1 |
;/************************************************************************** |
;/***************************************************************************** |
2 |
; * |
; * |
3 |
; * XVID MPEG-4 VIDEO CODEC |
; * XVID MPEG-4 VIDEO CODEC |
4 |
; * mmx sum of absolute difference |
; * mmx sum of absolute difference |
5 |
; * |
; * |
6 |
; * This program is free software; you can redistribute it and/or modify |
; * Copyright(C) 2002 Peter Ross <pross@xvid.org> |
7 |
; * it under the terms of the GNU General Public License as published by |
; * |
8 |
|
; * This file is part of XviD, a free MPEG-4 video encoder/decoder |
9 |
|
; * |
10 |
|
; * XviD is free software; you can redistribute it and/or modify it |
11 |
|
; * under the terms of the GNU General Public License as published by |
12 |
; * the Free Software Foundation; either version 2 of the License, or |
; * the Free Software Foundation; either version 2 of the License, or |
13 |
; * (at your option) any later version. |
; * (at your option) any later version. |
14 |
; * |
; * |
19 |
; * |
; * |
20 |
; * You should have received a copy of the GNU General Public License |
; * You should have received a copy of the GNU General Public License |
21 |
; * along with this program; if not, write to the Free Software |
; * along with this program; if not, write to the Free Software |
22 |
; * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
23 |
; * |
; * |
24 |
; *************************************************************************/ |
; * Under section 8 of the GNU General Public License, the copyright |
25 |
|
; * holders of XVID explicitly forbid distribution in the following |
26 |
;/************************************************************************** |
; * countries: |
27 |
|
; * |
28 |
|
; * - Japan |
29 |
|
; * - United States of America |
30 |
|
; * |
31 |
|
; * Linking XviD statically or dynamically with other modules is making a |
32 |
|
; * combined work based on XviD. Thus, the terms and conditions of the |
33 |
|
; * GNU General Public License cover the whole combination. |
34 |
; * |
; * |
35 |
; * History: |
; * As a special exception, the copyright holders of XviD give you |
36 |
|
; * permission to link XviD with independent modules that communicate with |
37 |
|
; * XviD solely through the VFW1.1 and DShow interfaces, regardless of the |
38 |
|
; * license terms of these independent modules, and to copy and distribute |
39 |
|
; * the resulting combined work under terms of your choice, provided that |
40 |
|
; * every copy of the combined work is accompanied by a complete copy of |
41 |
|
; * the source code of XviD (the version of XviD used to produce the |
42 |
|
; * combined work), being distributed under the terms of the GNU General |
43 |
|
; * Public License plus this exception. An independent module is a module |
44 |
|
; * which is not derived from or based on XviD. |
45 |
; * |
; * |
46 |
; * 23.07.2002 sad[16,8]bi_xmm; <pross@xvid.org> |
; * Note that people who make modified versions of XviD are not obligated |
47 |
; * 04.06.2002 cleanup -Skal- |
