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 1.53 2004-04-15 12:05:19 suxen_drol Exp $ |
* $Id: xvid.c,v 1.76 2008-11-27 00:47:03 Isibaar Exp $ |
23 |
* |
* |
24 |
****************************************************************************/ |
****************************************************************************/ |
25 |
|
|
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" |
|
#include "image/reduced.h" |
|
39 |
#include "utils/mem_transfer.h" |
#include "utils/mem_transfer.h" |
40 |
#include "utils/mbfunctions.h" |
#include "utils/mbfunctions.h" |
41 |
#include "quant/quant.h" |
#include "quant/quant.h" |
42 |
#include "motion/motion.h" |
#include "motion/motion.h" |
43 |
|
#include "motion/gmc.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" |
52 |
unsigned int xvid_debug = 0; /* xvid debug mask */ |
unsigned int xvid_debug = 0; /* xvid debug mask */ |
53 |
#endif |
#endif |
54 |
|
|
55 |
#if defined(ARCH_IS_IA32) && defined(_MSC_VER) |
#if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER) |
56 |
# include <windows.h> |
# include <windows.h> |
57 |
#elif defined(ARCH_IS_IA32) || defined(ARCH_IS_PPC) |
#elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC) |
58 |
# include <signal.h> |
# include <signal.h> |
59 |
# include <setjmp.h> |
# include <setjmp.h> |
60 |
|
|
77 |
* 0 : SIGILL was *not* signalled |
* 0 : SIGILL was *not* signalled |
78 |
* 1 : SIGILL was signalled |
* 1 : SIGILL was signalled |
79 |
*/ |
*/ |
80 |
#if defined(ARCH_IS_IA32) && defined(_MSC_VER) |
#if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER) |
81 |
static int |
static int |
82 |
sigill_check(void (*func)()) |
sigill_check(void (*func)()) |
83 |
{ |
{ |
90 |
} |
} |
91 |
return(0); |
return(0); |
92 |
} |
} |
93 |
#elif defined(ARCH_IS_IA32) || defined(ARCH_IS_PPC) |
#elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC) |
94 |
static int |
static int |
95 |
sigill_check(void (*func)()) |
sigill_check(void (*func)()) |
96 |
{ |
{ |
127 |
|
|
128 |
/* detect cpu flags */ |
/* detect cpu flags */ |
129 |
static unsigned int |
static unsigned int |
130 |
detect_cpu_flags() |
detect_cpu_flags(void) |
131 |
{ |
{ |
132 |
/* enable native assembly optimizations by default */ |
/* enable native assembly optimizations by default */ |
133 |
unsigned int cpu_flags = XVID_CPU_ASM; |
unsigned int cpu_flags = XVID_CPU_ASM; |
134 |
|
|
135 |
#if defined(ARCH_IS_IA32) |
#if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) |
136 |
cpu_flags |= check_cpu_features(); |
cpu_flags |= check_cpu_features(); |
137 |
if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) |
if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) |
138 |
cpu_flags &= ~XVID_CPU_SSE; |
cpu_flags &= ~XVID_CPU_SSE; |
139 |
|
|
140 |
if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) |
if ((cpu_flags & (XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41)) && sigill_check(sse2_os_trigger)) |
141 |
cpu_flags &= ~XVID_CPU_SSE2; |
cpu_flags &= ~(XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41); |
142 |
#endif |
#endif |
143 |
|
|
144 |
#if defined(ARCH_IS_PPC) |
#if defined(ARCH_IS_PPC) |
184 |
idct = idct_int32; |
idct = idct_int32; |
185 |
|
|
186 |
/* Only needed on PPC Altivec archs */ |
/* Only needed on PPC Altivec archs */ |
187 |
sadInit = 0; |
sadInit = NULL; |
188 |
|
|
189 |
/* Restore FPU context : emms_c is a nop functions */ |
/* Restore FPU context : emms_c is a nop functions */ |
190 |
emms = emms_c; |
emms = emms_c; |
211 |
transfer_8to16sub = transfer_8to16sub_c; |
transfer_8to16sub = transfer_8to16sub_c; |
212 |
transfer_8to16subro = transfer_8to16subro_c; |
transfer_8to16subro = transfer_8to16subro_c; |
213 |
transfer_8to16sub2 = transfer_8to16sub2_c; |
