1 |
;/***************************************************************************** |
;/************************************************************************** |
2 |
; * |
; * |
3 |
; * XVID MPEG-4 VIDEO CODEC |
; * XVID MPEG-4 VIDEO CODEC |
4 |
; * mmx 8x8 block-based halfpel interpolation |
; * mmx 8x8 block-based halfpel interpolation |
5 |
; * |
; * |
6 |
; * Copyright(C) 2002 Peter Ross <pross@xvid.org> |
; * This program is free software; you can redistribute it and/or modify |
7 |
; * |
; * it under the terms of the GNU General Public License as published by |
|
; * This file is part of XviD, a free MPEG-4 video encoder/decoder |
|
|
; * |
|
|
; * XviD is free software; you can redistribute it and/or modify it |
|
|
; * under the terms of the GNU General Public License as published by |
|
8 |
; * the Free Software Foundation; either version 2 of the License, or |
; * the Free Software Foundation; either version 2 of the License, or |
9 |
; * (at your option) any later version. |
; * (at your option) any later version. |
10 |
; * |
; * |
15 |
; * |
; * |
16 |
; * You should have received a copy of the GNU General Public License |
; * You should have received a copy of the GNU General Public License |
17 |
; * along with this program; if not, write to the Free Software |
; * along with this program; if not, write to the Free Software |
18 |
; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
; * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
|
; * |
|
|
; * Under section 8 of the GNU General Public License, the copyright |
|
|
; * holders of XVID explicitly forbid distribution in the following |
|
|
; * countries: |
|
|
; * |
|
|
; * - Japan |
|
|
; * - United States of America |
|
19 |
; * |
; * |
20 |
; * Linking XviD statically or dynamically with other modules is making a |
; *************************************************************************/ |
21 |
; * combined work based on XviD. Thus, the terms and conditions of the |
|
22 |
; * GNU General Public License cover the whole combination. |
;/************************************************************************** |
|
; * |
|
|
; * As a special exception, the copyright holders of XviD give you |
|
|
; * permission to link XviD with independent modules that communicate with |
|
|
; * XviD solely through the VFW1.1 and DShow interfaces, regardless of the |
|
|
; * license terms of these independent modules, and to copy and distribute |
|
|
; * the resulting combined work under terms of your choice, provided that |
