Parent Directory
|
Revision Log
Revision 1226 - (view) (download)
1 : | edgomez | 200 | /***************************************************************************** |
2 : | edgomez | 201 | * |
3 : | * XVID MPEG-4 VIDEO CODEC | ||
4 : | * - Native API implementation - | ||
5 : | * | ||
6 : | edgomez | 1054 | * Copyright(C) 2001-2003 Peter Ross <pross@xvid.org> |
7 : | * | ||
8 : | edgomez | 851 | * This program is free software ; you can redistribute it and/or modify |
9 : | * it under the terms of the GNU General Public License as published by | ||
10 : | * the Free Software Foundation ; either version 2 of the License, or | ||
11 : | edgomez | 201 | * (at your option) any later version. |
12 : | * | ||
13 : | * This program is distributed in the hope that it will be useful, | ||
14 : | edgomez | 851 | * but WITHOUT ANY WARRANTY ; without even the implied warranty of |
15 : | edgomez | 201 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
16 : | * GNU General Public License for more details. | ||
17 : | * | ||
18 : | * You should have received a copy of the GNU General Public License | ||
19 : | edgomez | 851 | * along with this program ; if not, write to the Free Software |
20 : | edgomez | 201 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
21 : | * | ||
22 : | edgomez | 1226 | * $Id: xvid.c,v 1.45.2.21 2003-11-29 17:26:12 edgomez Exp $ |
23 : | edgomez | 410 | * |
24 : | edgomez | 201 | ****************************************************************************/ |
25 : | chenm001 | 274 | |
26 : | edgomez | 851 | #include <stdio.h> |
27 : | #include <stdlib.h> | ||
28 : | #include <string.h> | ||
29 : | #include <time.h> | ||
30 : | |||
31 : | Isibaar | 3 | #include "xvid.h" |
32 : | #include "decoder.h" | ||
33 : | #include "encoder.h" | ||
34 : | #include "bitstream/cbp.h" | ||
35 : | #include "dct/idct.h" | ||
36 : | #include "dct/fdct.h" | ||
37 : | #include "image/colorspace.h" | ||
38 : | #include "image/interpolate8x8.h" | ||
39 : | edgomez | 851 | #include "image/reduced.h" |
40 : | Isibaar | 3 | #include "utils/mem_transfer.h" |
41 : | edgomez | 851 | #include "utils/mbfunctions.h" |
42 : | edgomez | 1174 | #include "quant/quant.h" |
43 : | ia64p | 299 | #include "motion/motion.h" |
44 : | Isibaar | 3 | #include "motion/sad.h" |
45 : | #include "utils/emms.h" | ||
46 : | #include "utils/timer.h" | ||
47 : | Isibaar | 100 | #include "bitstream/mbcoding.h" |
48 : | Isibaar | 1125 | #include "image/qpel.h" |
49 : | Isibaar | 3 | |
50 : | suxen_drol | 1031 | #if defined(_DEBUG) |
51 : | unsigned int xvid_debug = 0; /* xvid debug mask */ | ||
52 : | #endif | ||
53 : | |||
54 : | edgomez | 851 | #if defined(ARCH_IS_IA32) |
55 : | #if defined(_MSC_VER) | ||
56 : | # include <windows.h> | ||
57 : | suxen_drol | 311 | #else |
58 : | edgomez | 851 | # include <signal.h> |
59 : | # include <setjmp.h> | ||
60 : | suxen_drol | 311 | |
61 : | edgomez | 851 | static jmp_buf mark; |
62 : | suxen_drol | 311 | |
63 : | edgomez | 851 | static void |
64 : | sigill_handler(int signal) | ||
65 : | { | ||
66 : | longjmp(mark, 1); | ||
67 : | } | ||
68 : | suxen_drol | 311 | #endif |
69 : | |||
70 : | |||
71 : | /* | ||
72 : | edgomez | 874 | * Calls the funcptr, and returns whether SIGILL (illegal instruction) was |
73 : | * signalled | ||
74 : | * | ||
75 : | * Return values: | ||
76 : | * -1 : could not determine | ||
77 : | * 0 : SIGILL was *not* signalled | ||
78 : | * 1 : SIGILL was signalled | ||
79 : | */ | ||
80 : | suxen_drol | 311 | |
81 : | int | ||
82 : | sigill_check(void (*func)()) | ||
83 : | { | ||
84 : | edgomez | 851 | #if defined(_MSC_VER) |
85 : | suxen_drol | 311 | _try { |
86 : | func(); | ||
87 : | } | ||
88 : | _except(EXCEPTION_EXECUTE_HANDLER) { | ||
89 : | |||
90 : | if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION) | ||
91 : | return 1; | ||
