Parent Directory | Revision Log
Revision 651 - (view) (download)
1 : | chl | 434 | ;/***************************************************************************** |
2 : | Isibaar | 3 | ; * |
3 : | chl | 434 | ; * XVID MPEG-4 VIDEO CODEC |
4 : | ; * mmx yuyv/uyvy to yuv planar conversion | ||
5 : | Isibaar | 3 | ; * |
6 : | chl | 434 | ; * Copyright(C) 2002 Peter Ross <pross@xvid.org> |
7 : | Isibaar | 3 | ; * |
8 : | edgomez | 651 | ; * This file is part of XviD, a free MPEG-4 video encoder/decoder |
9 : | Isibaar | 3 | ; * |
10 : | edgomez | 651 | ; * 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 : | chl | 434 | ; * the Free Software Foundation; either version 2 of the License, or |
13 : | ; * (at your option) any later version. | ||
14 : | Isibaar | 3 | ; * |
15 : | chl | 434 | ; * This program is distributed in the hope that it will be useful, |
16 : | ; * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
17 : | ; * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
18 : | ; * GNU General Public License for more details. | ||
19 : | Isibaar | 3 | ; * |
20 : | chl | 434 | ; * You should have received a copy of the GNU General Public License |
21 : | ; * along with this program; if not, write to the Free Software | ||
22 : | ; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
23 : | Isibaar | 3 | ; * |
24 : | edgomez | 651 | ; * 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 : | ; * 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 : | ; * Note that people who make modified versions of XviD are not obligated | ||
47 : | ; * to grant this special exception for their modified versions; it is | ||
48 : | ; * 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: yuyv_to_yv12_mmx.asm,v 1.3 2002-11-17 00:20:30 edgomez Exp $ | ||
54 : | ; * | ||
55 : | chl | 434 | ; ****************************************************************************/ |
56 : | Isibaar | 3 | |
57 : | bits 32 | ||
58 : | |||
59 : | |||
60 : | section .data | ||
61 : | |||
62 : | %macro cglobal 1 | ||
63 : | %ifdef PREFIX | ||
64 : | global _%1 | ||
65 : | %define %1 _%1 | ||
66 : | %else | ||
67 : | global %1 | ||
68 : | %endif | ||
69 : | %endmacro | ||
70 : | |||
71 : | align 16 | ||
72 : | |||
73 : | |||
74 : | section .data | ||
75 : | |||
76 : | |||
77 : | ;=========================================================================== | ||
78 : | ; masks for extracting yuv components | ||
79 : | ;=========================================================================== | ||
80 : | ; y u y v y u y v | ||
81 : | |||
82 : | mask1 db 0xff, 0, 0xff, 0, 0xff, 0, 0xff, 0 | ||
83 : | mask2 db 0, 0xff, 0, 0xff, 0, 0xff, 0, 0xff | ||
84 : | |||
85 : | |||
86 : | section .text | ||
87 : | |||
88 : | ;=========================================================================== | ||
89 : | ; | ||
90 : | ; void yuyv_to_yv12_mmx(uint8_t * const y_out, | ||
91 : | ; uint8_t * const u_out, | ||
92 : | ; uint8_t * const v_out, | ||
93 : | ; const uint8_t * const src, | ||
94 : | ; const uint32_t width, | ||
95 : | ; const uint32_t height, | ||
96 : | ; const uint32_t stride); | ||
97 : | ; | ||
98 : | ; width must be multiple of 8 | ||
99 : | ; does not flip | ||
100 : | ; ~30% faster than plain c | ||
101 : | ; | ||
102 : | ;=========================================================================== | ||
103 : | |||
104 : | align 16 | ||
105 : | cglobal yuyv_to_yv12_mmx | ||
106 : | yuyv_to_yv12_mmx | ||
107 : | |||
108 : | push ebx | ||
109 : | push ecx | ||
110 : | push esi | ||
111 : | push edi | ||