|
|
; * every copy of the combined work is accompanied by a complete copy of |
|
|
; * the source code of XviD (the version of XviD used to produce the |
|
|
; * combined work), being distributed under the terms of the GNU General |
|
|
; * Public License plus this exception. An independent module is a module |
|
|
; * which is not derived from or based on XviD. |
|
23 |
; * |
; * |
24 |
; * Note that people who make modified versions of XviD are not obligated |
; * History: |
|
; * to grant this special exception for their modified versions; it is |
|
|
; * their choice whether to do so. The GNU General Public License gives |
|
|
; * permission to release a modified version without this exception; this |
|
|
; * exception also makes it possible to release a modified version which |
|
|
; * carries forward this exception. |
|
25 |
; * |
; * |
26 |
; * $Id: interpolate8x8_mmx.asm,v 1.11 2002-11-17 00:20:30 edgomez Exp $ |
; * 05.10.2002 added some qpel mmx code - Isibaar |
27 |
|
; * 06.07.2002 mmx cleanup - Isibaar |
28 |
|
; * 22.12.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
29 |
; * |
; * |
30 |
; ****************************************************************************/ |
; *************************************************************************/ |
31 |
|
|
32 |
|
|
33 |
bits 32 |
bits 32 |
34 |
|
|
46 |
align 16 |
align 16 |
47 |
|
|
48 |
;=========================================================================== |
;=========================================================================== |
49 |
|
; (16 - r) rounding table |
50 |
|
;=========================================================================== |
51 |
|
|
52 |
|
rounding_lowpass_mmx |
53 |
|
times 4 dw 16 |
54 |
|
times 4 dw 15 |
55 |
|
|
56 |
|
;=========================================================================== |
57 |
; (1 - r) rounding table |
; (1 - r) rounding table |
58 |
;=========================================================================== |
;=========================================================================== |
59 |
|
|
72 |
mmx_one |
mmx_one |
73 |
times 8 db 1 |
times 8 db 1 |
74 |
|
|
75 |
|
mmx_two |
76 |
|
times 8 db 2 |
77 |
|
|
78 |
|
mmx_three |
79 |
|
times 8 db 3 |
80 |
|
|
81 |
|
mmx_five |
82 |
|
times 4 dw 5 |
83 |
|
|
84 |
|
mmx_mask |
85 |
|
times 8 db 254 |
86 |
|
|
87 |
|
mmx_mask2 |
88 |
|
times 8 db 252 |
89 |
|
|
90 |
section .text |
section .text |
91 |
|
|
92 |
%macro CALC_AVG 6 |
%macro CALC_AVG 6 |
