Parent Directory
|
Revision Log
Revision 1.27 - (view) (download)
1 : | edgomez | 1.16 | /***************************************************************************** |
2 : | edgomez | 1.17 | * |
3 : | * XVID MPEG-4 VIDEO CODEC | ||
4 : | * - Native API implementation - | ||
5 : | * | ||
6 : | * This program is an implementation of a part of one or more MPEG-4 | ||
7 : | * Video tools as specified in ISO/IEC 14496-2 standard. Those intending | ||
8 : | * to use this software module in hardware or software products are | ||
9 : | * advised that its use may infringe existing patents or copyrights, and | ||
10 : | * any such use would be at such party's own risk. The original | ||
11 : | * developer of this software module and his/her company, and subsequent | ||
12 : | * editors and their companies, will have no liability for use of this | ||
13 : | * software or modifications or derivatives thereof. | ||
14 : | * | ||
15 : | * This program is free software ; you can redistribute it and/or modify | ||
16 : | * it under the terms of the GNU General Public License as published by | ||
17 : | * the Free Software Foundation ; either version 2 of the License, or | ||
18 : | * (at your option) any later version. | ||
19 : | * | ||
20 : | * This program is distributed in the hope that it will be useful, | ||
21 : | * but WITHOUT ANY WARRANTY ; without even the implied warranty of | ||
22 : | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
23 : | * GNU General Public License for more details. | ||
24 : | * | ||
25 : | * You should have received a copy of the GNU General Public License | ||
26 : | * along with this program ; if not, write to the Free Software | ||
27 : | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
28 : | * | ||
29 : | ****************************************************************************/ | ||
30 : | chenm001 | 1.27 | |
31 : | edgomez | 1.16 | /***************************************************************************** |
32 : | edgomez | 1.17 | * |
33 : | * History | ||
34 : | * | ||
35 : | suxen_drol | 1.21 | * - 23.06.2002 added XVID_CPU_CHKONLY |
36 : | edgomez | 1.17 | * - 17.03.2002 Added interpolate8x8_halfpel_hv_xmm |
37 : | * - 22.12.2001 API change: added xvid_init() - Isibaar | ||
38 : | * - 16.12.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> | ||
39 : | * | ||
40 : | chenm001 | 1.27 | * $Id: xvid.c,v 1.25 2002/07/07 13:21:34 Isibaar Exp $ |
41 : | edgomez | 1.17 | * |
42 : | ****************************************************************************/ | ||
43 : | Isibaar | 1.1 | |
44 : | #include "xvid.h" | ||
45 : | #include "decoder.h" | ||
46 : | #include "encoder.h" | ||
47 : | #include "bitstream/cbp.h" | ||
48 : | #include "dct/idct.h" | ||
49 : | #include "dct/fdct.h" | ||
50 : | #include "image/colorspace.h" | ||
51 : | #include "image/interpolate8x8.h" | ||
52 : | #include "utils/mem_transfer.h" | ||
53 : | #include "quant/quant_h263.h" | ||
54 : | #include "quant/quant_mpeg4.h" | ||
55 : | #include "motion/sad.h" | ||
56 : | #include "utils/emms.h" | ||
57 : | #include "utils/timer.h" | ||
58 : | Isibaar | 1.9 | #include "bitstream/mbcoding.h" |
59 : | Isibaar | 1.1 | |
60 : | edgomez | 1.16 | /***************************************************************************** |
61 : | * XviD Init Entry point | ||
62 : | * | ||
63 : | * Well this function initialize all internal function pointers according | ||
64 : | * to the CPU features forced by the library client or autodetected (depending | ||
65 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
66 : | * image colorspace transformation tables. | ||
