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) 2001-2003 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 |
* |
* |
|
****************************************************************************/ |
|
|
|
|
|
/***************************************************************************** |
|
|
* |
|
|
* 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> |
|
|
* |
|
22 |
* $Id$ |
* $Id$ |
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" |
39 |
#include "image/reduced.h" |
#include "image/reduced.h" |
40 |
#include "utils/mem_transfer.h" |
#include "utils/mem_transfer.h" |
41 |
#include "utils/mbfunctions.h" |
#include "utils/mbfunctions.h" |
42 |
#include "quant/quant_h263.h" |
#include "quant/quant.h" |
|
#include "quant/quant_mpeg4.h" |
|
43 |
#include "motion/motion.h" |
#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 |
|
|
50 |
#if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE) |
#if defined(_DEBUG) |
51 |
|
unsigned int xvid_debug = 0; /* xvid debug mask */ |
52 |
|
#endif |
53 |
|
|
54 |
#ifdef WIN32 |
#if defined(ARCH_IS_IA32) |
55 |
|
#if defined(_MSC_VER) |
56 |
#include <windows.h> |
#include <windows.h> |
57 |
#else |
#else |
58 |
#include <signal.h> |
#include <signal.h> |
59 |
#include <setjmp.h> |
#include <setjmp.h> |
|
#endif |
|
|
|
|
|
|
|
|
#ifndef WIN32 |
|
60 |
|
|
61 |
static jmp_buf mark; |
static jmp_buf mark; |
62 |
|
|
69 |
|
|
70 |
|
|
71 |
/* |
/* |
72 |
calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled |
* Calls the funcptr, and returns whether SIGILL (illegal instruction) was |
73 |
return values: |
* signalled |
74 |
-1 : could not determine |
* |
75 |
0 : SIGILL was *not* signalled |
* Return values: |
76 |
1 : SIGILL was signalled |
* -1 : could not determine |
77 |
|
* 0 : SIGILL was *not* signalled |
78 |
|
* 1 : SIGILL was signalled |
79 |
*/ |
*/ |
80 |
|
|
81 |
int |
int |
82 |
sigill_check(void (*func)()) |
sigill_check(void (*func)()) |
83 |
{ |
{ |
84 |
#ifdef WIN32 |
#if defined(_MSC_VER) |
85 |
_try { |
_try { |
86 |
func(); |
func(); |
87 |
} |
} |
115 |
} |
} |
116 |
#endif |
#endif |
117 |
|
|
118 |
|
|
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 |
/***************************************************************************** |
/***************************************************************************** |
146 |
* XviD Init Entry point |
* XviD Init Entry point |
147 |
* |
* |
158 |
|
|
159 |
|
|
160 |
static |
static |
161 |
int xvid_init_init(XVID_INIT_PARAM * init_param) |
int xvid_gbl_init(xvid_gbl_init_t * init) |
162 |
{ |
{ |
163 |
int cpu_flags; |
unsigned int cpu_flags; |
|
|
|
|
/* 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; |
|
164 |
|
|
165 |
/* Do we have to force CPU features ? */ |
if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */ |
166 |
if ((init_param->cpu_flags & XVID_CPU_FORCE)) { |
return XVID_ERR_VERSION; |
|
|
|
|
cpu_flags = init_param->cpu_flags; |
|
|
|
|
|
} else { |
|
|
|
|
|
cpu_flags = check_cpu_features(); |
|
|
|
|
|
#if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE) |
|
|
if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) |
|
|
cpu_flags &= ~XVID_CPU_SSE; |