319 |
pop esi |
pop esi |
320 |
|
|
321 |
ret |
ret |
322 |
|
|
323 |
|
;=========================================================================== |
324 |
|
; |
325 |
|
; void interpolate8x8_avg2_mmx(uint8_t const *dst, |
326 |
|
; const uint8_t * const src1, |
327 |
|
; const uint8_t * const src2, |
328 |
|
; const uint32_t stride, |
329 |
|
; const uint32_t rounding, |
330 |
|
; const uint32_t height); |
331 |
|
; |
332 |
|
;=========================================================================== |
333 |
|
|
334 |
|
%macro AVG2_MMX_RND0 0 |
335 |
|
movq mm0, [eax] ; src1 -> mm0 |
336 |
|
movq mm1, [ebx] ; src2 -> mm1 |
337 |
|
|
338 |
|
movq mm4, [eax+edx] |
339 |
|
movq mm5, [ebx+edx] |
340 |
|
|
341 |
|
movq mm2, mm0 ; src1 -> mm2 |
342 |
|
movq mm3, mm1 ; src2 -> mm3 |
343 |
|
|
344 |
|
pand mm2, mm7 ; isolate the lsb |
345 |
|
pand mm3, mm7 ; isolate the lsb |
346 |
|
|
347 |
|
por mm2, mm3 ; ODD(src1) OR ODD(src2) -> mm2 |
348 |
|
|
349 |
|
movq mm3, mm4 |
350 |
|
movq mm6, mm5 |
351 |
|
|
352 |
|
pand mm3, mm7 |
353 |
|
pand mm6, mm7 |
354 |
|
|
355 |
|
por mm3, mm6 |
356 |
|
|
357 |
|
pand mm0, [mmx_mask] |
358 |
|
pand mm1, [mmx_mask] |
359 |
|
pand mm4, [mmx_mask] |
360 |
|
pand mm5, [mmx_mask] |
361 |
|
|
362 |
|
psrlq mm0, 1 ; src1 / 2 |
363 |
|
psrlq mm1, 1 ; src2 / 2 |
364 |
|
|
365 |
|
psrlq mm4, 1 |
366 |
|
psrlq mm5, 1 |
367 |
|
|
368 |
|
paddb mm0, mm1 ; src1/2 + src2/2 -> mm0 |
369 |
|
paddb mm0, mm2 ; correct rounding error |
370 |
|
|
371 |
|
paddb mm4, mm5 |
372 |
|
paddb mm4, mm3 |
373 |
|
|
374 |
|
lea eax,[eax+2*edx] |
375 |
|
lea ebx,[ebx+2*edx] |
376 |
|
|
377 |
|
movq [ecx], mm0 ; (src1 + src2 + 1) / 2 -> dst |
378 |
|
movq [ecx+edx], mm4 |
379 |
|
%endmacro |
380 |
|
|
381 |
|
%macro AVG2_MMX_RND1 0 |
382 |
|
movq mm0, [eax] ; src1 -> mm0 |
383 |
|
movq mm1, [ebx] ; src2 -> mm1 |
384 |
|
|
385 |
|
movq mm4, [eax+edx] |
386 |
|
movq mm5, [ebx+edx] |
387 |
|
|
388 |
|
movq mm2, mm0 ; src1 -> mm2 |
389 |
|
movq mm3, mm1 ; src2 -> mm3 |
390 |
|
|
391 |
|
pand mm2, mm7 ; isolate the lsb |
392 |
|
pand mm3, mm7 ; isolate the lsb |
393 |
|
|
394 |
|
pand mm2, mm3 ; ODD(src1) AND ODD(src2) -> mm2 |
395 |
|
|
396 |
|
movq mm3, mm4 |
397 |
|
movq mm6, mm5 |
398 |
|
|
399 |
|
pand mm3, mm7 |
400 |
|
pand mm6, mm7 |
401 |
|
|
402 |
|
pand mm3, mm6 |
403 |
|
|
404 |
|
pand mm0, [mmx_mask] |
405 |
|
pand mm1, [mmx_mask] |
406 |
|
pand mm4, [mmx_mask] |
407 |
|
pand mm5, [mmx_mask] |
408 |
|
|
409 |
|
psrlq mm0, 1 ; src1 / 2 |
410 |
|