transfer_8to16sub2 = transfer_8to16sub2_c; |
214 |
|
transfer_8to16sub2ro = transfer_8to16sub2ro_c; |
215 |
transfer_16to8add = transfer_16to8add_c; |
transfer_16to8add = transfer_16to8add_c; |
216 |
transfer8x8_copy = transfer8x8_copy_c; |
transfer8x8_copy = transfer8x8_copy_c; |
217 |
|
transfer8x4_copy = transfer8x4_copy_c; |
218 |
|
|
219 |
/* Interlacing functions */ |
/* Interlacing functions */ |
220 |
MBFieldTest = MBFieldTest_c; |
MBFieldTest = MBFieldTest_c; |
224 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; |
225 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
226 |
|
|
227 |
|
interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_c; |
228 |
|
interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_c; |
229 |
|
interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c; |
230 |
|
|
231 |
|
interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c; |
232 |
|
interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c; |
233 |
|
interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c; |
234 |
|
interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c; |
235 |
|
|
236 |
interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; |
interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; |
237 |
interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; |
interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; |
238 |
interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; |
interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; |
250 |
/* postprocessing */ |
/* postprocessing */ |
251 |
image_brightness = image_brightness_c; |
image_brightness = image_brightness_c; |
252 |
|
|
|
/* reduced resolution */ |
|
|
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
|
|
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; |
|
|
vfilter_31 = xvid_VFilter_31_C; |
|
|
hfilter_31 = xvid_HFilter_31_C; |
|
|
filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; |
|
|
filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; |
|
|
|
|
253 |
/* Initialize internal colorspace transformation tables */ |
/* Initialize internal colorspace transformation tables */ |
254 |
colorspace_init(); |
colorspace_init(); |
255 |
|
|
257 |
yv12_to_yv12 = yv12_to_yv12_c; |
yv12_to_yv12 = yv12_to_yv12_c; |
258 |
rgb555_to_yv12 = rgb555_to_yv12_c; |
rgb555_to_yv12 = rgb555_to_yv12_c; |
259 |
rgb565_to_yv12 = rgb565_to_yv12_c; |
rgb565_to_yv12 = rgb565_to_yv12_c; |
260 |
|
rgb_to_yv12 = rgb_to_yv12_c; |
261 |
bgr_to_yv12 = bgr_to_yv12_c; |
bgr_to_yv12 = bgr_to_yv12_c; |
262 |
bgra_to_yv12 = bgra_to_yv12_c; |
bgra_to_yv12 = bgra_to_yv12_c; |
263 |
abgr_to_yv12 = abgr_to_yv12_c; |
abgr_to_yv12 = abgr_to_yv12_c; |
279 |
/* All colorspace transformation functions YV12->User format */ |
/* All colorspace transformation functions YV12->User format */ |
280 |
yv12_to_rgb555 = yv12_to_rgb555_c; |
yv12_to_rgb555 = yv12_to_rgb555_c; |
281 |
yv12_to_rgb565 = yv12_to_rgb565_c; |
yv12_to_rgb565 = yv12_to_rgb565_c; |
282 |
|
yv12_to_rgb = yv12_to_rgb_c; |
283 |
yv12_to_bgr = yv12_to_bgr_c; |
yv12_to_bgr = yv12_to_bgr_c; |
284 |
yv12_to_bgra = yv12_to_bgra_c; |
yv12_to_bgra = yv12_to_bgra_c; |
285 |
yv12_to_abgr = yv12_to_abgr_c; |
yv12_to_abgr = yv12_to_abgr_c; |
309 |
sse8_16bit = sse8_16bit_c; |
sse8_16bit = sse8_16bit_c; |
310 |
sse8_8bit = sse8_8bit_c; |
sse8_8bit = sse8_8bit_c; |
311 |
|
|
312 |
#if defined(ARCH_IS_IA32) |
init_GMC(cpu_flags); |
313 |
|
|
314 |
if ((cpu_flags & XVID_CPU_ASM)) { |
#if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) |
|
vfilter_31 = xvid_VFilter_31_x86; |
|
|
hfilter_31 = xvid_HFilter_31_x86; |
|
|
} |
|
315 |
|
|
316 |
if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || |
if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || |
317 |
(cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || |
(cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || |
318 |
(cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2)) |
(cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2) || |
319 |
|
(cpu_flags & XVID_CPU_SSE3) || (cpu_flags & XVID_CPU_SSE41)) |
320 |
{ |
{ |
321 |
/* Restore FPU context : emms_c is a nop functions */ |
/* Restore FPU context : emms_c is a nop functions */ |
322 |
emms = emms_mmx; |
emms = emms_mmx; |
337 |
quant_h263_inter = quant_h263_inter_mmx; |
quant_h263_inter = quant_h263_inter_mmx; |
338 |
dequant_h263_intra = dequant_h263_intra_mmx; |
dequant_h263_intra = dequant_h263_intra_mmx; |
339 |
dequant_h263_inter = dequant_h263_inter_mmx; |
dequant_h263_inter = dequant_h263_inter_mmx; |
|
|
|
340 |
quant_mpeg_intra = quant_mpeg_intra_mmx; |
quant_mpeg_intra = quant_mpeg_intra_mmx; |
341 |
quant_mpeg_inter = quant_mpeg_inter_mmx; |
quant_mpeg_inter = quant_mpeg_inter_mmx; |
342 |
dequant_mpeg_intra = dequant_mpeg_intra_mmx; |
dequant_mpeg_intra = dequant_mpeg_intra_mmx; |
343 |
dequant_mpeg_inter = dequant_mpeg_inter_mmx; |
dequant_mpeg_inter = dequant_mpeg_inter_mmx; |
344 |
|
|
345 |
|
|
346 |
/* Block related functions */ |
/* Block related functions */ |
347 |
transfer_8to16copy = transfer_8to16copy_mmx; |
transfer_8to16copy = transfer_8to16copy_mmx; |
348 |
transfer_16to8copy = transfer_16to8copy_mmx; |
transfer_16to8copy = transfer_16to8copy_mmx; |
351 |
transfer_8to16sub2 = transfer_8to16sub2_mmx; |
transfer_8to16sub2 = transfer_8to16sub2_mmx; |
352 |
transfer_16to8add = transfer_16to8add_mmx; |
transfer_16to8add = transfer_16to8add_mmx; |
353 |
transfer8x8_copy = transfer8x8_copy_mmx; |
transfer8x8_copy = transfer8x8_copy_mmx; |
354 |
|
transfer8x4_copy = transfer8x4_copy_mmx; |
355 |
|
|
356 |
/* Interlacing Functions */ |
/* Interlacing Functions */ |
357 |
MBFieldTest = MBFieldTest_mmx; |
MBFieldTest = MBFieldTest_mmx; |
361 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; |
362 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
363 |
|
|
364 |
|
interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_mmx; |
365 |
|
interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_mmx; |
366 |
|
interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx; |
367 |
|
|
368 |
|
interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx; |
369 |
|
interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx; |
370 |
|
interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx; |
371 |
|
interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx; |
372 |
|
|
373 |
interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; |
interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; |
374 |
interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; |
interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; |
375 |
|
|
379 |
/* postprocessing */ |
/* postprocessing */ |
380 |
image_brightness = image_brightness_mmx; |
image_brightness = image_brightness_mmx; |
381 |
|
|
|
/* reduced resolution */ |
|
|
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx; |
|
|
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx; |
|
|
hfilter_31 = xvid_HFilter_31_mmx; |
|
|
filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx; |
|
|
filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx; |
|
|
|
|
382 |
/* image input xxx_to_yv12 related functions */ |
/* image input xxx_to_yv12 related functions */ |
383 |
|
|
384 |
yv12_to_yv12 = yv12_to_yv12_mmx; |
yv12_to_yv12 = yv12_to_yv12_mmx; |
385 |
|
|
386 |
bgr_to_yv12 = bgr_to_yv12_mmx; |
bgr_to_yv12 = bgr_to_yv12_mmx; |
387 |
|
rgb_to_yv12 = rgb_to_yv12_mmx; |