; * to grant this special exception for their modified versions; it is |
48 |
; * 12.11.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
; * their choice whether to do so. The GNU General Public License gives |
49 |
|
; * permission to release a modified version without this exception; this |
50 |
|
; * exception also makes it possible to release a modified version which |
51 |
|
; * carries forward this exception. |
52 |
; * |
; * |
53 |
; *************************************************************************/ |
; * $Id: sad_mmx.asm,v 1.10 2002-11-17 00:32:06 edgomez Exp $ |
54 |
|
; * |
55 |
|
; ****************************************************************************/ |
56 |
|
|
57 |
bits 32 |
bits 32 |
58 |
|
|
59 |
%macro cglobal 1 |
%macro cglobal 1 |
60 |
|
%ifdef PREFIX |
61 |
global _%1 |
global _%1 |
62 |
%define %1 _%1 |
%define %1 _%1 |
63 |
%else |
%else |
83 |
; uint32_t sad16_mmx(const uint8_t * const cur, |
; uint32_t sad16_mmx(const uint8_t * const cur, |
84 |
; const uint8_t * const ref, |
; const uint8_t * const ref, |
85 |
; const uint32_t stride, |
; const uint32_t stride, |
86 |
|
; const uint32_t best_sad); |
87 |
|
; |
88 |
; (early termination ignore; slows this down) |
; (early termination ignore; slows this down) |
89 |
; |
; |
90 |
;=========================================================================== |
;=========================================================================== |
234 |
; |
; |
235 |
; uint32_t sad16bi_mmx(const uint8_t * const cur, |
; uint32_t sad16bi_mmx(const uint8_t * const cur, |
236 |
; const uint8_t * const ref1, |
; const uint8_t * const ref1, |
237 |
|
; const uint8_t * const ref2, |
238 |
|
; const uint32_t stride); |
239 |
|
; |
240 |
|
;=========================================================================== |
241 |
|
%macro SADBI_16x16_MMX 2 ; SADBI_16x16_MMX( int_ptr_offset, bool_increment_ptr ); |
242 |
|
|
243 |
|
movq mm0, [edx+%1] |
244 |
|
movq mm2, [ebx+%1] |
245 |
|
movq mm1, mm0 |
246 |
|
movq mm3, mm2 |
247 |
|
|
248 |
|
%if %2 != 0 |
249 |
|
add edx, ecx |
250 |
|
%endif |
251 |
|
|
252 |
|
punpcklbw mm0, mm7 |
253 |
|
punpckhbw mm1, mm7 |
254 |
|
punpcklbw mm2, mm7 |
255 |
|
punpckhbw mm3, mm7 |
256 |
|
|
257 |
|
%if %2 != 0 |
258 |
|
add ebx, ecx |
259 |
|
%endif |
260 |
|
|
261 |
|
paddusw mm0, mm2 ; mm01 = ref1 + ref2 |
262 |
|
paddusw mm1, mm3 |
263 |
|
paddusw mm0, [mmx_one] ; mm01 += 1 |
264 |
|
paddusw mm1, [mmx_one] |
265 |
|
psrlw mm0, 1 ; mm01 >>= 1 |
266 |
|
psrlw mm1, 1 |
267 |
|
|
268 |
|
movq mm2, [eax+%1] |
269 |
|
movq mm3, mm2 |
270 |
|
punpcklbw mm2, mm7 ; mm23 = src |
271 |
|
punpckhbw mm3, mm7 |
272 |
|
|
273 |
|
%if %2 != 0 |
274 |
|
add eax, ecx |
275 |
|
%endif |
276 |
|
|
277 |
|
movq mm4, mm0 |
278 |
|
movq mm5, mm1 |
279 |
|
psubusw mm0, mm2 |
280 |
|
psubusw mm1, mm3 |
281 |
|
psubusw mm2, mm4 |
282 |
|
psubusw mm3, mm5 |
283 |
|
por mm0, mm2 ; mm01 = ABS(mm01 - mm23) |
284 |
|
por mm1, mm3 |
285 |
|
|
286 |
|
paddusw mm6,mm0 ; mm6 += mm01 |
287 |
|
paddusw mm6,mm1 |
288 |
|
|
289 |
|
%endmacro |
290 |
|
|
291 |
|
align 16 |
292 |
|
sad16bi_mmx: |
293 |
|
push ebx |
294 |