psrlq mm1, 1 ; src2 / 2 |
411 |
|
|
412 |
|
psrlq mm4, 1 |
413 |
|
psrlq mm5, 1 |
414 |
|
|
415 |
|
paddb mm0, mm1 ; src1/2 + src2/2 -> mm0 |
416 |
|
paddb mm0, mm2 ; correct rounding error |
417 |
|
|
418 |
|
paddb mm4, mm5 |
419 |
|
paddb mm4, mm3 |
420 |
|
|
421 |
|
lea eax,[eax+2*edx] |
422 |
|
lea ebx,[ebx+2*edx] |
423 |
|
|
424 |
|
movq [ecx], mm0 ; (src1 + src2 + 1) / 2 -> dst |
425 |
|
movq [ecx+edx], mm4 |
426 |
|
%endmacro |
427 |
|
|
428 |
|
align 16 |
429 |
|
cglobal interpolate8x8_avg2_mmx |
430 |
|
interpolate8x8_avg2_mmx |
431 |
|
|
432 |
|
push ebx |
433 |
|
|
434 |
|
mov eax, [esp + 4 + 20] ; rounding |
435 |
|
test eax, eax |
436 |
|
|
437 |
|
jnz near .rounding1 |
438 |
|
|
439 |
|
mov eax, [esp + 4 + 24] ; height -> eax |
440 |
|
sub eax, 8 |
441 |
|
test eax, eax |
442 |
|
|
443 |
|
mov ecx, [esp + 4 + 4] ; dst -> edi |
444 |
|
mov eax, [esp + 4 + 8] ; src1 -> esi |
445 |
|
mov ebx, [esp + 4 + 12] ; src2 -> eax |
446 |
|
mov edx, [esp + 4 + 16] ; stride -> edx |
447 |
|
|
448 |
|
movq mm7, [mmx_one] |
449 |
|
|
450 |
|
jz near .start0 |
451 |
|
|
452 |
|
AVG2_MMX_RND0 |
453 |
|
lea ecx, [ecx+2*edx] |
454 |
|
|
455 |
|
.start0 |
456 |
|
|
457 |
|
AVG2_MMX_RND0 |
458 |
|
lea ecx, [ecx+2*edx] |
459 |
|
AVG2_MMX_RND0 |
460 |
|
lea ecx, [ecx+2*edx] |
461 |
|
AVG2_MMX_RND0 |
462 |
|
lea ecx, [ecx+2*edx] |
463 |
|
AVG2_MMX_RND0 |
464 |
|
|
465 |
|
pop ebx |
466 |
|
ret |
467 |
|
|
468 |
|
.rounding1 |
469 |
|
mov eax, [esp + 4 + 24] ; height -> eax |
470 |
|
sub eax, 8 |
471 |
|
test eax, eax |
472 |
|
|
473 |
|
mov ecx, [esp + 4 + 4] ; dst -> edi |
474 |
|
mov eax, [esp + 4 + 8] ; src1 -> esi |
475 |
|
mov ebx, [esp + 4 + 12] ; src2 -> eax |
476 |
|
mov edx, [esp + 4 + 16] ; stride -> edx |
477 |
|
|
478 |
|
movq mm7, [mmx_one] |
479 |
|
|
480 |
|
jz near .start1 |
481 |
|
|
482 |
|
AVG2_MMX_RND1 |
483 |
|
lea ecx, [ecx+2*edx] |
484 |
|
|
485 |
|
.start1 |
486 |
|
|
487 |
|
AVG2_MMX_RND1 |
488 |
|
lea ecx, [ecx+2*edx] |
489 |
|
AVG2_MMX_RND1 |
490 |
|
lea ecx, [ecx+2*edx] |
491 |
|
AVG2_MMX_RND1 |
492 |
|
lea ecx, [ecx+2*edx] |
493 |
|
AVG2_MMX_RND1 |
494 |
|
|
495 |
|
pop ebx |
496 |
|
ret |
497 |
|
|
498 |
|
|
499 |
|
;=========================================================================== |
500 |
|
; |
501 |
|
; void interpolate8x8_avg4_mmx(uint8_t const *dst, |
502 |
|
; const uint8_t * const src1, |
503 |
|
; const uint8_t * const src2, |
504 |
|
; const uint8_t * const src3, |
505 |
|
; const uint8_t * const src4, |
506 |
|
; const uint32_t stride, |
