*3]
    psubw                m0, m5
    psubw                m1, m5
    psubw                m2, m5
    psubw                m3, m5
    mova        [tmpq+32*0], m0
    mova        [tmpq+32*1], m1
    mova        [tmpq+32*2], m2
    mova        [tmpq+32*3], m3
    pmullw               m0, m4, [srcq+32*4]
    pmullw               m1, m4, [srcq+32*5]
    pmullw               m2, m4, [srcq+32*6]
    pmullw               m3, m4, [srcq+32*7]
    add                tmpq, 32*8
    add                srcq, strideq
    psubw                m0, m5
    psubw                m1, m5
    psubw                m2, m5
    psubw                m3, m5
    mova        [tmpq-32*4], m0
    mova        [tmpq-32*3], m1
    mova        [tmpq-32*2], m2
    mova        [tmpq-32*1], m3
    dec                  hd
    jg .prep_w128
    RET
.h:
    movd                xm5, mxyd
    mov                mxyd, r6m ; my
    vpbroadcastd         m4, [pw_16]
    vpbroadcastw         m5, xm5
    vpbroadcastd         m3, [pw_32766]
    psubw                m4, m5
    test          dword r7m, 0x800
    jnz .h_12bpc
    psllw                m4, 2
    psllw                m5, 2
.h_12bpc:
    test               mxyd, mxyd
    jnz .hv
    movzx                wd, word [r6+wq*2+table_offset(prep, _bilin_h)]
    add                  wq, r6
    lea            stride3q, [strideq*3]
    jmp                  wq
.h_w4:
    movu                xm1, [srcq+strideq*0]
    vinserti128          m1, [srcq+strideq*2], 1
    movu                xm2, [srcq+strideq*1]
    vinserti128          m2, [srcq+stride3q ], 1
    lea                srcq, [srcq+strideq*4]
    punpcklqdq           m0, m1, m2
    psrldq               m1, 2
    pslldq               m2, 6
    pmullw               m0, m4
    vpblendd             m1, m2, 0xcc
    pmullw               m1, m5
    psubw                m0, m3
    paddw                m0, m1
    psraw                m0, 2
    mova             [tmpq], m0
    add                tmpq, 32
    sub                  hd, 4
    jg .h_w4
    RET
.h_w8:
    movu                xm0, [srcq+strideq*0]
    vinserti128          m0, [srcq+strideq*1], 1
    movu                xm1, [srcq+strideq*0+2]
    vinserti128          m1, [srcq+strideq*1+2], 1
    lea                srcq, [srcq+strideq*2]
    pmullw               m0, m4
    pmullw               m1, m5
    psubw                m0, m3
    paddw                m0, m1
    psraw                m0, 2
    mova             [tmpq], m0
    add                tmpq, 32
    sub                  hd, 2
    jg .h_w8
    RET
.h_w16:
    pmullw               m0, m4, [srcq+strideq*0]
    pmullw               m1, m5, [srcq+strideq*0+2]
    psubw                m0, m3
    paddw                m0, m1
    pmullw               m1, m4, [srcq+strideq*1]
    pmullw               m2, m5, [srcq+strideq*1+2]
    lea                srcq, [srcq+strideq*2]
    psubw                m1, m3
    paddw                m1, m2
    psraw                m0, 2
    psraw                m1, 2
    mova        [tmpq+32*0], m0
    mova        [tmpq+32*1], m1
    add                tmpq, 32*2
    sub                  hd, 2
    jg .h_w16
    RET
.h_w32:
.h_w64:
.h_w128:
    movifnidn           t0d, org_w
.h_w32_loop0:
    mov                 r3d, t0d
.h_w32_loop:
    pmullw               m0, m4, [srcq+r3*2-32*1]
    pmullw               m1, m5, [srcq+r3*2-32*1+2]
    psubw                m0, m3
    paddw                m0, m1
    pmullw               m1, m4, [srcq+r3*2-32*2]
    pmullw               m2, m5, [srcq+r3*2-32*2+2]
    psubw                m1, m3
    paddw                m1, m2
    psraw                m0, 2
    psraw                m1, 2
    mova   [tmpq+r3*2-32*1], m0
    mova   [tmpq+r3*2-32*2], m1
    sub                 r3d, 32
    jg .h_w32_loop
    add                srcq, strideq
    lea                tmpq, [tmpq+t0*2]
    dec                  hd
    jg .h_w32_loop0
    RET
.v:
    movzx                wd, word [r6+wq*2+table_offset(prep, _bilin_v)]
    movd                xm5, mxyd
    vpbroadcastd         m4, [pw_16]
    vpbroadcastw         m5, xm5
    vpbroadcastd         m3, [pw_32766]
    add                  wq, r6
    lea            stride3q, [strideq*3]
    psubw                m4, m5
    test          dword r7m, 0x800
    jnz .v_12bpc
    psllw                m4, 2
    psllw                m5, 2
.v_12bpc:
    jmp                  wq
.v_w4:
    movq                xm0, [srcq+strideq*0]
.v_w4_loop:
    vpbroadcastq         m2, [srcq+strideq*2]
    vpbroadcastq        xm1, [srcq+strideq*1]
    vpblendd             m2, m0, 0x03 ; 0 2 2 2
    vpbroadcastq         m0, [srcq+stride3q ]
    lea                srcq, [srcq+strideq*4]
    vpblendd             m1, m0, 0xf0 ; 1 1 3 3
    vpbroadcastq         m0, [srcq+strideq*0]
    vpblendd             m1, m2, 0x33 ; 0 1 2 3
    vpblendd             m0, m2, 0x0c ; 4 2 4 4
    punpckhqdq           m2, m1, m0   ; 1 2 3 4
    pmullw               m1, m4
    pmullw               m2, m5
    psubw                m1, m3
    paddw                m1, m2
    psraw                m1, 2
    mova             [tmpq], m1
    add                tmpq, 32
    sub                  hd, 4
    jg .v_w4_loop
    RET
.v_w8:
    movu                xm0, [srcq+strideq*0]
.v_w8_loop:
    vbroadcasti128       m2, [srcq+strideq*1]
    lea                srcq, [srcq+strideq*2]
    vpblendd             m1, m0, m2, 0xf0 ; 0 1
    vbroadcasti128       m0, [srcq+strideq*0]
    vpblendd             m2, m0, 0xf0     ; 1 2
    pmullw               m1, m4
    pmullw               m2, m5
    psubw                m1, m3
    paddw                m1, m2
    psraw                m1, 2
    mova             [tmpq], m1
    add                tmpq, 32
    sub                  hd, 2
    jg .v_w8_loop
    RET
.v_w16:
    movu                 m0, [srcq+strideq*0]
.v_w16_loop:
    movu                 m2, [srcq+strideq*1]
    lea                srcq, [srcq+strideq*2]
    pmullw               m0, m4
    pmullw               m1, m5, m2
    psubw                m0, m3
    paddw                m1, m0
    movu                 m0, [srcq+strideq*0]
    psraw                m1, 2
    pmullw               m2, m4
    mova        [tmpq+32*0], m1
    pmullw               m1, m5, m0
    psubw                m2, m3
    paddw                m1, m2
    psraw                m1, 2
    mova        [tmpq+32*1], m1
    add                tmpq, 32*2
    sub                  hd, 2
    jg .v_w16_loop
    RET
.v_w32:
.v_w64:
.v_w128:
%if WIN64
    PUSH                 r7
%endif
    movifnidn           r7d, org_w
    add                 r7d, r7d
    mov                  r3, srcq
    lea                 r6d, [hq+r7*8-256]
    mov                  r5, tmpq
.v_w32_loop0:
    movu                 m0, [srcq+strideq*0]
.v_w32_loop:
    movu                 m2, [srcq+strideq*1]
    lea                srcq, [srcq+strideq*2]
    pmullw               m0, m4
    pmullw               m1, m5, m2
    psubw                m0, m3
    paddw                m1, m0
    movu                 m0, [srcq+strideq*0]
    psraw                m1, 2
    pmullw               m2, m4
    mova        [tmpq+r7*0], m1
    pmullw               m1, m5, m0
    psubw                m2, m3
    paddw                m1, m2
    psraw                m1, 2
    mova        [tmpq+r7*1], m1
    lea                tmpq, [tmpq+r7*2]
    sub                  hd, 2
    jg .v_w32_loop
    add                  r3, 32
    add                  r5, 32
    movzx                hd, r6b
    mov                srcq, r3
    mov                tmpq, r5
    sub                 r6d, 1<<8
    jg .v_w32_loop0
%if WIN64
    POP                  r7
%endif
    RET
.hv:
    WIN64_SPILL_XMM       7
    movzx                wd, word [r6+wq*2+table_offset(prep, _bilin_hv)]
    shl                mxyd, 11
    movd                xm6, mxyd
    add                  wq, r6
    lea            stride3q, [strideq*3]
    vpbroadcastw         m6, xm6
    jmp                  wq
.hv_w4:
    movu                xm1, [srcq+strideq*0]
%if WIN64
    movaps         [rsp+24], xmm7
%endif
    pmullw              xm0, xm4, xm1
    psrldq              xm1, 2
    pmullw              xm1, xm5
    psubw               xm0, xm3
    paddw               xm0, xm1
    psraw               xm0, 2
    vpbroadcastq         m0, xm0
.hv_w4_loop:
    movu                xm1, [srcq+strideq*1]
    vinserti128          m1, [srcq+stride3q ], 1
    movu                xm2, [srcq+strideq*2]
    lea                srcq, [srcq+strideq*4]
    vinserti128          m2, [srcq+strideq*0], 1
    punpcklqdq           m7, m1, m2
    psrldq               m1, 2
    pslldq               m2, 6
    pmullw               m7, m4
    vpblendd             m1, m2, 0xcc
    pmullw               m1, m5
    psubw                m7, m3
    paddw                m1, m7
    psraw                m1, 2         ; 1 2 3 4
    vpblendd             m0, m1, 0x3f
    vpermq               m2, m0, q2103 ; 0 1 2 3
    mova                 m0, m1
    psubw                m1, m2
    pmulhrsw             m1, m6
    paddw                m1, m2
    mova             [tmpq], m1
    add                tmpq, 32
    sub                  hd, 4
    jg .hv_w4_loop
%if WIN64
    movaps             xmm7, [rsp+24]
%endif
    RET
.hv_w8:
    pmullw              xm0, xm4, [srcq+strideq*0]
    pmullw              xm1, xm5, [srcq+strideq*0+2]
    psubw               xm0, xm3
    paddw               xm0, xm1
    psraw               xm0, 2
    vinserti128          m0, xm0, 1
.hv_w8_loop:
    movu                xm1, [srcq+strideq*1]
    movu                xm2, [srcq+strideq*1+2]
    lea                srcq, [srcq+strideq*2]
    vinserti128          m1, [srcq+strideq*0], 1
    vinserti128          m2, [srcq+strideq*0+2], 1
    pmullw               m1, m4
    pmullw               m2, m5
    psubw                m1, m3
    paddw                m1, m2
    psraw                m1, 2            ; 1 2
    vperm2i128           m2, m0, m1, 0x21 ; 0 1
    mova                 m0, m1
    psubw                m1, m2
    pmulhrsw             m1, m6
    paddw                m1, m2
    mova             [tmpq], m1
    add                tmpq, 32
    sub                  hd, 2
    jg .hv_w8_loop
    RET
.hv_w16:
.hv_w32:
.hv_w64:
.hv_w128:
%if WIN64
    PUSH                 r7
%endif
    movifnidn           r7d, org_w
    add                 r7d, r7d
    mov                  r3, srcq
    lea                 r6d, [hq+r7*8-256]
    mov                  r5, tmpq
.hv_w16_loop0:
    pmullw               m0, m4, [srcq]
    pmullw               m1, m5, [srcq+2]
    psubw                m0, m3
    paddw                m0, m1
    psraw                m0, 2
.hv_w16_loop:
    pmullw               m1, m4, [srcq+strideq*1]
    pmullw               m2, m5, [srcq+strideq*1+2]
    lea                srcq, [srcq+strideq*2]
    psubw                m1, m3
    paddw                m1, m2
    psraw                m1, 2
    psubw                m2, m1, m0
    pmulhrsw             m2, m6
    paddw                m2, m0
    mova        [tmpq+r7*0], m2
    pmullw               m0, m4, [srcq+strideq*0]
    pmullw               m2, m5, [srcq+strideq*0+2]
    psubw                m0, m3
    paddw                m0, m2
    psraw                m0, 2
    psubw                m2, m0, m1
    pmulhrsw             m2, m6
    paddw                m2, m1
    mova        [tmpq+r7*1], m2
    lea                tmpq, [tmpq+r7*2]
    sub                  hd, 2
    jg .hv_w16_loop
    add                  r3, 32
    add                  r5, 32
    movzx                hd, r6b
    mov                srcq, r3
    mov                tmpq, r5
    sub                 r6d, 1<<8
    jg .hv_w16_loop0
%if WIN64
    POP                  r7
%endif
    RET

; int8_t subpel_filters[5][15][8]
%assign FILTER_REGULAR (0*15 << 16) | 3*15
%assign FILTER_SMOOTH  (1*15 << 16) | 4*15
%assign FILTER_SHARP   (2*15 << 16) | 3*15

%macro FN 4 ; prefix, type, type_h, type_v
cglobal %1_%2_16bpc
    mov                 t0d, FILTER_%3
%ifidn %3, %4
    mov                 t1d, t0d
%else
    mov                 t1d, FILTER_%4
%endif
%ifnidn %2, regular ; skip the jump in the last filter
    jmp mangle(private_prefix %+ _%1_16bpc %+ SUFFIX)
%endif
%endmacro

%if WIN64
DECLARE_REG_TMP 4, 5
%else
DECLARE_REG_TMP 7, 8
%endif

%define PUT_8TAP_FN FN put_8tap,
PUT_8TAP_FN sharp,          SHARP,   SHARP
PUT_8TAP_FN sharp_smooth,   SHARP,   SMOOTH
PUT_8TAP_FN smooth_sharp,   SMOOTH,  SHARP
PUT_8TAP_FN smooth,         SMOOTH,  SMOOTH
PUT_8TAP_FN sharp_regular,  SHARP,   REGULAR
PUT_8TAP_FN regular_sharp,  REGULAR, SHARP
PUT_8TAP_FN smooth_regular, SMOOTH,  REGULAR
PUT_8TAP_FN regular_smooth, REGULAR, SMOOTH
PUT_8TAP_FN regular,        REGULAR, REGULAR