|
mov eax, [esp+4+ 4] ; Src |
295 |
|
mov edx, [esp+4+ 8] ; Ref1 |
296 |
|
mov ebx, [esp+4+12] ; Ref2 |
297 |
|
mov ecx, [esp+4+16] ; Stride |
298 |
|
|
299 |
|
pxor mm6, mm6 ; accum2 |
300 |
|
pxor mm7, mm7 |
301 |
|
.Loop |
302 |
|
SADBI_16x16_MMX 0, 0 |
303 |
|
SADBI_16x16_MMX 8, 1 |
304 |
|
SADBI_16x16_MMX 0, 0 |
305 |
|
SADBI_16x16_MMX 8, 1 |
306 |
|
SADBI_16x16_MMX 0, 0 |
307 |
|
SADBI_16x16_MMX 8, 1 |
308 |
|
SADBI_16x16_MMX 0, 0 |
309 |
|
SADBI_16x16_MMX 8, 1 |
310 |
|
SADBI_16x16_MMX 0, 0 |
311 |
|
SADBI_16x16_MMX 8, 1 |
312 |
|
SADBI_16x16_MMX 0, 0 |
313 |
|
SADBI_16x16_MMX 8, 1 |
314 |
|
SADBI_16x16_MMX 0, 0 |
315 |
|
SADBI_16x16_MMX 8, 1 |
316 |
|
SADBI_16x16_MMX 0, 0 |
317 |
|
SADBI_16x16_MMX 8, 1 |
318 |
|
|
319 |
|
SADBI_16x16_MMX 0, 0 |
320 |
|
SADBI_16x16_MMX 8, 1 |
321 |
|
SADBI_16x16_MMX 0, 0 |
322 |
|
SADBI_16x16_MMX 8, 1 |
323 |
|
SADBI_16x16_MMX 0, 0 |
324 |
|
SADBI_16x16_MMX 8, 1 |
325 |
|
SADBI_16x16_MMX 0, 0 |
326 |
|
SADBI_16x16_MMX 8, 1 |
327 |
|
SADBI_16x16_MMX 0, 0 |
328 |
|
SADBI_16x16_MMX 8, 1 |
329 |
|
SADBI_16x16_MMX 0, 0 |
330 |
|
SADBI_16x16_MMX 8, 1 |
331 |
|
SADBI_16x16_MMX 0, 0 |
332 |
|
SADBI_16x16_MMX 8, 1 |
333 |
|
SADBI_16x16_MMX 0, 0 |
334 |
|
SADBI_16x16_MMX 8, 1 |
335 |
|
|
336 |
|
pmaddwd mm6, [mmx_one] ; collapse |
337 |
|
movq mm7, mm6 |
338 |
|
psrlq mm7, 32 |
339 |
|
paddd mm6, mm7 |
340 |
|
|
341 |
|
movd eax, mm6 |
342 |
|
pop ebx |
343 |
|
ret |
344 |
|
|
345 |
|
;=========================================================================== |
346 |
|
; |
347 |
|
; uint32_t sad8bi_mmx(const uint8_t * const cur, |
348 |
|
; const uint8_t * const ref1, |
349 |
|
; const uint8_t * const ref2, |
350 |
|
; const uint32_t stride); |
351 |
|
; |
352 |
|
;=========================================================================== |
353 |
|
align 16 |
354 |
|
sad8bi_mmx: |
355 |
|
push ebx |
356 |
|
mov eax, [esp+4+ 4] ; Src |
357 |
|
mov edx, [esp+4+ 8] ; Ref1 |
358 |
|
mov ebx, [esp+4+12] ; Ref2 |
359 |
|
mov ecx, [esp+4+16] ; Stride |
360 |
|
|
361 |
|
pxor mm6, mm6 ; accum2 |
362 |
|
pxor mm7, mm7 |
363 |
|
.Loop |
364 |
|
SADBI_16x16_MMX 0, 1 |
365 |
|
SADBI_16x16_MMX 0, 1 |
366 |
|
SADBI_16x16_MMX 0, 1 |
367 |
|
SADBI_16x16_MMX 0, 1 |
368 |
|
SADBI_16x16_MMX 0, 1 |
369 |
|
SADBI_16x16_MMX 0, 1 |
370 |
|
SADBI_16x16_MMX 0, 1 |
371 |
|
SADBI_16x16_MMX 0, 1 |
372 |
|
|
373 |
|
pmaddwd mm6, [mmx_one] ; collapse |
374 |
|
movq mm7, mm6 |
375 |
|
psrlq mm7, 32 |
376 |
|
paddd mm6, mm7 |
377 |
|
|
378 |
|
movd eax, mm6 |
379 |
|
pop ebx |
380 |
|
ret |
381 |
|
|
382 |
|
|
383 |
|
|
384 |
|
|
385 |
|
;=========================================================================== |
386 |
|
; |
387 |
|
; uint32_t dev16_mmx(const uint8_t * const cur, |
388 |
|
; const uint32_t stride); |
389 |
|
; |
390 |
|
;=========================================================================== |
391 |
|
|
392 |
|
%macro MEAN_16x16_MMX 0 |
393 |
movq mm0, [eax] |
movq mm0, [eax] |
394 |
movq mm2, [eax+8] |
movq mm2, [eax+8] |
395 |
lea eax,[eax+ecx] |
lea eax,[eax+ecx] |
510 |
|
|
511 |
movd eax, mm6 |
movd eax, mm6 |
512 |
ret |
ret |
513 |
|
|