1 |
;/************************************************************************** |
;/***************************************************************************** |
2 |
; * |
; * |
3 |
; * XVID MPEG-4 VIDEO CODEC |
; * XVID MPEG-4 VIDEO CODEC |
4 |
; * xmm sum of absolute difference |
; * xmm (extended 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 |
; * Copyright(C) 2002 Michael Militzer <michael@xvid.org> |
8 |
|
; * Copyright(C) 2002 Pascal Massimino <skal@planet-d.net> |
9 |
|
; * |
10 |
|
; * This file is part of XviD, a free MPEG-4 video encoder/decoder |
11 |
|
; * |
12 |
|
; * XviD is free software; you can redistribute it and/or modify it |
13 |
|
; * under the terms of the GNU General Public License as published by |
14 |
; * the Free Software Foundation; either version 2 of the License, or |
; * the Free Software Foundation; either version 2 of the License, or |
15 |
; * (at your option) any later version. |
; * (at your option) any later version. |
16 |
; * |
; * |
21 |
; * |
; * |
22 |
; * You should have received a copy of the GNU General Public License |
; * You should have received a copy of the GNU General Public License |
23 |
; * along with this program; if not, write to the Free Software |
; * along with this program; if not, write to the Free Software |
24 |
; * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
25 |
; * |
; * |
26 |
; *************************************************************************/ |
; * Under section 8 of the GNU General Public License, the copyright |
27 |
|
; * holders of XVID explicitly forbid distribution in the following |
28 |
;/************************************************************************** |
; * countries: |
29 |
; * |
; * |
30 |
; * History: |
; * - Japan |
31 |
|
; * - United States of America |
32 |
; * |
; * |
33 |
; * 23.07.2002 sad8bi_xmm; <pross@xvid.org> |
; * Linking XviD statically or dynamically with other modules is making a |
34 |
; * 04.06.2002 rewrote some funcs (XMM mainly) -Skal- |
; * combined work based on XviD. Thus, the terms and conditions of the |
35 |
; * 17.11.2001 bugfix and small improvement for dev16_xmm, |
; * GNU General Public License cover the whole combination. |
36 |
; * removed terminate early in sad16_xmm (Isibaar) |
; * |
37 |
; * 12.11.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
; * As a special exception, the copyright holders of XviD give you |
38 |
|
; * permission to link XviD with independent modules that communicate with |
39 |
|
; * XviD solely through the VFW1.1 and DShow interfaces, regardless of the |
40 |
|
; * license terms of these independent modules, and to copy and distribute |
41 |
|
; * the resulting combined work under terms of your choice, provided that |
42 |
|
; * every copy of the combined work is accompanied by a complete copy of |
43 |
|
; * the source code of XviD (the version of XviD used to produce the |
44 |
|
; * combined work), being distributed under the terms of the GNU General |
45 |
|
; * Public License plus this exception. An independent module is a module |
46 |
|
; * which is not derived from or based on XviD. |
47 |
|
; * |
48 |
|
; * Note that people who make modified versions of XviD are not obligated |
49 |
|
; * to grant this special exception for their modified versions; it is |
50 |
|
; * their choice whether to do so. The GNU General Public License gives |
51 |
|
; * permission to release a modified version without this exception; this |
52 |
|
; * exception also makes it possible to release a modified version which |
53 |
|
; * carries forward this exception. |
54 |
|
; * |
55 |
|
; * $Id: sad_xmm.asm,v 1.5 2002-11-17 00:32:06 edgomez Exp $ |
56 |
; * |
; * |
57 |
; *************************************************************************/ |
; *************************************************************************/ |
58 |
|
|
59 |
bits 32 |
bits 32 |
60 |
|
|
61 |
%macro cglobal 1 |
%macro cglobal 1 |
62 |
%ifdef PREFIX |
%ifdef PREFIX |
63 |
global _%1 |
global _%1 |
84 |
; |
; |
85 |
; uint32_t sad16_xmm(const uint8_t * const cur, |
; uint32_t sad16_xmm(const uint8_t * const cur, |
86 |
; const uint8_t * const ref, |
; const uint8_t * const ref, |
87 |
|
; const uint32_t stride, |
88 |
; const uint32_t best_sad); |
; const uint32_t best_sad); |
89 |
cglobal sad8_xmm |
; |
90 |
;=========================================================================== |