|
|
|
|
|
if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) |
|
|
cpu_flags &= ~XVID_CPU_SSE2; |
|
|
#endif |
|
|
} |
|
|
|
|
|
if ((init_param->cpu_flags & XVID_CPU_CHKONLY)) |
|
|
{ |
|
|
init_param->cpu_flags = cpu_flags; |
|
|
return XVID_ERR_OK; |
|
|
} |
|
|
|
|
|
init_param->cpu_flags = cpu_flags; |
|
167 |
|
|
168 |
|
cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags(); |
169 |
|
|
170 |
/* Initialize the function pointers */ |
/* Initialize the function pointers */ |
171 |
idct_int32_init(); |
idct_int32_init(); |
181 |
/* Restore FPU context : emms_c is a nop functions */ |
/* Restore FPU context : emms_c is a nop functions */ |
182 |
emms = emms_c; |
emms = emms_c; |
183 |
|
|
184 |
|
/* Qpel stuff */ |
185 |
|
xvid_QP_Funcs = &xvid_QP_Funcs_C; |
186 |
|
xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C; |
187 |
|
xvid_Init_QP(); |
188 |
|
|
189 |
/* Quantization functions */ |
/* Quantization functions */ |
190 |
quant_intra = quant_intra_c; |
quant_h263_intra = quant_h263_intra_c; |
191 |
dequant_intra = dequant_intra_c; |
quant_h263_inter = quant_h263_inter_c; |
192 |
quant_inter = quant_inter_c; |
dequant_h263_intra = dequant_h263_intra_c; |
193 |
dequant_inter = dequant_inter_c; |
dequant_h263_inter = dequant_h263_inter_c; |
194 |
|
|
195 |
quant4_intra = quant4_intra_c; |
quant_mpeg_intra = quant_mpeg_intra_c; |
196 |
dequant4_intra = dequant4_intra_c; |
quant_mpeg_inter = quant_mpeg_inter_c; |
197 |
quant4_inter = quant4_inter_c; |
dequant_mpeg_intra = dequant_mpeg_intra_c; |
198 |
dequant4_inter = dequant4_inter_c; |
dequant_mpeg_inter = dequant_mpeg_inter_c; |
199 |
|
|
200 |
/* Block transfer related functions */ |
/* Block transfer related functions */ |
201 |
transfer_8to16copy = transfer_8to16copy_c; |
transfer_8to16copy = transfer_8to16copy_c; |
202 |
transfer_16to8copy = transfer_16to8copy_c; |
transfer_16to8copy = transfer_16to8copy_c; |
203 |
transfer_8to16sub = transfer_8to16sub_c; |
transfer_8to16sub = transfer_8to16sub_c; |
204 |
|
transfer_8to16subro = transfer_8to16subro_c; |
205 |
transfer_8to16sub2 = transfer_8to16sub2_c; |
transfer_8to16sub2 = transfer_8to16sub2_c; |
206 |
transfer_16to8add = transfer_16to8add_c; |
transfer_16to8add = transfer_16to8add_c; |
207 |
transfer8x8_copy = transfer8x8_copy_c; |
transfer8x8_copy = transfer8x8_copy_c; |
228 |
interpolate8x8_avg2 = interpolate8x8_avg2_c; |
interpolate8x8_avg2 = interpolate8x8_avg2_c; |
229 |
interpolate8x8_avg4 = interpolate8x8_avg4_c; |
interpolate8x8_avg4 = interpolate8x8_avg4_c; |
230 |
|
|
231 |
/* reduced resoltuion */ |
/* reduced resolution */ |
|
|
|
232 |
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
233 |
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; |
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; |
|
#ifdef ARCH_X86 |
|
|
vfilter_31 = xvid_VFilter_31_x86; |
|
|
hfilter_31 = xvid_HFilter_31_x86; |
|
|
#else |
|
234 |
vfilter_31 = xvid_VFilter_31_C; |
vfilter_31 = xvid_VFilter_31_C; |
235 |
hfilter_31 = xvid_HFilter_31_C; |
hfilter_31 = xvid_HFilter_31_C; |
|
#endif |
|
236 |
filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; |
filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; |
237 |
filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; |
filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; |
238 |
|
|
287 |
sad8bi = sad8bi_c; |
sad8bi = sad8bi_c; |
288 |
dev16 = dev16_c; |
dev16 = dev16_c; |
289 |
sad16v = sad16v_c; |
sad16v = sad16v_c; |
290 |
|
sse8_16bit = sse8_16bit_c; |
291 |
|
|
292 |
// Halfpel8_Refine = Halfpel8_Refine_c; |
#if defined(ARCH_IS_IA32) |
293 |
|
|
294 |
#ifdef ARCH_X86 |
if ((cpu_flags & XVID_CPU_ASM)) { |
295 |
|
vfilter_31 = xvid_VFilter_31_x86; |
296 |
|
hfilter_31 = xvid_HFilter_31_x86; |
297 |
|
} |
298 |
|
|
299 |
if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || |
if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || |
300 |
(cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || |
(cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || |
304 |
emms = emms_mmx; |
emms = emms_mmx; |
305 |
} |
} |
306 |
|
|
307 |
if ((cpu_flags & XVID_CPU_MMX) > 0) { |
if ((cpu_flags & XVID_CPU_MMX)) { |
308 |
|
|
309 |
/* Forward and Inverse Discrete Cosine Transformation functions */ |
/* Forward and Inverse Discrete Cosine Transformation functions */ |
310 |
fdct = fdct_mmx; |
fdct = fdct_mmx_skal; |
311 |
idct = idct_mmx; |
idct = idct_mmx; |
312 |
|
|
313 |
|
/* Qpel stuff */ |
314 |
|
xvid_QP_Funcs = &xvid_QP_Funcs_mmx; |
315 |
|
xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx; |
316 |
|
|
317 |
/* Quantization related functions */ |
/* Quantization related functions */ |
318 |
quant_intra = quant_intra_mmx; |
quant_h263_intra = quant_h263_intra_mmx; |
319 |
dequant_intra = dequant_intra_mmx; |
quant_h263_inter = quant_h263_inter_mmx; |
320 |
quant_inter = quant_inter_mmx; |
dequant_h263_intra = dequant_h263_intra_mmx; |
321 |
dequant_inter = dequant_inter_mmx; |
dequant_h263_inter = dequant_h263_inter_mmx; |
322 |
|
|
323 |
quant4_intra = quant4_intra_mmx; |
quant_mpeg_intra = quant_mpeg_intra_mmx; |
324 |
dequant4_intra = dequant4_intra_mmx; |
quant_mpeg_inter = quant_mpeg_inter_mmx; |
325 |
quant4_inter = quant4_inter_mmx; |
dequant_mpeg_intra = dequant_mpeg_intra_mmx; |
326 |
dequant4_inter = dequant4_inter_mmx; |
dequant_mpeg_inter = dequant_mpeg_inter_mmx; |
327 |
|
|
328 |
/* Block related functions */ |
/* Block related functions */ |
329 |
transfer_8to16copy = transfer_8to16copy_mmx; |
transfer_8to16copy = transfer_8to16copy_mmx; |
330 |
transfer_16to8copy = transfer_16to8copy_mmx; |
transfer_16to8copy = transfer_16to8copy_mmx; |
331 |
transfer_8to16sub = transfer_8to16sub_mmx; |
transfer_8to16sub = transfer_8to16sub_mmx; |
332 |
|
transfer_8to16subro = transfer_8to16subro_mmx; |
333 |
transfer_8to16sub2 = transfer_8to16sub2_mmx; |
transfer_8to16sub2 = transfer_8to16sub2_mmx; |
334 |
transfer_16to8add = transfer_16to8add_mmx; |
transfer_16to8add = transfer_16to8add_mmx; |
335 |
transfer8x8_copy = transfer8x8_copy_mmx; |
transfer8x8_copy = transfer8x8_copy_mmx; |
379 |
sad8bi = sad8bi_mmx; |
sad8bi = sad8bi_mmx; |
380 |
dev16 = dev16_mmx; |
dev16 = dev16_mmx; |
381 |
sad16v = sad16v_mmx; |
sad16v = sad16v_mmx; |
382 |
|
sse8_16bit = sse8_16bit_mmx; |
383 |
} |
} |
384 |
|
|
385 |