67 : | * | ||
68 : | * Returned value : XVID_ERR_OK | ||
69 : | * + API_VERSION in the input XVID_INIT_PARAM structure | ||
70 : | * + core build " " " " " | ||
71 : | * | ||
72 : | ****************************************************************************/ | ||
73 : | |||
74 : | edgomez | 1.15 | int |
75 : | xvid_init(void *handle, | ||
76 : | int opt, | ||
77 : | void *param1, | ||
78 : | void *param2) | ||
79 : | Isibaar | 1.1 | { |
80 : | int cpu_flags; | ||
81 : | XVID_INIT_PARAM *init_param; | ||
82 : | |||
83 : | init_param = (XVID_INIT_PARAM *) param1; | ||
84 : | |||
85 : | suxen_drol | 1.21 | /* Inform the client the API version */ |
86 : | init_param->api_version = API_VERSION; | ||
87 : | |||
88 : | /* Inform the client the core build - unused because we're still alpha */ | ||
89 : | init_param->core_build = 1000; | ||
90 : | |||
91 : | if ((init_param->cpu_flags & XVID_CPU_CHKONLY)) | ||
92 : | { | ||
93 : | chenm001 | 1.27 | init_param->cpu_flags = check_cpu_features(); |
94 : | suxen_drol | 1.21 | return XVID_ERR_OK; |
95 : | } | ||
96 : | |||
97 : | edgomez | 1.16 | /* Do we have to force CPU features ? */ |
98 : | if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0) { | ||
99 : | Isibaar | 1.1 | cpu_flags = init_param->cpu_flags; |
100 : | edgomez | 1.16 | } else { |
101 : | Isibaar | 1.1 | |
102 : | chenm001 | 1.27 | cpu_flags = check_cpu_features(); |
103 : | Isibaar | 1.1 | init_param->cpu_flags = cpu_flags; |
104 : | } | ||
105 : | |||
106 : | edgomez | 1.16 | /* Initialize the function pointers */ |
107 : | Isibaar | 1.1 | idct_int32_init(); |
108 : | Isibaar | 1.9 | init_vlc_tables(); |
109 : | |||
110 : | edgomez | 1.16 | /* Fixed Point Forward/Inverse DCT transformations */ |
111 : | Isibaar | 1.1 | fdct = fdct_int32; |
112 : | idct = idct_int32; | ||
113 : | |||
114 : | edgomez | 1.16 | /* Only needed on PPC Altivec archs */ |
115 : | canard | 1.10 | sadInit = 0; |
116 : | edgomez | 1.15 | |
117 : | edgomez | 1.16 | /* Restore FPU context : emms_c is a nop functions */ |
118 : | Isibaar | 1.1 | emms = emms_c; |
119 : | |||
120 : | edgomez | 1.16 | /* Quantization functions */ |
121 : | quant_intra = quant_intra_c; | ||
122 : | Isibaar | 1.1 | dequant_intra = dequant_intra_c; |
123 : | edgomez | 1.16 | quant_inter = quant_inter_c; |
124 : | Isibaar | 1.1 | dequant_inter = dequant_inter_c; |
125 : | |||
126 : | edgomez | 1.16 | quant4_intra = quant4_intra_c; |
127 : | Isibaar | 1.1 | dequant4_intra = dequant4_intra_c; |
128 : | edgomez | 1.16 | quant4_inter = quant4_inter_c; |
129 : | Isibaar | 1.1 | dequant4_inter = dequant4_inter_c; |
130 : | |||
131 : | edgomez | 1.16 | /* Block transfer related functions */ |
132 : | Isibaar | 1.1 | transfer_8to16copy = transfer_8to16copy_c; |
133 : | transfer_16to8copy = transfer_16to8copy_c; | ||
134 : | edgomez | 1.16 | transfer_8to16sub = transfer_8to16sub_c; |
135 : | suxen_drol | 1.11 | transfer_8to16sub2 = transfer_8to16sub2_c; |
136 : | edgomez | 1.16 | transfer_16to8add = transfer_16to8add_c; |
137 : | transfer8x8_copy = transfer8x8_copy_c; | ||
138 : | Isibaar | 1.1 | |
139 : | edgomez | 1.16 | /* Image interpolation related functions */ |
140 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; | ||
141 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; | ||
142 : | Isibaar | 1.1 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
143 : | |||
144 : | edgomez | 1.16 | /* Initialize internal colorspace transformation tables */ |
145 : | Isibaar | 1.1 | colorspace_init(); |
146 : | |||
147 : | edgomez | 1.16 | /* All colorspace transformation functions User Format->YV12 */ |