cglobal put_8tap_16bpc, 4, 9, 0, dst, ds, src, ss, w, h, mx, my
%define base r8-put_avx2
    imul                mxd, mxm, 0x010101
    add                 mxd, t0d ; 8tap_h, mx, 4tap_h
    imul                myd, mym, 0x010101
    add                 myd, t1d ; 8tap_v, my, 4tap_v
    lea                  r8, [put_avx2]
    movifnidn            wd, wm
    movifnidn            hd, hm
    test                mxd, 0xf00
    jnz .h
    test                myd, 0xf00
    jnz .v
    tzcnt                wd, wd
    movzx                wd, word [r8+wq*2+table_offset(put,)]
    add                  wq, r8
%if WIN64
    pop                  r8
%endif
    jmp                  wq
.h_w2:
    movzx               mxd, mxb
    sub                srcq, 2
    mova                xm2, [subpel_h_shuf2]
    vpbroadcastd        xm3, [base+subpel_filters+mxq*8+2]
    pmovsxbw            xm3, xm3
.h_w2_loop:
    movu                xm0, [srcq+ssq*0]
    movu                xm1, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    pshufb              xm0, xm2
    pshufb              xm1, xm2
    pmaddwd             xm0, xm3
    pmaddwd             xm1, xm3
    phaddd              xm0, xm1
    paddd               xm0, xm4
    psrad               xm0, 6
    packusdw            xm0, xm0
    pminsw              xm0, xm5
    movd       [dstq+dsq*0], xm0
    pextrd     [dstq+dsq*1], xm0, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .h_w2_loop
    RET
.h_w4:
    movzx               mxd, mxb
    sub                srcq, 2
    pmovsxbw            xm3, [base+subpel_filters+mxq*8]
    WIN64_SPILL_XMM       8
    vbroadcasti128       m6, [subpel_h_shufA]
    vbroadcasti128       m7, [subpel_h_shufB]
    pshufd              xm3, xm3, q2211
    vpbroadcastq         m2, xm3
    vpermq               m3, m3, q1111
.h_w4_loop:
    movu                xm1, [srcq+ssq*0]
    vinserti128          m1, [srcq+ssq*1], 1
    lea                srcq, [srcq+ssq*2]
    pshufb               m0, m1, m6 ; 0 1 1 2 2 3 3 4
    pshufb               m1, m7     ; 2 3 3 4 4 5 5 6
    pmaddwd              m0, m2
    pmaddwd              m1, m3
    paddd                m0, m4
    paddd                m0, m1
    psrad                m0, 6
    vextracti128        xm1, m0, 1
    packusdw            xm0, xm1
    pminsw              xm0, xm5
    movq       [dstq+dsq*0], xm0
    movhps     [dstq+dsq*1], xm0
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .h_w4_loop
    RET
.h:
    test                myd, 0xf00
    jnz .hv
    mov                 r7d, r8m
    vpbroadcastw         m5, r8m
    shr                 r7d, 11
    vpbroadcastd         m4, [base+put_8tap_h_rnd+r7*4]
    cmp                  wd, 4
    je .h_w4
    jl .h_w2
    WIN64_SPILL_XMM      13
    shr                 mxd, 16
    sub                srcq, 6
    vpbroadcastq         m0, [base+subpel_filters+mxq*8]
    vbroadcasti128       m6, [subpel_h_shufA]
    vbroadcasti128       m7, [subpel_h_shufB]
    punpcklbw            m0, m0
    psraw                m0, 8 ; sign-extend
    pshufd               m8, m0, q0000
    pshufd               m9, m0, q1111
    pshufd              m10, m0, q2222
    pshufd              m11, m0, q3333
    cmp                  wd, 8
    jg .h_w16
.h_w8:
%macro PUT_8TAP_H 5 ; dst/src+0, src+8, src+16, tmp[1-2]
    pshufb              m%4, m%1, m7   ; 2 3 3 4 4 5 5 6
    pshufb              m%1, m6        ; 0 1 1 2 2 3 3 4
    pmaddwd             m%5, m9, m%4   ; abcd1
    pmaddwd             m%1, m8        ; abcd0
    pshufb              m%2, m7        ; 6 7 7 8 8 9 9 a
    shufpd              m%4, m%2, 0x05 ; 4 5 5 6 6 7 7 8
    paddd               m%5, m4
    paddd               m%1, m%5
    pmaddwd             m%5, m11, m%2  ; abcd3
    paddd               m%1, m%5
    pmaddwd             m%5, m10, m%4  ; abcd2
    pshufb              m%3, m7        ; a b b c c d d e
    pmaddwd             m%4, m8        ; efgh0
    paddd               m%1, m%5
    pmaddwd             m%5, m9, m%2   ; efgh1
    shufpd              m%2, m%3, 0x05 ; 8 9 9 a a b b c
    pmaddwd             m%3, m11       ; efgh3
    pmaddwd             m%2, m10       ; efgh2
    paddd               m%4, m4
    paddd               m%4, m%5
    paddd               m%3, m%4
    paddd               m%2, m%3
    psrad               m%1, 6
    psrad               m%2, 6
    packusdw            m%1, m%2
    pminsw              m%1, m5
%endmacro
    movu                xm0, [srcq+ssq*0+ 0]
    vinserti128          m0, [srcq+ssq*1+ 0], 1
    movu                xm2, [srcq+ssq*0+16]
    vinserti128          m2, [srcq+ssq*1+16], 1
    lea                srcq, [srcq+ssq*2]
    shufpd               m1, m0, m2, 0x05
    PUT_8TAP_H            0, 1, 2, 3, 12
    mova         [dstq+dsq*0], xm0
    vextracti128 [dstq+dsq*1], m0, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .h_w8
    RET
.h_w16:
    mov                 r6d, wd
.h_w16_loop:
    movu                 m0, [srcq+r6*2-32]
    movu                 m1, [srcq+r6*2-24]
    movu                 m2, [srcq+r6*2-16]
    PUT_8TAP_H            0, 1, 2, 3, 12
    mova     [dstq+r6*2-32], m0
    sub                 r6d, 16
    jg .h_w16_loop
    add                srcq, ssq
    add                dstq, dsq
    dec                  hd
    jg .h_w16
    RET
.v:
    movzx               mxd, myb
    shr                 myd, 16
    cmp                  hd, 4
    cmovle              myd, mxd
    vpbroadcastq         m0, [base+subpel_filters+myq*8]
    WIN64_SPILL_XMM      15
    vpbroadcastd         m6, [pd_32]
    vpbroadcastw         m7, r8m
    lea                  r6, [ssq*3]
    sub                srcq, r6
    punpcklbw            m0, m0
    psraw                m0, 8 ; sign-extend
    pshufd               m8, m0, q0000
    pshufd               m9, m0, q1111
    pshufd              m10, m0, q2222
    pshufd              m11, m0, q3333
    cmp                  wd, 4
    jg .v_w8
    je .v_w4
.v_w2:
    movd                xm2, [srcq+ssq*0]
    pinsrd              xm2, [srcq+ssq*1], 1
    pinsrd              xm2, [srcq+ssq*2], 2
    pinsrd              xm2, [srcq+r6   ], 3 ; 0 1 2 3
    lea                srcq, [srcq+ssq*4]
    movd                xm3, [srcq+ssq*0]
    vpbroadcastd        xm1, [srcq+ssq*1]
    vpbroadcastd        xm0, [srcq+ssq*2]
    add                srcq, r6
    vpblendd            xm3, xm1, 0x02       ; 4 5
    vpblendd            xm1, xm0, 0x02       ; 5 6
    palignr             xm4, xm3, xm2, 4     ; 1 2 3 4
    punpcklwd           xm3, xm1             ; 45 56
    punpcklwd           xm1, xm2, xm4        ; 01 12
    punpckhwd           xm2, xm4             ; 23 34
.v_w2_loop:
    vpbroadcastd        xm4, [srcq+ssq*0]
    pmaddwd             xm5, xm8, xm1        ; a0 b0
    mova                xm1, xm2
    pmaddwd             xm2, xm9             ; a1 b1
    paddd               xm5, xm6
    paddd               xm5, xm2
    mova                xm2, xm3
    pmaddwd             xm3, xm10            ; a2 b2
    paddd               xm5, xm3
    vpblendd            xm3, xm0, xm4, 0x02  ; 6 7
    vpbroadcastd        xm0, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    vpblendd            xm4, xm0, 0x02       ; 7 8
    punpcklwd           xm3, xm4             ; 67 78
    pmaddwd             xm4, xm11, xm3       ; a3 b3
    paddd               xm5, xm4
    psrad               xm5, 6
    packusdw            xm5, xm5
    pminsw              xm5, xm7
    movd       [dstq+dsq*0], xm5
    pextrd     [dstq+dsq*1], xm5, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .v_w2_loop
    RET
.v_w4:
    movq                xm1, [srcq+ssq*0]
    vpbroadcastq         m0, [srcq+ssq*1]
    vpbroadcastq         m2, [srcq+ssq*2]
    vpbroadcastq         m4, [srcq+r6   ]
    lea                srcq, [srcq+ssq*4]
    vpbroadcastq         m3, [srcq+ssq*0]
    vpbroadcastq         m5, [srcq+ssq*1]
    vpblendd             m1, m0, 0x30
    vpblendd             m0, m2, 0x30
    punpcklwd            m1, m0      ; 01 12
    vpbroadcastq         m0, [srcq+ssq*2]
    add                srcq, r6
    vpblendd             m2, m4, 0x30
    vpblendd             m4, m3, 0x30
    punpcklwd            m2, m4      ; 23 34
    vpblendd             m3, m5, 0x30
    vpblendd             m5, m0, 0x30
    punpcklwd            m3, m5      ; 45 56
.v_w4_loop:
    vpbroadcastq         m4, [srcq+ssq*0]
    pmaddwd              m5, m8, m1  ; a0 b0
    mova                 m1, m2
    pmaddwd              m2, m9      ; a1 b1
    paddd                m5, m6
    paddd                m5, m2
    mova                 m2, m3
    pmaddwd              m3, m10     ; a2 b2
    paddd                m5, m3
    vpblendd             m3, m0, m4, 0x30
    vpbroadcastq         m0, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    vpblendd             m4, m0, 0x30
    punpcklwd            m3, m4      ; 67 78
    pmaddwd              m4, m11, m3 ; a3 b3
    paddd                m5, m4
    psrad                m5, 6
    vextracti128        xm4, m5, 1
    packusdw            xm5, xm4
    pminsw              xm5, xm7
    movq       [dstq+dsq*0], xm5
    movhps     [dstq+dsq*1], xm5
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .v_w4_loop
    RET
.v_w8:
    shl                  wd, 5
    mov                  r7, srcq
    mov                  r8, dstq
    lea                  wd, [hq+wq-256]
.v_w8_loop0:
    vbroadcasti128       m4, [srcq+ssq*0]
    vbroadcasti128       m5, [srcq+ssq*1]
    vbroadcasti128       m0, [srcq+r6   ]
    vbroadcasti128       m6, [srcq+ssq*2]
    lea                srcq, [srcq+ssq*4]
    vbroadcasti128       m1, [srcq+ssq*0]
    vbroadcasti128       m2, [srcq+ssq*1]
    vbroadcasti128       m3, [srcq+ssq*2]
    add                srcq, r6
    shufpd               m4, m0, 0x0c
    shufpd               m5, m1, 0x0c
    punpcklwd            m1, m4, m5 ; 01
    punpckhwd            m4, m5     ; 34
    shufpd               m6, m2, 0x0c
    punpcklwd            m2, m5, m6 ; 12
    punpckhwd            m5, m6     ; 45
    shufpd               m0, m3, 0x0c
    punpcklwd            m3, m6, m0 ; 23
    punpckhwd            m6, m0     ; 56
.v_w8_loop:
    vbroadcasti128      m14, [srcq+ssq*0]
    pmaddwd             m12, m8, m1  ; a0
    pmaddwd             m13, m8, m2  ; b0
    mova                 m1, m3
    mova                 m2, m4
    pmaddwd              m3, m9      ; a1
    pmaddwd              m4, m9      ; b1
    paddd               m12, m3
    paddd               m13, m4
    mova                 m3, m5
    mova                 m4, m6
    pmaddwd              m5, m10     ; a2
    pmaddwd              m6, m10     ; b2
    paddd               m12, m5
    vbroadcasti128       m5, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    paddd               m13, m6
    shufpd               m6, m0, m14, 0x0d
    shufpd               m0, m14, m5, 0x0c
    punpcklwd            m5, m6, m0  ; 67
    punpckhwd            m6, m0      ; 78
    pmaddwd             m14, m11, m5 ; a3
    paddd               m12, m14
    pmaddwd             m14, m11, m6 ; b3
    paddd               m13, m14
    psrad               m12, 5
    psrad               m13, 5
    packusdw            m12, m13
    pxor                m13, m13
    pavgw               m12, m13
    pminsw              m12, m7
    vpermq              m12, m12, q3120
    mova         [dstq+dsq*0], xm12
    vextracti128 [dstq+dsq*1], m12, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .v_w8_loop
    add                  r7, 16
    add                  r8, 16
    movzx                hd, wb
    mov                srcq, r7
    mov                dstq, r8
    sub                  wd, 1<<8
    jg .v_w8_loop0
    RET
.hv:
    WIN64_SPILL_XMM      16
    vpbroadcastw        m15, r8m
    cmp                  wd, 4
    jg .hv_w8
    movzx               mxd, mxb
    vpbroadcastd         m0, [base+subpel_filters+mxq*8+2]
    movzx               mxd, myb
    shr                 myd, 16
    cmp                  hd, 4
    cmovle              myd, mxd
    vpbroadcastq         m1, [base+subpel_filters+myq*8]
    vpbroadcastd         m6, [pd_512]
    lea                  r6, [ssq*3]
    sub                srcq, 2
    sub                srcq, r6
    pxor                 m7, m7
    punpcklbw            m7, m0
    punpcklbw            m1, m1
    psraw                m1, 8 ; sign-extend
    test          dword r8m, 0x800
    jz .hv_10bit
    psraw                m7, 2
    psllw                m1, 2
.hv_10bit:
    pshufd              m11, m1, q0000
    pshufd              m12, m1, q1111
    pshufd              m13, m1, q2222
    pshufd              m14, m1, q3333
    cmp                  wd, 4
    je .hv_w4
    vbroadcasti128       m9, [subpel_h_shuf2]
    vbroadcasti128       m1, [srcq+r6   ]    ; 3 3
    movu                xm3, [srcq+ssq*2]
    movu                xm0, [srcq+ssq*0]
    movu                xm2, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*4]
    vinserti128          m3, [srcq+ssq*0], 1 ; 2 4
    vinserti128          m0, [srcq+ssq*1], 1 ; 0 5
    vinserti128          m2, [srcq+ssq*2], 1 ; 1 6
    add                srcq, r6
    pshufb               m1, m9
    pshufb               m3, m9
    pshufb               m0, m9
    pshufb               m2, m9
    pmaddwd              m1, m7
    pmaddwd              m3, m7
    pmaddwd              m0, m7
    pmaddwd              m2, m7
    phaddd               m1, m3
    phaddd               m0, m2
    paddd                m1, m6
    paddd                m0, m6
    psrad                m1, 10
    psrad                m0, 10
    packssdw             m1, m0         ; 3 2 0 1
    vextracti128        xm0, m1, 1      ; 3 4 5 6
    pshufd              xm2, xm1, q1301 ; 2 3 1 2
    pshufd              xm3, xm0, q2121 ; 4 5 4 5
    punpckhwd           xm1, xm2        ; 01 12
    punpcklwd           xm2, xm0        ; 23 34
    punpckhwd           xm3, xm0        ; 45 56
.hv_w2_loop:
    movu                xm4, [srcq+ssq*0]
    movu                xm5, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    pshufb              xm4, xm9
    pshufb              xm5, xm9
    pmaddwd             xm4, xm7
    pmaddwd             xm5, xm7
    phaddd              xm4, xm5
    pmaddwd             xm5, xm11, xm1 ; a0 b0
    mova                xm1, xm2
    pmaddwd             xm2, xm12      ; a1 b1
    paddd               xm5, xm2
    mova                xm2, xm3
    pmaddwd             xm3, xm13      ; a2 b2
    paddd               xm5, xm3
    paddd               xm4, xm6
    psrad               xm4, 10
    packssdw            xm4, xm4
    palignr             xm3, xm4, xm0, 12
    mova                xm0, xm4
    punpcklwd           xm3, xm0       ; 67 78
    pmaddwd             xm4, xm14, xm3 ; a3 b3
    paddd               xm5, xm6
    paddd               xm5, xm4
    psrad               xm5, 10
    packusdw            xm5, xm5
    pminsw              xm5, xm15
    movd       [dstq+dsq*0], xm5
    pextrd     [dstq+dsq*1], xm5, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .hv_w2_loop
    RET
.hv_w4:
    vbroadcasti128       m9, [subpel_h_shufA]
    vbroadcasti128      m10, [subpel_h_shufB]
    pshufd               m8, m7, q1111
    pshufd               m7, m7, q0000
    movu                xm1, [srcq+ssq*0]
    vinserti128          m1, [srcq+ssq*1], 1     ; 0 1
    vbroadcasti128       m0, [srcq+r6   ]
    vinserti128          m2, m0, [srcq+ssq*2], 0 ; 2 3
    lea                srcq, [srcq+ssq*4]
    vinserti128          m0, [srcq+ssq*0], 1     ; 3 4
    movu                xm3, [srcq+ssq*1]
    vinserti128          m3, [srcq+ssq*2], 1     ; 5 6
    add                srcq, r6
    pshufb               m4, m1, m9
    pshufb               m1, m10
    pmaddwd              m4, m7
    pmaddwd              m1, m8
    pshufb               m5, m2, m9
    pshufb               m2, m10
    pmaddwd              m5, m7
    pmaddwd              m2, m8
    paddd                m4, m6
    paddd                m1, m4
    pshufb               m4, m0, m9
    pshufb               m0, m10
    pmaddwd              m4, m7
    pmaddwd              m0, m8
    paddd                m5, m6
    paddd                m2, m5
    pshufb               m5, m3, m9
    pshufb               m3, m10
    pmaddwd              m5, m7
    pmaddwd              m3, m8
    paddd                m4, m6
    paddd                m4, m0
    paddd                m5, m6
    paddd                m5, m3
    vperm2i128           m0, m1, m2, 0x21
    psrld                m1, 10
    psrld                m2, 10
    vperm2i128           m3, m4, m5, 0x21
    pslld                m4, 6
    pslld                m5, 6
    pblendw              m2, m4, 0xaa ; 23 34
    pslld                m0, 6
    pblendw              m1, m0, 0xaa ; 01 12
    psrld                m3, 10
    pblendw              m3, m5, 0xaa ; 45 56
    psrad                m0, m5, 16
.hv_w4_loop:
    movu                xm4, [srcq+ssq*0]
    vinserti128          m4, [srcq+ssq*1], 1
    lea                srcq, [srcq+ssq*2]
    pmaddwd              m5, m11, m1   ; a0 b0
    mova                 m1, m2
    pmaddwd              m2, m12       ; a1 b1
    paddd                m5, m6
    paddd                m5, m2
    mova                 m2, m3
    pmaddwd              m3, m13       ; a2 b2
    paddd                m5, m3
    pshufb               m3, m4, m9
    pshufb               m4, m10
    pmaddwd              m3, m7
    pmaddwd              m4, m8
    paddd                m3, m6
    paddd                m4, m3
    psrad                m4, 10
    packssdw             m0, m4        ; _ 7 6 8
    vpermq               m3, m0, q1122 ; _ 6 _ 7
    punpckhwd            m3, m0        ; 67 78
    mova                 m0, m4
    pmaddwd              m4, m14, m3   ; a3 b3
    paddd                m4, m5
    psrad                m4, 10
    vextracti128        xm5, m4, 1
    packusdw            xm4, xm5
    pminsw              xm4, xm15
    movq       [dstq+dsq*0], xm4
    movhps     [dstq+dsq*1], xm4
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .hv_w4_loop
    RET
.hv_w8:
    shr                 mxd, 16
    vpbroadcastq         m2, [base+subpel_filters+mxq*8]
    movzx               mxd, myb
    shr                 myd, 16
    cmp                  hd, 4
    cmovle              myd, mxd
    pmovsxbw            xm1, [base+subpel_filters+myq*8]
    shl                  wd, 5
    lea                  r6, [ssq*3]
    sub                srcq, 6
    sub                srcq, r6
    pxor                 m0, m0
    punpcklbw            m0, m2
    mov                  r7, srcq
    mov                  r8, dstq
    lea                  wd, [hq+wq-256]
    test          dword r8m, 0x800
    jz .hv_w8_10bit
    psraw                m0, 2
    psllw               xm1, 2
.hv_w8_10bit:
    pshufd              m11, m0, q0000
    pshufd              m12, m0, q1111
    pshufd              m13, m0, q2222
    pshufd              m14, m0, q3333
%if WIN64
    %define v_mul (rsp+stack_offset+40) ; r4m
%else
    %define v_mul (rsp-24) ; red zone
%endif
    mova            [v_mul], xm1
.hv_w8_loop0:
%macro PUT_8TAP_HV_H 3 ; dst/src+0, src+8, src+16
    pshufb               m2, m%1, m9   ; 2 3 3 4 4 5 5 6
    pshufb              m%1, m8        ; 0 1 1 2 2 3 3 4
    pmaddwd              m3, m12, m2
    pmaddwd             m%1, m11
    pshufb              m%2, m9        ; 6 7 7 8 8 9 9 a
    shufpd               m2, m%2, 0x05 ; 4 5 5 6 6 7 7 8
    paddd                m3, m10
    paddd               m%1, m3
    pmaddwd              m3, m14, m%2
    paddd               m%1, m3
    pmaddwd              m3, m13, m2
    pshufb              m%3, m9        ; a b b c c d d e
    pmaddwd              m2, m11
    paddd               m%1, m3
    pmaddwd              m3, m12, m%2
    shufpd              m%2, m%3, 0x05 ; 8 9 9 a a b b c
    pmaddwd             m%3, m14
    pmaddwd             m%2, m13
    paddd                m2, m10
    paddd                m2, m3
    paddd               m%3, m2
    paddd               m%2, m%3
    psrad               m%1, 10
    psrad               m%2, 10
    packssdw            m%1, m%2
%endmacro
    movu                xm4, [srcq+r6 *1+ 0]
    vbroadcasti128       m8, [subpel_h_shufA]
    movu                xm6, [srcq+r6 *1+ 8]
    vbroadcasti128       m9, [subpel_h_shufB]
    movu                xm0, [srcq+r6 *1+16]
    vpbroadcastd        m10, [pd_512]
    movu                xm5, [srcq+ssq*0+ 0]
    vinserti128          m5, [srcq+ssq*4+ 0], 1
    movu                xm1, [srcq+ssq*0+16]
    vinserti128          m1, [srcq+ssq*4+16], 1
    shufpd               m7, m5, m1, 0x05
    INIT_XMM avx2
    PUT_8TAP_HV_H         4, 6, 0    ; 3
    INIT_YMM avx2
    PUT_8TAP_HV_H         5, 7, 1    ; 0 4
    movu                xm0, [srcq+ssq*2+ 0]
    vinserti128          m0, [srcq+r6 *2+ 0], 1
    movu                xm1, [srcq+ssq*2+16]
    vinserti128          m1, [srcq+r6 *2+16], 1
    shufpd               m7, m0, m1, 0x05
    PUT_8TAP_HV_H         0, 7, 1    ; 2 6
    movu                xm6, [srcq+ssq*1+ 0]
    movu                xm1, [srcq+ssq*1+16]
    lea                srcq, [srcq+ssq*4]
    vinserti128          m6, [srcq+ssq*1+ 0], 1
    vinserti128          m1, [srcq+ssq*1+16], 1
    add                srcq, r6
    shufpd               m7, m6, m1, 0x05
    PUT_8TAP_HV_H         6, 7, 1    ; 1 5
    vpermq               m4, m4, q1100
    vpermq               m5, m5, q3120
    vpermq               m6, m6, q3120
    vpermq               m7, m0, q3120
    punpcklwd            m3, m7, m4  ; 23
    punpckhwd            m4, m5      ; 34
    punpcklwd            m1, m5, m6  ; 01
    punpckhwd            m5, m6      ; 45
    punpcklwd            m2, m6, m7  ; 12
    punpckhwd            m6, m7      ; 56
.hv_w8_loop:
    vpbroadcastd         m9, [v_mul+4*0]
    vpbroadcastd         m7, [v_mul+4*1]
    vpbroadcastd        m10, [v_mul+4*2]
    pmaddwd              m8, m9, m1  ; a0
    pmaddwd              m9, m2      ; b0
    mova                 m1, m3
    mova                 m2, m4
    pmaddwd              m3, m7      ; a1
    pmaddwd              m4, m7      ; b1
    paddd                m8, m3
    paddd                m9, m4
    mova                 m3, m5
    mova                 m4, m6
    pmaddwd              m5, m10     ; a2
    pmaddwd              m6, m10     ; b2
    paddd                m8, m5
    paddd                m9, m6
    movu                xm5, [srcq+ssq*0]
    vinserti128          m5, [srcq+ssq*1], 1
    vbroadcasti128       m7, [subpel_h_shufA]
    vbroadcasti128      m10, [subpel_h_shufB]
    movu                xm6, [srcq+ssq*0+16]
    vinserti128          m6, [srcq+ssq*1+16], 1
    vextracti128     [dstq], m0, 1
    pshufb               m0, m5, m7  ; 01
    pshufb               m5, m10     ; 23
    pmaddwd              m0, m11
    pmaddwd              m5, m12
    paddd                m0, m5
    pshufb               m5, m6, m7  ; 89
    pshufb               m6, m10     ; ab
    pmaddwd              m5, m13
    pmaddwd              m6, m14
    paddd                m6, m5
    movu                xm5, [srcq+ssq*0+8]
    vinserti128          m5, [srcq+ssq*1+8], 1
    lea                srcq, [srcq+ssq*2]
    pshufb               m7, m5, m7
    pshufb               m5, m10
    pmaddwd             m10, m13, m7
    pmaddwd              m7, m11
    paddd                m0, m10
    vpbroadcastd        m10, [pd_512]
    paddd                m6, m7
    pmaddwd              m7, m14, m5
    pmaddwd              m5, m12
    paddd                m0, m7
    paddd                m5, m6
    vbroadcasti128       m6, [dstq]
    paddd                m8, m10
    paddd                m9, m10
    paddd                m0, m10
    paddd                m5, m10
    vpbroadcastd        m10, [v_mul+4*3]
    psrad                m0, 10
    psrad                m5, 10
    packssdw             m0, m5
    vpermq               m7, m0, q3120 ; 7 8
    shufpd               m6, m7, 0x04  ; 6 7
    punpcklwd            m5, m6, m7    ; 67
    punpckhwd            m6, m7        ; 78
    pmaddwd              m7, m10, m5   ; a3
    pmaddwd             m10, m6        ; b3
    paddd                m7, m8
    paddd                m9, m10
    psrad                m7, 10
    psrad                m9, 10
    packusdw             m7, m9
    pminsw               m7, m15
    vpermq               m7, m7, q3120
    mova         [dstq+dsq*0], xm7
    vextracti128 [dstq+dsq*1], m7, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .hv_w8_loop
    add                  r7, 16
    add                  r8, 16
    movzx                hd, wb
    mov                srcq, r7
    mov                dstq, r8
    sub                  wd, 1<<8
    jg .hv_w8_loop0
    RET

%if WIN64
DECLARE_REG_TMP 6, 4
%else
DECLARE_REG_TMP 6, 7
%endif

%define PREP_8TAP_FN FN prep_8tap,
PREP_8TAP_FN sharp,          SHARP,   SHARP
PREP_8TAP_FN sharp_smooth,   SHARP,   SMOOTH
PREP_8TAP_FN smooth_sharp,   SMOOTH,  SHARP
PREP_8TAP_FN smooth,         SMOOTH,  SMOOTH
PREP_8TAP_FN sharp_regular,  SHARP,   REGULAR
PREP_8TAP_FN regular_sharp,  REGULAR, SHARP
PREP_8TAP_FN smooth_regular, SMOOTH,  REGULAR
PREP_8TAP_FN regular_smooth, REGULAR, SMOOTH
PREP_8TAP_FN regular,        REGULAR, REGULAR