/* these 3dnow functions are faster than mmx, but slower than xmm. */ |
/* these 3dnow functions are faster than mmx, but slower than xmm. */ |
386 |
if ((cpu_flags & XVID_CPU_3DNOW) > 0) { |
if ((cpu_flags & XVID_CPU_3DNOW)) { |
387 |
|
|
388 |
|
emms = emms_3dn; |
389 |
|
|
390 |
/* ME functions */ |
/* ME functions */ |
391 |
sad16bi = sad16bi_3dn; |
sad16bi = sad16bi_3dn; |
396 |
} |
} |
397 |
|
|
398 |
|
|
399 |
if ((cpu_flags & XVID_CPU_MMXEXT) > 0) { |
if ((cpu_flags & XVID_CPU_MMXEXT)) { |
400 |
|
|
401 |
/* Inverse DCT */ |
/* DCT */ |
402 |
|
fdct = fdct_xmm_skal; |
403 |
idct = idct_xmm; |
idct = idct_xmm; |
404 |
|
|
405 |
/* Interpolation */ |
/* Interpolation */ |
412 |
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm; |
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm; |
413 |
|
|
414 |
/* Quantization */ |
/* Quantization */ |
415 |
quant4_intra = quant4_intra_xmm; |
quant_mpeg_intra = quant_mpeg_intra_xmm; |
416 |
quant4_inter = quant4_inter_xmm; |
quant_mpeg_inter = quant_mpeg_inter_xmm; |
417 |
|
|
418 |
dequant_intra = dequant_intra_xmm; |
dequant_h263_intra = dequant_h263_intra_xmm; |
419 |
dequant_inter = dequant_inter_xmm; |
dequant_h263_inter = dequant_h263_inter_xmm; |
420 |
|
|
421 |
/* Buffer transfer */ |
/* Buffer transfer */ |
422 |
transfer_8to16sub2 = transfer_8to16sub2_xmm; |
transfer_8to16sub2 = transfer_8to16sub2_xmm; |
435 |
sad16v = sad16v_xmm; |
sad16v = sad16v_xmm; |
436 |
} |
} |
437 |
|
|
438 |
if ((cpu_flags & XVID_CPU_3DNOW) > 0) { |
if ((cpu_flags & XVID_CPU_3DNOW)) { |
439 |
|
|
440 |
/* Interpolation */ |
/* Interpolation */ |
441 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
443 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
444 |
} |
} |
445 |
|
|
446 |
if ((cpu_flags & XVID_CPU_3DNOWEXT) > 0) { |
if ((cpu_flags & XVID_CPU_3DNOWEXT)) { |
447 |
|
|
448 |
/* Inverse DCT */ |
/* Inverse DCT */ |
449 |
idct = idct_3dne; |
idct = idct_3dne; |
452 |
transfer_8to16copy = transfer_8to16copy_3dne; |
transfer_8to16copy = transfer_8to16copy_3dne; |
453 |
transfer_16to8copy = transfer_16to8copy_3dne; |
transfer_16to8copy = transfer_16to8copy_3dne; |
454 |
transfer_8to16sub = transfer_8to16sub_3dne; |
transfer_8to16sub = transfer_8to16sub_3dne; |
455 |
|
transfer_8to16subro = transfer_8to16subro_3dne; |
456 |
transfer_8to16sub2 = transfer_8to16sub2_3dne; |
transfer_8to16sub2 = transfer_8to16sub2_3dne; |
457 |
transfer_16to8add = transfer_16to8add_3dne; |
transfer_16to8add = transfer_16to8add_3dne; |
458 |
transfer8x8_copy = transfer8x8_copy_3dne; |
transfer8x8_copy = transfer8x8_copy_3dne; |
459 |
|
|
460 |
/* Quantization */ |
/* Quantization */ |
461 |
dequant4_intra = dequant4_intra_3dne; |
quant_h263_intra = quant_h263_intra_3dne; |
462 |
dequant4_inter = dequant4_inter_3dne; |
quant_h263_inter = quant_h263_inter_3dne; |
463 |
quant_intra = quant_intra_3dne; |
dequant_mpeg_intra = dequant_mpeg_intra_3dne; |
464 |
quant_inter = quant_inter_3dne; |
dequant_mpeg_inter = dequant_mpeg_inter_3dne; |
465 |
dequant_intra = dequant_intra_3dne; |
dequant_h263_intra = dequant_h263_intra_3dne; |
466 |
dequant_inter = dequant_inter_3dne; |
dequant_h263_inter = dequant_h263_inter_3dne; |