92 : | } | ||
93 : | return 0; | ||
94 : | #else | ||
95 : | void * old_handler; | ||
96 : | int jmpret; | ||
97 : | |||
98 : | |||
99 : | old_handler = signal(SIGILL, sigill_handler); | ||
100 : | if (old_handler == SIG_ERR) | ||
101 : | { | ||
102 : | return -1; | ||
103 : | } | ||
104 : | |||
105 : | jmpret = setjmp(mark); | ||
106 : | if (jmpret == 0) | ||
107 : | { | ||
108 : | func(); | ||
109 : | } | ||
110 : | |||
111 : | signal(SIGILL, old_handler); | ||
112 : | |||
113 : | return jmpret; | ||
114 : | #endif | ||
115 : | } | ||
116 : | #endif | ||
117 : | |||
118 : | edgomez | 851 | |
119 : | /* detect cpu flags */ | ||
120 : | static unsigned int | ||
121 : | detect_cpu_flags() | ||
122 : | { | ||
123 : | /* enable native assembly optimizations by default */ | ||
124 : | unsigned int cpu_flags = XVID_CPU_ASM; | ||
125 : | |||
126 : | #if defined(ARCH_IS_IA32) | ||
127 : | cpu_flags |= check_cpu_features(); | ||
128 : | if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) | ||
129 : | cpu_flags &= ~XVID_CPU_SSE; | ||
130 : | |||
131 : | if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) | ||
132 : | cpu_flags &= ~XVID_CPU_SSE2; | ||
133 : | #endif | ||
134 : | |||
135 : | #if defined(ARCH_IS_PPC) | ||
136 : | #if defined(ARCH_IS_PPC_ALTIVEC) | ||
137 : | cpu_flags |= XVID_CPU_ALTIVEC; | ||
138 : | #endif | ||
139 : | #endif | ||
140 : | |||
141 : | return cpu_flags; | ||
142 : | } | ||
143 : | |||
144 : | |||
145 : | edgomez | 200 | /***************************************************************************** |
146 : | * XviD Init Entry point | ||
147 : | * | ||
148 : | * Well this function initialize all internal function pointers according | ||
149 : | * to the CPU features forced by the library client or autodetected (depending | ||
150 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
151 : | * image colorspace transformation tables. | ||
152 : | edgomez | 1161 | * |
153 : | edgomez | 200 | * Returned value : XVID_ERR_OK |
154 : | * + API_VERSION in the input XVID_INIT_PARAM structure | ||
155 : | * + core build " " " " " | ||
156 : | * | ||
157 : | ****************************************************************************/ | ||
158 : | |||
159 : | edgomez | 851 | |
160 : | edgomez | 1161 | static |
161 : | suxen_drol | 890 | int xvid_gbl_init(xvid_gbl_init_t * init) |
162 : | Isibaar | 3 | { |
163 : | suxen_drol | 890 | unsigned int cpu_flags; |
164 : | Isibaar | 3 | |
165 : | edgomez | 1107 | if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */ |
166 : | suxen_drol | 890 | return XVID_ERR_VERSION; |
167 : | suxen_drol | 234 | |
168 : | suxen_drol | 890 | cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags(); |
169 : | suxen_drol | 234 | |
170 : | edgomez | 200 | /* Initialize the function pointers */ |
171 : | Isibaar | 3 | idct_int32_init(); |
172 : | Isibaar | 100 | init_vlc_tables(); |
173 : | |||
174 : | edgomez | 200 | /* Fixed Point Forward/Inverse DCT transformations */ |
175 : | Isibaar | 3 | fdct = fdct_int32; |
176 : | idct = idct_int32; | ||
177 : | |||
178 : | edgomez | 200 | /* Only needed on PPC Altivec archs */ |
179 : | canard | 115 | sadInit = 0; |
180 : | edgomez | 195 | |
181 : | edgomez | 200 | /* Restore FPU context : emms_c is a nop functions */ |
182 : | Isibaar | 3 | emms = emms_c; |
183 : | |||
184 : | Isibaar | 1125 | /* Qpel stuff */ |
185 : | xvid_QP_Funcs = &xvid_QP_Funcs_C; | ||
186 : | xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C; | ||
187 : | edgomez | 1127 | xvid_Init_QP(); |
188 : | Isibaar | 1125 | |
189 : | edgomez | 200 | /* Quantization functions */ |
190 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_c; |
191 : | quant_h263_inter = quant_h263_inter_c; | ||
192 : | dequant_h263_intra = dequant_h263_intra_c; | ||
193 : | dequant_h263_inter = dequant_h263_inter_c; | ||