507 |
|
; const uint32_t rounding); |
508 |
|
; |
509 |
|
;=========================================================================== |
510 |
|
|
511 |
|
%macro AVG4_MMX_RND0 0 |
512 |
|
movq mm0, [eax] ; src1 -> mm0 |
513 |
|
movq mm1, [ebx] ; src2 -> mm1 |
514 |
|
|
515 |
|
movq mm2, mm0 |
516 |
|
movq mm3, mm1 |
517 |
|
|
518 |
|
pand mm2, [mmx_three] |
519 |
|
pand mm3, [mmx_three] |
520 |
|
|
521 |
|
pand mm0, [mmx_mask2] |
522 |
|
pand mm1, [mmx_mask2] |
523 |
|
|
524 |
|
psrlq mm0, 2 |
525 |
|
psrlq mm1, 2 |
526 |
|
|
527 |
|
lea eax, [eax+edx] |
528 |
|
lea ebx, [ebx+edx] |
529 |
|
|
530 |
|
paddb mm0, mm1 |
531 |
|
paddb mm2, mm3 |
532 |
|
|
533 |
|
movq mm4, [esi] ; src3 -> mm0 |
534 |
|
movq mm5, [edi] ; src4 -> mm1 |
535 |
|
|
536 |
|
movq mm1, mm4 |
537 |
|
movq mm3, mm5 |
538 |
|
|
539 |
|
pand mm1, [mmx_three] |
540 |
|
pand mm3, [mmx_three] |
541 |
|
|
542 |
|
pand mm4, [mmx_mask2] |
543 |
|
pand mm5, [mmx_mask2] |
544 |
|
|
545 |
|
psrlq mm4, 2 |
546 |
|
psrlq mm5, 2 |
547 |
|
|
548 |
|
paddb mm4, mm5 |
549 |
|
paddb mm0, mm4 |
550 |
|
|
551 |
|
paddb mm1, mm3 |
552 |
|
paddb mm2, mm1 |
553 |
|
|
554 |
|
paddb mm2, [mmx_two] |
555 |
|
pand mm2, [mmx_mask2] |
556 |
|
|
557 |
|
psrlq mm2, 2 |
558 |
|
paddb mm0, mm2 |
559 |
|
|
560 |
|
lea esi, [esi+edx] |
561 |
|
lea edi, [edi+edx] |
562 |
|
|
563 |
|
movq [ecx], mm0 ; (src1 + src2 + src3 + src4 + 2) / 4 -> dst |
564 |
|
%endmacro |
565 |
|
|
566 |
|
%macro AVG4_MMX_RND1 0 |
567 |
|
movq mm0, [eax] ; src1 -> mm0 |
568 |
|
movq mm1, [ebx] ; src2 -> mm1 |
569 |
|
|
570 |
|
movq mm2, mm0 |
571 |
|
movq mm3, mm1 |
572 |
|
|
573 |
|
pand mm2, [mmx_three] |
574 |
|
pand mm3, [mmx_three] |
575 |
|
|
576 |
|
pand mm0, [mmx_mask2] |
577 |
|
pand mm1, [mmx_mask2] |
578 |
|
|
579 |
|
psrlq mm0, 2 |
580 |
|
psrlq mm1, 2 |
581 |
|
|
582 |
|
lea eax,[eax+edx] |
583 |
|
lea ebx,[ebx+edx] |
584 |
|
|
585 |
|
paddb mm0, mm1 |
586 |
|
paddb mm2, mm3 |
587 |
|
|
588 |
|
movq mm4, [esi] ; src3 -> mm0 |
589 |
|
movq mm5, [edi] ; src4 -> mm1 |
590 |
|
|
591 |
|
movq mm1, mm4 |
592 |
|
movq mm3, mm5 |
593 |
|
|
594 |
|
pand mm1, [mmx_three] |
595 |
|
pand mm3, [mmx_three] |
596 |
|
|
597 |
|
pand mm4, [mmx_mask2] |
598 |
|
pand mm5, [mmx_mask2] |
599 |
|
|
600 |
|
psrlq mm4, 2 |
601 |
|
psrlq mm5, 2 |
602 |
|
|
603 |
|
paddb mm4, mm5 |
604 |
|
paddb mm0, mm4 |
605 |
|
|
606 |
|
paddb mm1, mm3 |
607 |
|
paddb mm2, mm1 |
608 |
|
|
609 |
|
paddb mm2, [mmx_one] |
610 |
|
pand mm2, [mmx_mask2] |
611 |
|
|
612 |
|
psrlq mm2, 2 |
