3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* - Native API implementation - |
* - Native API implementation - |
5 |
* |
* |
6 |
* This program is an implementation of a part of one or more MPEG-4 |
* Copyright(C) 2001-2004 Peter Ross <pross@xvid.org> |
|
* Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
|
|
* to use this software module in hardware or software products are |
|
|
* advised that its use may infringe existing patents or copyrights, and |
|
|
* any such use would be at such party's own risk. The original |
|
|
* developer of this software module and his/her company, and subsequent |
|
|
* editors and their companies, will have no liability for use of this |
|
|
* software or modifications or derivatives thereof. |
|
7 |
* |
* |
8 |
* This program is free software ; you can redistribute it and/or modify |
* 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 |
* it under the terms of the GNU General Public License as published by |
19 |
* along with this program ; if not, write to the Free Software |
* along with this program ; if not, write to the Free Software |
20 |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
21 |
* |
* |
22 |
****************************************************************************/ |
* $Id: xvid.c,v 1.49 2004-04-01 11:11:28 suxen_drol Exp $ |
|
/***************************************************************************** |
|
|
* |
|
|
* History |
|
|
* |
|
|
* - 23.06.2002 added XVID_CPU_CHKONLY |
|
|
* - 17.03.2002 Added interpolate8x8_halfpel_hv_xmm |
|
|
* - 22.12.2001 API change: added xvid_init() - Isibaar |
|
|
* - 16.12.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
|
|
* |
|
|
* $Id: xvid.c,v 1.26 2002-07-09 01:09:33 chenm001 Exp $ |
|
23 |
* |
* |
24 |
****************************************************************************/ |
****************************************************************************/ |
25 |
|
|
26 |
|
#include <stdio.h> |
27 |
|
#include <stdlib.h> |
28 |
|
#include <string.h> |
29 |
|
#include <time.h> |
30 |
|
|
31 |
#include "xvid.h" |
#include "xvid.h" |
32 |
#include "decoder.h" |
#include "decoder.h" |
33 |
#include "encoder.h" |
#include "encoder.h" |
36 |
#include "dct/fdct.h" |
#include "dct/fdct.h" |
37 |
#include "image/colorspace.h" |
#include "image/colorspace.h" |
38 |
#include "image/interpolate8x8.h" |
#include "image/interpolate8x8.h" |
39 |
|
#include "image/reduced.h" |
40 |
#include "utils/mem_transfer.h" |
#include "utils/mem_transfer.h" |
41 |
#include "quant/quant_h263.h" |
#include "utils/mbfunctions.h" |
42 |
#include "quant/quant_mpeg4.h" |
#include "quant/quant.h" |
43 |
|
#include "motion/motion.h" |
44 |
#include "motion/sad.h" |
#include "motion/sad.h" |
45 |
#include "utils/emms.h" |
#include "utils/emms.h" |
46 |
#include "utils/timer.h" |
#include "utils/timer.h" |
47 |
#include "bitstream/mbcoding.h" |
#include "bitstream/mbcoding.h" |
48 |
|
#include "image/qpel.h" |
49 |
|
#include "image/postprocessing.h" |
50 |
|
|
51 |
|
#if defined(_DEBUG) |
52 |
|
unsigned int xvid_debug = 0; /* xvid debug mask */ |
53 |
|
#endif |
54 |
|
|
55 |
|
#if defined(ARCH_IS_IA32) |
56 |
|
#if defined(_MSC_VER) |
57 |
|
# include <windows.h> |
58 |
|
#else |
59 |
|
# include <signal.h> |
60 |
|
# include <setjmp.h> |
61 |
|
|
62 |
|
static jmp_buf mark; |
63 |
|
|
64 |
|
static void |
65 |
|
sigill_handler(int signal) |
66 |
|
{ |
67 |
|
longjmp(mark, 1); |
68 |
|
} |
69 |
|
#endif |
70 |
|
|
71 |
|
|
72 |
|
/* |
73 |
|
* Calls the funcptr, and returns whether SIGILL (illegal instruction) was |
74 |
|
* signalled |
75 |
|
* |
76 |
|
* Return values: |
77 |
|
* -1 : could not determine |
78 |
|
* 0 : SIGILL was *not* signalled |
79 |
|
* 1 : SIGILL was signalled |
80 |
|
*/ |
81 |
|
|
82 |
|
int |
83 |
|
sigill_check(void (*func)()) |
84 |
|
{ |
85 |
|
#if defined(_MSC_VER) |
86 |
|
_try { |
87 |
|
func(); |
88 |