194 : | Isibaar | 3 | |
195 : | edgomez | 1174 | quant_mpeg_intra = quant_mpeg_intra_c; |
196 : | quant_mpeg_inter = quant_mpeg_inter_c; | ||
197 : | dequant_mpeg_intra = dequant_mpeg_intra_c; | ||
198 : | dequant_mpeg_inter = dequant_mpeg_inter_c; | ||
199 : | Isibaar | 3 | |
200 : | edgomez | 200 | /* Block transfer related functions */ |
201 : | Isibaar | 3 | transfer_8to16copy = transfer_8to16copy_c; |
202 : | transfer_16to8copy = transfer_16to8copy_c; | ||
203 : | edgomez | 200 | transfer_8to16sub = transfer_8to16sub_c; |
204 : | edgomez | 851 | transfer_8to16subro = transfer_8to16subro_c; |
205 : | suxen_drol | 118 | transfer_8to16sub2 = transfer_8to16sub2_c; |
206 : | edgomez | 200 | transfer_16to8add = transfer_16to8add_c; |
207 : | transfer8x8_copy = transfer8x8_copy_c; | ||
208 : | Isibaar | 3 | |
209 : | edgomez | 851 | /* Interlacing functions */ |
210 : | MBFieldTest = MBFieldTest_c; | ||
211 : | |||
212 : | edgomez | 200 | /* Image interpolation related functions */ |
213 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; | ||
214 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; | ||
215 : | Isibaar | 3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
216 : | |||
217 : | edgomez | 851 | interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; |
218 : | interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; | ||
219 : | interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; | ||
220 : | |||
221 : | interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c; | ||
222 : | interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c; | ||
223 : | interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c; | ||
224 : | |||
225 : | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c; | ||
226 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c; | ||
227 : | |||
228 : | interpolate8x8_avg2 = interpolate8x8_avg2_c; | ||
229 : | interpolate8x8_avg4 = interpolate8x8_avg4_c; | ||
230 : | |||
231 : | edgomez | 1174 | /* reduced resolution */ |
232 : | edgomez | 851 | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
233 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; | ||
234 : | vfilter_31 = xvid_VFilter_31_C; | ||
235 : | hfilter_31 = xvid_HFilter_31_C; | ||
236 : | filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; | ||
237 : | filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; | ||
238 : | |||
239 : | edgomez | 200 | /* Initialize internal colorspace transformation tables */ |
240 : | Isibaar | 3 | colorspace_init(); |
241 : | |||
242 : | edgomez | 200 | /* All colorspace transformation functions User Format->YV12 */ |
243 : | edgomez | 851 | yv12_to_yv12 = yv12_to_yv12_c; |
244 : | rgb555_to_yv12 = rgb555_to_yv12_c; | ||
245 : | rgb565_to_yv12 = rgb565_to_yv12_c; | ||
246 : | bgr_to_yv12 = bgr_to_yv12_c; | ||
247 : | bgra_to_yv12 = bgra_to_yv12_c; | ||
248 : | abgr_to_yv12 = abgr_to_yv12_c; | ||
249 : | rgba_to_yv12 = rgba_to_yv12_c; | ||
250 : | yuyv_to_yv12 = yuyv_to_yv12_c; | ||
251 : | uyvy_to_yv12 = uyvy_to_yv12_c; | ||
252 : | Isibaar | 3 | |
253 : | edgomez | 851 | rgb555i_to_yv12 = rgb555i_to_yv12_c; |
254 : | rgb565i_to_yv12 = rgb565i_to_yv12_c; | ||
255 : | bgri_to_yv12 = bgri_to_yv12_c; | ||
256 : | bgrai_to_yv12 = bgrai_to_yv12_c; | ||
257 : | abgri_to_yv12 = abgri_to_yv12_c; | ||
258 : | rgbai_to_yv12 = rgbai_to_yv12_c; | ||
259 : | yuyvi_to_yv12 = yuyvi_to_yv12_c; | ||
260 : | uyvyi_to_yv12 = uyvyi_to_yv12_c; | ||
261 : | |||
262 : | |||
263 : | edgomez | 200 | /* All colorspace transformation functions YV12->User format */ |
264 : | edgomez | 851 | yv12_to_rgb555 = yv12_to_rgb555_c; |
265 : | yv12_to_rgb565 = yv12_to_rgb565_c; | ||
266 : | yv12_to_bgr = yv12_to_bgr_c; | ||
267 : | yv12_to_bgra = yv12_to_bgra_c; | ||
268 : | yv12_to_abgr = yv12_to_abgr_c; | ||
269 : | yv12_to_rgba = yv12_to_rgba_c; | ||