613 |
|
paddb mm0, mm2 |
614 |
|
|
615 |
|
lea esi,[esi+edx] |
616 |
|
lea edi,[edi+edx] |
617 |
|
|
618 |
|
movq [ecx], mm0 ; (src1 + src2 + src3 + src4 + 2) / 4 -> dst |
619 |
|
%endmacro |
620 |
|
|
621 |
|
align 16 |
622 |
|
cglobal interpolate8x8_avg4_mmx |
623 |
|
interpolate8x8_avg4_mmx |
624 |
|
|
625 |
|
push ebx |
626 |
|
push edi |
627 |
|
push esi |
628 |
|
|
629 |
|
mov eax, [esp + 12 + 28] ; rounding |
630 |
|
|
631 |
|
test eax, eax |
632 |
|
|
633 |
|
mov ecx, [esp + 12 + 4] ; dst -> edi |
634 |
|
mov eax, [esp + 12 + 8] ; src1 -> esi |
635 |
|
mov ebx, [esp + 12 + 12] ; src2 -> eax |
636 |
|
mov esi, [esp + 12 + 16] ; src3 -> esi |
637 |
|
mov edi, [esp + 12 + 20] ; src4 -> edi |
638 |
|
mov edx, [esp + 12 + 24] ; stride -> edx |
639 |
|
|
640 |
|
movq mm7, [mmx_one] |
641 |
|
|
642 |
|
jnz near .rounding1 |
643 |
|
|
644 |
|
AVG4_MMX_RND0 |
645 |
|
lea ecx, [ecx+edx] |
646 |
|
AVG4_MMX_RND0 |
647 |
|
lea ecx, [ecx+edx] |
648 |
|
AVG4_MMX_RND0 |
649 |
|
lea ecx, [ecx+edx] |
650 |
|
AVG4_MMX_RND0 |
651 |
|
lea ecx, [ecx+edx] |
652 |
|
AVG4_MMX_RND0 |
653 |
|
lea ecx, [ecx+edx] |
654 |
|
AVG4_MMX_RND0 |
655 |
|
lea ecx, [ecx+edx] |
656 |
|
AVG4_MMX_RND0 |
657 |
|
lea ecx, [ecx+edx] |
658 |
|
AVG4_MMX_RND0 |
659 |
|
|
660 |
|
pop esi |
661 |
|
pop edi |
662 |
|
pop ebx |
663 |
|
ret |
664 |
|
|
665 |
|
.rounding1 |
666 |
|
AVG4_MMX_RND1 |
667 |
|
lea ecx, [ecx+edx] |
668 |
|
AVG4_MMX_RND1 |
669 |
|
lea ecx, [ecx+edx] |
670 |
|
AVG4_MMX_RND1 |
671 |
|
lea ecx, [ecx+edx] |
672 |
|
AVG4_MMX_RND1 |
673 |
|
lea ecx, [ecx+edx] |
674 |
|
AVG4_MMX_RND1 |
675 |
|
lea ecx, [ecx+edx] |
676 |
|
AVG4_MMX_RND1 |
677 |
|
lea ecx, [ecx+edx] |
678 |
|
AVG4_MMX_RND1 |
679 |
|
lea ecx, [ecx+edx] |
680 |
|
AVG4_MMX_RND1 |
681 |
|
|
682 |
|
pop esi |
683 |
|
pop edi |
684 |
|
pop ebx |
685 |
|
ret |
686 |
|
|
687 |
|
|
688 |
|
;=========================================================================== |
689 |
|
; |
690 |
|
; void interpolate8x8_6tap_lowpass_h_mmx(uint8_t const *dst, |
691 |
|
; const uint8_t * const src, |
692 |
|
; const uint32_t stride, |
693 |
|
; const uint32_t rounding); |
694 |
|
; |
695 |
|
;=========================================================================== |
696 |
|
|
697 |
|
%macro LOWPASS_6TAP_H_MMX 0 |
698 |
|
movq mm0, [eax] |
699 |
|
movq mm2, [eax+1] |
700 |
|
|
701 |
|
movq mm1, mm0 |
702 |
|
movq mm3, mm2 |
703 |
|
|
704 |
|
punpcklbw mm0, mm7 |
705 |
|
punpcklbw mm2, mm7 |
706 |
|
|
707 |
|
punpckhbw mm1, mm7 |
708 |
|