cglobal prep_8tap_16bpc, 4, 8, 0, tmp, src, stride, w, h, mx, my
%define base r7-prep_avx2
    imul                mxd, mxm, 0x010101
    add                 mxd, t0d ; 8tap_h, mx, 4tap_h
    imul                myd, mym, 0x010101
    add                 myd, t1d ; 8tap_v, my, 4tap_v
    lea                  r7, [prep_avx2]
    movifnidn            hd, hm
    test                mxd, 0xf00
    jnz .h
    test                myd, 0xf00
    jnz .v
    tzcnt                wd, wd
    mov                 r6d, r7m ; bitdepth_max
    movzx                wd, word [r7+wq*2+table_offset(prep,)]
    vpbroadcastd         m5, [r7-prep_avx2+pw_8192]
    shr                 r6d, 11
    add                  wq, r7
    vpbroadcastd         m4, [base+prep_mul+r6*4]
    lea                  r6, [strideq*3]
%if WIN64
    pop                  r7
%endif
    jmp                  wq
.h_w4:
    movzx               mxd, mxb
    sub                srcq, 2
    pmovsxbw            xm0, [base+subpel_filters+mxq*8]
    vbroadcasti128       m3, [subpel_h_shufA]
    vbroadcasti128       m4, [subpel_h_shufB]
    WIN64_SPILL_XMM       8
    pshufd              xm0, xm0, q2211
    test          dword r7m, 0x800
    jnz .h_w4_12bpc
    psllw               xm0, 2
.h_w4_12bpc:
    vpbroadcastq         m6, xm0
    vpermq               m7, m0, q1111
.h_w4_loop:
    movu                xm1, [srcq+strideq*0]
    vinserti128          m1, [srcq+strideq*2], 1
    movu                xm2, [srcq+strideq*1]
    vinserti128          m2, [srcq+r6       ], 1
    lea                srcq, [srcq+strideq*4]
    pshufb               m0, m1, m3 ; 0 1 1 2 2 3 3 4
    pshufb               m1, m4     ; 2 3 3 4 4 5 5 6
    pmaddwd              m0, m6
    pmaddwd              m1, m7
    paddd                m0, m5
    paddd                m0, m1
    pshufb               m1, m2, m3
    pshufb               m2, m4
    pmaddwd              m1, m6
    pmaddwd              m2, m7
    paddd                m1, m5
    paddd                m1, m2
    psrad                m0, 4
    psrad                m1, 4
    packssdw             m0, m1
    mova             [tmpq], m0
    add                tmpq, 32
    sub                  hd, 4
    jg .h_w4_loop
    RET
.h:
    test                myd, 0xf00
    jnz .hv
    vpbroadcastd         m5, [prep_8tap_1d_rnd] ; 8 - (8192 << 4)
    lea                  r6, [strideq*3]
    cmp                  wd, 4
    je .h_w4
    shr                 mxd, 16
    sub                srcq, 6
    vpbroadcastq         m0, [base+subpel_filters+mxq*8]
    WIN64_SPILL_XMM      12
    vbroadcasti128       m6, [subpel_h_shufA]
    vbroadcasti128       m7, [subpel_h_shufB]
    punpcklbw            m0, m0
    psraw                m0, 8 ; sign-extend
    test          dword r7m, 0x800
    jnz .h_12bpc
    psllw                m0, 2
.h_12bpc:
    pshufd               m8, m0, q0000
    pshufd               m9, m0, q1111
    pshufd              m10, m0, q2222
    pshufd              m11, m0, q3333
    cmp                  wd, 8
    jg .h_w16
.h_w8:
%macro PREP_8TAP_H 5 ; dst/src+0, src+8, src+16, tmp[1-2]
    pshufb              m%4, m%1, m7   ; 2 3 3 4 4 5 5 6
    pshufb              m%1, m6        ; 0 1 1 2 2 3 3 4
    pmaddwd             m%5, m9, m%4   ; abcd1
    pmaddwd             m%1, m8        ; abcd0
    pshufb              m%2, m7        ; 6 7 7 8 8 9 9 a
    shufpd              m%4, m%2, 0x05 ; 4 5 5 6 6 7 7 8
    paddd               m%5, m5
    paddd               m%1, m%5
    pmaddwd             m%5, m11, m%2  ; abcd3
    paddd               m%1, m%5
    pmaddwd             m%5, m10, m%4  ; abcd2
    pshufb              m%3, m7        ; a b b c c d d e
    pmaddwd             m%4, m8        ; efgh0
    paddd               m%1, m%5
    pmaddwd             m%5, m9, m%2   ; efgh1
    shufpd              m%2, m%3, 0x05 ; 8 9 9 a a b b c
    pmaddwd             m%3, m11       ; efgh3
    pmaddwd             m%2, m10       ; efgh2
    paddd               m%4, m5
    paddd               m%4, m%5
    paddd               m%3, m%4
    paddd               m%2, m%3
    psrad               m%1, 4
    psrad               m%2, 4
    packssdw            m%1, m%2
%endmacro
    movu                xm0, [srcq+strideq*0+ 0]
    vinserti128          m0, [srcq+strideq*1+ 0], 1
    movu                xm2, [srcq+strideq*0+16]
    vinserti128          m2, [srcq+strideq*1+16], 1
    lea                srcq, [srcq+strideq*2]
    shufpd               m1, m0, m2, 0x05
    PREP_8TAP_H           0, 1, 2, 3, 4
    mova             [tmpq], m0
    add                tmpq, 32
    sub                  hd, 2
    jg .h_w8
    RET
.h_w16:
    add                  wd, wd
.h_w16_loop0:
    mov                 r6d, wd
.h_w16_loop:
    movu                 m0, [srcq+r6-32]
    movu                 m1, [srcq+r6-24]
    movu                 m2, [srcq+r6-16]
    PREP_8TAP_H           0, 1, 2, 3, 4
    mova       [tmpq+r6-32], m0
    sub                 r6d, 32
    jg .h_w16_loop
    add                srcq, strideq
    add                tmpq, wq
    dec                  hd
    jg .h_w16_loop0
    RET
.v:
    movzx               mxd, myb
    shr                 myd, 16
    cmp                  hd, 4
    cmovle              myd, mxd
    vpbroadcastq         m0, [base+subpel_filters+myq*8]
    WIN64_SPILL_XMM      15
    vpbroadcastd         m7, [prep_8tap_1d_rnd]
    lea                  r6, [strideq*3]
    sub                srcq, r6
    punpcklbw            m0, m0
    psraw                m0, 8 ; sign-extend
    test          dword r7m, 0x800
    jnz .v_12bpc
    psllw                m0, 2
.v_12bpc:
    pshufd               m8, m0, q0000
    pshufd               m9, m0, q1111
    pshufd              m10, m0, q2222
    pshufd              m11, m0, q3333
    cmp                  wd, 4
    jg .v_w8
.v_w4:
    movq                xm1, [srcq+strideq*0]
    vpbroadcastq         m0, [srcq+strideq*1]
    vpbroadcastq         m2, [srcq+strideq*2]
    vpbroadcastq         m4, [srcq+r6       ]
    lea                srcq, [srcq+strideq*4]
    vpbroadcastq         m3, [srcq+strideq*0]
    vpbroadcastq         m5, [srcq+strideq*1]
    vpblendd             m1, m0, 0x30
    vpblendd             m0, m2, 0x30
    punpcklwd            m1, m0      ; 01 12
    vpbroadcastq         m0, [srcq+strideq*2]
    add                srcq, r6
    vpblendd             m2, m4, 0x30
    vpblendd             m4, m3, 0x30
    punpcklwd            m2, m4      ; 23 34
    vpblendd             m3, m5, 0x30
    vpblendd             m5, m0, 0x30
    punpcklwd            m3, m5      ; 45 56
.v_w4_loop:
    vpbroadcastq         m4, [srcq+strideq*0]
    pmaddwd              m5, m8, m1  ; a0 b0
    mova                 m1, m2
    pmaddwd              m2, m9      ; a1 b1
    paddd                m5, m7
    paddd                m5, m2
    mova                 m2, m3
    pmaddwd              m3, m10     ; a2 b2
    paddd                m5, m3
    vpblendd             m3, m0, m4, 0x30
    vpbroadcastq         m0, [srcq+strideq*1]
    lea                srcq, [srcq+strideq*2]
    vpblendd             m4, m0, 0x30
    punpcklwd            m3, m4      ; 67 78
    pmaddwd              m4, m11, m3 ; a3 b3
    paddd                m5, m4
    psrad                m5, 4
    vextracti128        xm4, m5, 1
    packssdw            xm5, xm4
    mova             [tmpq], xm5
    add                tmpq, 16
    sub                  hd, 2
    jg .v_w4_loop
    RET
.v_w8:
%if WIN64
    push                 r8
%endif
    mov                 r8d, wd
    shl                  wd, 5
    mov                  r5, srcq
    mov                  r7, tmpq
    lea                  wd, [hq+wq-256]
.v_w8_loop0:
    vbroadcasti128       m4, [srcq+strideq*0]
    vbroadcasti128       m5, [srcq+strideq*1]
    vbroadcasti128       m0, [srcq+r6       ]
    vbroadcasti128       m6, [srcq+strideq*2]
    lea                srcq, [srcq+strideq*4]
    vbroadcasti128       m1, [srcq+strideq*0]
    vbroadcasti128       m2, [srcq+strideq*1]
    vbroadcasti128       m3, [srcq+strideq*2]
    add                srcq, r6
    shufpd               m4, m0, 0x0c
    shufpd               m5, m1, 0x0c
    punpcklwd            m1, m4, m5 ; 01
    punpckhwd            m4, m5     ; 34
    shufpd               m6, m2, 0x0c
    punpcklwd            m2, m5, m6 ; 12
    punpckhwd            m5, m6     ; 45
    shufpd               m0, m3, 0x0c
    punpcklwd            m3, m6, m0 ; 23
    punpckhwd            m6, m0     ; 56
.v_w8_loop:
    vbroadcasti128      m14, [srcq+strideq*0]
    pmaddwd             m12, m8, m1  ; a0
    pmaddwd             m13, m8, m2  ; b0
    mova                 m1, m3
    mova                 m2, m4
    pmaddwd              m3, m9      ; a1
    pmaddwd              m4, m9      ; b1
    paddd               m12, m7
    paddd               m13, m7
    paddd               m12, m3
    paddd               m13, m4
    mova                 m3, m5
    mova                 m4, m6
    pmaddwd              m5, m10     ; a2
    pmaddwd              m6, m10     ; b2
    paddd               m12, m5
    vbroadcasti128       m5, [srcq+strideq*1]
    lea                srcq, [srcq+strideq*2]
    paddd               m13, m6
    shufpd               m6, m0, m14, 0x0d
    shufpd               m0, m14, m5, 0x0c
    punpcklwd            m5, m6, m0  ; 67
    punpckhwd            m6, m0      ; 78
    pmaddwd             m14, m11, m5 ; a3
    paddd               m12, m14
    pmaddwd             m14, m11, m6 ; b3
    paddd               m13, m14
    psrad               m12, 4
    psrad               m13, 4
    packssdw            m12, m13
    vpermq              m12, m12, q3120
    mova         [tmpq+r8*0], xm12
    vextracti128 [tmpq+r8*2], m12, 1
    lea                tmpq, [tmpq+r8*4]
    sub                  hd, 2
    jg .v_w8_loop
    add                  r5, 16
    add                  r7, 16
    movzx                hd, wb
    mov                srcq, r5
    mov                tmpq, r7
    sub                  wd, 1<<8
    jg .v_w8_loop0
%if WIN64
    pop                  r8
%endif
    RET
.hv:
    WIN64_SPILL_XMM      16
    vpbroadcastd        m15, [prep_8tap_2d_rnd]
    cmp                  wd, 4
    jg .hv_w8
    movzx               mxd, mxb
    vpbroadcastd         m0, [base+subpel_filters+mxq*8+2]
    movzx               mxd, myb
    shr                 myd, 16
    cmp                  hd, 4
    cmovle              myd, mxd
    vpbroadcastq         m1, [base+subpel_filters+myq*8]
    lea                  r6, [strideq*3]
    sub                srcq, 2
    sub                srcq, r6
    pxor                 m7, m7
    punpcklbw            m7, m0
    punpcklbw            m1, m1
    psraw                m7, 4
    psraw                m1, 8
    test          dword r7m, 0x800
    jz .hv_w4_10bit
    psraw                m7, 2
.hv_w4_10bit:
    pshufd              m11, m1, q0000
    pshufd              m12, m1, q1111
    pshufd              m13, m1, q2222
    pshufd              m14, m1, q3333
.hv_w4:
    vbroadcasti128       m9, [subpel_h_shufA]
    vbroadcasti128      m10, [subpel_h_shufB]
    pshufd               m8, m7, q1111
    pshufd               m7, m7, q0000
    movu                xm1, [srcq+strideq*0]
    vinserti128          m1, [srcq+strideq*1], 1     ; 0 1
    vbroadcasti128       m0, [srcq+r6       ]
    vinserti128          m2, m0, [srcq+strideq*2], 0 ; 2 3
    lea                srcq, [srcq+strideq*4]
    vinserti128          m0, [srcq+strideq*0], 1     ; 3 4
    movu                xm3, [srcq+strideq*1]
    vinserti128          m3, [srcq+strideq*2], 1     ; 5 6
    add                srcq, r6
    pshufb               m4, m1, m9
    pshufb               m1, m10
    pmaddwd              m4, m7
    pmaddwd              m1, m8
    pshufb               m5, m2, m9
    pshufb               m2, m10
    pmaddwd              m5, m7
    pmaddwd              m2, m8
    paddd                m4, m15
    paddd                m1, m4
    pshufb               m4, m0, m9
    pshufb               m0, m10
    pmaddwd              m4, m7
    pmaddwd              m0, m8
    paddd                m5, m15
    paddd                m2, m5
    pshufb               m5, m3, m9
    pshufb               m3, m10
    pmaddwd              m5, m7
    pmaddwd              m3, m8
    paddd                m4, m15
    paddd                m4, m0
    paddd                m5, m15
    paddd                m5, m3
    vperm2i128           m0, m1, m2, 0x21
    psrld                m1, 6
    psrld                m2, 6
    vperm2i128           m3, m4, m5, 0x21
    pslld                m4, 10
    pslld                m5, 10
    pblendw              m2, m4, 0xaa ; 23 34
    pslld                m0, 10
    pblendw              m1, m0, 0xaa ; 01 12
    psrld                m3, 6
    pblendw              m3, m5, 0xaa ; 45 56
    psrad                m0, m5, 16
.hv_w4_loop:
    movu                xm4, [srcq+strideq*0]
    vinserti128          m4, [srcq+strideq*1], 1
    lea                srcq, [srcq+strideq*2]
    pmaddwd              m5, m11, m1   ; a0 b0
    mova                 m1, m2
    pmaddwd              m2, m12       ; a1 b1
    paddd                m5, m15
    paddd                m5, m2
    mova                 m2, m3
    pmaddwd              m3, m13       ; a2 b2
    paddd                m5, m3
    pshufb               m3, m4, m9
    pshufb               m4, m10
    pmaddwd              m3, m7
    pmaddwd              m4, m8
    paddd                m3, m15
    paddd                m4, m3
    psrad                m4, 6
    packssdw             m0, m4        ; _ 7 6 8
    vpermq               m3, m0, q1122 ; _ 6 _ 7
    punpckhwd            m3, m0        ; 67 78
    mova                 m0, m4
    pmaddwd              m4, m14, m3   ; a3 b3
    paddd                m4, m5
    psrad                m4, 6
    vextracti128        xm5, m4, 1
    packssdw            xm4, xm5
    mova             [tmpq], xm4
    add                tmpq, 16
    sub                  hd, 2
    jg .hv_w4_loop
    RET
.hv_w8:
    shr                 mxd, 16
    vpbroadcastq         m2, [base+subpel_filters+mxq*8]
    movzx               mxd, myb
    shr                 myd, 16
    cmp                  hd, 4
    cmovle              myd, mxd
    pmovsxbw            xm1, [base+subpel_filters+myq*8]
%if WIN64
    PUSH                 r8
%endif
    mov                 r8d, wd
    shl                  wd, 5
    lea                  r6, [strideq*3]
    sub                srcq, 6
    sub                srcq, r6
    mov                  r5, srcq
    mov                  r7, tmpq
    lea                  wd, [hq+wq-256]
    pxor                 m0, m0
    punpcklbw            m0, m2
    mova            [v_mul], xm1
    psraw                m0, 4
    test          dword r7m, 0x800
    jz .hv_w8_10bit
    psraw                m0, 2
.hv_w8_10bit:
    pshufd              m11, m0, q0000
    pshufd              m12, m0, q1111
    pshufd              m13, m0, q2222
    pshufd              m14, m0, q3333
.hv_w8_loop0:
%macro PREP_8TAP_HV_H 3 ; dst/src+0, src+8, src+16
    pshufb               m2, m%1, m9   ; 2 3 3 4 4 5 5 6
    pshufb              m%1, m8        ; 0 1 1 2 2 3 3 4
    pmaddwd              m3, m12, m2
    pmaddwd             m%1, m11
    pshufb              m%2, m9        ; 6 7 7 8 8 9 9 a
    shufpd               m2, m%2, 0x05 ; 4 5 5 6 6 7 7 8
    paddd                m3, m15
    paddd               m%1, m3
    pmaddwd              m3, m14, m%2
    paddd               m%1, m3
    pmaddwd              m3, m13, m2
    pshufb              m%3, m9        ; a b b c c d d e
    pmaddwd              m2, m11
    paddd               m%1, m3
    pmaddwd              m3, m12, m%2
    shufpd              m%2, m%3, 0x05 ; 8 9 9 a a b b c
    pmaddwd             m%3, m14
    pmaddwd             m%2, m13
    paddd                m2, m15
    paddd                m2, m3
    paddd                m2, m%3
    paddd                m2, m%2
    psrad               m%1, 6
    psrad                m2, 6
    packssdw            m%1, m2
%endmacro
    movu                xm4, [srcq+r6       + 0]
    vbroadcasti128       m8, [subpel_h_shufA]
    movu                xm6, [srcq+r6       + 8]
    vbroadcasti128       m9, [subpel_h_shufB]
    movu                xm0, [srcq+r6       +16]
    movu                xm5, [srcq+strideq*0+ 0]
    vinserti128          m5, [srcq+strideq*4+ 0], 1
    movu                xm1, [srcq+strideq*0+16]
    vinserti128          m1, [srcq+strideq*4+16], 1
    shufpd               m7, m5, m1, 0x05
    INIT_XMM avx2
    PREP_8TAP_HV_H        4, 6, 0    ; 3
    INIT_YMM avx2
    PREP_8TAP_HV_H        5, 7, 1    ; 0 4
    movu                xm0, [srcq+strideq*2+ 0]
    vinserti128          m0, [srcq+r6     *2+ 0], 1
    movu                xm1, [srcq+strideq*2+16]
    vinserti128          m1, [srcq+r6     *2+16], 1
    shufpd               m7, m0, m1, 0x05
    PREP_8TAP_HV_H        0, 7, 1    ; 2 6
    movu                xm6, [srcq+strideq*1+ 0]
    movu                xm1, [srcq+strideq*1+16]
    lea                srcq, [srcq+strideq*4]
    vinserti128          m6, [srcq+strideq*1+ 0], 1
    vinserti128          m1, [srcq+strideq*1+16], 1
    add                srcq, r6
    shufpd               m7, m6, m1, 0x05
    PREP_8TAP_HV_H        6, 7, 1    ; 1 5
    vpermq               m4, m4, q1100
    vpermq               m5, m5, q3120
    vpermq               m6, m6, q3120
    vpermq               m7, m0, q3120
    punpcklwd            m3, m7, m4  ; 23
    punpckhwd            m4, m5      ; 34
    punpcklwd            m1, m5, m6  ; 01
    punpckhwd            m5, m6      ; 45
    punpcklwd            m2, m6, m7  ; 12
    punpckhwd            m6, m7      ; 56
.hv_w8_loop:
    vpbroadcastd         m9, [v_mul+4*0]
    vpbroadcastd         m7, [v_mul+4*1]
    vpbroadcastd        m10, [v_mul+4*2]
    pmaddwd              m8, m9, m1  ; a0
    pmaddwd              m9, m2      ; b0
    mova                 m1, m3
    mova                 m2, m4
    pmaddwd              m3, m7      ; a1
    pmaddwd              m4, m7      ; b1
    paddd                m8, m15
    paddd                m9, m15
    paddd                m8, m3
    paddd                m9, m4
    mova                 m3, m5
    mova                 m4, m6
    pmaddwd              m5, m10     ; a2
    pmaddwd              m6, m10     ; b2
    paddd                m8, m5
    paddd                m9, m6
    movu                xm5, [srcq+strideq*0]
    vinserti128          m5, [srcq+strideq*1], 1
    vbroadcasti128       m7, [subpel_h_shufA]
    vbroadcasti128      m10, [subpel_h_shufB]
    movu                xm6, [srcq+strideq*0+16]
    vinserti128          m6, [srcq+strideq*1+16], 1
    vextracti128     [tmpq], m0, 1
    pshufb               m0, m5, m7  ; 01
    pshufb               m5, m10     ; 23
    pmaddwd              m0, m11
    pmaddwd              m5, m12
    paddd                m0, m15
    paddd                m0, m5
    pshufb               m5, m6, m7  ; 89
    pshufb               m6, m10     ; ab
    pmaddwd              m5, m13
    pmaddwd              m6, m14
    paddd                m5, m15
    paddd                m6, m5
    movu                xm5, [srcq+strideq*0+8]
    vinserti128          m5, [srcq+strideq*1+8], 1
    lea                srcq, [srcq+strideq*2]
    pshufb               m7, m5, m7
    pshufb               m5, m10
    pmaddwd             m10, m13, m7
    pmaddwd              m7, m11
    paddd                m0, m10
    paddd                m6, m7
    pmaddwd              m7, m14, m5
    pmaddwd              m5, m12
    paddd                m0, m7
    paddd                m5, m6
    vbroadcasti128       m6, [tmpq]
    vpbroadcastd        m10, [v_mul+4*3]
    psrad                m0, 6
    psrad                m5, 6
    packssdw             m0, m5
    vpermq               m7, m0, q3120 ; 7 8
    shufpd               m6, m7, 0x04  ; 6 7
    punpcklwd            m5, m6, m7    ; 67
    punpckhwd            m6, m7        ; 78
    pmaddwd              m7, m10, m5   ; a3
    pmaddwd             m10, m6        ; b3
    paddd                m7, m8
    paddd                m9, m10
    psrad                m7, 6
    psrad                m9, 6
    packssdw             m7, m9
    vpermq               m7, m7, q3120
    mova         [tmpq+r8*0], xm7
    vextracti128 [tmpq+r8*2], m7, 1
    lea                tmpq, [tmpq+r8*4]
    sub                  hd, 2
    jg .hv_w8_loop
    add                  r5, 16
    add                  r7, 16
    movzx                hd, wb
    mov                srcq, r5
    mov                tmpq, r7
    sub                  wd, 1<<8
    jg .hv_w8_loop0
%if WIN64
    POP                  r8
%endif
    RET

%macro movifprep 2
 %if isprep
    mov %1, %2
 %endif
%endmacro

%macro REMAP_REG 2
 %xdefine r%1  r%2
 %xdefine r%1q r%2q
 %xdefine r%1d r%2d
%endmacro

%macro MCT_8TAP_SCALED_REMAP_REGS_TO_PREV 0
 %if isprep
  %xdefine r14_save r14
  %assign %%i 14
  %rep 14
   %assign %%j %%i-1
   REMAP_REG %%i, %%j
   %assign %%i %%i-1
  %endrep
 %endif
%endmacro

%macro MCT_8TAP_SCALED_REMAP_REGS_TO_DEFAULT 0
 %if isprep
  %assign %%i 1
  %rep 13
   %assign %%j %%i+1
   REMAP_REG %%i, %%j
   %assign %%i %%i+1
  %endrep
  %xdefine r14 r14_save
  %undef r14_save
 %endif
%endmacro

%macro MC_8TAP_SCALED_RET 0-1 1 ; leave_mapping_unchanged
    MCT_8TAP_SCALED_REMAP_REGS_TO_DEFAULT
    RET
 %if %1
    MCT_8TAP_SCALED_REMAP_REGS_TO_PREV
 %endif
%endmacro

%macro MC_8TAP_SCALED_H 8-9 0 ; dst, tmp[0-6], load_hrnd
    movu               xm%1, [srcq+ r4*2]
    movu               xm%2, [srcq+ r6*2]
    movu               xm%3, [srcq+ r7*2]
    movu               xm%4, [srcq+ r9*2]
    vinserti128         m%1, [srcq+r10*2], 1
    vinserti128         m%2, [srcq+r11*2], 1
    vinserti128         m%3, [srcq+r13*2], 1
    vinserti128         m%4, [srcq+ rX*2], 1
    add                srcq, ssq
    movu               xm%5, [srcq+ r4*2]
    movu               xm%6, [srcq+ r6*2]
    movu               xm%7, [srcq+ r7*2]
    movu               xm%8, [srcq+ r9*2]
    vinserti128         m%5, [srcq+r10*2], 1
    vinserti128         m%6, [srcq+r11*2], 1
    vinserti128         m%7, [srcq+r13*2], 1
    vinserti128         m%8, [srcq+ rX*2], 1
    add                srcq, ssq
    pmaddwd             m%1, m12
    pmaddwd             m%2, m13
    pmaddwd             m%3, m14
    pmaddwd             m%4, m15
    pmaddwd             m%5, m12
    pmaddwd             m%6, m13
    pmaddwd             m%7, m14
    pmaddwd             m%8, m15
    phaddd              m%1, m%2
 %if %9
    mova                m10, [rsp+0x00]
 %endif
    phaddd              m%3, m%4
    phaddd              m%5, m%6
    phaddd              m%7, m%8
    phaddd              m%1, m%3
    phaddd              m%5, m%7
    paddd               m%1, m10
    paddd               m%5, m10
    psrad               m%1, xm11
    psrad               m%5, xm11
    packssdw            m%1, m%5
%endmacro