;=========================================================================== |
91 |
|
|
92 |
%macro SAD_16x16_SSE 0 |
%macro SAD_16x16_SSE 0 |
141 |
; |
; |
142 |
;=========================================================================== |
;=========================================================================== |
143 |
|
|
144 |
|
%macro SAD_8x8_SSE 0 |
145 |
|
movq mm0, [eax] |
146 |
|
movq mm1, [eax+ecx] |
147 |
|
|
148 |
|
psadbw mm0, [edx] |
149 |
|
psadbw mm1, [edx+ecx] |
150 |
|
add eax, ebx |
151 |
|
add edx, ebx |
152 |
|
|
153 |
|
paddusw mm5,mm0 |
154 |
|
paddusw mm6,mm1 |
155 |
|
%endmacro |
156 |
|
|
157 |
|
align 16 |
158 |
|
sad8_xmm: |
159 |
|
|
160 |
|
mov eax, [esp+ 4] ; Src1 |
161 |
|
mov edx, [esp+ 8] ; Src2 |
162 |
|
mov ecx, [esp+12] ; Stride |
163 |
|
push ebx |
164 |
|
lea ebx, [ecx+ecx] |
165 |
|
|
166 |
|
pxor mm5, mm5 ; accum1 |
167 |
|
pxor mm6, mm6 ; accum2 |
168 |
|
|
169 |
|
SAD_8x8_SSE |
170 |
|
SAD_8x8_SSE |
171 |
|
SAD_8x8_SSE |
172 |
|
|
173 |
|
movq mm0, [eax] |
174 |
|
movq mm1, [eax+ecx] |
175 |
|
psadbw mm0, [edx] |
176 |
|
psadbw mm1, [edx+ecx] |
177 |
|
|
178 |
|
pop ebx |
179 |
|
|
180 |
|
paddusw mm5,mm0 |
181 |
|
paddusw mm6,mm1 |
182 |
|
|
183 |
|
paddusw mm6,mm5 |
184 |
|
movd eax, mm6 |
185 |
|
|
186 |
|
ret |
187 |
|
|
188 |
|
|
189 |
|
;=========================================================================== |
190 |
|
; |
191 |
|
; uint32_t sad16bi_xmm(const uint8_t * const cur, |
192 |
|
; const uint8_t * const ref1, |
193 |
|
; const uint8_t * const ref2, |
194 |
|
; const uint32_t stride); |
195 |
|
; |
196 |
|
;=========================================================================== |
197 |
|
|
198 |
%macro SADBI_16x16_SSE 0 |
%macro SADBI_16x16_SSE 0 |
199 |
movq mm0, [eax] |
movq mm0, [eax] |
200 |
movq mm1, [eax+8] |
movq mm1, [eax+8] |
260 |
; |
; |
261 |
;=========================================================================== |
;=========================================================================== |
262 |
|
|
263 |
;=========================================================================== |
%macro SADBI_8x8_XMM 0 |
|
; |
|
|
; uint32_t sad8_xmm(const uint8_t * const cur, |
|
|
; const uint8_t * const ref, |
|
|
; const uint32_t stride); |
|
|
; |
|
|
;=========================================================================== |
|
|
|
|
|
%macro SAD_8x8_SSE 0 |
|
264 |
movq mm0, [eax] |
movq mm0, [eax] |
265 |
movq mm1, [eax+ecx] |
movq mm1, [eax+ecx] |
|
%macro MEAN_16x16_SSE 0 |
|
|
psadbw mm0, [edx] |
|
|
psadbw mm1, [edx+ecx] |
|
|
add eax, ebx |
|
|
add edx, ebx |
|
266 |
|
|
267 |
|
movq mm2, [edx] |
268 |
|
movq mm3, [edx+ecx] |
269 |
|
|
270 |
|
pavgb mm2, [ebx] |
271 |
|
lea edx, [edx+2*ecx] |
272 |
|
|
273 |
|
pavgb mm3, [ebx+ecx] |
274 |
|
lea ebx, [ebx+2*ecx] |
275 |
|
|
276 |
|
psadbw mm0, mm2 |
277 |
|
lea eax, [eax+2*ecx] |
278 |
|
|
279 |
|
psadbw mm1, mm3 |
280 |
paddusw mm5,mm0 |
paddusw mm5,mm0 |
281 |
|
|
282 |
paddusw mm6,mm1 |
paddusw mm6,mm1 |
283 |
%endmacro |
%endmacro |
284 |
|
|
285 |
align 16 |
align 16 |
286 |
sad8_xmm: |
sad8bi_xmm: |
|
|
|
|
mov eax, [esp+ 4] ; Src1 |
|
|
mov edx, [esp+ 8] ; Src2 |
|
|
mov ecx, [esp+12] ; Stride |
|
287 |
push ebx |
push ebx |
288 |
lea ebx, [ecx+ecx] |
mov eax, [esp+4+ 4] ; Src |
289 |
|
mov edx, [esp+4+ 8] ; Ref1 |
290 |
|
mov ebx, [esp+4+12] ; Ref2 |
291 |
|
mov ecx, [esp+4+16] ; Stride |
292 |
|
|
293 |
pxor mm5, mm5 ; accum1 |
pxor mm5, mm5 ; accum1 |
294 |
pxor mm6, mm6 ; accum2 |
pxor mm6, mm6 ; accum2 |
295 |
|
.Loop |
296 |
|
SADBI_8x8_XMM |
297 |
|
SADBI_8x8_XMM |
298 |
|
SADBI_8x8_XMM |
299 |
|
SADBI_8x8_XMM |
300 |
|
|
301 |
SAD_8x8_SSE |
paddusw mm6,mm5 |
302 |
SAD_8x8_SSE |
movd eax, mm6 |
|
SAD_8x8_SSE |
|
|
|
|
|
movq mm0, [eax] |
|
|
movq mm1, [eax+ecx] |
|
|
psadbw mm0, [edx] |
|
|
psadbw mm1, [edx+ecx] |
|
|
|
|
303 |
pop ebx |
pop ebx |
304 |
|
ret |
305 |
|
|
|
paddusw mm5,mm0 |
|
|
paddusw mm6,mm1 |
|
306 |
|
|
307 |
paddusw mm6,mm5 |
;=========================================================================== |
308 |
movd eax, mm6 |
; |
309 |
|
; uint32_t dev16_xmm(const uint8_t * const cur, |
310 |
|
; const uint32_t stride); |
311 |
|
; |
312 |
|
;=========================================================================== |
313 |
|
|
314 |
ret |
%macro MEAN_16x16_SSE 0 |
315 |
movq mm0, [eax] |
movq mm0, [eax] |
316 |
movq mm1, [eax+8] |
movq mm1, [eax+8] |
317 |
psadbw mm0, mm7 |
psadbw mm0, mm7 |