|
} |
89 |
|
_except(EXCEPTION_EXECUTE_HANDLER) { |
90 |
|
|
91 |
|
if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION) |
92 |
|
return 1; |
93 |
|
} |
94 |
|
return 0; |
95 |
|
#else |
96 |
|
void * old_handler; |
97 |
|
int jmpret; |
98 |
|
|
99 |
|
|
100 |
|
old_handler = signal(SIGILL, sigill_handler); |
101 |
|
if (old_handler == SIG_ERR) |
102 |
|
{ |
103 |
|
return -1; |
104 |
|
} |
105 |
|
|
106 |
|
jmpret = setjmp(mark); |
107 |
|
if (jmpret == 0) |
108 |
|
{ |
109 |
|
func(); |
110 |
|
} |
111 |
|
|
112 |
|
signal(SIGILL, old_handler); |
113 |
|
|
114 |
|
return jmpret; |
115 |
|
#endif |
116 |
|
} |
117 |
|
#endif |
118 |
|
|
119 |
|
|
120 |
|
/* detect cpu flags */ |
121 |
|
static unsigned int |
122 |
|
detect_cpu_flags() |
123 |
|
{ |
124 |
|
/* enable native assembly optimizations by default */ |
125 |
|
unsigned int cpu_flags = XVID_CPU_ASM; |
126 |
|
|
127 |
|
#if defined(ARCH_IS_IA32) |
128 |
|
cpu_flags |= check_cpu_features(); |
129 |
|
if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) |
130 |
|
cpu_flags &= ~XVID_CPU_SSE; |
131 |
|
|
132 |
|
if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) |
133 |
|
cpu_flags &= ~XVID_CPU_SSE2; |
134 |
|
#endif |
135 |
|
|
136 |
|
#if defined(ARCH_IS_PPC) |
137 |
|
#if defined(ARCH_IS_PPC_ALTIVEC) |
138 |
|
cpu_flags |= XVID_CPU_ALTIVEC; |
139 |
|
#endif |
140 |
|
#endif |
141 |
|
|
142 |
|
return cpu_flags; |
143 |
|
} |
144 |
|
|
145 |
|
|
146 |
/***************************************************************************** |
/***************************************************************************** |
147 |
* XviD Init Entry point |
* XviD Init Entry point |
157 |
* |
* |
158 |
****************************************************************************/ |
****************************************************************************/ |
159 |
|
|
|
int |
|
|
xvid_init(void *handle, |
|
|
int opt, |
|
|
void *param1, |
|
|
void *param2) |
|
|
{ |
|
|
int cpu_flags; |
|
|
XVID_INIT_PARAM *init_param; |
|
160 |
|
|
161 |
init_param = (XVID_INIT_PARAM *) param1; |
static |
162 |
|
int xvid_gbl_init(xvid_gbl_init_t * init) |
|
/* Inform the client the API version */ |
|
|
init_param->api_version = API_VERSION; |
|
|
|
|
|
/* Inform the client the core build - unused because we're still alpha */ |
|
|
init_param->core_build = 1000; |
|
|
|
|
|
printf("init_param->cpu_flags %x\n",init_param->cpu_flags);//NIC |
|
|
|
|
|
if ((init_param->cpu_flags & XVID_CPU_CHKONLY)) |
|
163 |
{ |
{ |
164 |
//init_param->cpu_flags = check_cpu_features();//nic |
unsigned int cpu_flags; |
|
return XVID_ERR_OK; |
|
|
} |
|
165 |
|
|
166 |
/* Do we have to force CPU features ? */ |
if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */ |
167 |
if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0) { |
return XVID_ERR_VERSION; |
|
cpu_flags = init_param->cpu_flags; |
|
|
} else { |
|
168 |
|
|
169 |
//cpu_flags = check_cpu_features();//nic |
cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags(); |
|
init_param->cpu_flags = cpu_flags; |
|
|
} |
|
170 |
|
|
171 |
/* Initialize the function pointers */ |
/* Initialize the function pointers */ |
172 |
idct_int32_init(); |
idct_int32_init(); |
182 |
/* Restore FPU context : emms_c is a nop functions */ |
/* Restore FPU context : emms_c is a nop functions */ |
183 |
emms = emms_c; |
emms = emms_c; |
184 |
|
|
185 |
|
/* Qpel stuff */ |
186 |
|
xvid_QP_Funcs = &xvid_QP_Funcs_C; |
187 |
|
xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C; |
188 |
|
xvid_Init_QP(); |
189 |
|
|
190 |
/* Quantization functions */ |
/* Quantization functions */ |
191 |
quant_intra = quant_intra_c; |
quant_h263_intra = quant_h263_intra_c; |
192 |
dequant_intra = dequant_intra_c; |
quant_h263_inter = quant_h263_inter_c; |
193 |
quant_inter = quant_inter_c; |
dequant_h263_intra = dequant_h263_intra_c; |
194 |
dequant_inter = dequant_inter_c; |
