3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* - Native API implementation - |
* - Native API implementation - |
5 |
* |
* |
|
* This program is an implementation of a part of one or more MPEG-4 |
|
|
* 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. |
|
|
* |
|
6 |
* This program is free software ; you can redistribute it and/or modify |
* This program is free software ; you can redistribute it and/or modify |
7 |
* 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 |
8 |
* the Free Software Foundation ; either version 2 of the License, or |
* the Free Software Foundation ; either version 2 of the License, or |
17 |
* along with this program ; if not, write to the Free Software |
* along with this program ; if not, write to the Free Software |
18 |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
19 |
* |
* |
|
****************************************************************************/ |
|
|
|
|
|
/***************************************************************************** |
|
|
* |
|
|
* 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> |
|
|
* |
|
20 |
* $Id$ |
* $Id$ |
21 |
* |
* |
22 |
****************************************************************************/ |
****************************************************************************/ |
44 |
#include "utils/emms.h" |
#include "utils/emms.h" |
45 |
#include "utils/timer.h" |
#include "utils/timer.h" |
46 |
#include "bitstream/mbcoding.h" |
#include "bitstream/mbcoding.h" |
47 |
|
#include "image/qpel.h" |
48 |
|
|
49 |
#if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE) |
#if defined(ARCH_IS_IA32) |
50 |
|
|
51 |
#ifdef WIN32 |
#if defined(_MSC_VER) |
52 |
#include <windows.h> |
#include <windows.h> |
53 |
#else |
#else |
54 |
#include <signal.h> |
#include <signal.h> |
55 |
#include <setjmp.h> |
#include <setjmp.h> |
|
#endif |
|
|
|
|
|
|
|
|
#ifndef WIN32 |
|
56 |
|
|
57 |
static jmp_buf mark; |
static jmp_buf mark; |
58 |
|
|
65 |
|
|
66 |
|
|
67 |
/* |
/* |
68 |
calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled |
* Calls the funcptr, and returns whether SIGILL (illegal instruction) was |
69 |
return values: |
* signalled |
70 |
-1 : could not determine |
* |
71 |
0 : SIGILL was *not* signalled |
* Return values: |
72 |
1 : SIGILL was signalled |
* -1 : could not determine |
73 |
|
* 0 : SIGILL was *not* signalled |
74 |
|
* 1 : SIGILL was signalled |
75 |
*/ |
*/ |
76 |
|
|
77 |
int |
int |
78 |
sigill_check(void (*func)()) |
sigill_check(void (*func)()) |
79 |
{ |
{ |
80 |
#ifdef WIN32 |
#if defined(_MSC_VER) |
81 |
_try { |
_try { |
82 |
func(); |
func(); |
83 |
} |
} |
111 |
} |
} |
112 |
#endif |
#endif |
113 |
|
|
114 |
|
|
115 |
|
/* detect cpu flags */ |
116 |
|
static unsigned int |
117 |
|
detect_cpu_flags() |
118 |
|
{ |
119 |
|
/* enable native assembly optimizations by default */ |
120 |
|
unsigned int cpu_flags = XVID_CPU_ASM; |
121 |
|
|
122 |
|
#if defined(ARCH_IS_IA32) |
123 |
|
cpu_flags |= check_cpu_features(); |
124 |
|
if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) |