112 : | push ebp ; STACK BASE = 20 | ||
113 : | |||
114 : | ; some global consants | ||
115 : | |||
116 : | mov ecx, [esp + 20 + 20] ; width | ||
117 : | mov eax, [esp + 20 + 28] ; stride | ||
118 : | sub eax, ecx ; eax = stride - width | ||
119 : | mov edx, eax | ||
120 : | add edx, [esp + 20 + 28] ; edx = y_dif + stride | ||
121 : | push edx ; [esp + 12] = y_dif | ||
122 : | |||
123 : | shr eax, 1 | ||
124 : | push eax ; [esp + 8] = uv_dif | ||
125 : | |||
126 : | shr ecx, 3 | ||
127 : | push ecx ; [esp + 4] = width/8 | ||
128 : | |||
129 : | sub esp, 4 ; [esp + 0] = tmp_height_counter | ||
130 : | ; STACK_BASE = 36 | ||
131 : | |||
132 : | movq mm6, [mask1] | ||
133 : | movq mm7, [mask2] | ||
134 : | |||
135 : | mov edi, [esp + 36 + 4] ; y_out | ||
136 : | mov ebx, [esp + 36 + 8] ; u_out | ||
137 : | mov edx, [esp + 36 + 12] ; v_out | ||
138 : | mov esi, [esp + 36 + 16] ; src | ||
139 : | |||
140 : | mov eax, [esp + 36 + 20] | ||
141 : | mov ebp, [esp + 36 + 24] | ||
142 : | mov ecx, [esp + 36 + 28] ; ecx = stride | ||
143 : | shr ebp, 1 ; ebp = height /= 2 | ||
144 : | add eax, eax ; eax = 2 * width | ||
145 : | |||
146 : | .yloop | ||
147 : | mov [esp], ebp | ||
148 : | mov ebp, [esp + 4] ; width/8 | ||
149 : | .xloop | ||
150 : | movq mm2, [esi] ; y 1st row | ||
151 : | movq mm3, [esi + 8] | ||
152 : | movq mm0, mm2 | ||
153 : | movq mm1, mm3 | ||
154 : | pand mm2, mm6 ; mask1 | ||
155 : | pand mm3, mm6 ; mask1 | ||
156 : | pand mm0, mm7 ; mask2 | ||
157 : | pand mm1, mm7 ; mask2 | ||
158 : | packuswb mm2, mm3 | ||
159 : | psrlq mm0, 8 | ||
160 : | psrlq mm1, 8 | ||
161 : | movq [edi], mm2 | ||
162 : | |||
163 : | movq mm4, [esi + eax] ; y 2nd row | ||
164 : | movq mm5, [esi + eax + 8] | ||
165 : | movq mm2, mm4 | ||
166 : | movq mm3, mm5 | ||
167 : | pand mm4, mm6 ; mask1 | ||
168 : | pand mm5, mm6 ; mask1 | ||
169 : | pand mm2, mm7 ; mask2 | ||
170 : | pand mm3, mm7 ; mask2 | ||
171 : | packuswb mm4, mm5 | ||
172 : | psrlq mm2, 8 | ||
173 : | psrlq mm3, 8 | ||
174 : | movq [edi + ecx], mm4 | ||
175 : | |||
176 : | paddw mm0, mm2 ; uv avg 1st & 2nd | ||
177 : | paddw mm1, mm3 | ||
178 : | psrlw mm0, 1 | ||
179 : | psrlw mm1, 1 | ||
180 : | packuswb mm0, mm1 | ||
181 : | movq mm2, mm0 | ||
182 : | pand mm0, mm6 ; mask1 | ||
183 : | pand mm2, mm7 ; mask2 | ||
184 : | packuswb mm0, mm0 | ||
185 : | psrlq mm2, 8 | ||
186 : | movd [ebx], mm0 | ||
187 : | packuswb mm2, mm2 | ||
188 : | movd [edx], mm2 | ||
189 : | |||
190 : | add esi, 16 | ||
191 : | add edi, 8 | ||
192 : | add ebx, 4 | ||
193 : | add edx, 4 | ||
194 : | dec ebp | ||
195 : | jnz near .xloop | ||
196 : | |||
197 : | mov ebp, [esp] | ||
198 : | |||
199 : | add esi, eax ; += width2 | ||
200 : | add edi, [esp + 12] ; += y_dif + stride | ||
201 : | add ebx, [esp + 8] ; += uv_dif | ||
202 : | add edx, [esp + 8] ; += uv_dif | ||
203 : | |||
204 : | dec ebp | ||
205 : | jnz near .yloop | ||
206 : | |||
207 : | emms | ||
208 : | |||
209 : | add esp, 16 | ||
210 : | pop ebp | ||
211 : | pop edi | ||
212 : | pop esi | ||
213 : | pop ecx | ||
214 : | pop ebx | ||
215 : | |||
216 : | ret | ||
217 : | |||
218 : | |||
219 : | ;=========================================================================== | ||
220 : | ; | ||
221 : | ; void uyvy_to_yv12_mmx(uint8_t * const y_out, | ||
222 : | ; uint8_t * const u_out, | ||
223 : | ; uint8_t * const v_out, | ||
224 : | ; const uint8_t * const src, | ||
225 : | ; const uint32_t width, | ||
226 : | ; const uint32_t height, | ||
227 : | ; const uint32_t stride); | ||
228 : | ; | ||
229 : | ; width must be multiple of 8 | ||