dequant_h263_inter = dequant_h263_inter_c; |
195 |
|
|
196 |
quant4_intra = quant4_intra_c; |
quant_mpeg_intra = quant_mpeg_intra_c; |
197 |
dequant4_intra = dequant4_intra_c; |
quant_mpeg_inter = quant_mpeg_inter_c; |
198 |
quant4_inter = quant4_inter_c; |
dequant_mpeg_intra = dequant_mpeg_intra_c; |
199 |
dequant4_inter = dequant4_inter_c; |
dequant_mpeg_inter = dequant_mpeg_inter_c; |
200 |
|
|
201 |
/* Block transfer related functions */ |
/* Block transfer related functions */ |
202 |
transfer_8to16copy = transfer_8to16copy_c; |
transfer_8to16copy = transfer_8to16copy_c; |
203 |
transfer_16to8copy = transfer_16to8copy_c; |
transfer_16to8copy = transfer_16to8copy_c; |
204 |
transfer_8to16sub = transfer_8to16sub_c; |
transfer_8to16sub = transfer_8to16sub_c; |
205 |
|
transfer_8to16subro = transfer_8to16subro_c; |
206 |
transfer_8to16sub2 = transfer_8to16sub2_c; |
transfer_8to16sub2 = transfer_8to16sub2_c; |
207 |
transfer_16to8add = transfer_16to8add_c; |
transfer_16to8add = transfer_16to8add_c; |
208 |
transfer8x8_copy = transfer8x8_copy_c; |
transfer8x8_copy = transfer8x8_copy_c; |
209 |
|
|
210 |
|
/* Interlacing functions */ |
211 |
|
MBFieldTest = MBFieldTest_c; |
212 |
|
|
213 |
/* Image interpolation related functions */ |
/* Image interpolation related functions */ |
214 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; |
215 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; |
216 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
217 |
|
|
218 |
|
interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; |
219 |
|
interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; |
220 |
|
interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; |
221 |
|
|
222 |
|
interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c; |
223 |
|
interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c; |
224 |
|
interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c; |
225 |
|
|
226 |
|
interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c; |
227 |
|
interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c; |
228 |
|
|
229 |
|
interpolate8x8_avg2 = interpolate8x8_avg2_c; |
230 |
|
interpolate8x8_avg4 = interpolate8x8_avg4_c; |
231 |
|
|
232 |
|
/* postprocessing */ |
233 |
|
image_brightness = image_brightness_c; |
234 |
|
|
235 |
|
/* reduced resolution */ |
236 |
|
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
237 |
|
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; |
238 |
|
vfilter_31 = xvid_VFilter_31_C; |
239 |
|
hfilter_31 = xvid_HFilter_31_C; |
240 |
|
filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; |
241 |
|
filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; |
242 |
|
|
243 |
/* Initialize internal colorspace transformation tables */ |
/* Initialize internal colorspace transformation tables */ |
244 |
colorspace_init(); |
colorspace_init(); |
245 |
|
|
246 |
/* All colorspace transformation functions User Format->YV12 */ |
/* All colorspace transformation functions User Format->YV12 */ |
247 |
|
yv12_to_yv12 = yv12_to_yv12_c; |
248 |
rgb555_to_yv12 = rgb555_to_yv12_c; |
rgb555_to_yv12 = rgb555_to_yv12_c; |
249 |
rgb565_to_yv12 = rgb565_to_yv12_c; |
rgb565_to_yv12 = rgb565_to_yv12_c; |
250 |
rgb24_to_yv12 = rgb24_to_yv12_c; |
bgr_to_yv12 = bgr_to_yv12_c; |
251 |
rgb32_to_yv12 = rgb32_to_yv12_c; |
bgra_to_yv12 = bgra_to_yv12_c; |
252 |
yuv_to_yv12 = yuv_to_yv12_c; |
abgr_to_yv12 = abgr_to_yv12_c; |
253 |
|
rgba_to_yv12 = rgba_to_yv12_c; |
254 |
|
argb_to_yv12 = argb_to_yv12_c; |
255 |
yuyv_to_yv12 = yuyv_to_yv12_c; |
yuyv_to_yv12 = yuyv_to_yv12_c; |
256 |
uyvy_to_yv12 = uyvy_to_yv12_c; |
uyvy_to_yv12 = uyvy_to_yv12_c; |
257 |
|
|
258 |
|
rgb555i_to_yv12 = rgb555i_to_yv12_c; |
259 |
|
rgb565i_to_yv12 = rgb565i_to_yv12_c; |
260 |
|
bgri_to_yv12 = bgri_to_yv12_c; |
261 |
|
bgrai_to_yv12 = bgrai_to_yv12_c; |
262 |
|
abgri_to_yv12 = abgri_to_yv12_c; |
263 |
|