125 |
|
cpu_flags &= ~XVID_CPU_SSE; |
126 |
|
|
127 |
|
if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) |
128 |
|
cpu_flags &= ~XVID_CPU_SSE2; |
129 |
|
#endif |
130 |
|
|
131 |
|
#if defined(ARCH_IS_PPC) |
132 |
|
#if defined(ARCH_IS_PPC_ALTIVEC) |
133 |
|
cpu_flags |= XVID_CPU_ALTIVEC; |
134 |
|
#endif |
135 |
|
#endif |
136 |
|
|
137 |
|
return cpu_flags; |
138 |
|
} |
139 |
|
|
140 |
|
|
141 |
/***************************************************************************** |
/***************************************************************************** |
142 |
* XviD Init Entry point |
* XviD Init Entry point |
143 |
* |
* |
171 |
|
|
172 |
} else { |
} else { |
173 |
|
|
174 |
cpu_flags = check_cpu_features(); |
cpu_flags = detect_cpu_flags(); |
|
|
|
|
#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 |
|
175 |
} |
} |
176 |
|
|
177 |
if ((init_param->cpu_flags & XVID_CPU_CHKONLY)) |
if ((init_param->cpu_flags & XVID_CPU_CHKONLY)) |
182 |
|
|
183 |
init_param->cpu_flags = cpu_flags; |
init_param->cpu_flags = cpu_flags; |
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_mmx(); |
189 |
|
|
190 |
/* Initialize the function pointers */ |
/* Initialize the function pointers */ |
191 |
idct_int32_init(); |
idct_int32_init(); |
216 |
transfer_8to16copy = transfer_8to16copy_c; |
transfer_8to16copy = transfer_8to16copy_c; |
217 |
transfer_16to8copy = transfer_16to8copy_c; |
transfer_16to8copy = transfer_16to8copy_c; |
218 |
transfer_8to16sub = transfer_8to16sub_c; |
transfer_8to16sub = transfer_8to16sub_c; |
219 |
|
transfer_8to16subro = transfer_8to16subro_c; |
220 |
transfer_8to16sub2 = transfer_8to16sub2_c; |
transfer_8to16sub2 = transfer_8to16sub2_c; |
221 |
transfer_16to8add = transfer_16to8add_c; |
transfer_16to8add = transfer_16to8add_c; |
222 |
transfer8x8_copy = transfer8x8_copy_c; |
transfer8x8_copy = transfer8x8_copy_c; |
244 |
interpolate8x8_avg4 = interpolate8x8_avg4_c; |
interpolate8x8_avg4 = interpolate8x8_avg4_c; |
245 |
|
|
246 |
/* reduced resoltuion */ |
/* reduced resoltuion */ |
|
|
|
247 |
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
248 |
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 |
|
249 |
vfilter_31 = xvid_VFilter_31_C; |
vfilter_31 = xvid_VFilter_31_C; |
250 |
hfilter_31 = xvid_HFilter_31_C; |
hfilter_31 = xvid_HFilter_31_C; |
|
#endif |
|
251 |
filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; |
filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; |
252 |
filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; |
filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; |
253 |
|
|
303 |
dev16 = dev16_c; |
dev16 = dev16_c; |
304 |
sad16v = sad16v_c; |
sad16v = sad16v_c; |
305 |
|
|
306 |
// Halfpel8_Refine = Halfpel8_Refine_c; |
/* Halfpel8_Refine = Halfpel8_Refine_c; */ |
307 |
|
|
308 |
#ifdef ARCH_X86 |
#if defined(ARCH_IS_IA32) |
309 |
|
|
310 |
|
if ((cpu_flags & XVID_CPU_ASM)) |
311 |
|
{ |
312 |
|
vfilter_31 = xvid_VFilter_31_x86; |
313 |
|
hfilter_31 = xvid_HFilter_31_x86; |
314 |
|
} |