148 : | Isibaar | 1.1 | rgb555_to_yv12 = rgb555_to_yv12_c; |
149 : | rgb565_to_yv12 = rgb565_to_yv12_c; | ||
150 : | edgomez | 1.16 | rgb24_to_yv12 = rgb24_to_yv12_c; |
151 : | rgb32_to_yv12 = rgb32_to_yv12_c; | ||
152 : | yuv_to_yv12 = yuv_to_yv12_c; | ||
153 : | yuyv_to_yv12 = yuyv_to_yv12_c; | ||
154 : | uyvy_to_yv12 = uyvy_to_yv12_c; | ||
155 : | Isibaar | 1.1 | |
156 : | edgomez | 1.16 | /* All colorspace transformation functions YV12->User format */ |
157 : | Isibaar | 1.1 | yv12_to_rgb555 = yv12_to_rgb555_c; |
158 : | yv12_to_rgb565 = yv12_to_rgb565_c; | ||
159 : | edgomez | 1.16 | yv12_to_rgb24 = yv12_to_rgb24_c; |
160 : | yv12_to_rgb32 = yv12_to_rgb32_c; | ||
161 : | yv12_to_yuv = yv12_to_yuv_c; | ||
162 : | yv12_to_yuyv = yv12_to_yuyv_c; | ||
163 : | yv12_to_uyvy = yv12_to_uyvy_c; | ||
164 : | Isibaar | 1.1 | |
165 : | edgomez | 1.16 | /* Functions used in motion estimation algorithms */ |
166 : | Isibaar | 1.1 | calc_cbp = calc_cbp_c; |
167 : | edgomez | 1.16 | sad16 = sad16_c; |
168 : | sad16bi = sad16bi_c; | ||
169 : | sad8 = sad8_c; | ||
170 : | dev16 = dev16_c; | ||
171 : | Isibaar | 1.1 | |
172 : | #ifdef ARCH_X86 | ||
173 : | edgomez | 1.15 | if ((cpu_flags & XVID_CPU_MMX) > 0) { |
174 : | edgomez | 1.16 | |
175 : | /* Forward and Inverse Discrete Cosine Transformation functions */ | ||
176 : | Isibaar | 1.1 | fdct = fdct_mmx; |
177 : | idct = idct_mmx; | ||
178 : | |||
179 : | edgomez | 1.16 | /* To restore FPU context after mmx use */ |
180 : | Isibaar | 1.1 | emms = emms_mmx; |
181 : | |||
182 : | edgomez | 1.16 | /* Quantization related functions */ |
183 : | quant_intra = quant_intra_mmx; | ||
184 : | Isibaar | 1.1 | dequant_intra = dequant_intra_mmx; |
185 : | edgomez | 1.16 | quant_inter = quant_inter_mmx; |
186 : | Isibaar | 1.1 | dequant_inter = dequant_inter_mmx; |
187 : | |||
188 : | edgomez | 1.16 | quant4_intra = quant4_intra_mmx; |
189 : | Isibaar | 1.1 | dequant4_intra = dequant4_intra_mmx; |
190 : | edgomez | 1.16 | quant4_inter = quant4_inter_mmx; |
191 : | Isibaar | 1.1 | dequant4_inter = dequant4_inter_mmx; |
192 : | |||
193 : | edgomez | 1.16 | /* Block related functions */ |
194 : | Isibaar | 1.1 | transfer_8to16copy = transfer_8to16copy_mmx; |
195 : | transfer_16to8copy = transfer_16to8copy_mmx; | ||
196 : | edgomez | 1.16 | transfer_8to16sub = transfer_8to16sub_mmx; |
197 : | edgomez | 1.22 | transfer_8to16sub2 = transfer_8to16sub2_mmx; |
198 : | edgomez | 1.16 | transfer_16to8add = transfer_16to8add_mmx; |
199 : | transfer8x8_copy = transfer8x8_copy_mmx; | ||
200 : | edgomez | 1.22 | |
201 : | edgomez | 1.16 | |
202 : | /* Image Interpolation related functions */ | ||
203 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; | ||
204 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; | ||
205 : | Isibaar | 1.1 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
206 : | |||
207 : | edgomez | 1.16 | /* Image RGB->YV12 related functions */ |
208 : | Isibaar | 1.1 | rgb24_to_yv12 = rgb24_to_yv12_mmx; |
209 : | rgb32_to_yv12 = rgb32_to_yv12_mmx; | ||
210 : | edgomez | 1.16 | yuv_to_yv12 = yuv_to_yv12_mmx; |
211 : | yuyv_to_yv12 = yuyv_to_yv12_mmx; | ||
212 : | uyvy_to_yv12 = uyvy_to_yv12_mmx; | ||
213 : | Isibaar | 1.1 | |
214 : | edgomez | 1.16 | /* Image YV12->RGB related functions */ |
215 : | Isibaar | 1.1 | yv12_to_rgb24 = yv12_to_rgb24_mmx; |
216 : | yv12_to_rgb32 = yv12_to_rgb32_mmx; | ||
217 : | edgomez | 1.16 | yv12_to_yuyv = yv12_to_yuyv_mmx; |
218 : | yv12_to_uyvy = yv12_to_uyvy_mmx; | ||
219 : | Isibaar | 1.1 | |
220 : | edgomez | 1.16 | /* Motion estimation related functions */ |