punpckhbw mm3, mm7 |
709 |
|
|
710 |
|
paddw mm0, mm2 |
711 |
|
paddw mm1, mm3 |
712 |
|
|
713 |
|
psllw mm0, 2 |
714 |
|
psllw mm1, 2 |
715 |
|
|
716 |
|
movq mm2, [eax-1] |
717 |
|
movq mm4, [eax+2] |
718 |
|
|
719 |
|
movq mm3, mm2 |
720 |
|
movq mm5, mm4 |
721 |
|
|
722 |
|
punpcklbw mm2, mm7 |
723 |
|
punpcklbw mm4, mm7 |
724 |
|
|
725 |
|
punpckhbw mm3, mm7 |
726 |
|
punpckhbw mm5, mm7 |
727 |
|
|
728 |
|
paddw mm2, mm4 |
729 |
|
paddw mm3, mm5 |
730 |
|
|
731 |
|
psubsw mm0, mm2 |
732 |
|
psubsw mm1, mm3 |
733 |
|
|
734 |
|
pmullw mm0, [mmx_five] |
735 |
|
pmullw mm1, [mmx_five] |
736 |
|
|
737 |
|
movq mm2, [eax-2] |
738 |
|
movq mm4, [eax+3] |
739 |
|
|
740 |
|
movq mm3, mm2 |
741 |
|
movq mm5, mm4 |
742 |
|
|
743 |
|
punpcklbw mm2, mm7 |
744 |
|
punpcklbw mm4, mm7 |
745 |
|
|
746 |
|
punpckhbw mm3, mm7 |
747 |
|
punpckhbw mm5, mm7 |
748 |
|
|
749 |
|
paddw mm2, mm4 |
750 |
|
paddw mm3, mm5 |
751 |
|
|
752 |
|
paddsw mm0, mm2 |
753 |
|
paddsw mm1, mm3 |
754 |
|
|
755 |
|
paddsw mm0, mm6 |
756 |
|
paddsw mm1, mm6 |
757 |
|
|
758 |
|
psraw mm0, 5 |
759 |
|
psraw mm1, 5 |
760 |
|
|
761 |
|
lea eax, [eax+edx] |
762 |
|
packuswb mm0, mm1 |
763 |
|
movq [ecx], mm0 |
764 |
|
%endmacro |
765 |
|
|
766 |
|
align 16 |
767 |
|
cglobal interpolate8x8_6tap_lowpass_h_mmx |
768 |
|
interpolate8x8_6tap_lowpass_h_mmx |
769 |
|
|
770 |
|
mov eax, [esp + 16] ; rounding |
771 |
|
|
772 |
|
movq mm6, [rounding_lowpass_mmx + eax * 8] |
773 |
|
|
774 |
|
mov ecx, [esp + 4] ; dst -> edi |
775 |
|
mov eax, [esp + 8] ; src -> esi |
776 |
|
mov edx, [esp + 12] ; stride -> edx |
777 |
|
|
778 |
|
pxor mm7, mm7 |
779 |
|
|
780 |
|
LOWPASS_6TAP_H_MMX |
781 |
|
lea ecx, [ecx+edx] |
782 |
|
LOWPASS_6TAP_H_MMX |
783 |
|
lea ecx, [ecx+edx] |
784 |
|
LOWPASS_6TAP_H_MMX |
785 |
|
lea ecx, [ecx+edx] |
786 |
|
LOWPASS_6TAP_H_MMX |
787 |
|
lea ecx, [ecx+edx] |
788 |
|
LOWPASS_6TAP_H_MMX |
789 |
|
lea ecx, [ecx+edx] |
790 |
|
LOWPASS_6TAP_H_MMX |
791 |
|
lea ecx, [ecx+edx] |
792 |
|
LOWPASS_6TAP_H_MMX |
793 |
|
lea ecx, [ecx+edx] |
794 |
|
LOWPASS_6TAP_H_MMX |
795 |
|
|
796 |
|
ret |
797 |
|
|
798 |
|
;=========================================================================== |
799 |
|
; |
800 |
|
; void interpolate8x8_6tap_lowpass_v_mmx(uint8_t const *dst, |
801 |
|
; const uint8_t * const src, |
802 |
|
; const uint32_t stride, |
803 |
|
; const uint32_t rounding); |
804 |
|
; |
805 |
|
;=========================================================================== |
806 |
|
|
807 |
|
%macro LOWPASS_6TAP_V_MMX 0 |