270 : | yv12_to_yuyv = yv12_to_yuyv_c; | ||
271 : | yv12_to_uyvy = yv12_to_uyvy_c; | ||
272 : | edgomez | 1161 | |
273 : | edgomez | 851 | yv12_to_rgb555i = yv12_to_rgb555i_c; |
274 : | yv12_to_rgb565i = yv12_to_rgb565i_c; | ||
275 : | yv12_to_bgri = yv12_to_bgri_c; | ||
276 : | yv12_to_bgrai = yv12_to_bgrai_c; | ||
277 : | yv12_to_abgri = yv12_to_abgri_c; | ||
278 : | yv12_to_rgbai = yv12_to_rgbai_c; | ||
279 : | yv12_to_yuyvi = yv12_to_yuyvi_c; | ||
280 : | yv12_to_uyvyi = yv12_to_uyvyi_c; | ||
281 : | Isibaar | 3 | |
282 : | edgomez | 200 | /* Functions used in motion estimation algorithms */ |
283 : | edgomez | 1205 | calc_cbp = calc_cbp_c; |
284 : | sad16 = sad16_c; | ||
285 : | sad8 = sad8_c; | ||
286 : | sad16bi = sad16bi_c; | ||
287 : | sad8bi = sad8bi_c; | ||
288 : | dev16 = dev16_c; | ||
289 : | sad16v = sad16v_c; | ||
290 : | sse8_16bit = sse8_16bit_c; | ||
291 : | edgomez | 1161 | |
292 : | edgomez | 874 | /* Halfpel8_Refine = Halfpel8_Refine_c; */ |
293 : | Isibaar | 3 | |
294 : | edgomez | 851 | #if defined(ARCH_IS_IA32) |
295 : | edgomez | 200 | |
296 : | edgomez | 1058 | if ((cpu_flags & XVID_CPU_ASM)) { |
297 : | edgomez | 851 | vfilter_31 = xvid_VFilter_31_x86; |
298 : | hfilter_31 = xvid_HFilter_31_x86; | ||
299 : | } | ||
300 : | |||
301 : | if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || | ||
302 : | (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || | ||
303 : | (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2)) | ||
304 : | { | ||
305 : | /* Restore FPU context : emms_c is a nop functions */ | ||
306 : | emms = emms_mmx; | ||
307 : | } | ||
308 : | |||
309 : | if ((cpu_flags & XVID_CPU_MMX)) { | ||
310 : | |||
311 : | edgomez | 200 | /* Forward and Inverse Discrete Cosine Transformation functions */ |
312 : | edgomez | 1190 | fdct = fdct_mmx_skal; |
313 : | Isibaar | 1066 | idct = idct_mmx; |
314 : | Isibaar | 3 | |
315 : | Isibaar | 1125 | /* Qpel stuff */ |
316 : | xvid_QP_Funcs = &xvid_QP_Funcs_mmx; | ||
317 : | xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx; | ||
318 : | |||
319 : | edgomez | 200 | /* Quantization related functions */ |
320 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_mmx; |
321 : | quant_h263_inter = quant_h263_inter_mmx; | ||
322 : | dequant_h263_intra = dequant_h263_intra_mmx; | ||
323 : | dequant_h263_inter = dequant_h263_inter_mmx; | ||
324 : | Isibaar | 3 | |
325 : | edgomez | 1174 | quant_mpeg_intra = quant_mpeg_intra_mmx; |
326 : | quant_mpeg_inter = quant_mpeg_inter_mmx; | ||
327 : | dequant_mpeg_intra = dequant_mpeg_intra_mmx; | ||
328 : | dequant_mpeg_inter = dequant_mpeg_inter_mmx; | ||
329 : | Isibaar | 3 | |
330 : | edgomez | 200 | /* Block related functions */ |
331 : | Isibaar | 3 | transfer_8to16copy = transfer_8to16copy_mmx; |
332 : | transfer_16to8copy = transfer_16to8copy_mmx; | ||
333 : | edgomez | 200 | transfer_8to16sub = transfer_8to16sub_mmx; |
334 : | edgomez | 851 | transfer_8to16subro = transfer_8to16subro_mmx; |
335 : | edgomez | 236 | transfer_8to16sub2 = transfer_8to16sub2_mmx; |
336 : | edgomez | 200 | transfer_16to8add = transfer_16to8add_mmx; |
337 : | transfer8x8_copy = transfer8x8_copy_mmx; | ||
338 : | Isibaar | 3 | |
339 : | edgomez | 851 | /* Interlacing Functions */ |
340 : | MBFieldTest = MBFieldTest_mmx; | ||
341 : | edgomez | 236 | |
342 : | edgomez | 200 | /* Image Interpolation related functions */ |
343 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; | ||
344 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; | ||
345 : | Isibaar | 3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
346 : | |||
347 : | edgomez | 851 | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; |
348 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; | ||
349 : | |||
350 : | interpolate8x8_avg2 = interpolate8x8_avg2_mmx; | ||