388 |
bgra_to_yv12 = bgra_to_yv12_mmx; |
bgra_to_yv12 = bgra_to_yv12_mmx; |
389 |
|
rgba_to_yv12 = rgba_to_yv12_mmx; |
390 |
yuyv_to_yv12 = yuyv_to_yv12_mmx; |
yuyv_to_yv12 = yuyv_to_yv12_mmx; |
391 |
uyvy_to_yv12 = uyvy_to_yv12_mmx; |
uyvy_to_yv12 = uyvy_to_yv12_mmx; |
392 |
|
|
422 |
|
|
423 |
yuyv_to_yv12 = yuyv_to_yv12_3dn; |
yuyv_to_yv12 = yuyv_to_yv12_3dn; |
424 |
uyvy_to_yv12 = uyvy_to_yv12_3dn; |
uyvy_to_yv12 = uyvy_to_yv12_3dn; |
425 |
|
|
426 |
} |
} |
427 |
|
|
428 |
|
|
437 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; |
438 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
439 |
|
|
440 |
/* reduced resolution */ |
interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_xmm; |
441 |
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm; |
interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_xmm; |
442 |
add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm; |
interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm; |
443 |
|
|
444 |
|
interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm; |
445 |
|
interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm; |
446 |
|
interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm; |
447 |
|
interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm; |
448 |
|
|
449 |
/* Quantization */ |
/* Quantization */ |
|
quant_mpeg_intra = quant_mpeg_intra_xmm; |
|
450 |
quant_mpeg_inter = quant_mpeg_inter_xmm; |
quant_mpeg_inter = quant_mpeg_inter_xmm; |
451 |
|
|
452 |
dequant_h263_intra = dequant_h263_intra_xmm; |
dequant_h263_intra = dequant_h263_intra_xmm; |
454 |
|
|
455 |
/* Buffer transfer */ |
/* Buffer transfer */ |
456 |
transfer_8to16sub2 = transfer_8to16sub2_xmm; |
transfer_8to16sub2 = transfer_8to16sub2_xmm; |
457 |
|
transfer_8to16sub2ro = transfer_8to16sub2ro_xmm; |
458 |
|
|
459 |
/* Colorspace transformation */ |
/* Colorspace transformation */ |
460 |
yv12_to_yv12 = yv12_to_yv12_xmm; |
/* yv12_to_yv12 = yv12_to_yv12_xmm; */ /* appears to be slow on many machines */ |
461 |
yuyv_to_yv12 = yuyv_to_yv12_xmm; |
yuyv_to_yv12 = yuyv_to_yv12_xmm; |
462 |
uyvy_to_yv12 = uyvy_to_yv12_xmm; |
uyvy_to_yv12 = uyvy_to_yv12_xmm; |
463 |
|
|
476 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
477 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn; |
478 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
479 |
|
|
480 |
|
interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn; |
481 |
|
interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn; |
482 |
|
interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn; |
483 |
} |
} |
484 |
|
|
485 |
if ((cpu_flags & XVID_CPU_3DNOWEXT)) { |
if ((cpu_flags & XVID_CPU_3DNOWEXT)) { |
491 |
transfer_8to16subro = transfer_8to16subro_3dne; |
transfer_8to16subro = transfer_8to16subro_3dne; |
492 |
transfer_16to8add = transfer_16to8add_3dne; |
transfer_16to8add = transfer_16to8add_3dne; |
493 |
transfer8x8_copy = transfer8x8_copy_3dne; |
transfer8x8_copy = transfer8x8_copy_3dne; |
494 |
|
transfer8x4_copy = transfer8x4_copy_3dne; |
495 |
|
|
496 |
if ((cpu_flags & XVID_CPU_MMXEXT)) { |
if ((cpu_flags & XVID_CPU_MMXEXT)) { |
497 |
/* Inverse DCT */ |
/* Inverse DCT */ |
505 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne; |
506 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; |
507 |
|
|
508 |
|
interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne; |
509 |
|
interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne; |
510 |
|
interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne; |
511 |
|
|
512 |
/* Quantization */ |
/* Quantization */ |
513 |
quant_h263_intra = quant_h263_intra_3dne; /* cmov only */ |
quant_h263_intra = quant_h263_intra_3dne; /* cmov only */ |
514 |