467 |
|
|
468 |
/* ME functions */ |
/* ME functions */ |
469 |
calc_cbp = calc_cbp_3dne; |
calc_cbp = calc_cbp_3dne; |
479 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; |
480 |
} |
} |
481 |
|
|
482 |
|
#if defined(EXPERIMENTAL_SSE2_CODE) /* mark the whole SSE2 stuff as experimental. At least on |
483 |
if ((cpu_flags & XVID_CPU_SSE2) > 0) { |
my P4, it crashes... */ |
484 |
#ifdef EXPERIMENTAL_SSE2_CODE |
if ((cpu_flags & XVID_CPU_SSE2)) { |
485 |
|
|
486 |
calc_cbp = calc_cbp_sse2; |
calc_cbp = calc_cbp_sse2; |
487 |
|
|
488 |
/* Quantization */ |
/* Quantization */ |
489 |
quant_intra = quant_intra_sse2; |
quant_h263_intra = quant_h263_intra_sse2; |
490 |
dequant_intra = dequant_intra_sse2; |
quant_h263_inter = quant_h263_inter_sse2; |
491 |
quant_inter = quant_inter_sse2; |
dequant_h263_intra = dequant_h263_intra_sse2; |
492 |
dequant_inter = dequant_inter_sse2; |
dequant_h263_inter = dequant_h263_inter_sse2; |
493 |
|
|
494 |
/* ME */ |
/* SAD operators */ |
495 |
sad16 = sad16_sse2; |
sad16 = sad16_sse2; |
496 |
dev16 = dev16_sse2; |
dev16 = dev16_sse2; |
497 |
|
|
498 |
/* Forward and Inverse DCT */ |
/* DCT operators */ |
499 |
idct = idct_sse2; |
fdct = fdct_sse2_skal; |
500 |
fdct = fdct_sse2; |
idct = idct_sse2_dmitry; |
|
#endif |
|
501 |
} |
} |
502 |
|
#endif |
503 |
#endif |
#endif |
504 |
|
|
505 |
#ifdef ARCH_IA64 |
#if defined(ARCH_IS_IA64) |
506 |
if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines? |
if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */ |
507 |
idct_ia64_init(); |
idct_ia64_init(); |
508 |
fdct = fdct_ia64; |
fdct = fdct_ia64; |
509 |
idct = idct_ia64; //not yet working, crashes |
idct = idct_ia64; /*not yet working, crashes */ |
510 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
511 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; |
512 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; |
514 |
sad16bi = sad16bi_ia64; |
sad16bi = sad16bi_ia64; |
515 |
sad8 = sad8_ia64; |
sad8 = sad8_ia64; |
516 |
dev16 = dev16_ia64; |
dev16 = dev16_ia64; |
517 |
// Halfpel8_Refine = Halfpel8_Refine_ia64; |
/* Halfpel8_Refine = Halfpel8_Refine_ia64; */ |
518 |
quant_intra = quant_intra_ia64; |
quant_h263_intra = quant_h263_intra_ia64; |
519 |
dequant_intra = dequant_intra_ia64; |
quant_h263_inter = quant_h263_inter_ia64; |
520 |
quant_inter = quant_inter_ia64; |
dequant_h263_intra = dequant_h263_intra_ia64; |
521 |
dequant_inter = dequant_inter_ia64; |
dequant_h263_inter = dequant_h263_inter_ia64; |
522 |
transfer_8to16copy = transfer_8to16copy_ia64; |
transfer_8to16copy = transfer_8to16copy_ia64; |
523 |
transfer_16to8copy = transfer_16to8copy_ia64; |
transfer_16to8copy = transfer_16to8copy_ia64; |
524 |
transfer_8to16sub = transfer_8to16sub_ia64; |
transfer_8to16sub = transfer_8to16sub_ia64; |
525 |
transfer_8to16sub2 = transfer_8to16sub2_ia64; |
transfer_8to16sub2 = transfer_8to16sub2_ia64; |
526 |
transfer_16to8add = transfer_16to8add_ia64; |
transfer_16to8add = transfer_16to8add_ia64; |
527 |
transfer8x8_copy = transfer8x8_copy_ia64; |
transfer8x8_copy = transfer8x8_copy_ia64; |
|
DEBUG("Using IA-64 assembler routines.\n"); |