351 : | interpolate8x8_avg4 = interpolate8x8_avg4_mmx; | ||
352 : | |||
353 : | /* reduced resolution */ | ||
354 : | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx; | ||
355 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx; | ||
356 : | hfilter_31 = xvid_HFilter_31_mmx; | ||
357 : | filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx; | ||
358 : | filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx; | ||
359 : | |||
360 : | /* image input xxx_to_yv12 related functions */ | ||
361 : | yv12_to_yv12 = yv12_to_yv12_mmx; | ||
362 : | bgr_to_yv12 = bgr_to_yv12_mmx; | ||
363 : | bgra_to_yv12 = bgra_to_yv12_mmx; | ||
364 : | edgomez | 200 | yuyv_to_yv12 = yuyv_to_yv12_mmx; |
365 : | uyvy_to_yv12 = uyvy_to_yv12_mmx; | ||
366 : | Isibaar | 3 | |
367 : | edgomez | 851 | /* image output yv12_to_xxx related functions */ |
368 : | yv12_to_bgr = yv12_to_bgr_mmx; | ||
369 : | yv12_to_bgra = yv12_to_bgra_mmx; | ||
370 : | edgomez | 200 | yv12_to_yuyv = yv12_to_yuyv_mmx; |
371 : | yv12_to_uyvy = yv12_to_uyvy_mmx; | ||
372 : | edgomez | 1161 | |
373 : | edgomez | 851 | yv12_to_yuyvi = yv12_to_yuyvi_mmx; |
374 : | yv12_to_uyvyi = yv12_to_uyvyi_mmx; | ||
375 : | Isibaar | 3 | |
376 : | edgomez | 200 | /* Motion estimation related functions */ |
377 : | Isibaar | 3 | calc_cbp = calc_cbp_mmx; |
378 : | edgomez | 200 | sad16 = sad16_mmx; |
379 : | sad8 = sad8_mmx; | ||
380 : | suxen_drol | 329 | sad16bi = sad16bi_mmx; |
381 : | sad8bi = sad8bi_mmx; | ||
382 : | edgomez | 200 | dev16 = dev16_mmx; |
383 : | edgomez | 851 | sad16v = sad16v_mmx; |
384 : | edgomez | 1205 | sse8_16bit = sse8_16bit_mmx; |
385 : | Isibaar | 3 | } |
386 : | |||
387 : | suxen_drol | 329 | /* these 3dnow functions are faster than mmx, but slower than xmm. */ |
388 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_3DNOW)) { |
389 : | suxen_drol | 329 | |
390 : | edgomez | 851 | emms = emms_3dn; |
391 : | |||
392 : | suxen_drol | 329 | /* ME functions */ |
393 : | sad16bi = sad16bi_3dn; | ||
394 : | sad8bi = sad8bi_3dn; | ||
395 : | edgomez | 851 | |
396 : | yuyv_to_yv12 = yuyv_to_yv12_3dn; | ||
397 : | uyvy_to_yv12 = uyvy_to_yv12_3dn; | ||
398 : | suxen_drol | 329 | } |
399 : | |||
400 : | |||
401 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_MMXEXT)) { |
402 : | edgomez | 200 | |
403 : | edgomez | 1190 | /* DCT */ |
404 : | fdct = fdct_xmm_skal; | ||
405 : | Isibaar | 3 | idct = idct_xmm; |
406 : | edgomez | 200 | |
407 : | /* Interpolation */ | ||
408 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm; | ||
409 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; | ||
410 : | h | 38 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
411 : | edgomez | 200 | |
412 : | edgomez | 851 | /* reduced resolution */ |
413 : | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm; | ||
414 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm; | ||
415 : | |||
416 : | chenm001 | 274 | /* Quantization */ |
417 : | edgomez | 1174 | quant_mpeg_intra = quant_mpeg_intra_xmm; |
418 : | quant_mpeg_inter = quant_mpeg_inter_xmm; | ||
419 : | edgomez | 851 | |
420 : | edgomez | 1174 | dequant_h263_intra = dequant_h263_intra_xmm; |
421 : | dequant_h263_inter = dequant_h263_inter_xmm; | ||
422 : | chenm001 | 274 | |
423 : | edgomez | 218 | /* Buffer transfer */ |
424 : | transfer_8to16sub2 = transfer_8to16sub2_xmm; | ||
425 : | |||
426 : | edgomez | 200 | /* Colorspace transformation */ |
427 : | edgomez | 851 | yv12_to_yv12 = yv12_to_yv12_xmm; |
428 : | yuyv_to_yv12 = yuyv_to_yv12_xmm; | ||
429 : | uyvy_to_yv12 = uyvy_to_yv12_xmm; | ||
430 : | Isibaar | 3 | |
431 : | edgomez | 200 | /* ME functions */ |
432 : | Isibaar | 3 | sad16 = sad16_xmm; |
433 : | suxen_drol | 329 | sad8 = sad8_xmm; |
434 : | chenm001 | 274 | sad16bi = sad16bi_xmm; |
435 : | suxen_drol | 329 | sad8bi = sad8bi_xmm; |