%macro MC_8TAP_SCALED 1
%ifidn %1, put
 %assign isput  1
 %assign isprep 0
cglobal put_8tap_scaled_16bpc, 4, 14, 16, 0xe0, dst, ds, src, ss, w, h, mx, my, dx, dy, pxmax
 %xdefine base_reg r12
    mov                 r7d, pxmaxm
%else
 %assign isput  0
 %assign isprep 1
cglobal prep_8tap_scaled_16bpc, 4, 14, 16, 0xe0, tmp, src, ss, w, h, mx, my, dx, dy, pxmax
  %define tmp_stridem qword [rsp+0xd0]
 %xdefine base_reg r11
%endif
    lea            base_reg, [%1_8tap_scaled_16bpc_avx2]
%define base base_reg-%1_8tap_scaled_16bpc_avx2
    tzcnt                wd, wm
    vpbroadcastd         m8, dxm
%if isprep && UNIX64
    movd               xm10, mxd
    vpbroadcastd        m10, xm10
    mov                 r5d, t0d
 DECLARE_REG_TMP 5, 7
    mov                 r6d, pxmaxm
%else
    vpbroadcastd        m10, mxm
 %if isput
    vpbroadcastw        m11, pxmaxm
 %else
    mov                 r6d, pxmaxm
 %endif
%endif
    mov                 dyd, dym
%if isput
 %if WIN64
    mov                 r8d, hm
  DEFINE_ARGS dst, ds, src, ss, w, _, _, my, h, dy, ss3
  %define hm r5m
  %define dxm r8m
 %else
  DEFINE_ARGS dst, ds, src, ss, w, h, _, my, dx, dy, ss3
  %define hm r6m
 %endif
 %define dsm [rsp+0x98]
 %define rX r1
 %define rXd r1d
%else ; prep
 %if WIN64
    mov                 r7d, hm
  DEFINE_ARGS tmp, src, ss, w, _, _, my, h, dy, ss3
  %define hm r4m
  %define dxm r7m
 %else
  DEFINE_ARGS tmp, src, ss, w, h, _, my, dx, dy, ss3
  %define hm [rsp+0x98]
 %endif
 MCT_8TAP_SCALED_REMAP_REGS_TO_PREV
 %define rX r14
 %define rXd r14d
%endif
    shr                 r7d, 11
    vpbroadcastd         m6, [base+pd_0x3ff]
    vpbroadcastd        m12, [base+s_8tap_h_rnd+r7*4]
    movd                xm7, [base+s_8tap_h_sh+r7*4]
%if isput
    vpbroadcastd        m13, [base+put_s_8tap_v_rnd+r7*4]
    pinsrd              xm7, [base+put_s_8tap_v_sh+r7*4], 2
%else
    vpbroadcastd        m13, [base+pd_m524256]
%endif
    pxor                 m9, m9
    lea                ss3q, [ssq*3]
    movzx               r7d, t1b
    shr                 t1d, 16
    cmp                  hd, 6
    cmovs               t1d, r7d
    sub                srcq, ss3q
    cmp                 dyd, 1024
    je .dy1
    cmp                 dyd, 2048
    je .dy2
    movzx                wd, word [base+%1_8tap_scaled_avx2_table+wq*2]
    add                  wq, base_reg
    jmp                  wq
%if isput
.w2:
    mov                 myd, mym
    movzx               t0d, t0b
    sub                srcq, 2
    movd               xm15, t0d
    punpckldq            m8, m9, m8
    paddd               m10, m8 ; mx+dx*[0,1]
    vpbroadcastd       xm14, [base+pq_0x40000000+2]
    vpbroadcastd       xm15, xm15
    pand                xm8, xm10, xm6
    psrld               xm8, 6
    paddd              xm15, xm8
    movd                r4d, xm15
    pextrd              r6d, xm15, 1
    vbroadcasti128       m5, [base+bdct_lb_q]
    vbroadcasti128       m6, [base+subpel_s_shuf2]
    vpbroadcastd       xm15, [base+subpel_filters+r4*8+2]
    vpbroadcastd        xm4, [base+subpel_filters+r6*8+2]
    pcmpeqd             xm8, xm9
    psrld               m10, 10
    paddd               m10, m10
    movu                xm0, [srcq+ssq*0]
    movu                xm1, [srcq+ssq*1]
    movu                xm2, [srcq+ssq*2]
    movu                xm3, [srcq+ss3q ]
    lea                srcq, [srcq+ssq*4]
    pshufb              m10, m5
    paddb               m10, m6
    vpblendd           xm15, xm4, 0xa
    pblendvb           xm15, xm14, xm8
    pmovsxbw            m15, xm15
    vinserti128          m0, [srcq+ssq*0], 1 ; 0 4
    vinserti128          m1, [srcq+ssq*1], 1 ; 1 5
    vinserti128          m2, [srcq+ssq*2], 1 ; 2 6
    vinserti128          m3, [srcq+ss3q ], 1 ; 3 7
    lea                srcq, [srcq+ssq*4]
    REPX    {pshufb x, m10}, m0, m1, m2, m3
    REPX   {pmaddwd x, m15}, m0, m1, m2, m3
    phaddd               m0, m1
    phaddd               m2, m3
    paddd                m0, m12
    paddd                m2, m12
    psrad                m0, xm7
    psrad                m2, xm7
    packssdw             m0, m2             ; 0 1 2 3  4 5 6 7
    vextracti128        xm1, m0, 1
    palignr             xm2, xm1, xm0, 4    ; 1 2 3 4
    punpcklwd           xm3, xm0, xm2       ; 01 12
    punpckhwd           xm0, xm2            ; 23 34
    pshufd              xm4, xm1, q0321     ; 5 6 7 _
    punpcklwd           xm2, xm1, xm4       ; 45 56
    punpckhwd           xm4, xm1, xm4       ; 67 __
.w2_loop:
    and                 myd, 0x3ff
    mov                 r6d, 64 << 24
    mov                 r4d, myd
    shr                 r4d, 6
    lea                 r4d, [t1+r4]
    cmovnz              r6q, [base+subpel_filters+r4*8]
    movq               xm14, r6q
    pmovsxbw           xm14, xm14
    pshufd              xm8, xm14, q0000
    pshufd              xm9, xm14, q1111
    pmaddwd             xm5, xm3, xm8
    pmaddwd             xm6, xm0, xm9
    pshufd              xm8, xm14, q2222
    pshufd             xm14, xm14, q3333
    paddd               xm5, xm6
    pmaddwd             xm6, xm2, xm8
    pmaddwd             xm8, xm4, xm14
    psrldq              xm9, xm7, 8
    paddd               xm5, xm6
    paddd               xm5, xm13
    paddd               xm5, xm8
    psrad               xm5, xm9
    packusdw            xm5, xm5
    pminsw              xm5, xm11
    movd             [dstq], xm5
    add                dstq, dsq
    dec                  hd
    jz .ret
    add                 myd, dyd
    test                myd, ~0x3ff
    jz .w2_loop
    movu                xm5, [srcq]
    test                myd, 0x400
    jz .w2_skip_line
    add                srcq, ssq
    shufps              xm3, xm0, q1032     ; 01 12
    shufps              xm0, xm2, q1032     ; 23 34
    shufps              xm2, xm4, q1032     ; 45 56
    pshufb              xm5, xm10
    pmaddwd             xm5, xm15
    phaddd              xm5, xm5
    paddd               xm5, xm12
    psrad               xm5, xm7
    packssdw            xm5, xm5
    palignr             xm1, xm5, xm1, 12
    punpcklqdq          xm1, xm1            ; 6 7 6 7
    punpcklwd           xm4, xm1, xm5       ; 67 __
    jmp .w2_loop
.w2_skip_line:
    movu                xm6, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    mova                xm3, xm0            ; 01 12
    mova                xm0, xm2            ; 23 34
    pshufb              xm5, xm10
    pshufb              xm6, xm10
    pmaddwd             xm5, xm15
    pmaddwd             xm6, xm15
    phaddd              xm5, xm6
    paddd               xm5, xm12
    psrad               xm5, xm7
    packssdw            xm5, xm5            ; 6 7 6 7
    palignr             xm1, xm5, xm1, 8    ; 4 5 6 7
    pshufd              xm5, xm1, q0321     ; 5 6 7 _
    punpcklwd           xm2, xm1, xm5       ; 45 56
    punpckhwd           xm4, xm1, xm5       ; 67 __
    jmp .w2_loop
%endif
.w4:
    mov                 myd, mym
    mova         [rsp+0x00], m12
%if isput
    mova         [rsp+0x20], xm13
%else
    SWAP                m11, m13
%endif
    mova         [rsp+0x30], xm7
    vbroadcasti128       m7, [base+rescale_mul]
    movzx               t0d, t0b
    sub                srcq, 2
    movd               xm15, t0d
    pmaddwd              m8, m7
    vpbroadcastq         m2, [base+pq_0x40000000+1]
    vpbroadcastd       xm15, xm15
    SWAP                m13, m10
    paddd               m13, m8 ; mx+dx*[0-3]
    pand                 m6, m13
    psrld                m6, 6
    paddd              xm15, xm6
    movd                r4d, xm15
    pextrd              r6d, xm15, 1
    pextrd             r11d, xm15, 2
    pextrd             r13d, xm15, 3
    vbroadcasti128       m5, [base+bdct_lb_q+ 0]
    vbroadcasti128       m1, [base+bdct_lb_q+16]
    vbroadcasti128       m0, [base+subpel_s_shuf2]
    vpbroadcastd       xm14, [base+subpel_filters+r4*8+2]
    vpbroadcastd        xm7, [base+subpel_filters+r6*8+2]
    vpbroadcastd       xm15, [base+subpel_filters+r11*8+2]
    vpbroadcastd        xm8, [base+subpel_filters+r13*8+2]
    pcmpeqd              m6, m9
    punpckldq           m10, m6, m6
    punpckhdq            m6, m6
    psrld               m13, 10
    paddd               m13, m13
    vpblendd           xm14, xm7, 0xa
    vpblendd           xm15, xm8, 0xa
    pmovsxbw            m14, xm14
    pmovsxbw            m15, xm15
    pblendvb            m14, m2, m10
    pblendvb            m15, m2, m6
    pextrd               r4, xm13, 2
    pshufb              m12, m13, m5
    pshufb              m13, m1
    lea                  r6, [r4+ssq*1]
    lea                 r11, [r4+ssq*2]
    lea                 r13, [r4+ss3q ]
    movu                xm7, [srcq+ssq*0]
    movu                xm9, [srcq+ssq*1]
    movu                xm8, [srcq+ssq*2]
    movu               xm10, [srcq+ss3q ]
    movu                xm1, [srcq+r4   ]
    movu                xm3, [srcq+r6   ]
    movu                xm2, [srcq+r11  ]
    movu                xm4, [srcq+r13  ]
    lea                srcq, [srcq+ssq*4]
    vinserti128          m7, [srcq+ssq*0], 1
    vinserti128          m9, [srcq+ssq*1], 1
    vinserti128          m8, [srcq+ssq*2], 1
    vinserti128         m10, [srcq+ss3q ], 1
    vinserti128          m1, [srcq+r4   ], 1
    vinserti128          m3, [srcq+r6   ], 1
    vinserti128          m2, [srcq+r11  ], 1
    vinserti128          m4, [srcq+r13  ], 1
    lea                srcq, [srcq+ssq*4]
    vpbroadcastb         m5, xm13
    psubb               m13, m5
    paddb               m12, m0
    paddb               m13, m0
    REPX    {pshufb x, m12}, m7, m9, m8, m10
    REPX   {pmaddwd x, m14}, m7, m9, m8, m10
    REPX    {pshufb x, m13}, m1, m2, m3, m4
    REPX   {pmaddwd x, m15}, m1, m2, m3, m4
    mova                 m5, [rsp+0x00]
    movd                xm6, [rsp+0x30]
    phaddd               m7, m1
    phaddd               m9, m3
    phaddd               m8, m2
    phaddd              m10, m4
    REPX      {paddd x, m5}, m7, m9, m8, m10
    REPX     {psrad x, xm6}, m7, m9, m8, m10
    packssdw             m7, m9                 ; 0 1  4 5
    packssdw             m8, m10                ; 2 3  6 7
    vextracti128        xm9, m7, 1              ; 4 5
    vextracti128        xm3, m8, 1              ; 6 7
    shufps              xm4, xm7, xm8, q1032    ; 1 2
    shufps              xm5, xm8, xm9, q1032    ; 3 4
    shufps              xm6, xm9, xm3, q1032    ; 5 6
    psrldq             xm10, xm3, 8             ; 7 _
    punpcklwd           xm0, xm7, xm4   ; 01
    punpckhwd           xm7, xm4        ; 12
    punpcklwd           xm1, xm8, xm5   ; 23
    punpckhwd           xm8, xm5        ; 34
    punpcklwd           xm2, xm9, xm6   ; 45
    punpckhwd           xm9, xm6        ; 56
    punpcklwd           xm3, xm10       ; 67
    mova         [rsp+0x40], xm7
    mova         [rsp+0x50], xm8
    mova         [rsp+0x60], xm9
.w4_loop:
    and                 myd, 0x3ff
    mov                r11d, 64 << 24
    mov                r13d, myd
    shr                r13d, 6
    lea                r13d, [t1+r13]
    cmovnz             r11q, [base+subpel_filters+r13*8]
    movq                xm9, r11q
    pmovsxbw            xm9, xm9
    pshufd              xm7, xm9, q0000
    pshufd              xm8, xm9, q1111
    pmaddwd             xm4, xm0, xm7
    pmaddwd             xm5, xm1, xm8
    pshufd              xm7, xm9, q2222
    pshufd              xm9, xm9, q3333
    pmaddwd             xm6, xm2, xm7
    pmaddwd             xm8, xm3, xm9
%if isput
    mova                xm7, [rsp+0x20]
    movd                xm9, [rsp+0x38]
%else
    SWAP                 m7, m11
%endif
    paddd               xm4, xm5
    paddd               xm6, xm8
    paddd               xm4, xm6
    paddd               xm4, xm7
%if isput
    psrad               xm4, xm9
    packusdw            xm4, xm4
    pminuw              xm4, xm11
    movq             [dstq], xm4
    add                dstq, dsq
%else
    SWAP                m11, m7
    psrad               xm4, 6
    packssdw            xm4, xm4
    movq             [tmpq], xm4
    add                tmpq, 8
%endif
    dec                  hd
    jz .ret
    add                 myd, dyd
    test                myd, ~0x3ff
    jz .w4_loop
    mova                xm8, [rsp+0x00]
    movd                xm9, [rsp+0x30]
    movu                xm4, [srcq]
    movu                xm5, [srcq+r4]
    test                myd, 0x400
    jz .w4_skip_line
    mova                xm0, [rsp+0x40]
    mova         [rsp+0x40], xm1
    mova                xm1, [rsp+0x50]
    mova         [rsp+0x50], xm2
    mova                xm2, [rsp+0x60]
    mova         [rsp+0x60], xm3
    pshufb              xm4, xm12
    pshufb              xm5, xm13
    pmaddwd             xm4, xm14
    pmaddwd             xm5, xm15
    phaddd              xm4, xm5
    paddd               xm4, xm8
    psrad               xm4, xm9
    packssdw            xm4, xm4
    punpcklwd           xm3, xm10, xm4
    mova               xm10, xm4
    add                srcq, ssq
    jmp .w4_loop
.w4_skip_line:
    movu                xm6, [srcq+ssq*1]
    movu                xm7, [srcq+r6]
    movu                 m0, [rsp+0x50]
    pshufb              xm4, xm12
    pshufb              xm6, xm12
    pshufb              xm5, xm13
    pshufb              xm7, xm13
    pmaddwd             xm4, xm14
    pmaddwd             xm6, xm14
    pmaddwd             xm5, xm15
    pmaddwd             xm7, xm15
    mova         [rsp+0x40], m0
    phaddd              xm4, xm5
    phaddd              xm6, xm7
    paddd               xm4, xm8
    paddd               xm6, xm8
    psrad               xm4, xm9
    psrad               xm6, xm9
    packssdw            xm4, xm6
    punpcklwd           xm9, xm10, xm4
    mova         [rsp+0x60], xm9
    psrldq             xm10, xm4, 8
    mova                xm0, xm1
    mova                xm1, xm2
    mova                xm2, xm3
    punpcklwd           xm3, xm4, xm10
    lea                srcq, [srcq+ssq*2]
    jmp .w4_loop
    SWAP                m10, m13
%if isprep
    SWAP                m13, m11
%endif
.w8:
    mov    dword [rsp+0x80], 1
    movifprep   tmp_stridem, 16
    jmp .w_start
.w16:
    mov    dword [rsp+0x80], 2
    movifprep   tmp_stridem, 32
    jmp .w_start
.w32:
    mov    dword [rsp+0x80], 4
    movifprep   tmp_stridem, 64
    jmp .w_start
.w64:
    mov    dword [rsp+0x80], 8
    movifprep   tmp_stridem, 128
    jmp .w_start
.w128:
    mov    dword [rsp+0x80], 16
    movifprep   tmp_stridem, 256
.w_start:
    SWAP                m10, m12, m1
    SWAP                m11, m7
    ; m1=mx, m7=pxmax, m10=h_rnd, m11=h_sh, m12=free
%if isput
    movifnidn           dsm, dsq
    mova         [rsp+0xb0], xm7
%endif
    mova         [rsp+0x00], m10
    mova         [rsp+0x20], m13
    shr                 t0d, 16
    sub                srcq, 6
    pmaddwd              m8, [base+rescale_mul2]
    movd               xm15, t0d
    mov          [rsp+0x84], t0d
    mov          [rsp+0x88], srcq
    mov          [rsp+0x90], r0q ; dstq / tmpq
%if UNIX64
    mov                  hm, hd
%endif
    shl           dword dxm, 3 ; dx*8
    vpbroadcastd        m15, xm15
    paddd                m1, m8 ; mx+dx*[0-7]
    jmp .hloop
.hloop_prep:
    dec    dword [rsp+0x80]
    jz .ret
    add    qword [rsp+0x90], 16
    mov                  hd, hm
    vpbroadcastd         m8, dxm
    vpbroadcastd         m6, [base+pd_0x3ff]
    paddd                m1, m8, [rsp+0x40]
    vpbroadcastd        m15, [rsp+0x84]
    pxor                 m9, m9
    mov                srcq, [rsp+0x88]
    mov                 r0q, [rsp+0x90] ; dstq / tmpq
.hloop:
    vpbroadcastq        xm2, [base+pq_0x40000000]
    pand                 m5, m1, m6
    psrld                m5, 6
    paddd               m15, m5
    pcmpeqd              m5, m9
    vextracti128        xm7, m15, 1
    movq                 r6, xm15
    pextrq               r9, xm15, 1
    movq                r11, xm7
    pextrq               rX, xm7, 1
    mov                 r4d, r6d
    shr                  r6, 32
    mov                 r7d, r9d
    shr                  r9, 32
    mov                r10d, r11d
    shr                 r11, 32
    mov                r13d, rXd
    shr                  rX, 32
    mova         [rsp+0x40], m1
    movq               xm12, [base+subpel_filters+ r4*8]
    movq               xm13, [base+subpel_filters+ r6*8]
    movhps             xm12, [base+subpel_filters+ r7*8]
    movhps             xm13, [base+subpel_filters+ r9*8]
    movq               xm14, [base+subpel_filters+r10*8]
    movq               xm15, [base+subpel_filters+r11*8]
    movhps             xm14, [base+subpel_filters+r13*8]
    movhps             xm15, [base+subpel_filters+ rX*8]
    psrld                m1, 10
    vextracti128        xm7, m1, 1
    vextracti128        xm6, m5, 1
    movq         [rsp+0xa0], xm1
    movq         [rsp+0xa8], xm7
    movq                 r6, xm1
    pextrq              r11, xm1, 1
    movq                 r9, xm7
    pextrq               rX, xm7, 1
    mov                 r4d, r6d
    shr                  r6, 32
    mov                r10d, r11d
    shr                 r11, 32
    mov                 r7d, r9d
    shr                  r9, 32
    mov                r13d, rXd
    shr                  rX, 32
    pshufd              xm4, xm5, q2200
    pshufd              xm5, xm5, q3311
    pshufd              xm7, xm6, q2200
    pshufd              xm6, xm6, q3311
    pblendvb           xm12, xm2, xm4
    pblendvb           xm13, xm2, xm5
    pblendvb           xm14, xm2, xm7
    pblendvb           xm15, xm2, xm6
    pmovsxbw            m12, xm12
    pmovsxbw            m13, xm13
    pmovsxbw            m14, xm14
    pmovsxbw            m15, xm15
    MC_8TAP_SCALED_H 0, 1, 2, 3, 4, 5, 6, 7 ; 0a 1a 0b 1b
    mova        [rsp+0x60], m0
    MC_8TAP_SCALED_H 1, 2, 3, 4, 5, 6, 7, 8 ; 2a 3a 2b 3b
    MC_8TAP_SCALED_H 2, 3, 4, 5, 6, 7, 8, 9 ; 4a 5a 4b 5b
    MC_8TAP_SCALED_H 3, 4, 5, 6, 7, 8, 9, 0 ; 6a 7a 6b 7b
    mova                 m0, [rsp+0x60]
    vbroadcasti128       m9, [base+subpel_s_shuf8]
    mov                 myd, mym
    mov                 dyd, dym
    pshufb               m0, m9     ; 01a 01b
    pshufb               m1, m9     ; 23a 23b
    pshufb               m2, m9     ; 45a 45b
    pshufb               m3, m9     ; 67a 67b
.vloop:
    and                 myd, 0x3ff
    mov                 r6d, 64 << 24
    mov                 r4d, myd
    shr                 r4d, 6
    lea                 r4d, [t1+r4]
    cmovnz              r6q, [base+subpel_filters+r4*8]
    movq                xm9, r6q
    punpcklqdq          xm9, xm9
    pmovsxbw             m9, xm9
    pshufd               m8, m9, q0000
    pshufd               m7, m9, q1111
    pmaddwd              m4, m0, m8
    pmaddwd              m5, m1, m7
    pshufd               m8, m9, q2222
    pshufd               m9, m9, q3333
    pmaddwd              m6, m2, m8
    pmaddwd              m7, m3, m9
%if isput
    psrldq              xm8, xm11, 8
%endif
    paddd                m4, [rsp+0x20]
    paddd                m6, m7
    paddd                m4, m5
    paddd                m4, m6
%if isput
    psrad                m4, xm8
    vextracti128        xm5, m4, 1
    packusdw            xm4, xm5
    pminsw              xm4, [rsp+0xb0]
    mova             [dstq], xm4
    add                dstq, dsm
%else
    psrad                m4, 6
    vextracti128        xm5, m4, 1
    packssdw            xm4, xm5
    mova             [tmpq], xm4
    add                tmpq, tmp_stridem
%endif
    dec                  hd
    jz .hloop_prep
    add                 myd, dyd
    test                myd, ~0x3ff
    jz .vloop
    test                myd, 0x400
    mov          [rsp+0x60], myd
    mov                 r4d, [rsp+0xa0]
    mov                 r6d, [rsp+0xa4]
    mov                 r7d, [rsp+0xa8]
    mov                 r9d, [rsp+0xac]
    jz .skip_line
    vbroadcasti128       m9, [base+wswap]
    movu                xm4, [srcq+ r4*2]
    movu                xm5, [srcq+ r6*2]
    movu                xm6, [srcq+ r7*2]
    movu                xm7, [srcq+ r9*2]
    vinserti128          m4, [srcq+r10*2], 1
    vinserti128          m5, [srcq+r11*2], 1
    vinserti128          m6, [srcq+r13*2], 1
    vinserti128          m7, [srcq+ rX*2], 1
    add                srcq, ssq
    mov                 myd, [rsp+0x60]
    mov                 dyd, dym
    pshufb               m0, m9
    pshufb               m1, m9
    pshufb               m2, m9
    pshufb               m3, m9
    pmaddwd              m4, m12
    pmaddwd              m5, m13
    pmaddwd              m6, m14
    pmaddwd              m7, m15
    phaddd               m4, m5
    phaddd               m6, m7
    phaddd               m4, m6
    paddd                m4, m10
    psrad                m4, xm11
    pslld                m4, 16
    pblendw              m0, m1, 0xaa
    pblendw              m1, m2, 0xaa
    pblendw              m2, m3, 0xaa
    pblendw              m3, m4, 0xaa
    jmp .vloop
.skip_line:
    mova                 m0, m1
    mova                 m1, m2
    mova                 m2, m3
    MC_8TAP_SCALED_H      3, 10, 4, 5, 6, 7, 8, 9, 1
    vbroadcasti128       m9, [base+subpel_s_shuf8]
    mov                 myd, [rsp+0x60]
    mov                 dyd, dym
    pshufb               m3, m9
    jmp .vloop
    SWAP                 m1, m12, m10
    SWAP                 m7, m11
.dy1:
    movzx                wd, word [base+%1_8tap_scaled_avx2_dy1_table+wq*2]
    add                  wq, base_reg
    jmp                  wq
%if isput
.dy1_w2:
    mov                 myd, mym
    movzx               t0d, t0b
    sub                srcq, 2
    movd               xm15, t0d
    punpckldq            m8, m9, m8
    paddd               m10, m8 ; mx+dx*[0-1]
    vpbroadcastd       xm14, [base+pq_0x40000000+2]
    vpbroadcastd       xm15, xm15
    pand                xm8, xm10, xm6
    psrld               xm8, 6
    paddd              xm15, xm8
    movd                r4d, xm15
    pextrd              r6d, xm15, 1
    vbroadcasti128       m5, [base+bdct_lb_q]
    vbroadcasti128       m6, [base+subpel_s_shuf2]
    vpbroadcastd        m15, [base+subpel_filters+r4*8+2]
    vpbroadcastd         m4, [base+subpel_filters+r6*8+2]
    pcmpeqd             xm8, xm9
    psrld               m10, 10
    paddd               m10, m10
    movu                xm0, [srcq+ssq*0]
    movu                xm1, [srcq+ssq*1]
    movu                xm2, [srcq+ssq*2]
    movu                xm3, [srcq+ss3q ]
    lea                srcq, [srcq+ssq*4]
    shr                 myd, 6
    mov                 r4d, 64 << 24
    lea                 myd, [t1+myq]
    cmovnz              r4q, [base+subpel_filters+myq*8]
    pshufb              m10, m5
    paddb               m10, m6
    vpblendd           xm15, xm4, 0xa
    pblendvb           xm15, xm14, xm8
    pmovsxbw            m15, xm15
    vinserti128          m0, [srcq+ssq*0], 1
    vinserti128          m1, [srcq+ssq*1], 1
    vinserti128          m2, [srcq+ssq*2], 1
    add                srcq, ss3q
    movq                xm6, r4q
    pmovsxbw            xm6, xm6
    pshufd              xm8, xm6, q0000
    pshufd              xm9, xm6, q1111
    pshufd             xm14, xm6, q2222
    pshufd              xm6, xm6, q3333
    REPX    {pshufb x, m10}, m0, m1, m2
    pshufb              xm3, xm10
    REPX   {pmaddwd x, m15}, m0, m1, m2
    pmaddwd             xm3, xm15
    phaddd               m0, m1
    phaddd               m2, m3
    paddd                m0, m12
    paddd                m2, m12
    psrad                m0, xm7
    psrad                m2, xm7
    packssdw             m0, m2
    vextracti128        xm1, m0, 1
    palignr             xm2, xm1, xm0, 4
    pshufd              xm4, xm1, q2121
    punpcklwd           xm3, xm0, xm2       ; 01 12
    punpckhwd           xm0, xm2            ; 23 34
    punpcklwd           xm2, xm1, xm4       ; 45 56
.dy1_w2_loop:
    movu                xm1, [srcq+ssq*0]
    movu                xm5, [srcq+ssq*1]
    lea                srcq, [srcq+ssq*2]
    pshufb              xm1, xm10
    pshufb              xm5, xm10
    pmaddwd             xm1, xm15
    pmaddwd             xm5, xm15
    phaddd              xm1, xm5
    pmaddwd             xm5, xm3, xm8
    mova                xm3, xm0
    pmaddwd             xm0, xm9
    paddd               xm1, xm12
    psrad               xm1, xm7
    packssdw            xm1, xm1
    paddd               xm5, xm0
    mova                xm0, xm2
    pmaddwd             xm2, xm14
    paddd               xm5, xm2
    palignr             xm2, xm1, xm4, 12
    punpcklwd           xm2, xm1            ; 67 78
    pmaddwd             xm4, xm2, xm6
    paddd               xm5, xm13
    paddd               xm5, xm4
    mova                xm4, xm1
    psrldq              xm1, xm7, 8
    psrad               xm5, xm1
    packusdw            xm5, xm5
    pminsw              xm5, xm11
    movd       [dstq+dsq*0], xm5
    pextrd     [dstq+dsq*1], xm5, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .dy1_w2_loop
    RET
%endif
.dy1_w4:
    mov                 myd, mym
%if isput
    mova         [rsp+0x50], xm11
%endif
    mova         [rsp+0x00], m12
    mova         [rsp+0x20], m13
    mova         [rsp+0x40], xm7
    vbroadcasti128       m7, [base+rescale_mul]
    movzx               t0d, t0b
    sub                srcq, 2
    movd               xm15, t0d
    pmaddwd              m8, m7
    vpbroadcastq         m2, [base+pq_0x40000000+1]
    vpbroadcastd       xm15, xm15
    SWAP                m13, m10
    paddd               m13, m8 ; mx+dx*[0-3]
    pand                 m6, m13
    psrld                m6, 6
    paddd              xm15, xm6
    movd                r4d, xm15
    pextrd              r6d, xm15, 1
    pextrd             r11d, xm15, 2
    pextrd             r13d, xm15, 3
    vbroadcasti128       m5, [base+bdct_lb_q+ 0]
    vbroadcasti128       m1, [base+bdct_lb_q+16]
    vbroadcasti128       m4, [base+subpel_s_shuf2]
    vpbroadcastd       xm14, [base+subpel_filters+r4*8+2]
    vpbroadcastd        xm7, [base+subpel_filters+r6*8+2]
    vpbroadcastd       xm15, [base+subpel_filters+r11*8+2]
    vpbroadcastd        xm8, [base+subpel_filters+r13*8+2]
    pcmpeqd              m6, m9
    punpckldq           m10, m6, m6
    punpckhdq            m6, m6
    psrld               m13, 10
    paddd               m13, m13
    vpblendd           xm14, xm7, 0xa
    vpblendd           xm15, xm8, 0xa
    pmovsxbw            m14, xm14
    pmovsxbw            m15, xm15
    pblendvb            m14, m2, m10
    pblendvb            m15, m2, m6
    pextrd               r4, xm13, 2
    pshufb              m12, m13, m5
    pshufb              m13, m1
    lea                  r6, [r4+ssq*2]
    lea                 r11, [r4+ssq*1]
    lea                 r13, [r4+ss3q ]
    movu                xm0, [srcq+ssq*0]
    movu                xm7, [srcq+r4   ]
    movu                xm1, [srcq+ssq*2]
    movu                xm8, [srcq+r6   ]
    vinserti128          m0, [srcq+ssq*1], 1 ; 0 1
    vinserti128          m7, [srcq+r11  ], 1
    vinserti128          m1, [srcq+ss3q ], 1 ; 2 3
    vinserti128          m8, [srcq+r13  ], 1