|
528 |
} |
} |
529 |
#endif |
#endif |
530 |
|
|
531 |
#ifdef ARCH_PPC |
#if defined(ARCH_IS_PPC) |
532 |
#ifdef ARCH_PPC_ALTIVEC |
if ((cpu_flags & XVID_CPU_ASM)) |
533 |
|
{ |
534 |
|
calc_cbp = calc_cbp_ppc; |
535 |
|
} |
536 |
|
|
537 |
|
if ((cpu_flags & XVID_CPU_ALTIVEC)) |
538 |
|
{ |
539 |
calc_cbp = calc_cbp_altivec; |
calc_cbp = calc_cbp_altivec; |
540 |
fdct = fdct_altivec; |
fdct = fdct_altivec; |
541 |
idct = idct_altivec; |
idct = idct_altivec; |
543 |
sad16 = sad16_altivec; |
sad16 = sad16_altivec; |
544 |
sad8 = sad8_altivec; |
sad8 = sad8_altivec; |
545 |
dev16 = dev16_altivec; |
dev16 = dev16_altivec; |
546 |
#else |
} |
|
calc_cbp = calc_cbp_ppc; |
|
547 |
#endif |
#endif |
548 |
|
|
549 |
|
#if defined(_DEBUG) |
550 |
|
xvid_debug = init->debug; |
551 |
#endif |
#endif |
552 |
|
|
553 |
return XVID_ERR_OK; |
return 0; |
554 |
} |
} |
555 |
|
|
556 |
|
|
557 |
|
static int |
558 |
|
xvid_gbl_info(xvid_gbl_info_t * info) |
559 |
|
{ |
560 |
|
if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */ |
561 |
|
return XVID_ERR_VERSION; |
562 |
|
|
563 |
|
info->actual_version = XVID_VERSION; |
564 |
|
info->build = "xvid-1.0.0"; |
565 |
|
info->cpu_flags = detect_cpu_flags(); |
566 |
|
|
567 |
|
#if defined(_SMP) && defined(WIN32) |
568 |
|
info->num_threads = pthread_num_processors_np();; |
569 |
|
#else |
570 |
|
info->num_threads = 0; |
571 |
|
#endif |
572 |
|
|
573 |
|
return 0; |
574 |
|
} |
575 |
|
|
576 |
|
|
577 |
static int |
static int |
578 |
xvid_init_convert(XVID_INIT_CONVERTINFO* convert) |
xvid_gbl_convert(xvid_gbl_convert_t* convert) |
579 |
{ |
{ |
580 |
// const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); |
int width; |
581 |
const int width = convert->width; |
int height; |
582 |
const int height = convert->height; |
int width2; |
583 |
const int width2 = convert->width/2; |
int height2; |
|
const int height2 = convert->height/2; |
|
584 |
IMAGE img; |
IMAGE img; |
585 |
|
|
586 |
switch (convert->input.colorspace & ~XVID_CSP_VFLIP) |
if (XVID_VERSION_MAJOR(convert->version) != 1) /* v1.x.x */ |
587 |
|
return XVID_ERR_VERSION; |
588 |
|
|
589 |
|
#if 0 |
590 |
|
const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); |
591 |
|
#endif |
592 |
|
width = convert->width; |
593 |
|
height = convert->height; |
594 |
|
width2 = convert->width/2; |
595 |
|
height2 = convert->height/2; |
596 |
|
|
597 |
|
switch (convert->input.csp & ~XVID_CSP_VFLIP) |
598 |
{ |
{ |
599 |
case XVID_CSP_YV12 : |
case XVID_CSP_YV12 : |
600 |
img.y = convert->input.y; |
img.y = convert->input.plane[0]; |
601 |
img.v = (uint8_t*)convert->input.y + width*height; |
img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height; |
602 |
img.u = (uint8_t*)convert->input.y + width*height + width2*height2; |
img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2; |
603 |
image_output(&img, width, height, width, |
image_output(&img, width, height, width, |
604 |
convert->output.y, convert->output.y_stride, |
(uint8_t**)convert->output.plane, convert->output.stride, |
605 |
convert->output.colorspace, convert->interlacing); |
convert->output.csp, convert->interlacing); |
606 |
break; |
break; |