rgbai_to_yv12 = rgbai_to_yv12_c; |
264 |
|
argbi_to_yv12 = argbi_to_yv12_c; |
265 |
|
yuyvi_to_yv12 = yuyvi_to_yv12_c; |
266 |
|
uyvyi_to_yv12 = uyvyi_to_yv12_c; |
267 |
|
|
268 |
/* All colorspace transformation functions YV12->User format */ |
/* All colorspace transformation functions YV12->User format */ |
269 |
yv12_to_rgb555 = yv12_to_rgb555_c; |
yv12_to_rgb555 = yv12_to_rgb555_c; |
270 |
yv12_to_rgb565 = yv12_to_rgb565_c; |
yv12_to_rgb565 = yv12_to_rgb565_c; |
271 |
yv12_to_rgb24 = yv12_to_rgb24_c; |
yv12_to_bgr = yv12_to_bgr_c; |
272 |
yv12_to_rgb32 = yv12_to_rgb32_c; |
yv12_to_bgra = yv12_to_bgra_c; |
273 |
yv12_to_yuv = yv12_to_yuv_c; |
yv12_to_abgr = yv12_to_abgr_c; |
274 |
|
yv12_to_rgba = yv12_to_rgba_c; |
275 |
|
yv12_to_argb = yv12_to_argb_c; |
276 |
yv12_to_yuyv = yv12_to_yuyv_c; |
yv12_to_yuyv = yv12_to_yuyv_c; |
277 |
yv12_to_uyvy = yv12_to_uyvy_c; |
yv12_to_uyvy = yv12_to_uyvy_c; |
278 |
|
|
279 |
|
yv12_to_rgb555i = yv12_to_rgb555i_c; |
280 |
|
yv12_to_rgb565i = yv12_to_rgb565i_c; |
281 |
|
yv12_to_bgri = yv12_to_bgri_c; |
282 |
|
yv12_to_bgrai = yv12_to_bgrai_c; |
283 |
|
yv12_to_abgri = yv12_to_abgri_c; |
284 |
|
yv12_to_rgbai = yv12_to_rgbai_c; |
285 |
|
yv12_to_argbi = yv12_to_argbi_c; |
286 |
|
yv12_to_yuyvi = yv12_to_yuyvi_c; |
287 |
|
yv12_to_uyvyi = yv12_to_uyvyi_c; |
288 |
|
|
289 |
/* Functions used in motion estimation algorithms */ |
/* Functions used in motion estimation algorithms */ |
290 |
calc_cbp = calc_cbp_c; |
calc_cbp = calc_cbp_c; |
291 |
sad16 = sad16_c; |
sad16 = sad16_c; |
|
sad16bi = sad16bi_c; |
|
292 |
sad8 = sad8_c; |
sad8 = sad8_c; |
293 |
|
sad16bi = sad16bi_c; |
294 |
|
sad8bi = sad8bi_c; |
295 |
dev16 = dev16_c; |
dev16 = dev16_c; |
296 |
|
sad16v = sad16v_c; |
297 |
|
sse8_16bit = sse8_16bit_c; |
298 |
|
|
299 |
|
#if defined(ARCH_IS_IA32) |
300 |
|
|
301 |
|
if ((cpu_flags & XVID_CPU_ASM)) { |
302 |
|
vfilter_31 = xvid_VFilter_31_x86; |
303 |
|
hfilter_31 = xvid_HFilter_31_x86; |
304 |
|
} |
305 |
|
|
306 |
|
if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || |
307 |
|
(cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || |
308 |
|
(cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2)) |
309 |
|
{ |
310 |
|
/* Restore FPU context : emms_c is a nop functions */ |
311 |
|
emms = emms_mmx; |
312 |
|
} |
313 |
|
|
314 |
#ifdef ARCH_X86 |
if ((cpu_flags & XVID_CPU_MMX)) { |
|
if ((cpu_flags & XVID_CPU_MMX) > 0) { |
|
315 |
|
|
316 |
/* Forward and Inverse Discrete Cosine Transformation functions */ |
/* Forward and Inverse Discrete Cosine Transformation functions */ |
317 |
fdct = fdct_mmx; |
fdct = fdct_mmx_skal; |
318 |
idct = idct_mmx; |
idct = idct_mmx; |
319 |
|
|
320 |
/* To restore FPU context after mmx use */ |
/* Qpel stuff */ |
321 |
emms = emms_mmx; |
xvid_QP_Funcs = &xvid_QP_Funcs_mmx; |
322 |
|
xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx; |
323 |
|
|
324 |
/* Quantization related functions */ |
/* Quantization related functions */ |
325 |
quant_intra = quant_intra_mmx; |
quant_h263_intra = quant_h263_intra_mmx; |
326 |
dequant_intra = dequant_intra_mmx; |
quant_h263_inter = quant_h263_inter_mmx; |
327 |
quant_inter = quant_inter_mmx; |
dequant_h263_intra = dequant_h263_intra_mmx; |
328 |
dequant_inter = dequant_inter_mmx; |
dequant_h263_inter = dequant_h263_inter_mmx; |
329 |
|
|
330 |
quant4_intra = quant4_intra_mmx; |
quant_mpeg_intra = quant_mpeg_intra_mmx; |
331 |
dequant4_intra = dequant4_intra_mmx; |
quant_mpeg_inter = quant_mpeg_inter_mmx; |
332 |
quant4_inter = quant4_inter_mmx; |
dequant_mpeg_intra = dequant_mpeg_intra_mmx; |
333 |
dequant4_inter = dequant4_inter_mmx; |
dequant_mpeg_inter = dequant_mpeg_inter_mmx; |
334 |
|
|
335 |
/* Block related functions */ |
/* Block related functions */ |
336 |
transfer_8to16copy = transfer_8to16copy_mmx; |
transfer_8to16copy = transfer_8to16copy_mmx; |
337 |
transfer_16to8copy = transfer_16to8copy_mmx; |