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) || |
321 |
emms = emms_mmx; |
emms = emms_mmx; |
322 |
} |
} |
323 |
|
|
324 |
if ((cpu_flags & XVID_CPU_MMX) > 0) { |
if ((cpu_flags & XVID_CPU_MMX)) { |
325 |
|
|
326 |
|
/* Qpel stuff */ |
327 |
|
xvid_QP_Funcs = &xvid_QP_Funcs_mmx; |
328 |
|
xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx; |
329 |
|
xvid_Init_QP_mmx(); |
330 |
|
|
331 |
/* Forward and Inverse Discrete Cosine Transformation functions */ |
/* Forward and Inverse Discrete Cosine Transformation functions */ |
332 |
fdct = fdct_mmx; |
fdct = fdct_mmx; |
347 |
transfer_8to16copy = transfer_8to16copy_mmx; |
transfer_8to16copy = transfer_8to16copy_mmx; |
348 |
transfer_16to8copy = transfer_16to8copy_mmx; |
transfer_16to8copy = transfer_16to8copy_mmx; |
349 |
transfer_8to16sub = transfer_8to16sub_mmx; |
transfer_8to16sub = transfer_8to16sub_mmx; |
350 |
|
transfer_8to16subro = transfer_8to16subro_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; |
397 |
sad8bi = sad8bi_mmx; |
sad8bi = sad8bi_mmx; |
398 |
dev16 = dev16_mmx; |
dev16 = dev16_mmx; |
399 |
sad16v = sad16v_mmx; |
sad16v = sad16v_mmx; |
|
|
|
400 |
} |
} |
401 |
|
|
402 |
/* these 3dnow functions are faster than mmx, but slower than xmm. */ |
/* these 3dnow functions are faster than mmx, but slower than xmm. */ |
403 |
if ((cpu_flags & XVID_CPU_3DNOW) > 0) { |
if ((cpu_flags & XVID_CPU_3DNOW)) { |
404 |
|
|
405 |
|
emms = emms_3dn; |
406 |
|
|
407 |
/* ME functions */ |
/* ME functions */ |
408 |
sad16bi = sad16bi_3dn; |
sad16bi = sad16bi_3dn; |
413 |
} |
} |
414 |
|
|
415 |
|
|
416 |
if ((cpu_flags & XVID_CPU_MMXEXT) > 0) { |
if ((cpu_flags & XVID_CPU_MMXEXT)) { |
417 |
|
|
418 |
/* Inverse DCT */ |
/* Inverse DCT */ |
419 |
idct = idct_xmm; |
idct = idct_xmm; |
451 |
sad16v = sad16v_xmm; |
sad16v = sad16v_xmm; |
452 |
} |
} |
453 |
|
|
454 |
if ((cpu_flags & XVID_CPU_3DNOW) > 0) { |
if ((cpu_flags & XVID_CPU_3DNOW)) { |
455 |
|
|
456 |
/* Interpolation */ |
/* Interpolation */ |
457 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
459 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
460 |
} |
} |
461 |
|
|
462 |
if ((cpu_flags & XVID_CPU_3DNOWEXT) > 0) { |
if ((cpu_flags & XVID_CPU_3DNOWEXT)) { |
463 |
|
|
464 |
/* Inverse DCT */ |
/* Inverse DCT */ |
465 |
idct = idct_3dne; |
idct = idct_3dne; |
468 |
transfer_8to16copy = transfer_8to16copy_3dne; |
transfer_8to16copy = transfer_8to16copy_3dne; |
469 |
transfer_16to8copy = transfer_16to8copy_3dne; |
transfer_16to8copy = transfer_16to8copy_3dne; |
470 |
transfer_8to16sub = transfer_8to16sub_3dne; |
transfer_8to16sub = transfer_8to16sub_3dne; |
471 |
|
transfer_8to16subro = transfer_8to16subro_3dne; |
472 |
transfer_8to16sub2 = transfer_8to16sub2_3dne; |
transfer_8to16sub2 = transfer_8to16sub2_3dne; |
473 |
transfer_16to8add = transfer_16to8add_3dne; |
transfer_16to8add = transfer_16to8add_3dne; |
474 |
transfer8x8_copy = transfer8x8_copy_3dne; |
transfer8x8_copy = transfer8x8_copy_3dne; |
496 |
} |
} |
497 |
|
|
498 |
|
|
499 |
if ((cpu_flags & XVID_CPU_SSE2) > 0) { |