230 : | ; does not flip | ||
231 : | ; ~30% faster than plain c | ||
232 : | ; | ||
233 : | ;=========================================================================== | ||
234 : | |||
235 : | align 16 | ||
236 : | cglobal uyvy_to_yv12_mmx | ||
237 : | uyvy_to_yv12_mmx | ||
238 : | |||
239 : | push ebx | ||
240 : | push ecx | ||
241 : | push esi | ||
242 : | push edi | ||
243 : | push ebp ; STACK BASE = 20 | ||
244 : | |||
245 : | ; some global consants | ||
246 : | |||
247 : | mov ecx, [esp + 20 + 20] ; width | ||
248 : | mov eax, [esp + 20 + 28] ; stride | ||
249 : | sub eax, ecx ; eax = stride - width | ||
250 : | mov edx, eax | ||
251 : | add edx, [esp + 20 + 28] ; edx = y_dif + stride | ||
252 : | push edx ; [esp + 12] = y_dif | ||
253 : | |||
254 : | shr eax, 1 | ||
255 : | push eax ; [esp + 8] = uv_dif | ||
256 : | |||
257 : | shr ecx, 3 | ||
258 : | push ecx ; [esp + 4] = width/8 | ||
259 : | |||
260 : | sub esp, 4 ; [esp + 0] = tmp_height_counter | ||
261 : | ; STACK_BASE = 36 | ||
262 : | |||
263 : | movq mm6, [mask1] | ||
264 : | movq mm7, [mask2] | ||
265 : | |||
266 : | mov edi, [esp + 36 + 4] ; y_out | ||
267 : | mov ebx, [esp + 36 + 8] ; u_out | ||
268 : | mov edx, [esp + 36 + 12] ; v_out | ||
269 : | mov esi, [esp + 36 + 16] ; src | ||
270 : | |||
271 : | mov eax, [esp + 36 + 20] | ||
272 : | mov ebp, [esp + 36 + 24] | ||
273 : | mov ecx, [esp + 36 + 28] ; ecx = stride | ||
274 : | shr ebp, 1 ; ebp = height /= 2 | ||
275 : | add eax, eax ; eax = 2 * width | ||
276 : | |||
277 : | .yloop | ||
278 : | mov [esp], ebp | ||
279 : | mov ebp, [esp + 4] ; width/8 | ||
280 : | .xloop | ||
281 : | movq mm2, [esi] ; y 1st row | ||
282 : | movq mm3, [esi + 8] | ||
283 : | movq mm0, mm2 | ||
284 : | movq mm1, mm3 | ||
285 : | pand mm2, mm7 ; mask2 | ||
286 : | pand mm3, mm7 ; mask2 | ||
287 : | psrlq mm2, 8 | ||
288 : | psrlq mm3, 8 | ||
289 : | pand mm0, mm6 ; mask1 | ||
290 : | pand mm1, mm6 ; mask1 | ||
291 : | packuswb mm2, mm3 | ||
292 : | movq [edi], mm2 | ||
293 : | |||
294 : | movq mm4, [esi + eax] ; y 2nd row | ||
295 : | movq mm5, [esi + eax + 8] | ||
296 : | movq mm2, mm4 | ||
297 : | movq mm3, mm5 | ||
298 : | pand mm4, mm7 ; mask2 | ||
299 : | pand mm5, mm7 ; mask2 | ||
300 : | psrlq mm4, 8 | ||
301 : | psrlq mm5, 8 | ||
302 : | pand mm2, mm6 ; mask1 | ||
303 : | pand mm3, mm6 ; mask1 | ||
304 : | packuswb mm4, mm5 | ||
305 : | movq [edi + ecx], mm4 | ||
306 : | |||
307 : | paddw mm0, mm2 ; uv avg 1st & 2nd | ||
308 : | paddw mm1, mm3 | ||
309 : | psrlw mm0, 1 | ||
310 : | psrlw mm1, 1 | ||
311 : | packuswb mm0, mm1 | ||
312 : | movq mm2, mm0 | ||
313 : | pand mm0, mm6 ; mask1 | ||
314 : | pand mm2, mm7 ; mask2 | ||
315 : | packuswb mm0, mm0 | ||
316 : | psrlq mm2, 8 | ||
317 : | movd [ebx], mm0 | ||
318 : | packuswb mm2, mm2 | ||
319 : | movd [edx], mm2 | ||
320 : | |||
321 : | add esi, 16 | ||
322 : | add edi, 8 | ||
323 : | add ebx, 4 | ||
324 : | add edx, 4 | ||
325 : | dec ebp | ||
326 : | jnz near .xloop | ||
327 : | |||
328 : | mov ebp, [esp] | ||
329 : | |||
330 : | add esi, eax ; += width2 | ||
331 : | add edi, [esp + 12] ; += y_dif + stride | ||
332 : | add ebx, [esp + 8] ; += uv_dif | ||
333 : | add edx, [esp + 8] ; += uv_dif | ||
334 : | |||
335 : | dec ebp | ||
336 : | jnz near .yloop | ||
337 : | |||
338 : | emms | ||
339 : | |||
340 : | add esp, 16 | ||
341 : | pop ebp | ||
342 : | pop edi | ||
343 : | pop esi | ||
344 : | pop ecx | ||
345 : | pop ebx | ||
346 : | |||
347 : | ret |
No admin address has been configured | ViewVC Help |
Powered by ViewVC 1.0.4 |