transfer_16to8copy = transfer_16to8copy_mmx; |
338 |
transfer_8to16sub = transfer_8to16sub_mmx; |
transfer_8to16sub = transfer_8to16sub_mmx; |
339 |
|
transfer_8to16subro = transfer_8to16subro_mmx; |
340 |
transfer_8to16sub2 = transfer_8to16sub2_mmx; |
transfer_8to16sub2 = transfer_8to16sub2_mmx; |
341 |
transfer_16to8add = transfer_16to8add_mmx; |
transfer_16to8add = transfer_16to8add_mmx; |
342 |
transfer8x8_copy = transfer8x8_copy_mmx; |
transfer8x8_copy = transfer8x8_copy_mmx; |
343 |
|
|
344 |
|
/* Interlacing Functions */ |
345 |
|
MBFieldTest = MBFieldTest_mmx; |
346 |
|
|
347 |
/* Image Interpolation related functions */ |
/* Image Interpolation related functions */ |
348 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; |
349 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; |
350 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
351 |
|
|
352 |
/* Image RGB->YV12 related functions */ |
interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; |
353 |
rgb24_to_yv12 = rgb24_to_yv12_mmx; |
interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; |
354 |
rgb32_to_yv12 = rgb32_to_yv12_mmx; |
|
355 |
yuv_to_yv12 = yuv_to_yv12_mmx; |
interpolate8x8_avg2 = interpolate8x8_avg2_mmx; |
356 |
|
interpolate8x8_avg4 = interpolate8x8_avg4_mmx; |
357 |
|
|
358 |
|
/* postprocessing */ |
359 |
|
image_brightness = image_brightness_mmx; |
360 |
|
|
361 |
|
/* reduced resolution */ |
362 |
|
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx; |
363 |
|
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx; |
364 |
|
hfilter_31 = xvid_HFilter_31_mmx; |
365 |
|
filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx; |
366 |
|
filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx; |
367 |
|
|
368 |
|
/* image input xxx_to_yv12 related functions */ |
369 |
|
yv12_to_yv12 = yv12_to_yv12_mmx; |
370 |
|
bgr_to_yv12 = bgr_to_yv12_mmx; |
371 |
|
bgra_to_yv12 = bgra_to_yv12_mmx; |
372 |
yuyv_to_yv12 = yuyv_to_yv12_mmx; |
yuyv_to_yv12 = yuyv_to_yv12_mmx; |
373 |
uyvy_to_yv12 = uyvy_to_yv12_mmx; |
uyvy_to_yv12 = uyvy_to_yv12_mmx; |
374 |
|
|
375 |
/* Image YV12->RGB related functions */ |
/* image output yv12_to_xxx related functions */ |
376 |
yv12_to_rgb24 = yv12_to_rgb24_mmx; |
yv12_to_bgr = yv12_to_bgr_mmx; |
377 |
yv12_to_rgb32 = yv12_to_rgb32_mmx; |
yv12_to_bgra = yv12_to_bgra_mmx; |
378 |
yv12_to_yuyv = yv12_to_yuyv_mmx; |
yv12_to_yuyv = yv12_to_yuyv_mmx; |
379 |
yv12_to_uyvy = yv12_to_uyvy_mmx; |
yv12_to_uyvy = yv12_to_uyvy_mmx; |
380 |
|
|
381 |
|
yv12_to_yuyvi = yv12_to_yuyvi_mmx; |
382 |
|
yv12_to_uyvyi = yv12_to_uyvyi_mmx; |
383 |
|
|
384 |
/* Motion estimation related functions */ |
/* Motion estimation related functions */ |
385 |
calc_cbp = calc_cbp_mmx; |
calc_cbp = calc_cbp_mmx; |
386 |
sad16 = sad16_mmx; |
sad16 = sad16_mmx; |
387 |
sad8 = sad8_mmx; |
sad8 = sad8_mmx; |
388 |
|
sad16bi = sad16bi_mmx; |
389 |
|
sad8bi = sad8bi_mmx; |
390 |
dev16 = dev16_mmx; |
dev16 = dev16_mmx; |
391 |
|
sad16v = sad16v_mmx; |
392 |
|
sse8_16bit = sse8_16bit_mmx; |
393 |
|
} |
394 |
|
|
395 |
|
/* these 3dnow functions are faster than mmx, but slower than xmm. */ |
396 |
|
if ((cpu_flags & XVID_CPU_3DNOW)) { |
397 |
|
|
398 |
|
emms = emms_3dn; |
399 |
|
|
400 |
|
/* ME functions */ |
401 |
|
sad16bi = sad16bi_3dn; |
402 |
|
sad8bi = sad8bi_3dn; |
403 |
|
|
404 |
|
yuyv_to_yv12 = yuyv_to_yv12_3dn; |
405 |
|
uyvy_to_yv12 = uyvy_to_yv12_3dn; |
406 |
} |
} |
407 |
|
|
|
if ((cpu_flags & XVID_CPU_MMXEXT) > 0) { |
|
408 |
|
|
409 |
/* Inverse DCT */ |
if ((cpu_flags & XVID_CPU_MMXEXT)) { |
410 |
|
|
411 |
|
/* DCT */ |
412 |
|
fdct = fdct_xmm_skal; |
413 |
idct = idct_xmm; |
idct = idct_xmm; |
414 |
|
|
415 |
/* Interpolation */ |
/* Interpolation */ |
417 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; |
418 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
419 |
|
|
420 |
|
/* reduced resolution */ |