quant_h263_inter = quant_h263_inter_3dne; |
quant_h263_inter = quant_h263_inter_3dne; |
518 |
dequant_h263_inter = dequant_h263_inter_3dne; |
dequant_h263_inter = dequant_h263_inter_3dne; |
519 |
|
|
520 |
/* ME functions */ |
/* ME functions */ |
|
calc_cbp = calc_cbp_3dne; |
|
|
|
|
521 |
sad16 = sad16_3dne; |
sad16 = sad16_3dne; |
522 |
sad8 = sad8_3dne; |
sad8 = sad8_3dne; |
523 |
sad16bi = sad16bi_3dne; |
sad16bi = sad16bi_3dne; |
540 |
sad16 = sad16_sse2; |
sad16 = sad16_sse2; |
541 |
dev16 = dev16_sse2; |
dev16 = dev16_sse2; |
542 |
|
|
543 |
/* DCT operators |
/* DCT operators */ |
|
* no iDCT because it's not "Walken matching" */ |
|
544 |
fdct = fdct_sse2_skal; |
fdct = fdct_sse2_skal; |
545 |
|
idct = idct_sse2_skal; /* Is now IEEE1180 and Walken compliant. */ |
546 |
|
|
547 |
|
/* postprocessing */ |
548 |
|
image_brightness = image_brightness_sse2; |
549 |
|
|
550 |
} |
} |
551 |
|
|
552 |
|
if ((cpu_flags & XVID_CPU_SSE3)) { |
553 |
|
|
554 |
|
/* SAD operators */ |
555 |
|
sad16 = sad16_sse3; |
556 |
|
dev16 = dev16_sse3; |
557 |
|
} |
558 |
|
|
559 |
#endif /* ARCH_IS_IA32 */ |
#endif /* ARCH_IS_IA32 */ |
560 |
|
|
561 |
#if defined(ARCH_IS_IA64) |
#if defined(ARCH_IS_IA64) |
614 |
interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c; |
interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c; |
615 |
interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c; |
interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c; |
616 |
|
|
617 |
interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_altivec_c; |
interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c; |
618 |
|
interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c; |
619 |
|
interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c; |
620 |
|
interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c; |
621 |
|
|
622 |
/* Colorspace conversion */ |
/* Colorspace conversion */ |
623 |
bgra_to_yv12 = bgra_to_yv12_altivec_c; |
bgra_to_yv12 = bgra_to_yv12_altivec_c; |
636 |
quant_h263_inter = quant_h263_inter_altivec_c; |
quant_h263_inter = quant_h263_inter_altivec_c; |
637 |
dequant_h263_intra = dequant_h263_intra_altivec_c; |
dequant_h263_intra = dequant_h263_intra_altivec_c; |
638 |
dequant_h263_inter = dequant_h263_inter_altivec_c; |
dequant_h263_inter = dequant_h263_inter_altivec_c; |
639 |
|
|
640 |
|
dequant_mpeg_intra = dequant_mpeg_intra_altivec_c; |
641 |
|
dequant_mpeg_inter = dequant_mpeg_inter_altivec_c; |
642 |
|
|
643 |
|
/* Qpel stuff */ |
644 |
|
xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C; |
645 |
|
xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C; |
646 |
} |
} |
647 |
#endif |
#endif |
648 |
|
|
661 |
return XVID_ERR_VERSION; |
return XVID_ERR_VERSION; |
662 |
|
|
663 |
info->actual_version = XVID_VERSION; |
info->actual_version = XVID_VERSION; |
664 |
info->build = "xvid-1.1-cvshead"; |
info->build = "xvid-1.2.0-dev"; |
665 |
info->cpu_flags = detect_cpu_flags(); |
info->cpu_flags = detect_cpu_flags(); |
666 |
|
info->num_threads = 0; |
667 |
|
|
668 |
#if defined(_SMP) && defined(WIN32) |
#if defined(_WIN32) |
669 |
info->num_threads = pthread_num_processors_np();; |
{ |
670 |
|
DWORD dwProcessAffinityMask, dwSystemAffinityMask; |
671 |
|
if (GetProcessAffinityMask(GetCurrentProcess(), (PDWORD_PTR) &dwProcessAffinityMask, (PDWORD_PTR) &dwSystemAffinityMask)) { |
672 |
|
int i; |
673 |
|
for(i=0; i<32; i++) { |
674 |
|
if ((dwProcessAffinityMask & (1<<i))) |
675 |
|
info->num_threads++; |
676 |
|
} |
677 |
|
} |
678 |
|
} |
679 |
#else |
#else |
680 |
info->num_threads = 0; |
|
681 |
|
#include <unistd.h> |
682 |
|
info->num_threads = sysconf(_SC_NPROCESSORS_CONF); |
683 |
|
|
684 |
#endif |
#endif |
685 |
|
|
686 |
return 0; |
return 0; |