if ((cpu_flags & XVID_CPU_SSE2)) { |
|
#ifdef EXPERIMENTAL_SSE2_CODE |
|
500 |
|
|
501 |
calc_cbp = calc_cbp_sse2; |
calc_cbp = calc_cbp_sse2; |
502 |
|
|
506 |
quant_inter = quant_inter_sse2; |
quant_inter = quant_inter_sse2; |
507 |
dequant_inter = dequant_inter_sse2; |
dequant_inter = dequant_inter_sse2; |
508 |
|
|
509 |
/* ME */ |
#if defined(EXPERIMENTAL_SSE2_CODE) |
510 |
|
/* ME; slower than xmm */ |
511 |
sad16 = sad16_sse2; |
sad16 = sad16_sse2; |
512 |
dev16 = dev16_sse2; |
dev16 = dev16_sse2; |
513 |
|
#endif |
514 |
/* Forward and Inverse DCT */ |
/* Forward and Inverse DCT */ |
515 |
idct = idct_sse2; |
idct = idct_sse2; |
516 |
fdct = fdct_sse2; |
fdct = fdct_sse2; |
|
#endif |
|
517 |
} |
} |
|
|
|
518 |
#endif |
#endif |
519 |
|
|
520 |
#ifdef ARCH_IA64 |
#if defined(ARCH_IS_IA64) |
521 |
if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines? |
if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */ |
522 |
idct_ia64_init(); |
idct_ia64_init(); |
523 |
fdct = fdct_ia64; |
fdct = fdct_ia64; |
524 |
idct = idct_ia64; //not yet working, crashes |
idct = idct_ia64; /*not yet working, crashes */ |
525 |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
526 |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; |
interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; |
527 |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; |
interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; |
529 |
sad16bi = sad16bi_ia64; |
sad16bi = sad16bi_ia64; |
530 |
sad8 = sad8_ia64; |
sad8 = sad8_ia64; |
531 |
dev16 = dev16_ia64; |
dev16 = dev16_ia64; |
532 |
// Halfpel8_Refine = Halfpel8_Refine_ia64; |
/* Halfpel8_Refine = Halfpel8_Refine_ia64; */ |
533 |
quant_intra = quant_intra_ia64; |
quant_intra = quant_intra_ia64; |
534 |
dequant_intra = dequant_intra_ia64; |
dequant_intra = dequant_intra_ia64; |
535 |
quant_inter = quant_inter_ia64; |
quant_inter = quant_inter_ia64; |
540 |
transfer_8to16sub2 = transfer_8to16sub2_ia64; |
transfer_8to16sub2 = transfer_8to16sub2_ia64; |
541 |
transfer_16to8add = transfer_16to8add_ia64; |
transfer_16to8add = transfer_16to8add_ia64; |
542 |
transfer8x8_copy = transfer8x8_copy_ia64; |
transfer8x8_copy = transfer8x8_copy_ia64; |
543 |
DEBUG("Using IA-64 assembler routines.\n"); |
DPRINTF(DPRINTF_DEBUG, "Using IA-64 assembler routines."); |
544 |
} |
} |
545 |
#endif |
#endif |
546 |
|
|
547 |
#ifdef ARCH_PPC |
#if defined(ARCH_IS_PPC) |
548 |
#ifdef ARCH_PPC_ALTIVEC |
if ((cpu_flags & XVID_CPU_ASM)) |
549 |
|
{ |
550 |
|
calc_cbp = calc_cbp_ppc; |
551 |
|
} |
552 |
|
|
553 |
|
if ((cpu_flags & XVID_CPU_ALTIVEC)) |
554 |
|
{ |
555 |
calc_cbp = calc_cbp_altivec; |
calc_cbp = calc_cbp_altivec; |
556 |
fdct = fdct_altivec; |
fdct = fdct_altivec; |
557 |
idct = idct_altivec; |
idct = idct_altivec; |
559 |
sad16 = sad16_altivec; |
sad16 = sad16_altivec; |
560 |
sad8 = sad8_altivec; |
sad8 = sad8_altivec; |
561 |
dev16 = dev16_altivec; |
dev16 = dev16_altivec; |
562 |
#else |
} |
|
calc_cbp = calc_cbp_ppc; |
|
|
#endif |
|
563 |
#endif |
#endif |
564 |
|
|