421 |
|
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm; |
422 |
|
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm; |
423 |
|
|
424 |
|
/* Quantization */ |
425 |
|
quant_mpeg_intra = quant_mpeg_intra_xmm; |
426 |
|
quant_mpeg_inter = quant_mpeg_inter_xmm; |
427 |
|
|
428 |
|
dequant_h263_intra = dequant_h263_intra_xmm; |
429 |
|
dequant_h263_inter = dequant_h263_inter_xmm; |
430 |
|
|
431 |
/* Buffer transfer */ |
/* Buffer transfer */ |
432 |
transfer_8to16sub2 = transfer_8to16sub2_xmm; |
transfer_8to16sub2 = transfer_8to16sub2_xmm; |
433 |
|
|
434 |
/* Colorspace transformation */ |
/* Colorspace transformation */ |
435 |
yuv_to_yv12 = yuv_to_yv12_xmm; |
yv12_to_yv12 = yv12_to_yv12_xmm; |
436 |
|
yuyv_to_yv12 = yuyv_to_yv12_xmm; |
437 |
|
uyvy_to_yv12 = uyvy_to_yv12_xmm; |
438 |
|
|
439 |
/* ME functions */ |
/* ME functions */ |
440 |
sad16 = sad16_xmm; |
sad16 = sad16_xmm; |
441 |
sad8 = sad8_xmm; |
sad8 = sad8_xmm; |
442 |
|
sad16bi = sad16bi_xmm; |
443 |
|
sad8bi = sad8bi_xmm; |
444 |
dev16 = dev16_xmm; |
dev16 = dev16_xmm; |
445 |
|
sad16v = sad16v_xmm; |
446 |
} |
} |
447 |
|
|
448 |
if ((cpu_flags & XVID_CPU_3DNOW) > 0) { |
if ((cpu_flags & XVID_CPU_3DNOW)) { |
449 |
|
|
450 |
/* Interpolation */ |
/* Interpolation */ |
451 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
453 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
454 |
} |
} |
455 |
|
|
456 |
if ((cpu_flags & XVID_CPU_SSE2) > 0) { |
if ((cpu_flags & XVID_CPU_3DNOWEXT)) { |
457 |
#ifdef EXPERIMENTAL_SSE2_CODE |
|
458 |
|
/* Inverse DCT */ |
459 |
|
idct = idct_3dne; |
460 |
|
|
461 |
|
/* Buffer transfer */ |
462 |
|
transfer_8to16copy = transfer_8to16copy_3dne; |
463 |
|
transfer_16to8copy = transfer_16to8copy_3dne; |
464 |
|
transfer_8to16sub = transfer_8to16sub_3dne; |
465 |
|
transfer_8to16subro = transfer_8to16subro_3dne; |
466 |
|
transfer_8to16sub2 = transfer_8to16sub2_3dne; |
467 |
|
transfer_16to8add = transfer_16to8add_3dne; |
468 |
|
transfer8x8_copy = transfer8x8_copy_3dne; |
469 |
|
|
470 |
/* Quantization */ |
/* Quantization */ |
471 |
quant_intra = quant_intra_sse2; |
quant_h263_intra = quant_h263_intra_3dne; |
472 |
dequant_intra = dequant_intra_sse2; |
quant_h263_inter = quant_h263_inter_3dne; |
473 |
quant_inter = quant_inter_sse2; |
dequant_mpeg_intra = dequant_mpeg_intra_3dne; |
474 |
dequant_inter = dequant_inter_sse2; |
dequant_mpeg_inter = dequant_mpeg_inter_3dne; |
475 |
|
dequant_h263_intra = dequant_h263_intra_3dne; |
476 |
|
dequant_h263_inter = dequant_h263_inter_3dne; |
477 |
|
|
478 |
|
/* ME functions */ |
479 |
|
calc_cbp = calc_cbp_3dne; |
480 |
|
sad16 = sad16_3dne; |
481 |
|
sad8 = sad8_3dne; |
482 |
|
sad16bi = sad16bi_3dne; |
483 |
|
sad8bi = sad8bi_3dne; |
484 |
|
dev16 = dev16_3dne; |
485 |
|
|
486 |
|
/* Interpolation */ |
487 |
|
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne; |
488 |
|
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne; |
489 |
|
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; |
490 |
|
} |
491 |
|
|
492 |
|
if ((cpu_flags & XVID_CPU_SSE2)) { |
493 |
|
|
|
/* ME */ |
|
494 |
calc_cbp = calc_cbp_sse2; |
calc_cbp = calc_cbp_sse2; |
495 |
|
|
496 |
|
/* Quantization */ |
497 |
|
quant_h263_intra = quant_h263_intra_sse2; |
498 |
|
quant_h263_inter = quant_h263_inter_sse2; |
499 |
|
dequant_h263_intra = dequant_h263_intra_sse2; |
500 |
|
dequant_h263_inter = dequant_h263_inter_sse2; |
501 |
|
|
502 |
|
/* SAD operators */ |
503 |
sad16 = sad16_sse2; |
sad16 = sad16_sse2; |
504 |
dev16 = dev16_sse2; |
dev16 = dev16_sse2; |
505 |
|
|
506 |
/* Forward and Inverse DCT */ |
/* DCT operators |
507 |
idct = idct_sse2; |
* no iDCT because it's not "Walken matching" */ |
508 |
fdct = fdct_sse2; |
fdct = fdct_sse2_skal; |
|
#endif |
|
509 |
} |
} |
510 |
|
#endif /* ARCH_IS_IA32 */ |
511 |
|
|
512 |
#endif |
#if defined(ARCH_IS_IA64) |
513 |
|