808 |
|
movq mm0, [eax] |
809 |
|
movq mm2, [eax+edx] |
810 |
|
|
811 |
|
movq mm1, mm0 |
812 |
|
movq mm3, mm2 |
813 |
|
|
814 |
|
punpcklbw mm0, mm7 |
815 |
|
punpcklbw mm2, mm7 |
816 |
|
|
817 |
|
punpckhbw mm1, mm7 |
818 |
|
punpckhbw mm3, mm7 |
819 |
|
|
820 |
|
paddw mm0, mm2 |
821 |
|
paddw mm1, mm3 |
822 |
|
|
823 |
|
psllw mm0, 2 |
824 |
|
psllw mm1, 2 |
825 |
|
|
826 |
|
movq mm4, [eax+2*edx] |
827 |
|
sub eax, ebx |
828 |
|
movq mm2, [eax+2*edx] |
829 |
|
|
830 |
|
movq mm3, mm2 |
831 |
|
movq mm5, mm4 |
832 |
|
|
833 |
|
punpcklbw mm2, mm7 |
834 |
|
punpcklbw mm4, mm7 |
835 |
|
|
836 |
|
punpckhbw mm3, mm7 |
837 |
|
punpckhbw mm5, mm7 |
838 |
|
|
839 |
|
paddw mm2, mm4 |
840 |
|
paddw mm3, mm5 |
841 |
|
|
842 |
|
psubsw mm0, mm2 |
843 |
|
psubsw mm1, mm3 |
844 |
|
|
845 |
|
pmullw mm0, [mmx_five] |
846 |
|
pmullw mm1, [mmx_five] |
847 |
|
|
848 |
|
movq mm2, [eax+edx] |
849 |
|
movq mm4, [eax+2*ebx] |
850 |
|
|
851 |
|
movq mm3, mm2 |
852 |
|
movq mm5, mm4 |
853 |
|
|
854 |
|
punpcklbw mm2, mm7 |
855 |
|
punpcklbw mm4, mm7 |
856 |
|
|
857 |
|
punpckhbw mm3, mm7 |
858 |
|
punpckhbw mm5, mm7 |
859 |
|
|
860 |
|
paddw mm2, mm4 |
861 |
|
paddw mm3, mm5 |
862 |
|
|
863 |
|
paddsw mm0, mm2 |
864 |
|
paddsw mm1, mm3 |
865 |
|
|
866 |
|
paddsw mm0, mm6 |
867 |
|
paddsw mm1, mm6 |
868 |
|
|
869 |
|
psraw mm0, 5 |
870 |
|
psraw mm1, 5 |
871 |
|
|
872 |
|
lea eax, [eax+4*edx] |
873 |
|
packuswb mm0, mm1 |
874 |
|
movq [ecx], mm0 |
875 |
|
%endmacro |
876 |
|
|
877 |
|
align 16 |
878 |
|
cglobal interpolate8x8_6tap_lowpass_v_mmx |
879 |
|
interpolate8x8_6tap_lowpass_v_mmx |
880 |
|
|
881 |
|
push ebx |
882 |
|
|
883 |
|
mov eax, [esp + 4 + 16] ; rounding |
884 |
|
|
885 |
|
movq mm6, [rounding_lowpass_mmx + eax * 8] |
886 |
|
|
887 |
|
mov ecx, [esp + 4 + 4] ; dst -> edi |
888 |
|
mov eax, [esp + 4 + 8] ; src -> esi |
889 |
|
mov edx, [esp + 4 + 12] ; stride -> edx |
890 |
|
|
891 |
|
mov ebx, edx |
892 |
|
shl ebx, 1 |
893 |
|
add ebx, edx |
894 |
|
|
895 |
|
pxor mm7, mm7 |
896 |
|
|
897 |
|
LOWPASS_6TAP_V_MMX |
898 |
|
lea ecx, [ecx+edx] |
899 |
|
LOWPASS_6TAP_V_MMX |
900 |
|
lea ecx, [ecx+edx] |
901 |
|
LOWPASS_6TAP_V_MMX |
902 |
|
lea ecx, [ecx+edx] |
903 |
|
LOWPASS_6TAP_V_MMX |
904 |
|
lea ecx, [ecx+edx] |
905 |
|
LOWPASS_6TAP_V_MMX |
906 |
|
lea ecx, [ecx+edx] |
907 |
|
LOWPASS_6TAP_V_MMX |
908 |
|
lea ecx, [ecx+edx] |
909 |
|
LOWPASS_6TAP_V_MMX |
910 |
|
lea ecx, [ecx+edx] |
911 |
|
LOWPASS_6TAP_V_MMX |
912 |
|
|
913 |
|
pop ebx |
914 |
|
ret |