    lea                srcq, [srcq+ssq*4]
    movu                xm2, [srcq+ssq*0]
    movu                xm9, [srcq+r4   ]
    movu                xm3, [srcq+ssq*2]    ; 6 _
    movu               xm10, [srcq+r6   ]
    vinserti128          m2, [srcq+ssq*1], 1 ; 4 5
    vinserti128          m9, [srcq+r11  ], 1
    lea                srcq, [srcq+ss3q ]
    vpbroadcastb         m5, xm13
    psubb               m13, m5
    paddb               m12, m4
    paddb               m13, m4
    mova                 m5, [rsp+0x00]
    movd                xm6, [rsp+0x40]
    pshufb               m0, m12
    pshufb               m1, m12
    pmaddwd              m0, m14
    pmaddwd              m1, m14
    pshufb               m7, m13
    pshufb               m8, m13
    pmaddwd              m7, m15
    pmaddwd              m8, m15
    pshufb               m2, m12
    pshufb              xm3, xm12
    pmaddwd              m2, m14
    pmaddwd             xm3, xm14
    pshufb               m9, m13
    pshufb             xm10, xm13
    pmaddwd              m9, m15
    pmaddwd            xm10, xm15
    phaddd               m0, m7
    phaddd               m1, m8
    phaddd               m2, m9
    phaddd              xm3, xm10
    paddd                m0, m5
    paddd                m1, m5
    paddd                m2, m5
    paddd               xm3, xm5
    psrad                m0, xm6
    psrad                m1, xm6
    psrad                m2, xm6
    psrad               xm3, xm6
    vperm2i128           m4, m0, m1, 0x21 ; 1 2
    vperm2i128           m5, m1, m2, 0x21 ; 3 4
    vperm2i128           m6, m2, m3, 0x21 ; 5 6
    shr                 myd, 6
    mov                r13d, 64 << 24
    lea                 myd, [t1+myq]
    cmovnz             r13q, [base+subpel_filters+myq*8]
    pslld                m4, 16
    pslld                m5, 16
    pslld                m6, 16
    pblendw              m0, m4, 0xaa ; 01 12
    pblendw              m1, m5, 0xaa ; 23 34
    pblendw              m2, m6, 0xaa ; 45 56
    movq               xm10, r13q
    punpcklqdq         xm10, xm10
    pmovsxbw            m10, xm10
    pshufd               m7, m10, q0000
    pshufd               m8, m10, q1111
    pshufd               m9, m10, q2222
    pshufd              m10, m10, q3333
.dy1_w4_loop:
    movu               xm11, [srcq+ssq*0]
    movu                xm6, [srcq+r4   ]
    vinserti128         m11, [srcq+ssq*1], 1
    vinserti128          m6, [srcq+r11  ], 1
    lea                srcq, [srcq+ssq*2]
    pmaddwd              m4, m0, m7
    pmaddwd              m5, m1, m8
    pshufb              m11, m12
    pshufb               m6, m13
    pmaddwd             m11, m14
    pmaddwd              m6, m15
    paddd                m4, [rsp+0x20]
    phaddd              m11, m6
    pmaddwd              m6, m2, m9
    paddd               m11, [rsp+0x00]
    psrad               m11, [rsp+0x40]
    mova                 m0, m1
    mova                 m1, m2
    paddd                m5, m6
    paddd                m4, m5
    vinserti128          m2, m3, xm11, 1
    pslld                m3, m11, 16
    pblendw              m2, m3, 0xaa   ; 67 78
    pmaddwd              m5, m2, m10
    vextracti128        xm3, m11, 1
    paddd                m4, m5
%if isput
    psrad                m4, [rsp+0x48]
    vextracti128        xm5, m4, 1
    packusdw            xm4, xm5
    pminsw              xm4, [rsp+0x50]
    movq       [dstq+dsq*0], xm4
    movhps     [dstq+dsq*1], xm4
    lea                dstq, [dstq+dsq*2]
%else
    psrad                m4, 6
    vextracti128        xm5, m4, 1
    packssdw            xm4, xm5
    mova             [tmpq], xm4
    add                tmpq, 16
%endif
    sub                  hd, 2
    jg .dy1_w4_loop
    MC_8TAP_SCALED_RET
    SWAP                 m10, m13
.dy1_w8:
    mov    dword [rsp+0xa0], 1
    movifprep   tmp_stridem, 16
    jmp .dy1_w_start
.dy1_w16:
    mov    dword [rsp+0xa0], 2
    movifprep   tmp_stridem, 32
    jmp .dy1_w_start
.dy1_w32:
    mov    dword [rsp+0xa0], 4
    movifprep   tmp_stridem, 64
    jmp .dy1_w_start
.dy1_w64:
    mov    dword [rsp+0xa0], 8
    movifprep   tmp_stridem, 128
    jmp .dy1_w_start
.dy1_w128:
    mov    dword [rsp+0xa0], 16
    movifprep   tmp_stridem, 256
.dy1_w_start:
    SWAP                m10, m12, m1
    SWAP                m11, m7
    ; m1=mx, m7=pxmax, m10=h_rnd, m11=h_sh, m12=free
    mov                 myd, mym
%if isput
 %define dsm [rsp+0xb8]
    movifnidn           dsm, dsq
    mova         [rsp+0xc0], xm7
%else
 %if UNIX64
  %define hm [rsp+0xb8]
 %endif
%endif
    mova         [rsp+0x00], m10
    mova         [rsp+0x20], m13
    mova         [rsp+0x40], xm11
    shr                 t0d, 16
    sub                srcq, 6
    shr                 myd, 6
    mov                 r4d, 64 << 24
    lea                 myd, [t1+myq]
    cmovnz              r4q, [base+subpel_filters+myq*8]
    pmaddwd              m8, [base+rescale_mul2]
    movd               xm15, t0d
    mov          [rsp+0xa4], t0d
    mov          [rsp+0xa8], srcq
    mov          [rsp+0xb0], r0q ; dstq / tmpq
%if UNIX64
    mov                  hm, hd
%endif
    shl           dword dxm, 3 ; dx*8
    vpbroadcastd        m15, xm15
    paddd                m1, m8 ; mx+dx*[0-7]
    movq                xm0, r4q
    pmovsxbw            xm0, xm0
    mova         [rsp+0x50], xm0
    jmp .dy1_hloop
.dy1_hloop_prep:
    dec    dword [rsp+0xa0]
    jz .ret
    add    qword [rsp+0xb0], 16
    mov                  hd, hm
    vpbroadcastd         m8, dxm
    vpbroadcastd         m6, [base+pd_0x3ff]
    paddd                m1, m8, [rsp+0x60]
    vpbroadcastd        m15, [rsp+0xa4]
    pxor                 m9, m9
    mov                srcq, [rsp+0xa8]
    mov                 r0q, [rsp+0xb0] ; dstq / tmpq
    mova                m10, [rsp+0x00]
    mova               xm11, [rsp+0x40]
.dy1_hloop:
    vpbroadcastq        xm2, [base+pq_0x40000000]
    pand                 m5, m1, m6
    psrld                m5, 6
    paddd               m15, m5
    pcmpeqd              m5, m9
    vextracti128        xm7, m15, 1
    movq                 r6, xm15
    pextrq               r9, xm15, 1
    movq                r11, xm7
    pextrq               rX, xm7, 1
    mov                 r4d, r6d
    shr                  r6, 32
    mov                 r7d, r9d
    shr                  r9, 32
    mov                r10d, r11d
    shr                 r11, 32
    mov                r13d, rXd
    shr                  rX, 32
    mova         [rsp+0x60], m1
    movq               xm12, [base+subpel_filters+ r4*8]
    movq               xm13, [base+subpel_filters+ r6*8]
    movhps             xm12, [base+subpel_filters+ r7*8]
    movhps             xm13, [base+subpel_filters+ r9*8]
    movq               xm14, [base+subpel_filters+r10*8]
    movq               xm15, [base+subpel_filters+r11*8]
    movhps             xm14, [base+subpel_filters+r13*8]
    movhps             xm15, [base+subpel_filters+ rX*8]
    psrld                m1, 10
    vextracti128        xm7, m1, 1
    vextracti128        xm6, m5, 1
    movq                 r6, xm1
    pextrq              r11, xm1, 1
    movq                 r9, xm7
    pextrq               rX, xm7, 1
    mov                 r4d, r6d
    shr                  r6, 32
    mov                r10d, r11d
    shr                 r11, 32
    mov                 r7d, r9d
    shr                  r9, 32
    mov                r13d, rXd
    shr                  rX, 32
    pshufd              xm4, xm5, q2200
    pshufd              xm5, xm5, q3311
    pshufd              xm7, xm6, q2200
    pshufd              xm6, xm6, q3311
    pblendvb           xm12, xm2, xm4
    pblendvb           xm13, xm2, xm5
    pblendvb           xm14, xm2, xm7
    pblendvb           xm15, xm2, xm6
    pmovsxbw            m12, xm12
    pmovsxbw            m13, xm13
    pmovsxbw            m14, xm14
    pmovsxbw            m15, xm15
    MC_8TAP_SCALED_H 0, 1, 2, 3, 4, 5, 6, 7 ; 0a 1a 0b 1b
    mova         [rsp+0x80], m0
    MC_8TAP_SCALED_H 1, 2, 3, 4, 5, 6, 7, 8 ; 2a 3a 2b 3b
    MC_8TAP_SCALED_H 2, 3, 4, 5, 6, 7, 8, 9 ; 4a 5a 4b 5b
    MC_8TAP_SCALED_H 3, 4, 5, 6, 7, 8, 9, 0 ; 6a 7a 6b 7b
    mova                 m0, [rsp+0x80]
    vbroadcasti128       m7, [base+subpel_s_shuf8]
    vpbroadcastd         m8, [rsp+0x50]
    vpbroadcastd         m9, [rsp+0x54]
    vpbroadcastd        m10, [rsp+0x58]
    vpbroadcastd        m11, [rsp+0x5c]
    pshufb               m0, m7     ; 01a 01b
    pshufb               m1, m7     ; 23a 23b
    pshufb               m2, m7     ; 45a 45b
    pshufb               m3, m7     ; 67a 67b
.dy1_vloop:
    pmaddwd              m4, m0, m8
    pmaddwd              m5, m1, m9
    pmaddwd              m6, m2, m10
    pmaddwd              m7, m3, m11
    paddd                m4, [rsp+0x20]
    paddd                m6, m7
    paddd                m4, m5
    paddd                m4, m6
%if isput
    psrad                m4, [rsp+0x48]
    vextracti128        xm5, m4, 1
    packusdw            xm4, xm5
    pminsw              xm4, [rsp+0xc0]
    mova             [dstq], xm4
    add                dstq, dsm
%else
    psrad                m4, 6
    vextracti128        xm5, m4, 1
    packssdw            xm4, xm5
    mova             [tmpq], xm4
    add                tmpq, tmp_stridem
%endif
    dec                  hd
    jz .dy1_hloop_prep
    vbroadcasti128       m7, [base+wswap]
    pshufb               m0, m7
    pshufb               m1, m7
    pshufb               m2, m7
    pshufb               m3, m7
    movu                xm4, [srcq+ r4*2]
    movu                xm5, [srcq+ r6*2]
    movu                xm6, [srcq+ r7*2]
    movu                xm7, [srcq+ r9*2]
    vinserti128          m4, [srcq+r10*2], 1
    vinserti128          m5, [srcq+r11*2], 1
    vinserti128          m6, [srcq+r13*2], 1
    vinserti128          m7, [srcq+ rX*2], 1
    add                srcq, ssq
    pmaddwd              m4, m12
    pmaddwd              m5, m13
    pmaddwd              m6, m14
    pmaddwd              m7, m15
    phaddd               m4, m5
    phaddd               m6, m7
    phaddd               m4, m6
    paddd                m4, [rsp+0x00]
    psrad                m4, [rsp+0x40]
    pslld                m4, 16
    pblendw              m0, m1, 0xaa
    pblendw              m1, m2, 0xaa
    pblendw              m2, m3, 0xaa
    pblendw              m3, m4, 0xaa
    jmp .dy1_vloop
    SWAP                 m1, m12, m10
    SWAP                 m7, m11
.dy2:
    movzx                wd, word [base+%1_8tap_scaled_avx2_dy2_table+wq*2]
    add                  wq, base_reg
    jmp                  wq
%if isput
.dy2_w2:
    mov                 myd, mym
    movzx               t0d, t0b
    sub                srcq, 2
    movd               xm15, t0d
    punpckldq            m8, m9, m8
    paddd               m10, m8 ; mx+dx*[0-1]
    vpbroadcastd       xm14, [base+pq_0x40000000+2]
    vpbroadcastd       xm15, xm15
    pand                xm8, xm10, xm6
    psrld               xm8, 6
    paddd              xm15, xm8
    movd                r4d, xm15
    pextrd              r6d, xm15, 1
    vbroadcasti128       m5, [base+bdct_lb_q]
    vbroadcasti128       m6, [base+subpel_s_shuf2]
    vpbroadcastd       xm15, [base+subpel_filters+r4*8+2]
    vpbroadcastd        xm4, [base+subpel_filters+r6*8+2]
    pcmpeqd             xm8, xm9
    psrld               m10, 10
    paddd               m10, m10
    movu                xm0, [srcq+ssq*0]
    movu                xm1, [srcq+ssq*2]
    movu                xm2, [srcq+ssq*4]
    pshufb              m10, m5
    paddb               m10, m6
    vpblendd           xm15, xm4, 0xa
    pblendvb           xm15, xm14, xm8
    pmovsxbw            m15, xm15
    vinserti128          m0, [srcq+ssq*1], 1 ; 0 1
    vinserti128          m1, [srcq+ss3q ], 1 ; 2 3
    lea                srcq, [srcq+ssq*4]
    vinserti128          m2, [srcq+ssq*1], 1 ; 4 5
    lea                srcq, [srcq+ssq*2]
    shr                 myd, 6
    mov                 r4d, 64 << 24
    lea                 myd, [t1+myq]
    cmovnz              r4q, [base+subpel_filters+myq*8]
    pshufb               m0, m10
    pshufb               m1, m10
    pshufb               m2, m10
    pmaddwd              m0, m15
    pmaddwd              m1, m15
    pmaddwd              m2, m15
    movq                xm6, r4q
    pmovsxbw            xm6, xm6
    phaddd               m0, m1
    phaddd               m1, m2
    paddd                m0, m12
    paddd                m1, m12
    psrad                m0, xm7
    psrad                m1, xm7
    packssdw             m0, m1             ; 0 2 2 4  1 3 3 5
    vextracti128        xm1, m0, 1
    pshufd              xm8, xm6, q0000
    pshufd              xm9, xm6, q1111
    pshufd             xm14, xm6, q2222
    pshufd              xm6, xm6, q3333
    punpcklwd           xm2, xm0, xm1       ; 01 23
    punpckhwd           xm1, xm0, xm1       ; 23 45
.dy2_w2_loop:
    movu                xm3, [srcq+ssq*0]
    movu                xm5, [srcq+ssq*2]
    vinserti128          m3, [srcq+ssq*1], 1 ; 6 7
    vinserti128          m5, [srcq+ss3q ], 1 ; 8 9
    lea                srcq, [srcq+ssq*4]
    pmaddwd             xm4, xm2, xm8
    pmaddwd             xm1, xm9
    pshufb               m3, m10
    pshufb               m5, m10
    pmaddwd              m3, m15
    pmaddwd              m5, m15
    phaddd               m3, m5
    paddd               xm4, xm1
    paddd                m3, m12
    psrad                m3, xm7
    packssdw             m3, m3
    pshufd               m3, m3, q2100
    palignr              m0, m3, m0, 12     ; 4 6 6 8  5 7 7 9
    vextracti128        xm1, m0, 1
    punpcklwd           xm2, xm0, xm1       ; 45 67
    punpckhwd           xm1, xm0, xm1       ; 67 89
    pmaddwd             xm3, xm2, xm14
    pmaddwd             xm5, xm1, xm6
    paddd               xm4, xm13
    paddd               xm4, xm3
    psrldq              xm3, xm7, 8
    paddd               xm4, xm5
    psrad               xm4, xm3
    packusdw            xm4, xm4
    pminsw              xm4, xm11
    movd       [dstq+dsq*0], xm4
    pextrd     [dstq+dsq*1], xm4, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .dy2_w2_loop
    RET
%endif
.dy2_w4:
    mov                 myd, mym
%if isput
    mova         [rsp+0x50], xm11
%endif
    mova         [rsp+0x00], m12
    mova         [rsp+0x20], m13
    mova         [rsp+0x40], xm7
    vbroadcasti128       m7, [base+rescale_mul]
    movzx               t0d, t0b
    sub                srcq, 2
    movd               xm15, t0d
    pmaddwd              m8, m7
    vpbroadcastq         m2, [base+pq_0x40000000+1]
    vpbroadcastd       xm15, xm15
    SWAP                m13, m10
    paddd               m13, m8 ; mx+dx*[0-3]
    pand                 m6, m13
    psrld                m6, 6
    paddd              xm15, xm6
    movd                r4d, xm15
    pextrd              r6d, xm15, 1
    pextrd             r11d, xm15, 2
    pextrd             r13d, xm15, 3
    vbroadcasti128       m5, [base+bdct_lb_q+ 0]
    vbroadcasti128       m1, [base+bdct_lb_q+16]
    vbroadcasti128       m4, [base+subpel_s_shuf2]
    vpbroadcastd       xm14, [base+subpel_filters+r4*8+2]
    vpbroadcastd        xm7, [base+subpel_filters+r6*8+2]
    vpbroadcastd       xm15, [base+subpel_filters+r11*8+2]
    vpbroadcastd        xm8, [base+subpel_filters+r13*8+2]
    shr                 myd, 6
    mov                r13d, 64 << 24
    lea                 myd, [t1+myq]
    cmovnz             r13q, [base+subpel_filters+myq*8]
    pcmpeqd              m6, m9
    punpckldq           m11, m6, m6
    punpckhdq            m6, m6
    psrld               m13, 10
    paddd               m13, m13
    vpblendd           xm14, xm7, 0xa
    vpblendd           xm15, xm8, 0xa
    pmovsxbw            m14, xm14
    pmovsxbw            m15, xm15
    movq               xm10, r13q
    pblendvb            m14, m2, m11
    pblendvb            m15, m2, m6
    pextrd               r4, xm13, 2
    pshufb              m12, m13, m5
    pshufb              m13, m1
    lea                  r6, [r4+ssq*1]
    lea                 r11, [r4+ssq*2]
    lea                 r13, [r4+ss3q ]
    movu                xm0, [srcq+ssq*0]
    movu                xm7, [srcq+r4   ]
    movu                xm1, [srcq+ssq*1]
    movu                xm8, [srcq+r6   ]
    vinserti128          m0, [srcq+ssq*2], 1 ; 0 2
    vinserti128          m7, [srcq+r11  ], 1
    vinserti128          m1, [srcq+ss3q ], 1 ; 1 3
    vinserti128          m8, [srcq+r13  ], 1
    lea                srcq, [srcq+ssq*4]
    movu                xm2, [srcq+ssq*0]
    movu                xm9, [srcq+r4   ]
    vinserti128          m2, [srcq+ssq*1], 1 ; 4 5
    vinserti128          m9, [srcq+r6   ], 1
    lea                srcq, [srcq+ssq*2]
    vpbroadcastb         m5, xm13
    psubb               m13, m5
    paddb               m12, m4
    paddb               m13, m4
    mova                 m5, [rsp+0x00]
    movd                xm6, [rsp+0x40]
    pshufb               m0, m12
    pshufb               m1, m12
    pshufb               m2, m12
    pmaddwd              m0, m14
    pmaddwd              m1, m14
    pmaddwd              m2, m14
    pshufb               m7, m13
    pshufb               m8, m13
    pshufb               m9, m13
    pmaddwd              m7, m15
    pmaddwd              m8, m15
    pmaddwd              m9, m15
    punpcklqdq         xm10, xm10
    pmovsxbw            m10, xm10
    phaddd               m0, m7
    phaddd               m1, m8
    phaddd               m2, m9
    paddd                m0, m5
    paddd                m1, m5
    paddd                m2, m5
    psrad                m0, xm6
    psrad                m1, xm6
    psrad                m2, xm6
    vperm2i128           m3, m0, m2, 0x21 ; 2 4
    vperm2i128           m2, m1, 0x13     ; 3 5
    pshufd               m7, m10, q0000
    pshufd               m8, m10, q1111
    pshufd               m9, m10, q2222
    pshufd              m10, m10, q3333
    packssdw             m0, m3 ; 0 2  2 4
    packssdw             m1, m2 ; 1 3  3 5
    punpckhwd            m2, m0, m1 ; 23 45
    punpcklwd            m0, m1     ; 01 23
.dy2_w4_loop:
    movu                xm1, [srcq+ssq*0]
    movu                xm6, [srcq+r4   ]
    movu                xm3, [srcq+ssq*1]
    movu               xm11, [srcq+r6   ]
    vinserti128          m1, [srcq+ssq*2], 1 ; 6 8
    vinserti128          m6, [srcq+r11  ], 1
    vinserti128          m3, [srcq+ss3q ], 1 ; 7 9
    vinserti128         m11, [srcq+r13  ], 1
    lea                srcq, [srcq+ssq*4]
    pmaddwd              m4, m0, m7
    pmaddwd              m5, m2, m8
    pshufb               m1, m12
    pshufb               m3, m12
    pmaddwd              m1, m14
    pmaddwd              m3, m14
    mova                 m0, [rsp+0x00]
    pshufb               m6, m13
    pshufb              m11, m13
    pmaddwd              m6, m15
    pmaddwd             m11, m15
    paddd                m4, m5
    movd                xm5, [rsp+0x40]
    phaddd               m1, m6
    phaddd               m3, m11
    paddd                m1, m0
    paddd                m3, m0
    psrad                m1, xm5
    psrad                m3, xm5
    pslld                m3, 16
    pblendw              m1, m3, 0xaa     ; 67 89
    vperm2i128           m0, m2, m1, 0x21 ; 45 67
    paddd                m4, [rsp+0x20]
    mova                 m2, m1
    pmaddwd              m5, m0, m9
    pmaddwd              m6, m2, m10
    paddd                m4, m5
    paddd                m4, m6
%if isput
    psrad                m4, [rsp+0x48]
    vextracti128        xm5, m4, 1
    packusdw            xm4, xm5
    pminsw              xm4, [rsp+0x50]
    movq       [dstq+dsq*0], xm4
    movhps     [dstq+dsq*1], xm4
    lea                dstq, [dstq+dsq*2]
%else
    psrad                m4, 6
    vextracti128        xm5, m4, 1
    packssdw            xm4, xm5
    mova             [tmpq], xm4
    add                tmpq, 16
%endif
    sub                  hd, 2
    jg .dy2_w4_loop
    MC_8TAP_SCALED_RET
    SWAP                m10, m13
.dy2_w8:
    mov    dword [rsp+0xa0], 1
    movifprep   tmp_stridem, 16
    jmp .dy2_w_start
.dy2_w16:
    mov    dword [rsp+0xa0], 2
    movifprep   tmp_stridem, 32
    jmp .dy2_w_start
.dy2_w32:
    mov    dword [rsp+0xa0], 4
    movifprep   tmp_stridem, 64
    jmp .dy2_w_start
.dy2_w64:
    mov    dword [rsp+0xa0], 8
    movifprep   tmp_stridem, 128
    jmp .dy2_w_start
.dy2_w128:
    mov    dword [rsp+0xa0], 16
    movifprep   tmp_stridem, 256
.dy2_w_start:
    SWAP                m10, m12, m1
    SWAP                m11, m7
    ; m1=mx, m7=pxmax, m10=h_rnd, m11=h_sh, m12=free
    mov                 myd, mym
%if isput
    movifnidn           dsm, dsq
    mova         [rsp+0xc0], xm7
%endif
    mova         [rsp+0x00], m10
    mova         [rsp+0x20], m13
    mova         [rsp+0x40], xm11
    shr                 t0d, 16
    sub                srcq, 6
    shr                 myd, 6
    mov                 r4d, 64 << 24
    lea                 myd, [t1+myq]
    cmovnz              r4q, [base+subpel_filters+myq*8]
    pmaddwd              m8, [base+rescale_mul2]
    movd               xm15, t0d
    mov          [rsp+0xa4], t0d
    mov          [rsp+0xa8], srcq
    mov          [rsp+0xb0], r0q ; dstq / tmpq
%if UNIX64
    mov                  hm, hd
%endif
    shl           dword dxm, 3 ; dx*8
    vpbroadcastd        m15, xm15
    paddd                m1, m8 ; mx+dx*[0-7]
    movq                xm0, r4q
    pmovsxbw            xm0, xm0
    mova         [rsp+0x50], xm0
    jmp .dy2_hloop
.dy2_hloop_prep:
    dec    dword [rsp+0xa0]
    jz .ret
    add    qword [rsp+0xb0], 16
    mov                  hd, hm
    vpbroadcastd         m8, dxm
    vpbroadcastd         m6, [base+pd_0x3ff]
    paddd                m1, m8, [rsp+0x60]
    vpbroadcastd        m15, [rsp+0xa4]
    pxor                 m9, m9
    mov                srcq, [rsp+0xa8]
    mov                 r0q, [rsp+0xb0] ; dstq / tmpq
    mova                m10, [rsp+0x00]
    mova               xm11, [rsp+0x40]
.dy2_hloop:
    vpbroadcastq        xm2, [base+pq_0x40000000]
    pand                 m5, m1, m6
    psrld                m5, 6
    paddd               m15, m5
    pcmpeqd              m5, m9
    vextracti128        xm7, m15, 1
    movq                 r6, xm15
    pextrq               r9, xm15, 1
    movq                r11, xm7
    pextrq               rX, xm7, 1
    mov                 r4d, r6d
    shr                  r6, 32
    mov                 r7d, r9d
    shr                  r9, 32
    mov                r10d, r11d
    shr                 r11, 32
    mov                r13d, rXd
    shr                  rX, 32
    mova         [rsp+0x60], m1
    movq               xm12, [base+subpel_filters+ r4*8]
    movq               xm13, [base+subpel_filters+ r6*8]
    movhps             xm12, [base+subpel_filters+ r7*8]
    movhps             xm13, [base+subpel_filters+ r9*8]
    movq               xm14, [base+subpel_filters+r10*8]
    movq               xm15, [base+subpel_filters+r11*8]
    movhps             xm14, [base+subpel_filters+r13*8]
    movhps             xm15, [base+subpel_filters+ rX*8]
    psrld                m1, 10
    vextracti128        xm7, m1, 1
    vextracti128        xm6, m5, 1
    movq                 r6, xm1
    pextrq              r11, xm1, 1
    movq                 r9, xm7
    pextrq               rX, xm7, 1
    mov                 r4d, r6d
    shr                  r6, 32
    mov                r10d, r11d
    shr                 r11, 32
    mov                 r7d, r9d
    shr                  r9, 32
    mov                r13d, rXd
    shr                  rX, 32
    pshufd              xm4, xm5, q2200
    pshufd              xm5, xm5, q3311
    pshufd              xm7, xm6, q2200
    pshufd              xm6, xm6, q3311
    pblendvb           xm12, xm2, xm4
    pblendvb           xm13, xm2, xm5
    pblendvb           xm14, xm2, xm7
    pblendvb           xm15, xm2, xm6
    pmovsxbw            m12, xm12
    pmovsxbw            m13, xm13
    pmovsxbw            m14, xm14
    pmovsxbw            m15, xm15
    MC_8TAP_SCALED_H 0, 1, 2, 3, 4, 5, 6, 7 ; 0a 1a 0b 1b
    mova         [rsp+0x80], m0
    MC_8TAP_SCALED_H 1, 2, 3, 4, 5, 6, 7, 8 ; 2a 3a 2b 3b
    MC_8TAP_SCALED_H 2, 3, 4, 5, 6, 7, 8, 9 ; 4a 5a 4b 5b
    MC_8TAP_SCALED_H 3, 4, 5, 6, 7, 8, 9, 0 ; 6a 7a 6b 7b
    mova                 m0, [rsp+0x80]
    vbroadcasti128       m7, [base+subpel_s_shuf8]
    vpbroadcastd         m8, [rsp+0x50]
    vpbroadcastd         m9, [rsp+0x54]
    vpbroadcastd        m10, [rsp+0x58]
    vpbroadcastd        m11, [rsp+0x5c]
    pshufb               m0, m7     ; 01a 01b
    pshufb               m1, m7     ; 23a 23b
    pshufb               m2, m7     ; 45a 45b
    pshufb               m3, m7     ; 67a 67b
.dy2_vloop:
    pmaddwd              m4, m0, m8
    pmaddwd              m5, m1, m9
    pmaddwd              m6, m2, m10
    pmaddwd              m7, m3, m11
    paddd                m4, [rsp+0x20]
    paddd                m6, m7
    paddd                m4, m5
    paddd                m4, m6
%if isput
    psrad                m4, [rsp+0x48]
    vextracti128        xm5, m4, 1
    packusdw            xm4, xm5
    pminsw              xm4, [rsp+0xc0]
    mova             [dstq], xm4
    add                dstq, dsm
%else
    psrad                m4, 6
    vextracti128        xm5, m4, 1
    packssdw            xm4, xm5
    mova             [tmpq], xm4
    add                tmpq, tmp_stridem
%endif
    dec                  hd
    jz .dy2_hloop_prep
    mova                 m0, m1
    mova                 m1, m2
    mova                 m2, m3
    movu                xm3, [srcq+ r4*2]
    movu                xm4, [srcq+ r6*2]
    movu                xm5, [srcq+ r7*2]
    movu                xm6, [srcq+ r9*2]
    vinserti128          m3, [srcq+r10*2], 1
    vinserti128          m4, [srcq+r11*2], 1
    vinserti128          m5, [srcq+r13*2], 1
    vinserti128          m6, [srcq+ rX*2], 1
    add                srcq, ssq
    pmaddwd              m3, m12
    pmaddwd              m4, m13
    pmaddwd              m5, m14
    pmaddwd              m6, m15
    phaddd               m3, m4
    phaddd               m5, m6
    phaddd               m3, m5
    movu                xm4, [srcq+ r4*2]
    movu                xm5, [srcq+ r6*2]
    movu                xm6, [srcq+ r7*2]
    movu                xm7, [srcq+ r9*2]
    vinserti128          m4, [srcq+r10*2], 1
    vinserti128          m5, [srcq+r11*2], 1
    vinserti128          m6, [srcq+r13*2], 1
    vinserti128          m7, [srcq+ rX*2], 1
    add                srcq, ssq
    pmaddwd              m4, m12
    pmaddwd              m5, m13
    pmaddwd              m6, m14
    pmaddwd              m7, m15
    phaddd               m4, m5
    phaddd               m6, m7
    mova                 m5, [rsp+0x00]
    movd                xm7, [rsp+0x40]
    phaddd               m4, m6
    paddd                m3, m5
    paddd                m4, m5
    psrad                m3, xm7
    psrad                m4, xm7
    pslld                m4, 16
    pblendw              m3, m4, 0xaa
    jmp .dy2_vloop
.ret:
    MC_8TAP_SCALED_RET 0
%undef isput
%undef isprep
%endmacro