436 : | Isibaar | 3 | dev16 = dev16_xmm; |
437 : | edgomez | 851 | sad16v = sad16v_xmm; |
438 : | Isibaar | 3 | } |
439 : | |||
440 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_3DNOW)) { |
441 : | edgomez | 200 | |
442 : | /* Interpolation */ | ||
443 : | Isibaar | 3 | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
444 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn; | ||
445 : | h | 40 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
446 : | Isibaar | 3 | } |
447 : | |||
448 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_3DNOWEXT)) { |
449 : | edgomez | 200 | |
450 : | edgomez | 851 | /* Inverse DCT */ |
451 : | idct = idct_3dne; | ||
452 : | |||
453 : | /* Buffer transfer */ | ||
454 : | transfer_8to16copy = transfer_8to16copy_3dne; | ||
455 : | transfer_16to8copy = transfer_16to8copy_3dne; | ||
456 : | transfer_8to16sub = transfer_8to16sub_3dne; | ||
457 : | transfer_8to16subro = transfer_8to16subro_3dne; | ||
458 : | transfer_8to16sub2 = transfer_8to16sub2_3dne; | ||
459 : | transfer_16to8add = transfer_16to8add_3dne; | ||
460 : | transfer8x8_copy = transfer8x8_copy_3dne; | ||
461 : | |||
462 : | /* Quantization */ | ||
463 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_3dne; |
464 : | quant_h263_inter = quant_h263_inter_3dne; | ||
465 : | dequant_mpeg_intra = dequant_mpeg_intra_3dne; | ||
466 : | dequant_mpeg_inter = dequant_mpeg_inter_3dne; | ||
467 : | dequant_h263_intra = dequant_h263_intra_3dne; | ||
468 : | dequant_h263_inter = dequant_h263_inter_3dne; | ||
469 : | edgomez | 851 | |
470 : | /* ME functions */ | ||
471 : | calc_cbp = calc_cbp_3dne; | ||
472 : | sad16 = sad16_3dne; | ||
473 : | sad8 = sad8_3dne; | ||
474 : | sad16bi = sad16bi_3dne; | ||
475 : | sad8bi = sad8bi_3dne; | ||
476 : | dev16 = dev16_3dne; | ||
477 : | |||
478 : | /* Interpolation */ | ||
479 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne; | ||
480 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne; | ||
481 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; | ||
482 : | } | ||
483 : | |||
484 : | Isibaar | 1125 | #if defined(EXPERIMENTAL_SSE2_CODE) /* mark the whole SSE2 stuff as experimental. At least on |
485 : | my P4, it crashes... */ | ||
486 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_SSE2)) { |
487 : | |||
488 : | chenm001 | 274 | calc_cbp = calc_cbp_sse2; |
489 : | |||
490 : | edgomez | 200 | /* Quantization */ |
491 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_sse2; |
492 : | quant_h263_inter = quant_h263_inter_sse2; | ||
493 : | dequant_h263_intra = dequant_h263_intra_sse2; | ||
494 : | dequant_h263_inter = dequant_h263_inter_sse2; | ||
495 : | h | 135 | |
496 : | edgomez | 1195 | /* SAD operators */ |
497 : | edgomez | 200 | sad16 = sad16_sse2; |
498 : | dev16 = dev16_sse2; | ||
499 : | edgomez | 1195 | |
500 : | /* DCT operators */ | ||
501 : | fdct = fdct_sse2_skal; | ||
502 : | edgomez | 1197 | idct = idct_sse2_dmitry; |
503 : | h | 126 | } |
504 : | Isibaar | 3 | #endif |
505 : | Isibaar | 1125 | #endif |
506 : | edgomez | 200 | |
507 : | edgomez | 851 | #if defined(ARCH_IS_IA64) |
508 : | edgomez | 874 | if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */ |
509 : | Isibaar | 209 | idct_ia64_init(); |
510 : | fdct = fdct_ia64; | ||
511 : | edgomez | 874 | idct = idct_ia64; /*not yet working, crashes */ |
512 : | Isibaar | 209 | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
513 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; | ||
514 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; | ||
515 : | sad16 = sad16_ia64; | ||
516 : | sad16bi = sad16bi_ia64; | ||
517 : | sad8 = sad8_ia64; | ||
518 : | dev16 = dev16_ia64; | ||
519 : | edgomez | 874 | /* Halfpel8_Refine = Halfpel8_Refine_ia64; */ |
520 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_ia64; |
521 : | quant_h263_inter = quant_h263_inter_ia64; | ||