221 : | Isibaar | 1.1 | calc_cbp = calc_cbp_mmx; |
222 : | edgomez | 1.16 | sad16 = sad16_mmx; |
223 : | sad8 = sad8_mmx; | ||
224 : | dev16 = dev16_mmx; | ||
225 : | Isibaar | 1.1 | |
226 : | } | ||
227 : | |||
228 : | edgomez | 1.15 | if ((cpu_flags & XVID_CPU_MMXEXT) > 0) { |
229 : | edgomez | 1.16 | |
230 : | /* Inverse DCT */ | ||
231 : | Isibaar | 1.1 | idct = idct_xmm; |
232 : | edgomez | 1.16 | |
233 : | /* Interpolation */ | ||
234 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm; | ||
235 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; | ||
236 : | h | 1.3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
237 : | Isibaar | 1.25 | |
238 : | chenm001 | 1.27 | /* Quantization */ |
239 : | dequant_intra = dequant_intra_xmm; | ||
240 : | dequant_inter = dequant_inter_xmm; | ||
241 : | |||
242 : | edgomez | 1.19 | /* Buffer transfer */ |
243 : | transfer_8to16sub2 = transfer_8to16sub2_xmm; | ||
244 : | edgomez | 1.16 | |
245 : | /* Colorspace transformation */ | ||
246 : | Isibaar | 1.1 | yuv_to_yv12 = yuv_to_yv12_xmm; |
247 : | |||
248 : | edgomez | 1.16 | /* ME functions */ |
249 : | Isibaar | 1.1 | sad16 = sad16_xmm; |
250 : | chenm001 | 1.27 | sad16bi = sad16bi_xmm; |
251 : | edgomez | 1.16 | sad8 = sad8_xmm; |
252 : | Isibaar | 1.1 | dev16 = dev16_xmm; |
253 : | |||
254 : | } | ||
255 : | |||
256 : | edgomez | 1.15 | if ((cpu_flags & XVID_CPU_3DNOW) > 0) { |
257 : | edgomez | 1.16 | |
258 : | /* Interpolation */ | ||
259 : | Isibaar | 1.1 | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
260 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn; | ||
261 : | h | 1.4 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
262 : | Isibaar | 1.1 | } |
263 : | |||
264 : | edgomez | 1.15 | if ((cpu_flags & XVID_CPU_SSE2) > 0) { |
265 : | Isibaar | 1.14 | #ifdef EXPERIMENTAL_SSE2_CODE |
266 : | edgomez | 1.16 | |
267 : | chenm001 | 1.27 | calc_cbp = calc_cbp_sse2; |
268 : | |||
269 : | edgomez | 1.16 | /* Quantization */ |
270 : | quant_intra = quant_intra_sse2; | ||
271 : | Isibaar | 1.14 | dequant_intra = dequant_intra_sse2; |
272 : | edgomez | 1.16 | quant_inter = quant_inter_sse2; |
273 : | Isibaar | 1.14 | dequant_inter = dequant_inter_sse2; |
274 : | h | 1.13 | |
275 : | edgomez | 1.16 | /* ME */ |
276 : | sad16 = sad16_sse2; | ||
277 : | dev16 = dev16_sse2; | ||
278 : | |||
279 : | /* Forward and Inverse DCT */ | ||
280 : | idct = idct_sse2; | ||
281 : | Isibaar | 1.14 | fdct = fdct_sse2; |
282 : | #endif | ||
283 : | h | 1.12 | } |
284 : | edgomez | 1.16 | |
285 : | Isibaar | 1.1 | #endif |
286 : | Isibaar | 1.18 | |
287 : | #ifdef ARCH_IA64 | ||
288 : | if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines? | ||
289 : | idct_ia64_init(); | ||
290 : | fdct = fdct_ia64; | ||
291 : | idct = idct_ia64; //not yet working, crashes | ||
292 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; | ||
293 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; | ||
294 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; | ||
295 : | sad16 = sad16_ia64; | ||
296 : | sad16bi = sad16bi_ia64; | ||
297 : | sad8 = sad8_ia64; | ||
298 : | dev16 = dev16_ia64; | ||
299 : | quant_intra = quant_intra_ia64; | ||
300 : | dequant_intra = dequant_intra_ia64; | ||
301 : | quant_inter = quant_inter_ia64; | ||
302 : | dequant_inter = dequant_inter_ia64; | ||
303 : | transfer_8to16copy = transfer_8to16copy_ia64; | ||
304 : | transfer_16to8copy = transfer_16to8copy_ia64; | ||
305 : | transfer_8to16sub = transfer_8to16sub_ia64; | ||
306 : | transfer_8to16sub2 = transfer_8to16sub2_ia64; | ||
307 : | transfer_16to8add = transfer_16to8add_ia64; | ||