if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */ |
|
#ifdef ARCH_IA64 |
|
|
if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines? |
|
514 |
idct_ia64_init(); |
idct_ia64_init(); |
515 |
fdct = fdct_ia64; |
fdct = fdct_ia64; |
516 |
idct = idct_ia64; //not yet working, crashes |
idct = idct_ia64; /*not yet working, crashes */ |
517 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
518 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; |
519 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; |
521 |
sad16bi = sad16bi_ia64; |
sad16bi = sad16bi_ia64; |
522 |
sad8 = sad8_ia64; |
sad8 = sad8_ia64; |
523 |
dev16 = dev16_ia64; |
dev16 = dev16_ia64; |
524 |
quant_intra = quant_intra_ia64; |
/* Halfpel8_Refine = Halfpel8_Refine_ia64; */ |
525 |
dequant_intra = dequant_intra_ia64; |
quant_h263_intra = quant_h263_intra_ia64; |
526 |
quant_inter = quant_inter_ia64; |
quant_h263_inter = quant_h263_inter_ia64; |
527 |
dequant_inter = dequant_inter_ia64; |
dequant_h263_intra = dequant_h263_intra_ia64; |
528 |
|
dequant_h263_inter = dequant_h263_inter_ia64; |
529 |
transfer_8to16copy = transfer_8to16copy_ia64; |
transfer_8to16copy = transfer_8to16copy_ia64; |
530 |
transfer_16to8copy = transfer_16to8copy_ia64; |
transfer_16to8copy = transfer_16to8copy_ia64; |
531 |
transfer_8to16sub = transfer_8to16sub_ia64; |
transfer_8to16sub = transfer_8to16sub_ia64; |
532 |
transfer_8to16sub2 = transfer_8to16sub2_ia64; |
transfer_8to16sub2 = transfer_8to16sub2_ia64; |
533 |
transfer_16to8add = transfer_16to8add_ia64; |
transfer_16to8add = transfer_16to8add_ia64; |
534 |
transfer8x8_copy = transfer8x8_copy_ia64; |
transfer8x8_copy = transfer8x8_copy_ia64; |
|
DEBUG("Using IA-64 assembler routines.\n"); |
|
535 |
} |
} |
536 |
#endif |
#endif |
537 |
|
|
538 |
#ifdef ARCH_PPC |
#if defined(ARCH_IS_PPC) |
539 |
#ifdef ARCH_PPC_ALTIVEC |
if ((cpu_flags & XVID_CPU_ASM)) |
540 |
|
{ |
541 |
|
calc_cbp = calc_cbp_ppc; |
542 |
|
} |
543 |
|
|
544 |
|
if ((cpu_flags & XVID_CPU_ALTIVEC)) |
545 |
|
{ |
546 |
calc_cbp = calc_cbp_altivec; |
calc_cbp = calc_cbp_altivec; |
547 |
fdct = fdct_altivec; |
fdct = fdct_altivec; |
548 |
idct = idct_altivec; |
idct = idct_altivec; |
550 |
sad16 = sad16_altivec; |
sad16 = sad16_altivec; |
551 |
sad8 = sad8_altivec; |
sad8 = sad8_altivec; |
552 |
dev16 = dev16_altivec; |
dev16 = dev16_altivec; |
553 |
|
} |
554 |
|
#endif |
555 |
|
|
556 |
|
#if defined(_DEBUG) |
557 |
|
xvid_debug = init->debug; |
558 |
|
#endif |
559 |
|
|
560 |
|
return 0; |
561 |
|
} |
562 |
|
|
563 |
|
|
564 |
|
static int |
565 |
|
xvid_gbl_info(xvid_gbl_info_t * info) |
566 |
|
{ |
567 |
|
if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */ |
568 |
|
return XVID_ERR_VERSION; |
569 |
|
|
570 |
|
info->actual_version = XVID_VERSION; |
571 |
|
info->build = "xvid-1.0.0"; |
572 |
|
info->cpu_flags = detect_cpu_flags(); |
573 |
|
|
574 |
|
#if defined(_SMP) && defined(WIN32) |
575 |
|
info->num_threads = pthread_num_processors_np();; |
576 |
#else |
#else |
577 |
calc_cbp = calc_cbp_ppc; |
info->num_threads = 0; |
578 |
#endif |
#endif |
579 |
|
|
580 |
|
return 0; |
581 |
|
} |
582 |
|
|
583 |
|
|
584 |
|
static int |
585 |
|
xvid_gbl_convert(xvid_gbl_convert_t* convert) |
586 |
|
{ |
587 |
|
int width; |
588 |
|
int height; |
589 |
|
int width2; |
590 |
|
int height2; |
591 |
|
IMAGE img; |
592 |
|
|
593 |
|
if (XVID_VERSION_MAJOR(convert->version) != 1) /* v1.x.x */ |
594 |
|
return XVID_ERR_VERSION; |
595 |
|
|
596 |
|
#if 0 |
597 |
|
const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); |
598 |
#endif |
#endif |
599 |
|
width = convert->width; |
600 |
|
height = convert->height; |
601 |
|
width2 = convert->width/2; |
602 |
|
height2 = convert->height/2; |
603 |
|
|
604 |
|
switch (convert->input.csp & ~XVID_CSP_VFLIP) |
605 |
|
{ |
606 |
|
case XVID_CSP_YV12 : |
607 |
|
img.y = convert->input.plane[0]; |
608 |
|