522 : | dequant_h263_intra = dequant_h263_intra_ia64; | ||
523 : | dequant_h263_inter = dequant_h263_inter_ia64; | ||
524 : | Isibaar | 209 | transfer_8to16copy = transfer_8to16copy_ia64; |
525 : | transfer_16to8copy = transfer_16to8copy_ia64; | ||
526 : | transfer_8to16sub = transfer_8to16sub_ia64; | ||
527 : | transfer_8to16sub2 = transfer_8to16sub2_ia64; | ||
528 : | transfer_16to8add = transfer_16to8add_ia64; | ||
529 : | transfer8x8_copy = transfer8x8_copy_ia64; | ||
530 : | } | ||
531 : | edgomez | 1161 | #endif |
532 : | Isibaar | 209 | |
533 : | edgomez | 851 | #if defined(ARCH_IS_PPC) |
534 : | if ((cpu_flags & XVID_CPU_ASM)) | ||
535 : | { | ||
536 : | calc_cbp = calc_cbp_ppc; | ||
537 : | } | ||
538 : | |||
539 : | if ((cpu_flags & XVID_CPU_ALTIVEC)) | ||
540 : | { | ||
541 : | calc_cbp = calc_cbp_altivec; | ||
542 : | fdct = fdct_altivec; | ||
543 : | idct = idct_altivec; | ||
544 : | sadInit = sadInit_altivec; | ||
545 : | sad16 = sad16_altivec; | ||
546 : | sad8 = sad8_altivec; | ||
547 : | dev16 = dev16_altivec; | ||
548 : | } | ||
549 : | canard | 52 | #endif |
550 : | edgomez | 851 | |
551 : | suxen_drol | 1031 | #if defined(_DEBUG) |
552 : | xvid_debug = init->debug; | ||
553 : | #endif | ||
554 : | |||
555 : | return 0; | ||
556 : | edgomez | 851 | } |
557 : | |||
558 : | |||
559 : | suxen_drol | 890 | static int |
560 : | xvid_gbl_info(xvid_gbl_info_t * info) | ||
561 : | { | ||
562 : | edgomez | 1107 | if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */ |
563 : | suxen_drol | 890 | return XVID_ERR_VERSION; |
564 : | edgomez | 851 | |
565 : | suxen_drol | 890 | info->actual_version = XVID_VERSION; |
566 : | edgomez | 1226 | info->build = "xvid-1.0.0"; |
567 : | suxen_drol | 890 | info->cpu_flags = detect_cpu_flags(); |
568 : | |||
569 : | #if defined(_SMP) && defined(WIN32) | ||
570 : | info->num_threads = pthread_num_processors_np();; | ||
571 : | #else | ||
572 : | info->num_threads = 0; | ||
573 : | #endif | ||
574 : | |||
575 : | return 0; | ||
576 : | } | ||
577 : | |||
578 : | |||
579 : | edgomez | 851 | static int |
580 : | suxen_drol | 890 | xvid_gbl_convert(xvid_gbl_convert_t* convert) |
581 : | edgomez | 851 | { |
582 : | suxen_drol | 890 | int width; |
583 : | int height; | ||
584 : | int width2; | ||
585 : | int height2; | ||
586 : | edgomez | 851 | IMAGE img; |
587 : | |||
588 : | edgomez | 1107 | if (XVID_VERSION_MAJOR(convert->version) != 1) /* v1.x.x */ |
589 : | suxen_drol | 890 | return XVID_ERR_VERSION; |
590 : | |||
591 : | edgomez | 1053 | #if 0 |
592 : | const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); | ||
593 : | #endif | ||
594 : | suxen_drol | 890 | width = convert->width; |
595 : | height = convert->height; | ||
596 : | width2 = convert->width/2; | ||
597 : | height2 = convert->height/2; | ||
598 : | |||
599 : | switch (convert->input.csp & ~XVID_CSP_VFLIP) | ||
600 : | edgomez | 851 | { |
601 : | case XVID_CSP_YV12 : | ||
602 : | suxen_drol | 890 | img.y = convert->input.plane[0]; |
603 : | edgomez | 1161 | img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height; |
604 : | suxen_drol | 890 | img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2; |
605 : | edgomez | 851 | image_output(&img, width, height, width, |
606 : | suxen_drol | 890 | (uint8_t**)convert->output.plane, convert->output.stride, |
607 : | convert->output.csp, convert->interlacing); | ||
608 : | edgomez | 851 | break; |
609 : | |||
610 : | default : | ||
611 : | return XVID_ERR_FORMAT; | ||
612 : | } | ||
613 : | |||
614 : | |||
615 : | emms(); | ||
616 : | suxen_drol | 890 | return 0; |
617 : | edgomez | 851 | } |
618 : | |||
619 : | suxen_drol | 890 | /***************************************************************************** |
620 : | * XviD Global Entry point | ||
621 : | * | ||