308 : | transfer8x8_copy = transfer8x8_copy_ia64; | ||
309 : | DEBUG("Using IA-64 assembler routines.\n"); | ||
310 : | } | ||
311 : | #endif | ||
312 : | edgomez | 1.16 | |
313 : | canard | 1.5 | #ifdef ARCH_PPC |
314 : | canard | 1.6 | #ifdef ARCH_PPC_ALTIVEC |
315 : | calc_cbp = calc_cbp_altivec; | ||
316 : | canard | 1.7 | fdct = fdct_altivec; |
317 : | idct = idct_altivec; | ||
318 : | canard | 1.10 | sadInit = sadInit_altivec; |
319 : | canard | 1.8 | sad16 = sad16_altivec; |
320 : | sad8 = sad8_altivec; | ||
321 : | dev16 = dev16_altivec; | ||
322 : | canard | 1.6 | #else |
323 : | canard | 1.5 | calc_cbp = calc_cbp_ppc; |
324 : | canard | 1.6 | #endif |
325 : | canard | 1.5 | #endif |
326 : | Isibaar | 1.1 | |
327 : | return XVID_ERR_OK; | ||
328 : | } | ||
329 : | |||
330 : | edgomez | 1.16 | /***************************************************************************** |
331 : | * XviD Native decoder entry point | ||
332 : | * | ||
333 : | * This function is just a wrapper to all the option cases. | ||
334 : | * | ||
335 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
336 : | * else returns the wrapped function result | ||
337 : | * | ||
338 : | ****************************************************************************/ | ||
339 : | |||
340 : | edgomez | 1.15 | int |
341 : | xvid_decore(void *handle, | ||
342 : | int opt, | ||
343 : | void *param1, | ||
344 : | void *param2) | ||
345 : | Isibaar | 1.1 | { |
346 : | edgomez | 1.15 | switch (opt) { |
347 : | case XVID_DEC_DECODE: | ||
348 : | return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1); | ||
349 : | |||
350 : | case XVID_DEC_CREATE: | ||
351 : | chenm001 | 1.27 | return decoder_create((XVID_DEC_PARAM *) param1); |
352 : | edgomez | 1.15 | |
353 : | case XVID_DEC_DESTROY: | ||
354 : | return decoder_destroy((DECODER *) handle); | ||
355 : | Isibaar | 1.1 | |
356 : | default: | ||
357 : | edgomez | 1.15 | return XVID_ERR_FAIL; |
358 : | } | ||
359 : | Isibaar | 1.1 | } |
360 : | |||
361 : | edgomez | 1.16 | |
362 : | /***************************************************************************** | ||
363 : | * XviD Native encoder entry point | ||
364 : | * | ||
365 : | * This function is just a wrapper to all the option cases. | ||
366 : | * | ||
367 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
368 : | * else returns the wrapped function result | ||
369 : | * | ||
370 : | ****************************************************************************/ | ||
371 : | Isibaar | 1.1 | |
372 : | edgomez | 1.15 | int |
373 : | xvid_encore(void *handle, | ||
374 : | int opt, | ||
375 : | void *param1, | ||
376 : | void *param2) | ||
377 : | Isibaar | 1.1 | { |
378 : | edgomez | 1.15 | switch (opt) { |
379 : | case XVID_ENC_ENCODE: | ||
380 : | suxen_drol | 1.20 | #ifdef BFRAMES |
381 : | if (((Encoder *) handle)->mbParam.max_bframes >= 0) | ||
382 : | return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1, | ||
383 : | (XVID_ENC_STATS *) param2); | ||
384 : | else | ||
385 : | #endif | ||
386 : | edgomez | 1.15 | return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
387 : | (XVID_ENC_STATS *) param2); | ||
388 : | |||
389 : | case XVID_ENC_CREATE: | ||
390 : | return encoder_create((XVID_ENC_PARAM *) param1); | ||
391 : | |||
392 : | case XVID_ENC_DESTROY: | ||
393 : | return encoder_destroy((Encoder *) handle); | ||
394 : | Isibaar | 1.1 | |
395 : | default: | ||
396 : | edgomez | 1.15 | return XVID_ERR_FAIL; |
397 : | } | ||
398 : | Isibaar | 1.1 | } |
No admin address has been configured | ViewVC Help |
Powered by ViewVC 1.0.4 |