%macro BILIN_SCALED_FN 1
cglobal %1_bilin_scaled_16bpc
    mov                 t0d, (5*15 << 16) | 5*15
    mov                 t1d, t0d
    jmp mangle(private_prefix %+ _%1_8tap_scaled_16bpc %+ SUFFIX)
%endmacro

%if WIN64
DECLARE_REG_TMP 6, 5
%else
DECLARE_REG_TMP 6, 8
%endif

%define PUT_8TAP_SCALED_FN FN put_8tap_scaled,
BILIN_SCALED_FN put
PUT_8TAP_SCALED_FN sharp,          SHARP,   SHARP
PUT_8TAP_SCALED_FN sharp_smooth,   SHARP,   SMOOTH
PUT_8TAP_SCALED_FN smooth_sharp,   SMOOTH,  SHARP
PUT_8TAP_SCALED_FN smooth,         SMOOTH,  SMOOTH
PUT_8TAP_SCALED_FN sharp_regular,  SHARP,   REGULAR
PUT_8TAP_SCALED_FN regular_sharp,  REGULAR, SHARP
PUT_8TAP_SCALED_FN smooth_regular, SMOOTH,  REGULAR
PUT_8TAP_SCALED_FN regular_smooth, REGULAR, SMOOTH
PUT_8TAP_SCALED_FN regular,        REGULAR, REGULAR
MC_8TAP_SCALED put

%if WIN64
DECLARE_REG_TMP 5, 4
%else
DECLARE_REG_TMP 6, 7
%endif

%define PREP_8TAP_SCALED_FN FN prep_8tap_scaled,
BILIN_SCALED_FN prep
PREP_8TAP_SCALED_FN sharp,          SHARP,   SHARP
PREP_8TAP_SCALED_FN sharp_smooth,   SHARP,   SMOOTH
PREP_8TAP_SCALED_FN smooth_sharp,   SMOOTH,  SHARP
PREP_8TAP_SCALED_FN smooth,         SMOOTH,  SMOOTH
PREP_8TAP_SCALED_FN sharp_regular,  SHARP,   REGULAR
PREP_8TAP_SCALED_FN regular_sharp,  REGULAR, SHARP
PREP_8TAP_SCALED_FN smooth_regular, SMOOTH,  REGULAR
PREP_8TAP_SCALED_FN regular_smooth, REGULAR, SMOOTH
PREP_8TAP_SCALED_FN regular,        REGULAR, REGULAR
MC_8TAP_SCALED prep

%macro WARP_V 5 ; dst, 01, 23, 45, 67
    lea               tmp1d, [myq+deltaq*4]
    lea               tmp2d, [myq+deltaq*1]
    shr                 myd, 10
    shr               tmp1d, 10
    movq                xm8, [filterq+myq  *8]
    vinserti128          m8, [filterq+tmp1q*8], 1 ; a e
    lea               tmp1d, [tmp2q+deltaq*4]
    lea                 myd, [tmp2q+deltaq*1]
    shr               tmp2d, 10
    shr               tmp1d, 10
    movq                xm0, [filterq+tmp2q*8]
    vinserti128          m0, [filterq+tmp1q*8], 1 ; b f
    lea               tmp1d, [myq+deltaq*4]
    lea               tmp2d, [myq+deltaq*1]
    shr                 myd, 10
    shr               tmp1d, 10
    movq                xm9, [filterq+myq  *8]
    vinserti128          m9, [filterq+tmp1q*8], 1 ; c g
    lea               tmp1d, [tmp2q+deltaq*4]
    lea                 myd, [tmp2q+gammaq]       ; my += gamma
    punpcklwd            m8, m0
    shr               tmp2d, 10
    shr               tmp1d, 10
    movq                xm0, [filterq+tmp2q*8]
    vinserti128          m0, [filterq+tmp1q*8], 1 ; d h
    punpcklwd            m0, m9, m0
    punpckldq            m9, m8, m0
    punpckhdq            m0, m8, m0
    punpcklbw            m8, m11, m9 ; a0 a1 b0 b1 c0 c1 d0 d1 << 8
    punpckhbw            m9, m11, m9 ; a2 a3 b2 b3 c2 c3 d2 d3 << 8
    pmaddwd             m%2, m8
    pmaddwd              m9, m%3
    punpcklbw            m8, m11, m0 ; a4 a5 b4 b5 c4 c5 d4 d5 << 8
    punpckhbw            m0, m11, m0 ; a6 a7 b6 b7 c6 c7 d6 d7 << 8
    pmaddwd              m8, m%4
    pmaddwd              m0, m%5
    paddd                m9, m%2
    mova                m%2, m%3
    paddd                m0, m8
    mova                m%3, m%4
    mova                m%4, m%5
    paddd               m%1, m0, m9
%endmacro

cglobal warp_affine_8x8t_16bpc, 4, 14, 16, tmp, ts
    mov                 r6d, r7m
    lea                  r9, [$$]
    shr                 r6d, 11
    vpbroadcastd        m13, [r9-$$+warp8x8_shift+r6*4]
    vpbroadcastd        m14, [warp8x8t_rnd]
    call mangle(private_prefix %+ _warp_affine_8x8_16bpc_avx2).main
    jmp .start
.loop:
    call mangle(private_prefix %+ _warp_affine_8x8_16bpc_avx2).main2
    lea                tmpq, [tmpq+tsq*4]
.start:
    paddd                m7, m14
    paddd                m0, m14
    psrad                m7, 15
    psrad                m0, 15
    packssdw             m7, m0
    vpermq               m7, m7, q3120
    mova         [tmpq+tsq*0], xm7
    vextracti128 [tmpq+tsq*2], m7, 1
    dec                 r4d
    jg .loop
.end:
    RET

cglobal warp_affine_8x8_16bpc, 4, 14, 16, dst, ds, src, ss, abcd, mx, tmp2, \
                                          alpha, beta, filter, tmp1, delta, \
                                          my, gamma
    mov                 r6d, r7m
    lea             filterq, [$$]
    shr                 r6d, 11
    vpbroadcastd        m13, [filterq-$$+warp8x8_shift+r6*4]
    vpbroadcastd        m14, [filterq-$$+warp8x8_rnd  +r6*4]
    vpbroadcastw        m15, r7m ; pixel_max
    call .main
    jmp .start
.loop:
    call .main2
    lea                dstq, [dstq+dsq*2]
.start:
    psrad                m7, 16
    psrad                m0, 16
    packusdw             m7, m0
    pmulhrsw             m7, m14
    pminsw               m7, m15
    vpermq               m7, m7, q3120
    mova         [dstq+dsq*0], xm7
    vextracti128 [dstq+dsq*1], m7, 1
    dec                 r4d
    jg .loop
.end:
    RET
ALIGN function_align
.main:
    ; Stack args offset by one (r4m -> r5m etc.) due to call
%if WIN64
    mov               abcdq, r5m
    mov                 mxd, r6m
%endif
    movsx            alphad, word [abcdq+2*0]
    movsx             betad, word [abcdq+2*1]
    vpbroadcastd        m12, [pd_32768]
    pxor                m11, m11
    add             filterq, mc_warp_filter-$$
    lea               tmp1q, [ssq*3]
    add                 mxd, 512+(64<<10)
    lea               tmp2d, [alphaq*3]
    sub                srcq, tmp1q    ; src -= src_stride*3
    sub               betad, tmp2d    ; beta -= alpha*3
    mov                 myd, r7m
    call .h
    psrld                m1, m0, 16
    call .h
    pblendw              m1, m0, 0xaa ; 01
    psrld                m2, m0, 16
    call .h
    pblendw              m2, m0, 0xaa ; 12
    psrld                m3, m0, 16
    call .h
    pblendw              m3, m0, 0xaa ; 23
    psrld                m4, m0, 16
    call .h
    pblendw              m4, m0, 0xaa ; 34
    psrld                m5, m0, 16
    call .h
    pblendw              m5, m0, 0xaa ; 45
    psrld                m6, m0, 16
    call .h
    pblendw              m6, m0, 0xaa ; 56
    movsx            deltad, word [abcdq+2*2]
    movsx            gammad, word [abcdq+2*3]
    add                 myd, 512+(64<<10)
    mov                 r4d, 4
    lea               tmp1d, [deltaq*3]
    sub              gammad, tmp1d    ; gamma -= delta*3
.main2:
    call .h
    psrld                m7, m6, 16
    pblendw              m7, m0, 0xaa ; 67
    WARP_V                7, 1, 3, 5, 7
    call .h
    psrld               m10, m5, 16
    pblendw             m10, m0, 0xaa ; 78
    WARP_V                0, 2, 4, 6, 10
    ret
ALIGN function_align
.h:
    lea               tmp1d, [mxq+alphaq*4]
    lea               tmp2d, [mxq+alphaq*1]
    movu               xm10, [srcq-6]
    vinserti128         m10, [srcq+2], 1
    shr                 mxd, 10 ; 0
    shr               tmp1d, 10 ; 4
    movq                xm0, [filterq+mxq  *8]
    vinserti128          m0, [filterq+tmp1q*8], 1
    lea               tmp1d, [tmp2q+alphaq*4]
    lea                 mxd, [tmp2q+alphaq*1]
    movu                xm8, [srcq-4]
    vinserti128          m8, [srcq+4], 1
    shr               tmp2d, 10 ; 1
    shr               tmp1d, 10 ; 5
    movq                xm9, [filterq+tmp2q*8]
    vinserti128          m9, [filterq+tmp1q*8], 1
    lea               tmp1d, [mxq+alphaq*4]
    lea               tmp2d, [mxq+alphaq*1]
    shr                 mxd, 10 ; 2
    shr               tmp1d, 10 ; 6
    punpcklbw            m0, m11, m0
    pmaddwd              m0, m10
    movu               xm10, [srcq-2]
    vinserti128         m10, [srcq+6], 1
    punpcklbw            m9, m11, m9
    pmaddwd              m9, m8
    movq                xm8, [filterq+mxq  *8]
    vinserti128          m8, [filterq+tmp1q*8], 1
    lea               tmp1d, [tmp2q+alphaq*4]
    lea                 mxd, [tmp2q+betaq] ; mx += beta
    phaddd               m0, m9 ; 0 1   4 5
    movu                xm9, [srcq+0]
    vinserti128          m9, [srcq+8], 1
    shr               tmp2d, 10 ; 3
    shr               tmp1d, 10 ; 7
    punpcklbw            m8, m11, m8
    pmaddwd              m8, m10
    movq               xm10, [filterq+tmp2q*8]
    vinserti128         m10, [filterq+tmp1q*8], 1
    punpcklbw           m10, m11, m10
    pmaddwd              m9, m10
    add                srcq, ssq
    phaddd               m8, m9 ; 2 3   6 7
    phaddd               m0, m8 ; 0 1 2 3   4 5 6 7
    vpsllvd              m0, m13
    paddd                m0, m12 ; rounded 14-bit result in upper 16 bits of dword
    ret