img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height; |
609 |
|
img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2; |
610 |
|
image_output(&img, width, height, width, |
611 |
|
(uint8_t**)convert->output.plane, convert->output.stride, |
612 |
|
convert->output.csp, convert->interlacing); |
613 |
|
break; |
614 |
|
|
615 |
|
default : |
616 |
|
return XVID_ERR_FORMAT; |
617 |
|
} |
618 |
|
|
619 |
|
|
620 |
|
emms(); |
621 |
|
return 0; |
622 |
|
} |
623 |
|
|
624 |
|
/***************************************************************************** |
625 |
|
* XviD Global Entry point |
626 |
|
* |
627 |
|
* Well this function initialize all internal function pointers according |
628 |
|
* to the CPU features forced by the library client or autodetected (depending |
629 |
|
* on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all |
630 |
|
* image colorspace transformation tables. |
631 |
|
* |
632 |
|
****************************************************************************/ |
633 |
|
|
634 |
|
|
635 |
|
int |
636 |
|
xvid_global(void *handle, |
637 |
|
int opt, |
638 |
|
void *param1, |
639 |
|
void *param2) |
640 |
|
{ |
641 |
|
switch(opt) |
642 |
|
{ |
643 |
|
case XVID_GBL_INIT : |
644 |
|
return xvid_gbl_init((xvid_gbl_init_t*)param1); |
645 |
|
|
646 |
|
case XVID_GBL_INFO : |
647 |
|
return xvid_gbl_info((xvid_gbl_info_t*)param1); |
648 |
|
|
649 |
|
case XVID_GBL_CONVERT : |
650 |
|
return xvid_gbl_convert((xvid_gbl_convert_t*)param1); |
651 |
|
|
652 |
return XVID_ERR_OK; |
default : |
653 |
|
return XVID_ERR_FAIL; |
654 |
|
} |
655 |
} |
} |
656 |
|
|
657 |
/***************************************************************************** |
/***************************************************************************** |
671 |
void *param2) |
void *param2) |
672 |
{ |
{ |
673 |
switch (opt) { |
switch (opt) { |
|
case XVID_DEC_DECODE: |
|
|
return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1); |
|
|
|
|
674 |
case XVID_DEC_CREATE: |
case XVID_DEC_CREATE: |
675 |
/* *************************************************************************** |
return decoder_create((xvid_dec_create_t *) param1); |
|
NIC uso il secondo parametro 'param2' ma in realta` non so bene per cosa e` |
|
|
stato pensato ..... e quindi in futuro potrebbe essere un problema |
|
|
*************************************************************************** */ |
|
|
if(param2!=NULL) |
|
|
return IM1_decoder_create((XVID_DEC_PARAM *) param1,(XVID_DEC_FRAME *) param2); |
|
|
else |
|
|
return decoder_create((XVID_DEC_PARAM *) param1); //NIC commentata |
|
676 |
|
|
677 |
case XVID_DEC_DESTROY: |
case XVID_DEC_DESTROY: |
678 |
return decoder_destroy((DECODER *) handle); |
return decoder_destroy((DECODER *) handle); |
679 |
|
|
680 |
|
case XVID_DEC_DECODE: |
681 |
|
return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2); |
682 |
|
|
683 |
default: |
default: |
684 |
return XVID_ERR_FAIL; |
return XVID_ERR_FAIL; |
685 |
} |
} |
704 |
{ |
{ |
705 |
switch (opt) { |
switch (opt) { |
706 |
case XVID_ENC_ENCODE: |
case XVID_ENC_ENCODE: |
707 |
#ifdef BFRAMES |
|
708 |
if (((Encoder *) handle)->mbParam.max_bframes >= 0) |
return enc_encode((Encoder *) handle, |
709 |
return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
(xvid_enc_frame_t *) param1, |
710 |
(XVID_ENC_STATS *) param2); |
(xvid_enc_stats_t *) param2); |
|
else |
|
|
#endif |
|
|
return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
|
|
(XVID_ENC_STATS *) param2); |
|
711 |
|
|
712 |
case XVID_ENC_CREATE: |
case XVID_ENC_CREATE: |
713 |
return encoder_create((XVID_ENC_PARAM *) param1); |
return enc_create((xvid_enc_create_t *) param1); |
714 |
|
|
715 |
case XVID_ENC_DESTROY: |
case XVID_ENC_DESTROY: |
716 |
return encoder_destroy((Encoder *) handle); |
return enc_destroy((Encoder *) handle); |
717 |
|
|
718 |
default: |
default: |
719 |
return XVID_ERR_FAIL; |
return XVID_ERR_FAIL; |