607 |
|
|
608 |
default : |
default : |
611 |
|
|
612 |
|
|
613 |
emms(); |
emms(); |
614 |
return XVID_ERR_OK; |
return 0; |
615 |
} |
} |
616 |
|
|
617 |
|
/***************************************************************************** |
618 |
|
* XviD Global Entry point |
619 |
|
* |
620 |
|
* Well this function initialize all internal function pointers according |
621 |
|
* to the CPU features forced by the library client or autodetected (depending |
622 |
|
* on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all |
623 |
|
* image colorspace transformation tables. |
624 |
|
* |
625 |
|
****************************************************************************/ |
626 |
|
|
627 |
|
|
628 |
int |
int |
629 |
xvid_init(void *handle, |
xvid_global(void *handle, |
630 |
int opt, |
int opt, |
631 |
void *param1, |
void *param1, |
632 |
void *param2) |
void *param2) |
633 |
{ |
{ |
634 |
switch(opt) |
switch(opt) |
635 |
{ |
{ |
636 |
case XVID_INIT_INIT : |
case XVID_GBL_INIT : |
637 |
return xvid_init_init((XVID_INIT_PARAM*)param1); |
return xvid_gbl_init((xvid_gbl_init_t*)param1); |
638 |
|
|
639 |
|
case XVID_GBL_INFO : |
640 |
|
return xvid_gbl_info((xvid_gbl_info_t*)param1); |
641 |
|
|
642 |
case XVID_INIT_CONVERT : |
case XVID_GBL_CONVERT : |
643 |
return xvid_init_convert((XVID_INIT_CONVERTINFO*)param1); |
return xvid_gbl_convert((xvid_gbl_convert_t*)param1); |
644 |
|
|
645 |
default : |
default : |
646 |
return XVID_ERR_FAIL; |
return XVID_ERR_FAIL; |
664 |
void *param2) |
void *param2) |
665 |
{ |
{ |
666 |
switch (opt) { |
switch (opt) { |
|
case XVID_DEC_DECODE: |
|
|
return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1, (XVID_DEC_STATS*) param2); |
|
|
|
|
667 |
case XVID_DEC_CREATE: |
case XVID_DEC_CREATE: |
668 |
return decoder_create((XVID_DEC_PARAM *) param1); |
return decoder_create((xvid_dec_create_t *) param1); |
669 |
|
|
670 |
case XVID_DEC_DESTROY: |
case XVID_DEC_DESTROY: |
671 |
return decoder_destroy((DECODER *) handle); |
return decoder_destroy((DECODER *) handle); |
672 |
|
|
673 |
|
case XVID_DEC_DECODE: |
674 |
|
return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2); |
675 |
|
|
676 |
default: |
default: |
677 |
return XVID_ERR_FAIL; |
return XVID_ERR_FAIL; |
678 |
} |
} |
698 |
switch (opt) { |
switch (opt) { |
699 |
case XVID_ENC_ENCODE: |
case XVID_ENC_ENCODE: |
700 |
|
|
701 |
if (((Encoder *) handle)->mbParam.max_bframes >= 0) |
return enc_encode((Encoder *) handle, |
702 |
return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
(xvid_enc_frame_t *) param1, |
703 |
(XVID_ENC_STATS *) param2); |
(xvid_enc_stats_t *) param2); |
|
else |
|
|
return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
|
|
(XVID_ENC_STATS *) param2); |
|
704 |
|
|
705 |
case XVID_ENC_CREATE: |
case XVID_ENC_CREATE: |
706 |
return encoder_create((XVID_ENC_PARAM *) param1); |
return enc_create((xvid_enc_create_t *) param1); |
707 |
|
|
708 |
case XVID_ENC_DESTROY: |
case XVID_ENC_DESTROY: |
709 |
return encoder_destroy((Encoder *) handle); |
return enc_destroy((Encoder *) handle); |
710 |
|
|
711 |
default: |
default: |
712 |
return XVID_ERR_FAIL; |
return XVID_ERR_FAIL; |