%macro BIDIR_FN 0
    call .main
    lea            stride3q, [strideq*3]
    jmp                  wq
.w4:
    movq   [dstq          ], xm0
    movhps [dstq+strideq*1], xm0
    vextracti128        xm0, m0, 1
    movq   [dstq+strideq*2], xm0
    movhps [dstq+stride3q ], xm0
    cmp                  hd, 4
    je .ret
    lea                dstq, [dstq+strideq*4]
    movq   [dstq          ], xm1
    movhps [dstq+strideq*1], xm1
    vextracti128        xm1, m1, 1
    movq   [dstq+strideq*2], xm1
    movhps [dstq+stride3q ], xm1
    cmp                  hd, 8
    je .ret
    lea                dstq, [dstq+strideq*4]
    movq   [dstq          ], xm2
    movhps [dstq+strideq*1], xm2
    vextracti128        xm2, m2, 1
    movq   [dstq+strideq*2], xm2
    movhps [dstq+stride3q ], xm2
    lea                dstq, [dstq+strideq*4]
    movq   [dstq          ], xm3
    movhps [dstq+strideq*1], xm3
    vextracti128        xm3, m3, 1
    movq   [dstq+strideq*2], xm3
    movhps [dstq+stride3q ], xm3
.ret:
    RET
.w8:
    mova         [dstq+strideq*0], xm0
    vextracti128 [dstq+strideq*1], m0, 1
    mova         [dstq+strideq*2], xm1
    vextracti128 [dstq+stride3q ], m1, 1
    cmp                  hd, 4
    jne .w8_loop_start
    RET
.w8_loop:
    call .main
    lea                dstq, [dstq+strideq*4]
    mova         [dstq+strideq*0], xm0
    vextracti128 [dstq+strideq*1], m0, 1
    mova         [dstq+strideq*2], xm1
    vextracti128 [dstq+stride3q ], m1, 1
.w8_loop_start:
    lea                dstq, [dstq+strideq*4]
    mova         [dstq+strideq*0], xm2
    vextracti128 [dstq+strideq*1], m2, 1
    mova         [dstq+strideq*2], xm3
    vextracti128 [dstq+stride3q ], m3, 1
    sub                  hd, 8
    jg .w8_loop
    RET
.w16_loop:
    call .main
    lea                dstq, [dstq+strideq*4]
.w16:
    mova   [dstq+strideq*0], m0
    mova   [dstq+strideq*1], m1
    mova   [dstq+strideq*2], m2
    mova   [dstq+stride3q ], m3
    sub                  hd, 4
    jg .w16_loop
    RET
.w32_loop:
    call .main
    lea                dstq, [dstq+strideq*2]
.w32:
    mova [dstq+strideq*0+32*0], m0
    mova [dstq+strideq*0+32*1], m1
    mova [dstq+strideq*1+32*0], m2
    mova [dstq+strideq*1+32*1], m3
    sub                  hd, 2
    jg .w32_loop
    RET
.w64_loop:
    call .main
    add                dstq, strideq
.w64:
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    mova        [dstq+32*2], m2
    mova        [dstq+32*3], m3
    dec                  hd
    jg .w64_loop
    RET
.w128_loop:
    call .main
    add                dstq, strideq
.w128:
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    mova        [dstq+32*2], m2
    mova        [dstq+32*3], m3
    call .main
    mova        [dstq+32*4], m0
    mova        [dstq+32*5], m1
    mova        [dstq+32*6], m2
    mova        [dstq+32*7], m3
    dec                  hd
    jg .w128_loop
    RET
%endmacro

%if WIN64
DECLARE_REG_TMP 5
%else
DECLARE_REG_TMP 7
%endif

cglobal avg_16bpc, 4, 7, 6, dst, stride, tmp1, tmp2, w, h, stride3
%define base r6-avg_avx2_table
    lea                  r6, [avg_avx2_table]
    tzcnt                wd, wm
    mov                 t0d, r6m ; pixel_max
    movsxd               wq, [r6+wq*4]
    shr                 t0d, 11
    vpbroadcastd         m4, [base+bidir_rnd+t0*4]
    vpbroadcastd         m5, [base+bidir_mul+t0*4]
    movifnidn            hd, hm
    add                  wq, r6
    BIDIR_FN
ALIGN function_align
.main:
    mova                 m0, [tmp1q+32*0]
    paddsw               m0, [tmp2q+32*0]
    mova                 m1, [tmp1q+32*1]
    paddsw               m1, [tmp2q+32*1]
    mova                 m2, [tmp1q+32*2]
    paddsw               m2, [tmp2q+32*2]
    mova                 m3, [tmp1q+32*3]
    paddsw               m3, [tmp2q+32*3]
    add               tmp1q, 32*4
    add               tmp2q, 32*4
    pmaxsw               m0, m4
    pmaxsw               m1, m4
    pmaxsw               m2, m4
    pmaxsw               m3, m4
    psubsw               m0, m4
    psubsw               m1, m4
    psubsw               m2, m4
    psubsw               m3, m4
    pmulhw               m0, m5
    pmulhw               m1, m5
    pmulhw               m2, m5
    pmulhw               m3, m5
    ret

cglobal w_avg_16bpc, 4, 7, 9, dst, stride, tmp1, tmp2, w, h, stride3
    lea                  r6, [w_avg_avx2_table]
    tzcnt                wd, wm
    mov                 t0d, r6m ; weight
    vpbroadcastw         m8, r7m ; pixel_max
    vpbroadcastd         m7, [r6-w_avg_avx2_table+pd_65538]
    movsxd               wq, [r6+wq*4]
    paddw                m7, m8
    add                  wq, r6
    lea                 r6d, [t0-16]
    shl                 t0d, 16
    sub                 t0d, r6d ; 16-weight, weight
    pslld                m7, 7
    rorx                r6d, t0d, 30 ; << 2
    test          dword r7m, 0x800
    cmovz               r6d, t0d
    movifnidn            hd, hm
    movd                xm6, r6d
    vpbroadcastd         m6, xm6
    BIDIR_FN
ALIGN function_align
.main:
    mova                 m4, [tmp1q+32*0]
    mova                 m0, [tmp2q+32*0]
    punpckhwd            m5, m0, m4
    punpcklwd            m0, m4
    mova                 m4, [tmp1q+32*1]
    mova                 m1, [tmp2q+32*1]
    pmaddwd              m5, m6
    pmaddwd              m0, m6
    paddd                m5, m7
    paddd                m0, m7
    psrad                m5, 8
    psrad                m0, 8
    packusdw             m0, m5
    punpckhwd            m5, m1, m4
    punpcklwd            m1, m4
    mova                 m4, [tmp1q+32*2]
    mova                 m2, [tmp2q+32*2]
    pmaddwd              m5, m6
    pmaddwd              m1, m6
    paddd                m5, m7
    paddd                m1, m7
    psrad                m5, 8
    psrad                m1, 8
    packusdw             m1, m5
    punpckhwd            m5, m2, m4
    punpcklwd            m2, m4
    mova                 m4, [tmp1q+32*3]
    mova                 m3, [tmp2q+32*3]
    add               tmp1q, 32*4
    add               tmp2q, 32*4
    pmaddwd              m5, m6
    pmaddwd              m2, m6
    paddd                m5, m7
    paddd                m2, m7
    psrad                m5, 8
    psrad                m2, 8
    packusdw             m2, m5
    punpckhwd            m5, m3, m4
    punpcklwd            m3, m4
    pmaddwd              m5, m6
    pmaddwd              m3, m6
    paddd                m5, m7
    paddd                m3, m7
    psrad                m5, 8
    psrad                m3, 8
    packusdw             m3, m5
    pminsw               m0, m8
    pminsw               m1, m8
    pminsw               m2, m8
    pminsw               m3, m8
    ret

cglobal mask_16bpc, 4, 8, 11, dst, stride, tmp1, tmp2, w, h, mask, stride3
%define base r7-mask_avx2_table
    lea                  r7, [mask_avx2_table]
    tzcnt                wd, wm
    mov                 r6d, r7m ; pixel_max
    movifnidn            hd, hm
    shr                 r6d, 11
    movsxd               wq, [r7+wq*4]
    vpbroadcastd         m8, [base+pw_64]
    vpbroadcastd         m9, [base+bidir_rnd+r6*4]
    vpbroadcastd        m10, [base+bidir_mul+r6*4]
    mov               maskq, maskmp
    add                  wq, r7
    BIDIR_FN
ALIGN function_align
.main:
%macro MASK 1
    pmovzxbw             m5, [maskq+16*%1]
    mova                m%1, [tmp1q+32*%1]
    mova                 m6, [tmp2q+32*%1]
    punpckhwd            m4, m%1, m6
    punpcklwd           m%1, m6
    psubw                m7, m8, m5
    punpckhwd            m6, m5, m7 ; m, 64-m
    punpcklwd            m5, m7
    pmaddwd              m4, m6     ; tmp1 * m + tmp2 * (64-m)
    pmaddwd             m%1, m5
    psrad                m4, 5
    psrad               m%1, 5
    packssdw            m%1, m4
    pmaxsw              m%1, m9
    psubsw              m%1, m9
    pmulhw              m%1, m10
%endmacro
    MASK                  0
    MASK                  1
    MASK                  2
    MASK                  3
    add               maskq, 16*4
    add               tmp1q, 32*4
    add               tmp2q, 32*4
    ret

cglobal w_mask_420_16bpc, 4, 8, 16, dst, stride, tmp1, tmp2, w, h, mask, stride3
%define base r7-w_mask_420_avx2_table
    lea                  r7, [w_mask_420_avx2_table]
    tzcnt                wd, wm
    mov                 r6d, r8m ; pixel_max
    movd                xm0, r7m ; sign
    movifnidn            hd, hm
    shr                 r6d, 11
    movsxd               wq, [r7+wq*4]
    vpbroadcastd        m10, [base+pw_27615] ; ((64 - 38) << 10) + 1023 - 32
    vpbroadcastd        m11, [base+pw_64]
    vpbroadcastd        m12, [base+bidir_rnd+r6*4]
    vpbroadcastd        m13, [base+bidir_mul+r6*4]
    movd               xm14, [base+pw_2]
    mov               maskq, maskmp
    psubw              xm14, xm0
    vpbroadcastw        m14, xm14
    add                  wq, r7
    call .main
    lea            stride3q, [strideq*3]
    jmp                  wq
.w4:
    phaddd               m4, m5
    paddw                m4, m14
    psrlw                m4, 2
    packuswb             m4, m4
    vextracti128        xm5, m4, 1
    punpcklwd           xm4, xm5
    movq   [dstq+strideq*0], xm0
    movhps [dstq+strideq*1], xm0
    vextracti128        xm0, m0, 1
    movq   [dstq+strideq*2], xm0
    movhps [dstq+stride3q ], xm0
    mova            [maskq], xm4
    cmp                  hd, 8
    jl .w4_end
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm1
    movhps [dstq+strideq*1], xm1
    vextracti128        xm1, m1, 1
    movq   [dstq+strideq*2], xm1
    movhps [dstq+stride3q ], xm1
    je .w4_end
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm2
    movhps [dstq+strideq*1], xm2
    vextracti128        xm2, m2, 1
    movq   [dstq+strideq*2], xm2
    movhps [dstq+stride3q ], xm2
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm3
    movhps [dstq+strideq*1], xm3
    vextracti128        xm3, m3, 1
    movq   [dstq+strideq*2], xm3
    movhps [dstq+stride3q ], xm3
.w4_end:
    RET
.w8_loop:
    call .main
    lea                dstq, [dstq+strideq*4]
    add               maskq, 16
.w8:
    vperm2i128           m6, m4, m5, 0x21
    vpblendd             m4, m5, 0xf0
    paddw                m4, m14
    paddw                m4, m6
    psrlw                m4, 2
    vextracti128        xm5, m4, 1
    packuswb            xm4, xm5
    mova         [dstq+strideq*0], xm0
    vextracti128 [dstq+strideq*1], m0, 1
    mova         [dstq+strideq*2], xm1
    vextracti128 [dstq+stride3q ], m1, 1
    mova            [maskq], xm4
    sub                  hd, 8
    jl .w8_end
    lea                dstq, [dstq+strideq*4]
    mova         [dstq+strideq*0], xm2
    vextracti128 [dstq+strideq*1], m2, 1
    mova         [dstq+strideq*2], xm3
    vextracti128 [dstq+stride3q ], m3, 1
    jg .w8_loop
.w8_end:
    RET
.w16_loop:
    call .main
    lea                dstq, [dstq+strideq*4]
    add               maskq, 16
.w16:
    punpcklqdq           m6, m4, m5
    punpckhqdq           m4, m5
    paddw                m6, m14
    paddw                m4, m6
    psrlw                m4, 2
    vextracti128        xm5, m4, 1
    packuswb            xm4, xm5
    pshufd              xm4, xm4, q3120
    mova   [dstq+strideq*0], m0
    mova   [dstq+strideq*1], m1
    mova   [dstq+strideq*2], m2
    mova   [dstq+stride3q ], m3
    mova            [maskq], xm4
    sub                  hd, 4
    jg .w16_loop
    RET
.w32_loop:
    call .main
    lea                dstq, [dstq+strideq*4]
    add               maskq, 32
.w32:
    paddw                m4, m14
    paddw                m4, m5
    psrlw               m15, m4, 2
    mova [dstq+strideq*0+32*0], m0
    mova [dstq+strideq*0+32*1], m1
    mova [dstq+strideq*1+32*0], m2
    mova [dstq+strideq*1+32*1], m3
    call .main
    mova                 m6, [deint_shuf]
    paddw                m4, m14
    paddw                m4, m5
    psrlw                m4, 2
    packuswb            m15, m4
    vpermd               m4, m6, m15
    mova [dstq+strideq*2+32*0], m0
    mova [dstq+strideq*2+32*1], m1
    mova [dstq+stride3q +32*0], m2
    mova [dstq+stride3q +32*1], m3
    mova            [maskq], m4
    sub                  hd, 4
    jg .w32_loop
    RET
.w64_loop:
    call .main
    lea                dstq, [dstq+strideq*2]
    add               maskq, 32
.w64:
    paddw                m4, m14
    paddw               m15, m14, m5
    mova [dstq+strideq*0+32*0], m0
    mova [dstq+strideq*0+32*1], m1
    mova [dstq+strideq*0+32*2], m2
    mova [dstq+strideq*0+32*3], m3
    mova            [maskq], m4 ; no available registers
    call .main
    paddw                m4, [maskq]
    mova                 m6, [deint_shuf]
    paddw                m5, m15
    psrlw                m4, 2
    psrlw                m5, 2
    packuswb             m4, m5 ; 0 2 4 6   1 3 5 7
    vpermd               m4, m6, m4
    mova [dstq+strideq*1+32*0], m0
    mova [dstq+strideq*1+32*1], m1
    mova [dstq+strideq*1+32*2], m2
    mova [dstq+strideq*1+32*3], m3
    mova            [maskq], m4
    sub                  hd, 2
    jg .w64_loop
    RET
.w128_loop:
    call .main
    lea                dstq, [dstq+strideq*2]
    add               maskq, 64
.w128:
    paddw                m4, m14
    paddw                m5, m14
    mova [dstq+strideq*0+32*0], m0
    mova [dstq+strideq*0+32*1], m1
    mova [dstq+strideq*0+32*2], m2
    mova [dstq+strideq*0+32*3], m3
    mova       [maskq+32*0], m4
    mova     [dstq+strideq], m5
    call .main
    paddw                m4, m14
    paddw               m15, m14, m5
    mova [dstq+strideq*0+32*4], m0
    mova [dstq+strideq*0+32*5], m1
    mova [dstq+strideq*0+32*6], m2
    mova [dstq+strideq*0+32*7], m3
    mova       [maskq+32*1], m4
    call .main
    paddw                m4, [maskq+32*0]
    paddw                m5, [dstq+strideq]
    mova                 m6, [deint_shuf]
    psrlw                m4, 2
    psrlw                m5, 2
    packuswb             m4, m5
    vpermd               m4, m6, m4
    mova [dstq+strideq*1+32*0], m0
    mova [dstq+strideq*1+32*1], m1
    mova [dstq+strideq*1+32*2], m2
    mova [dstq+strideq*1+32*3], m3
    mova       [maskq+32*0], m4
    call .main
    paddw                m4, [maskq+32*1]
    mova                 m6, [deint_shuf]
    paddw                m5, m15
    psrlw                m4, 2
    psrlw                m5, 2
    packuswb             m4, m5
    vpermd               m4, m6, m4
    mova [dstq+strideq*1+32*4], m0
    mova [dstq+strideq*1+32*5], m1
    mova [dstq+strideq*1+32*6], m2
    mova [dstq+strideq*1+32*7], m3
    mova       [maskq+32*1], m4
    sub                  hd, 2
    jg .w128_loop
    RET
ALIGN function_align
.main:
%macro W_MASK 2-6 11, 12, 13 ; dst/src1, mask/src2, pw_64, rnd, mul
    mova                m%1, [tmp1q+32*%1]
    mova                m%2, [tmp2q+32*%1]
    punpcklwd            m8, m%2, m%1
    punpckhwd            m9, m%2, m%1
    psubsw              m%1, m%2
    pabsw               m%1, m%1
    psubusw              m7, m10, m%1
    psrlw                m7, 10       ; 64-m
    psubw               m%2, m%3, m7  ; m
    punpcklwd           m%1, m7, m%2
    punpckhwd            m7, m%2
    pmaddwd             m%1, m8
    pmaddwd              m7, m9
    psrad               m%1, 5
    psrad                m7, 5
    packssdw            m%1, m7
    pmaxsw              m%1, m%4
    psubsw              m%1, m%4
    pmulhw              m%1, m%5
%endmacro
    W_MASK                0, 4
    W_MASK                1, 5
    phaddw               m4, m5
    W_MASK                2, 5
    W_MASK                3, 6
    phaddw               m5, m6
    add               tmp1q, 32*4
    add               tmp2q, 32*4
    ret

cglobal w_mask_422_16bpc, 4, 8, 16, dst, stride, tmp1, tmp2, w, h, mask, stride3
%define base r7-w_mask_422_avx2_table
    lea                  r7, [w_mask_422_avx2_table]
    tzcnt                wd, wm
    mov                 r6d, r8m ; pixel_max
    vpbroadcastb        m14, r7m ; sign
    movifnidn            hd, hm
    shr                 r6d, 11
    movsxd               wq, [r7+wq*4]
    vpbroadcastd        m10, [base+pw_27615]
    vpbroadcastd        m11, [base+pw_64]
    vpbroadcastd        m12, [base+bidir_rnd+r6*4]
    vpbroadcastd        m13, [base+bidir_mul+r6*4]
    mova                m15, [base+deint_shuf]
    mov               maskq, maskmp
    add                  wq, r7
    call .main
    lea            stride3q, [strideq*3]
    jmp                  wq
.w4:
    movq   [dstq+strideq*0], xm0
    movhps [dstq+strideq*1], xm0
    vextracti128        xm0, m0, 1
    movq   [dstq+strideq*2], xm0
    movhps [dstq+stride3q ], xm0
    cmp                  hd, 8
    jl .w4_end
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm1
    movhps [dstq+strideq*1], xm1
    vextracti128        xm1, m1, 1
    movq   [dstq+strideq*2], xm1
    movhps [dstq+stride3q ], xm1
    je .w4_end
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm2
    movhps [dstq+strideq*1], xm2
    vextracti128        xm2, m2, 1
    movq   [dstq+strideq*2], xm2
    movhps [dstq+stride3q ], xm2
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm3
    movhps [dstq+strideq*1], xm3
    vextracti128        xm3, m3, 1
    movq   [dstq+strideq*2], xm3
    movhps [dstq+stride3q ], xm3
.w4_end:
    RET
.w8_loop:
    call .main
    lea                dstq, [dstq+strideq*4]
.w8:
    mova         [dstq+strideq*0], xm0
    vextracti128 [dstq+strideq*1], m0, 1
    mova         [dstq+strideq*2], xm1
    vextracti128 [dstq+stride3q ], m1, 1
    sub                  hd, 8
    jl .w8_end
    lea                dstq, [dstq+strideq*4]
    mova         [dstq+strideq*0], xm2
    vextracti128 [dstq+strideq*1], m2, 1
    mova         [dstq+strideq*2], xm3
    vextracti128 [dstq+stride3q ], m3, 1
    jg .w8_loop
.w8_end:
    RET
.w16_loop:
    call .main
    lea                dstq, [dstq+strideq*4]
.w16:
    mova   [dstq+strideq*0], m0
    mova   [dstq+strideq*1], m1
    mova   [dstq+strideq*2], m2
    mova   [dstq+stride3q ], m3
    sub                  hd, 4
    jg .w16_loop
    RET
.w32_loop:
    call .main
    lea                dstq, [dstq+strideq*2]
.w32:
    mova [dstq+strideq*0+32*0], m0
    mova [dstq+strideq*0+32*1], m1
    mova [dstq+strideq*1+32*0], m2
    mova [dstq+strideq*1+32*1], m3
    sub                  hd, 2
    jg .w32_loop
    RET
.w64_loop:
    call .main
    add                dstq, strideq
.w64:
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    mova        [dstq+32*2], m2
    mova        [dstq+32*3], m3
    dec                  hd
    jg .w64_loop
    RET
.w128_loop:
    call .main
    add                dstq, strideq
.w128:
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    mova        [dstq+32*2], m2
    mova        [dstq+32*3], m3
    call .main
    mova        [dstq+32*4], m0
    mova        [dstq+32*5], m1
    mova        [dstq+32*6], m2
    mova        [dstq+32*7], m3
    dec                  hd
    jg .w128_loop
    RET
ALIGN function_align
.main:
    W_MASK                0, 4
    W_MASK                1, 5
    phaddw               m4, m5
    W_MASK                2, 5
    W_MASK                3, 6
    phaddw               m5, m6
    add               tmp1q, 32*4
    add               tmp2q, 32*4
    packuswb             m4, m5
    pxor                 m5, m5
    psubb                m4, m14
    pavgb                m4, m5
    vpermd               m4, m15, m4
    mova            [maskq], m4
    add               maskq, 32
    ret

cglobal w_mask_444_16bpc, 4, 8, 11, dst, stride, tmp1, tmp2, w, h, mask, stride3
%define base r7-w_mask_444_avx2_table
    lea                  r7, [w_mask_444_avx2_table]
    tzcnt                wd, wm
    mov                 r6d, r8m ; pixel_max
    movifnidn            hd, hm
    shr                 r6d, 11
    movsxd               wq, [r7+wq*4]
    vpbroadcastd        m10, [base+pw_27615]
    vpbroadcastd         m4, [base+pw_64]
    vpbroadcastd         m5, [base+bidir_rnd+r6*4]
    vpbroadcastd         m6, [base+bidir_mul+r6*4]
    mov               maskq, maskmp
    add                  wq, r7
    call .main
    lea            stride3q, [strideq*3]
    jmp                  wq
.w4:
    movq   [dstq+strideq*0], xm0
    movhps [dstq+strideq*1], xm0
    vextracti128        xm0, m0, 1
    movq   [dstq+strideq*2], xm0
    movhps [dstq+stride3q ], xm0
    cmp                  hd, 8
    jl .w4_end
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm1
    movhps [dstq+strideq*1], xm1
    vextracti128        xm1, m1, 1
    movq   [dstq+strideq*2], xm1
    movhps [dstq+stride3q ], xm1
    je .w4_end
    call .main
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm0
    movhps [dstq+strideq*1], xm0
    vextracti128        xm0, m0, 1
    movq   [dstq+strideq*2], xm0
    movhps [dstq+stride3q ], xm0
    lea                dstq, [dstq+strideq*4]
    movq   [dstq+strideq*0], xm1
    movhps [dstq+strideq*1], xm1
    vextracti128        xm1, m1, 1
    movq   [dstq+strideq*2], xm1
    movhps [dstq+stride3q ], xm1
.w4_end:
    RET
.w8_loop:
    call .main
    lea                dstq, [dstq+strideq*4]
.w8:
    mova         [dstq+strideq*0], xm0
    vextracti128 [dstq+strideq*1], m0, 1
    mova         [dstq+strideq*2], xm1
    vextracti128 [dstq+stride3q ], m1, 1
    sub                  hd, 4
    jg .w8_loop
.w8_end:
    RET
.w16_loop:
    call .main
    lea                dstq, [dstq+strideq*2]
.w16:
    mova   [dstq+strideq*0], m0
    mova   [dstq+strideq*1], m1
    sub                  hd, 2
    jg .w16_loop
    RET
.w32_loop:
    call .main
    add                dstq, strideq
.w32:
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    dec                  hd
    jg .w32_loop
    RET
.w64_loop:
    call .main
    add                dstq, strideq
.w64:
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    call .main
    mova        [dstq+32*2], m0
    mova        [dstq+32*3], m1
    dec                  hd
    jg .w64_loop
    RET
.w128_loop:
    call .main
    add                dstq, strideq
.w128:
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    call .main
    mova        [dstq+32*2], m0
    mova        [dstq+32*3], m1
    call .main
    mova        [dstq+32*4], m0
    mova        [dstq+32*5], m1
    call .main
    mova        [dstq+32*6], m0
    mova        [dstq+32*7], m1
    dec                  hd
    jg .w128_loop
    RET
ALIGN function_align
.main:
    W_MASK                0, 2, 4, 5, 6
    W_MASK                1, 3, 4, 5, 6
    packuswb             m2, m3
    vpermq               m2, m2, q3120
    add               tmp1q, 32*2
    add               tmp2q, 32*2
    mova            [maskq], m2
    add               maskq, 32
    ret