622 : | * Well this function initialize all internal function pointers according | ||
623 : | * to the CPU features forced by the library client or autodetected (depending | ||
624 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
625 : | * image colorspace transformation tables. | ||
626 : | edgomez | 1161 | * |
627 : | suxen_drol | 890 | ****************************************************************************/ |
628 : | |||
629 : | |||
630 : | edgomez | 851 | int |
631 : | suxen_drol | 890 | xvid_global(void *handle, |
632 : | edgomez | 851 | int opt, |
633 : | void *param1, | ||
634 : | void *param2) | ||
635 : | { | ||
636 : | switch(opt) | ||
637 : | { | ||
638 : | suxen_drol | 890 | case XVID_GBL_INIT : |
639 : | return xvid_gbl_init((xvid_gbl_init_t*)param1); | ||
640 : | edgomez | 851 | |
641 : | suxen_drol | 890 | case XVID_GBL_INFO : |
642 : | return xvid_gbl_info((xvid_gbl_info_t*)param1); | ||
643 : | edgomez | 851 | |
644 : | suxen_drol | 890 | case XVID_GBL_CONVERT : |
645 : | return xvid_gbl_convert((xvid_gbl_convert_t*)param1); | ||
646 : | |||
647 : | edgomez | 851 | default : |
648 : | return XVID_ERR_FAIL; | ||
649 : | } | ||
650 : | } | ||
651 : | |||
652 : | edgomez | 200 | /***************************************************************************** |
653 : | * XviD Native decoder entry point | ||
654 : | * | ||
655 : | * This function is just a wrapper to all the option cases. | ||
656 : | * | ||
657 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
658 : | * else returns the wrapped function result | ||
659 : | * | ||
660 : | ****************************************************************************/ | ||
661 : | |||
662 : | edgomez | 195 | int |
663 : | xvid_decore(void *handle, | ||
664 : | int opt, | ||
665 : | void *param1, | ||
666 : | void *param2) | ||
667 : | Isibaar | 3 | { |
668 : | edgomez | 195 | switch (opt) { |
669 : | case XVID_DEC_CREATE: | ||
670 : | suxen_drol | 890 | return decoder_create((xvid_dec_create_t *) param1); |
671 : | Isibaar | 3 | |
672 : | edgomez | 195 | case XVID_DEC_DESTROY: |
673 : | return decoder_destroy((DECODER *) handle); | ||
674 : | |||
675 : | suxen_drol | 890 | case XVID_DEC_DECODE: |
676 : | return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2); | ||
677 : | |||
678 : | Isibaar | 3 | default: |
679 : | edgomez | 195 | return XVID_ERR_FAIL; |
680 : | } | ||
681 : | Isibaar | 3 | } |
682 : | |||
683 : | |||
684 : | edgomez | 200 | /***************************************************************************** |
685 : | * XviD Native encoder entry point | ||
686 : | * | ||
687 : | * This function is just a wrapper to all the option cases. | ||
688 : | * | ||
689 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
690 : | * else returns the wrapped function result | ||
691 : | * | ||
692 : | ****************************************************************************/ | ||
693 : | |||
694 : | edgomez | 195 | int |
695 : | xvid_encore(void *handle, | ||
696 : | int opt, | ||
697 : | void *param1, | ||
698 : | void *param2) | ||
699 : | Isibaar | 3 | { |
700 : | edgomez | 195 | switch (opt) { |
701 : | case XVID_ENC_ENCODE: | ||
702 : | edgomez | 851 | |
703 : | suxen_drol | 890 | return enc_encode((Encoder *) handle, |
704 : | (xvid_enc_frame_t *) param1, | ||
705 : | (xvid_enc_stats_t *) param2); | ||
706 : | Isibaar | 3 | |
707 : | edgomez | 195 | case XVID_ENC_CREATE: |
708 : | suxen_drol | 948 | return enc_create((xvid_enc_create_t *) param1); |
709 : | Isibaar | 3 | |
710 : | edgomez | 195 | case XVID_ENC_DESTROY: |
711 : | suxen_drol | 890 | return enc_destroy((Encoder *) handle); |
712 : | edgomez | 195 | |
713 : | Isibaar | 3 | default: |
714 : | edgomez | 195 | return XVID_ERR_FAIL; |
715 : | } | ||
716 : | Isibaar | 3 | } |
No admin address has been configured | ViewVC Help |
Powered by ViewVC 1.0.4 |