; (a * (64 - m) + b * m + 32) >> 6
; = (((b - a) * m + 32) >> 6) + a
; = (((b - a) * (m << 9) + 16384) >> 15) + a
;   except m << 9 overflows int16_t when m == 64 (which is possible),
;   but if we negate m it works out (-64 << 9 == -32768).
; = (((a - b) * (m * -512) + 16384) >> 15) + a
cglobal blend_16bpc, 3, 7, 7, dst, ds, tmp, w, h, mask
%define base r6-blend_avx2_table
    lea                  r6, [blend_avx2_table]
    tzcnt                wd, wm
    movifnidn            hd, hm
    movsxd               wq, [r6+wq*4]
    movifnidn         maskq, maskmp
    vpbroadcastd         m6, [base+pw_m512]
    add                  wq, r6
    lea                  r6, [dsq*3]
    jmp                  wq
.w4:
    pmovzxbw             m3, [maskq]
    movq                xm0, [dstq+dsq*0]
    movhps              xm0, [dstq+dsq*1]
    vpbroadcastq         m1, [dstq+dsq*2]
    vpbroadcastq         m2, [dstq+r6   ]
    vpblendd             m0, m1, 0x30
    vpblendd             m0, m2, 0xc0
    psubw                m1, m0, [tmpq]
    add               maskq, 16
    add                tmpq, 32
    pmullw               m3, m6
    pmulhrsw             m1, m3
    paddw                m0, m1
    vextracti128        xm1, m0, 1
    movq       [dstq+dsq*0], xm0
    movhps     [dstq+dsq*1], xm0
    movq       [dstq+dsq*2], xm1
    movhps     [dstq+r6   ], xm1
    lea                dstq, [dstq+dsq*4]
    sub                  hd, 4
    jg .w4
    RET
.w8:
    pmovzxbw             m4, [maskq+16*0]
    pmovzxbw             m5, [maskq+16*1]
    mova                xm0, [dstq+dsq*0]
    vinserti128          m0, [dstq+dsq*1], 1
    mova                xm1, [dstq+dsq*2]
    vinserti128          m1, [dstq+r6   ], 1
    psubw                m2, m0, [tmpq+32*0]
    psubw                m3, m1, [tmpq+32*1]
    add               maskq, 16*2
    add                tmpq, 32*2
    pmullw               m4, m6
    pmullw               m5, m6
    pmulhrsw             m2, m4
    pmulhrsw             m3, m5
    paddw                m0, m2
    paddw                m1, m3
    mova         [dstq+dsq*0], xm0
    vextracti128 [dstq+dsq*1], m0, 1
    mova         [dstq+dsq*2], xm1
    vextracti128 [dstq+r6   ], m1, 1
    lea                dstq, [dstq+dsq*4]
    sub                  hd, 4
    jg .w8
    RET
.w16:
    pmovzxbw             m4, [maskq+16*0]
    pmovzxbw             m5, [maskq+16*1]
    mova                 m0,     [dstq+dsq*0]
    psubw                m2, m0, [tmpq+ 32*0]
    mova                 m1,     [dstq+dsq*1]
    psubw                m3, m1, [tmpq+ 32*1]
    add               maskq, 16*2
    add                tmpq, 32*2
    pmullw               m4, m6
    pmullw               m5, m6
    pmulhrsw             m2, m4
    pmulhrsw             m3, m5
    paddw                m0, m2
    paddw                m1, m3
    mova       [dstq+dsq*0], m0
    mova       [dstq+dsq*1], m1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .w16
    RET
.w32:
    pmovzxbw             m4, [maskq+16*0]
    pmovzxbw             m5, [maskq+16*1]
    mova                 m0,     [dstq+32*0]
    psubw                m2, m0, [tmpq+32*0]
    mova                 m1,     [dstq+32*1]
    psubw                m3, m1, [tmpq+32*1]
    add               maskq, 16*2
    add                tmpq, 32*2
    pmullw               m4, m6
    pmullw               m5, m6
    pmulhrsw             m2, m4
    pmulhrsw             m3, m5
    paddw                m0, m2
    paddw                m1, m3
    mova        [dstq+32*0], m0
    mova        [dstq+32*1], m1
    add                dstq, dsq
    dec                  hd
    jg .w32
    RET

INIT_XMM avx2
cglobal blend_v_16bpc, 3, 6, 6, dst, ds, tmp, w, h
%define base r5-blend_v_avx2_table
    lea                  r5, [blend_v_avx2_table]
    tzcnt                wd, wm
    movifnidn            hd, hm
    movsxd               wq, [r5+wq*4]
    add                  wq, r5
    jmp                  wq
.w2:
    vpbroadcastd         m2, [base+obmc_masks_avx2+2*2]
.w2_loop:
    movd                 m0, [dstq+dsq*0]
    pinsrd               m0, [dstq+dsq*1], 1
    movq                 m1, [tmpq]
    add                tmpq, 4*2
    psubw                m1, m0, m1
    pmulhrsw             m1, m2
    paddw                m0, m1
    movd       [dstq+dsq*0], m0
    pextrd     [dstq+dsq*1], m0, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .w2_loop
    RET
.w4:
    vpbroadcastq         m2, [base+obmc_masks_avx2+4*2]
.w4_loop:
    movq                 m0, [dstq+dsq*0]
    movhps               m0, [dstq+dsq*1]
    psubw                m1, m0, [tmpq]
    add                tmpq, 8*2
    pmulhrsw             m1, m2
    paddw                m0, m1
    movq       [dstq+dsq*0], m0
    movhps     [dstq+dsq*1], m0
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .w4_loop
    RET
INIT_YMM avx2
.w8:
    vbroadcasti128       m2, [base+obmc_masks_avx2+8*2]
.w8_loop:
    mova                xm0, [dstq+dsq*0]
    vinserti128          m0, [dstq+dsq*1], 1
    psubw                m1, m0, [tmpq]
    add                tmpq, 16*2
    pmulhrsw             m1, m2
    paddw                m0, m1
    mova         [dstq+dsq*0], xm0
    vextracti128 [dstq+dsq*1], m0, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .w8_loop
    RET
.w16:
    mova                 m4, [base+obmc_masks_avx2+16*2]
.w16_loop:
    mova                 m0,     [dstq+dsq*0]
    psubw                m2, m0, [tmpq+ 32*0]
    mova                 m1,     [dstq+dsq*1]
    psubw                m3, m1, [tmpq+ 32*1]
    add                tmpq, 32*2
    pmulhrsw             m2, m4
    pmulhrsw             m3, m4
    paddw                m0, m2
    paddw                m1, m3
    mova       [dstq+dsq*0], m0
    mova       [dstq+dsq*1], m1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .w16_loop
    RET
.w32:
%if WIN64
    movaps         [rsp+ 8], xmm6
    movaps         [rsp+24], xmm7
%endif
    mova                 m6, [base+obmc_masks_avx2+32*2]
    vbroadcasti128       m7, [base+obmc_masks_avx2+32*3]
.w32_loop:
    mova                 m0,     [dstq+dsq*0+32*0]
    psubw                m3, m0, [tmpq      +32*0]
    mova                xm2,     [dstq+dsq*0+32*1]
    mova                xm5,     [tmpq      +32*1]
    mova                 m1,     [dstq+dsq*1+32*0]
    psubw                m4, m1, [tmpq      +32*2]
    vinserti128          m2,     [dstq+dsq*1+32*1], 1
    vinserti128          m5,     [tmpq      +32*3], 1
    add                tmpq, 32*4
    psubw                m5, m2, m5
    pmulhrsw             m3, m6
    pmulhrsw             m4, m6
    pmulhrsw             m5, m7
    paddw                m0, m3
    paddw                m1, m4
    paddw                m2, m5
    mova         [dstq+dsq*0+32*0], m0
    mova         [dstq+dsq*1+32*0], m1
    mova         [dstq+dsq*0+32*1], xm2
    vextracti128 [dstq+dsq*1+32*1], m2, 1
    lea                dstq, [dstq+dsq*2]
    sub                  hd, 2
    jg .w32_loop
%if WIN64
    movaps             xmm6, [rsp+ 8]
    movaps             xmm7, [rsp+24]
%endif
    RET

%macro BLEND_H_ROW 2-3 0; dst_off, tmp_off, inc_tmp
    mova                 m0,     [dstq+32*(%1+0)]
    psubw                m2, m0, [tmpq+32*(%2+0)]
    mova                 m1,     [dstq+32*(%1+1)]
    psubw                m3, m1, [tmpq+32*(%2+1)]
%if %3
    add                tmpq, 32*%3
%endif
    pmulhrsw             m2, m4
    pmulhrsw             m3, m4
    paddw                m0, m2
    paddw                m1, m3
    mova   [dstq+32*(%1+0)], m0
    mova   [dstq+32*(%1+1)], m1
%endmacro

INIT_XMM avx2
cglobal blend_h_16bpc, 3, 6, 6, dst, ds, tmp, w, h, mask
%define base r5-blend_h_avx2_table
    lea                  r5, [blend_h_avx2_table]
    tzcnt                wd, wm
    mov                  hd, hm
    movsxd               wq, [r5+wq*4]
    add                  wq, r5
    lea               maskq, [base+obmc_masks_avx2+hq*2]
    lea                  hd, [hq*3]
    shr                  hd, 2 ; h * 3/4
    lea               maskq, [maskq+hq*2]
    neg                  hq
    jmp                  wq
.w2:
    movd                 m0, [dstq+dsq*0]
    pinsrd               m0, [dstq+dsq*1], 1
    movd                 m2, [maskq+hq*2]
    movq                 m1, [tmpq]
    add                tmpq, 4*2
    punpcklwd            m2, m2
    psubw                m1, m0, m1
    pmulhrsw             m1, m2
    paddw                m0, m1
    movd       [dstq+dsq*0], m0
    pextrd     [dstq+dsq*1], m0, 1
    lea                dstq, [dstq+dsq*2]
    add                  hq, 2
    jl .w2
    RET
.w4:
    mova                 m3, [blend_shuf]
.w4_loop:
    movq                 m0, [dstq+dsq*0]
    movhps               m0, [dstq+dsq*1]
    movd                 m2, [maskq+hq*2]
    psubw                m1, m0, [tmpq]
    add                tmpq, 8*2
    pshufb               m2, m3
    pmulhrsw             m1, m2
    paddw                m0, m1
    movq       [dstq+dsq*0], m0
    movhps     [dstq+dsq*1], m0
    lea                dstq, [dstq+dsq*2]
    add                  hq, 2
    jl .w4_loop
    RET
INIT_YMM avx2
.w8:
    vbroadcasti128       m3, [blend_shuf]
    shufpd               m3, m3, 0x0c
.w8_loop:
    mova                xm0, [dstq+dsq*0]
    vinserti128          m0, [dstq+dsq*1], 1
    vpbroadcastd         m2, [maskq+hq*2]
    psubw                m1, m0, [tmpq]
    add                tmpq, 16*2
    pshufb               m2, m3
    pmulhrsw             m1, m2
    paddw                m0, m1
    mova         [dstq+dsq*0], xm0
    vextracti128 [dstq+dsq*1], m0, 1
    lea                dstq, [dstq+dsq*2]
    add                  hq, 2
    jl .w8_loop
    RET
.w16:
    vpbroadcastw         m4, [maskq+hq*2]
    vpbroadcastw         m5, [maskq+hq*2+2]
    mova                 m0,     [dstq+dsq*0]
    psubw                m2, m0, [tmpq+ 32*0]
    mova                 m1,     [dstq+dsq*1]
    psubw                m3, m1, [tmpq+ 32*1]
    add                tmpq, 32*2
    pmulhrsw             m2, m4
    pmulhrsw             m3, m5
    paddw                m0, m2
    paddw                m1, m3
    mova       [dstq+dsq*0], m0
    mova       [dstq+dsq*1], m1
    lea                dstq, [dstq+dsq*2]
    add                  hq, 2
    jl .w16
    RET
.w32:
    vpbroadcastw         m4, [maskq+hq*2]
    BLEND_H_ROW           0, 0, 2
    add                dstq, dsq
    inc                  hq
    jl .w32
    RET
.w64:
    vpbroadcastw         m4, [maskq+hq*2]
    BLEND_H_ROW           0, 0
    BLEND_H_ROW           2, 2, 4
    add                dstq, dsq
    inc                  hq
    jl .w64
    RET
.w128:
    vpbroadcastw         m4, [maskq+hq*2]
    BLEND_H_ROW           0,  0
    BLEND_H_ROW           2,  2, 8
    BLEND_H_ROW           4, -4
    BLEND_H_ROW           6, -2
    add                dstq, dsq
    inc                  hq
    jl .w128
    RET

cglobal emu_edge_16bpc, 10, 13, 1, bw, bh, iw, ih, x, y, dst, dstride, src, sstride, \
                                   bottomext, rightext
    ; we assume that the buffer (stride) is larger than width, so we can
    ; safely overwrite by a few bytes

    ; ref += iclip(y, 0, ih - 1) * PXSTRIDE(ref_stride)
    xor                r12d, r12d
    lea                 r10, [ihq-1]
    cmp                  yq, ihq
    cmovs               r10, yq
    test                 yq, yq
    cmovs               r10, r12
    imul                r10, sstrideq
    add                srcq, r10

    ; ref += iclip(x, 0, iw - 1)
    lea                 r10, [iwq-1]
    cmp                  xq, iwq
    cmovs               r10, xq
    test                 xq, xq
    cmovs               r10, r12
    lea                srcq, [srcq+r10*2]

    ; bottom_ext = iclip(y + bh - ih, 0, bh - 1)
    lea          bottomextq, [yq+bhq]
    sub          bottomextq, ihq
    lea                  r3, [bhq-1]
    cmovs        bottomextq, r12

    DEFINE_ARGS bw, bh, iw, ih, x, topext, dst, dstride, src, sstride, \
                bottomext, rightext

    ; top_ext = iclip(-y, 0, bh - 1)
    neg             topextq
    cmovs           topextq, r12
    cmp          bottomextq, bhq
    cmovns       bottomextq, r3
    cmp             topextq, bhq
    cmovg           topextq, r3

    ; right_ext = iclip(x + bw - iw, 0, bw - 1)
    lea           rightextq, [xq+bwq]
    sub           rightextq, iwq
    lea                  r2, [bwq-1]
    cmovs         rightextq, r12

    DEFINE_ARGS bw, bh, iw, ih, leftext, topext, dst, dstride, src, sstride, \
                bottomext, rightext

    ; left_ext = iclip(-x, 0, bw - 1)
    neg            leftextq
    cmovs          leftextq, r12
    cmp           rightextq, bwq
    cmovns        rightextq, r2
    cmp            leftextq, bwq
    cmovns         leftextq, r2

    DEFINE_ARGS bw, centerh, centerw, dummy, leftext, topext, \
                dst, dstride, src, sstride, bottomext, rightext

    ; center_h = bh - top_ext - bottom_ext
    lea                  r3, [bottomextq+topextq]
    sub            centerhq, r3

    ; blk += top_ext * PXSTRIDE(dst_stride)
    mov                  r2, topextq
    imul                 r2, dstrideq
    add                dstq, r2
    mov                 r9m, dstq

    ; center_w = bw - left_ext - right_ext
    mov            centerwq, bwq
    lea                  r3, [rightextq+leftextq]
    sub            centerwq, r3

%macro v_loop 3 ; need_left_ext, need_right_ext, suffix
.v_loop_%3:
%if %1
    ; left extension
    xor                  r3, r3
    vpbroadcastw         m0, [srcq]
.left_loop_%3:
    mova        [dstq+r3*2], m0
    add                  r3, 16
    cmp                  r3, leftextq
    jl .left_loop_%3

    ; body
    lea                 r12, [dstq+leftextq*2]
%endif
    xor                  r3, r3
.body_loop_%3:
    movu                 m0, [srcq+r3*2]
%if %1
    movu         [r12+r3*2], m0
%else
    movu        [dstq+r3*2], m0
%endif
    add                  r3, 16
    cmp                  r3, centerwq
    jl .body_loop_%3

%if %2
    ; right extension
%if %1
    lea                 r12, [r12+centerwq*2]
%else
    lea                 r12, [dstq+centerwq*2]
%endif
    xor                  r3, r3
    vpbroadcastw         m0, [srcq+centerwq*2-2]
.right_loop_%3:
    movu         [r12+r3*2], m0
    add                  r3, 16
    cmp                  r3, rightextq
    jl .right_loop_%3

%endif
    add                dstq, dstrideq
    add                srcq, sstrideq
    dec            centerhq
    jg .v_loop_%3
%endmacro

    test           leftextq, leftextq
    jnz .need_left_ext
    test          rightextq, rightextq
    jnz .need_right_ext
    v_loop                0, 0, 0
    jmp .body_done

.need_left_ext:
    test          rightextq, rightextq
    jnz .need_left_right_ext
    v_loop                1, 0, 1
    jmp .body_done

.need_left_right_ext:
    v_loop                1, 1, 2
    jmp .body_done

.need_right_ext:
    v_loop                0, 1, 3

.body_done:
    ; bottom edge extension
    test         bottomextq, bottomextq
    jz .top
    mov                srcq, dstq
    sub                srcq, dstrideq
    xor                  r1, r1
.bottom_x_loop:
    mova                 m0, [srcq+r1*2]
    lea                  r3, [dstq+r1*2]
    mov                  r4, bottomextq
.bottom_y_loop:
    mova               [r3], m0
    add                  r3, dstrideq
    dec                  r4
    jg .bottom_y_loop
    add                  r1, 16
    cmp                  r1, bwq
    jl .bottom_x_loop

.top:
    ; top edge extension
    test            topextq, topextq
    jz .end
    mov                srcq, r9m
    mov                dstq, dstm
    xor                  r1, r1
.top_x_loop:
    mova                 m0, [srcq+r1*2]
    lea                  r3, [dstq+r1*2]
    mov                  r4, topextq
.top_y_loop:
    mova               [r3], m0
    add                  r3, dstrideq
    dec                  r4
    jg .top_y_loop
    add                  r1, 16
    cmp                  r1, bwq
    jl .top_x_loop

.end:
    RET

cglobal resize_16bpc, 6, 12, 16, dst, dst_stride, src, src_stride, \
                                 dst_w, h, src_w, dx, mx0, pxmax
    sub          dword mx0m, 4<<14
    sub        dword src_wm, 8
    vpbroadcastd         m5, dxm
    vpbroadcastd         m8, mx0m
    vpbroadcastd         m6, src_wm
 DEFINE_ARGS dst, dst_stride, src, src_stride, dst_w, h, x, _, _, pxmax
    LEA                  r7, $$
%define base r7-$$
    vpbroadcastd         m3, [base+pd_64]
    vpbroadcastw        xm7, pxmaxm
    pmaddwd              m2, m5, [base+rescale_mul] ; dx*[0,1,2,3,4,5,6,7]
    pslld                m5, 3                      ; dx*8
    pslld                m6, 14
    paddd                m8, m2                     ; mx+[0..7]*dx
.loop_y:
    xor                  xd, xd
    mova                 m4, m8             ; per-line working version of mx
.loop_x:
    vpbroadcastd        m10, [base+pd_63]
    pxor                 m2, m2
    pmaxsd               m0, m4, m2
    psrad                m9, m4, 8          ; filter offset (unmasked)
    pminsd               m0, m6             ; iclip(mx, 0, src_w-8)
    psubd                m1, m4, m0         ; pshufb offset
    psrad                m0, 14             ; clipped src_x offset
    psrad                m1, 14             ; pshufb edge_emu offset
    pand                 m9, m10            ; filter offset (masked)
    ; load source pixels
    movd                r8d, xm0
    pextrd              r9d, xm0, 1
    pextrd             r10d, xm0, 2
    pextrd             r11d, xm0, 3
    vextracti128        xm0, m0, 1
    movu               xm10, [srcq+r8*2]
    movu               xm11, [srcq+r9*2]
    movu               xm12, [srcq+r10*2]
    movu               xm13, [srcq+r11*2]
    movd                r8d, xm0
    pextrd              r9d, xm0, 1
    pextrd             r10d, xm0, 2
    pextrd             r11d, xm0, 3
    vinserti128         m10, [srcq+r8*2], 1
    vinserti128         m11, [srcq+r9*2], 1
    vinserti128         m12, [srcq+r10*2], 1
    vinserti128         m13, [srcq+r11*2], 1
    ptest                m1, m1
    jz .filter
    movq                 r9, xm1
    pextrq              r11, xm1, 1
    movsxd               r8, r9d
    sar                  r9, 32
    movsxd              r10, r11d
    sar                 r11, 32
    vextracti128        xm1, m1, 1
    movu               xm14, [base+resize_shuf+8+r8*2]
    movu               xm15, [base+resize_shuf+8+r9*2]
    movu                xm0, [base+resize_shuf+8+r10*2]
    movu                xm2, [base+resize_shuf+8+r11*2]
    movq                 r9, xm1
    pextrq              r11, xm1, 1
    movsxd               r8, r9d
    sar                  r9, 32
    movsxd              r10, r11d
    sar                 r11, 32
    vinserti128         m14, [base+resize_shuf+8+r8*2], 1
    vinserti128         m15, [base+resize_shuf+8+r9*2], 1
    vinserti128          m0, [base+resize_shuf+8+r10*2], 1
    vinserti128          m2, [base+resize_shuf+8+r11*2], 1
    pshufb              m10, m14
    pshufb              m11, m15
    pshufb              m12, m0
    pshufb              m13, m2
.filter:
    movd                r8d, xm9
    pextrd              r9d, xm9, 1
    pextrd             r10d, xm9, 2
    pextrd             r11d, xm9, 3
    vextracti128        xm9, m9, 1
    movq               xm14, [base+resize_filter+r8*8]
    movq               xm15, [base+resize_filter+r9*8]
    movq                xm0, [base+resize_filter+r10*8]
    movq                xm2, [base+resize_filter+r11*8]
    movd                r8d, xm9
    pextrd              r9d, xm9, 1
    pextrd             r10d, xm9, 2
    pextrd             r11d, xm9, 3
    movhps             xm14, [base+resize_filter+r8*8]
    movhps             xm15, [base+resize_filter+r9*8]
    movhps              xm0, [base+resize_filter+r10*8]
    movhps              xm2, [base+resize_filter+r11*8]
    pmovsxbw            m14, xm14
    pmovsxbw            m15, xm15
    pmovsxbw             m0, xm0
    pmovsxbw             m2, xm2
    pmaddwd             m10, m14
    pmaddwd             m11, m15
    pmaddwd             m12, m0
    pmaddwd             m13, m2
    phaddd              m10, m11
    phaddd              m12, m13
    phaddd              m10, m12
    psubd               m10, m3, m10
    psrad               m10, 7
    vextracti128        xm0, m10, 1
    packusdw           xm10, xm0
    pminsw             xm10, xm7
    mova        [dstq+xq*2], xm10
    paddd                m4, m5
    add                  xd, 8
    cmp                  xd, dst_wd
    jl .loop_x
    add                dstq, dst_strideq
    add                srcq, src_strideq
    dec                  hd
    jg .loop_y
    RET

%endif ; ARCH_X86_64