565 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
570 |
static int |
static int |
571 |
xvid_init_convert(XVID_INIT_CONVERTINFO* convert) |
xvid_init_convert(XVID_INIT_CONVERTINFO* convert) |
572 |
{ |
{ |
573 |
// const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); |
/* |
574 |
|
const int flip1 = |
575 |
|
(convert->input.colorspace & XVID_CSP_VFLIP) ^ |
576 |
|
(convert->output.colorspace & XVID_CSP_VFLIP); |
577 |
|
*/ |
578 |
const int width = convert->width; |
const int width = convert->width; |
579 |
const int height = convert->height; |
const int height = convert->height; |
580 |
const int width2 = convert->width/2; |
const int width2 = convert->width/2; |
660 |
#define TEST_FDCT (TEST_FORWARD) |
#define TEST_FDCT (TEST_FORWARD) |
661 |
#define TEST_IDCT (0) |
#define TEST_IDCT (0) |
662 |
|
|
663 |
int test_transform(void * funcA, void * funcB, const char * nameB, |
static int test_transform(void * funcA, void * funcB, const char * nameB, |
664 |
int test, int flags) |
int test, int flags) |
665 |
{ |
{ |
666 |
int i; |
int i; |
740 |
#define TEST_DEQUANT_INTRA (TEST_INTRA) |
#define TEST_DEQUANT_INTRA (TEST_INTRA) |
741 |
#define TEST_DEQUANT_INTER (0) |
#define TEST_DEQUANT_INTER (0) |
742 |
|
|
743 |
int test_quant(void * funcA, void * funcB, const char * nameB, |
static int test_quant(void * funcA, void * funcB, const char * nameB, |
744 |
int test, int flags) |
int test, int flags) |
745 |
{ |
{ |
746 |
int q,i; |
int q,i; |
747 |
int64_t timeSTART; |
int64_t timeSTART; |
748 |
int64_t timeA = 0; |
int64_t timeA = 0; |
749 |
int64_t timeB = 0; |
int64_t timeB = 0; |
750 |
int retA, retB; |
int retA = 0, retB = 0; |
751 |
DECLARE_ALIGNED_MATRIX(arrayX, 1, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(arrayX, 1, 64, int16_t, CACHE_LINE); |
752 |
DECLARE_ALIGNED_MATRIX(arrayA, 1, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(arrayA, 1, 64, int16_t, CACHE_LINE); |
753 |
DECLARE_ALIGNED_MATRIX(arrayB, 1, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(arrayB, 1, 64, int16_t, CACHE_LINE); |
821 |
|
|
822 |
int xvid_init_test(int flags) |
int xvid_init_test(int flags) |
823 |
{ |
{ |
824 |
|
#if defined(ARCH_IS_IA32) |
825 |
int cpu_flags; |
int cpu_flags; |
826 |
|
#endif |
827 |
|
|
828 |
srand(time(0)); |
printf("XviD tests\n\n"); |
829 |
|
|
830 |
printf("xvid_init_test\n"); |
#if defined(ARCH_IS_IA32) |
831 |
|
cpu_flags = detect_cpu_flags(); |
832 |
|
#endif |
833 |
|
|
|
#if defined(ARCH_X86) |
|
|
cpu_flags = check_cpu_features(); |
|
834 |
idct_int32_init(); |
idct_int32_init(); |
835 |
emms_mmx(); |
emms(); |
836 |
|
|
837 |
|
srand(time(0)); |
838 |
|
|
839 |
|
/* fDCT test */ |
840 |
printf("--- fdct ---\n"); |
printf("--- fdct ---\n"); |
841 |
test_transform(fdct_int32, fdct_int32, "c", TEST_FDCT, flags); |
test_transform(fdct_int32, fdct_int32, "c", TEST_FDCT, flags); |
842 |
|
|
843 |
|
#if defined(ARCH_IS_IA32) |
844 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
845 |
test_transform(fdct_int32, fdct_mmx, "mmx", TEST_FDCT, flags); |
test_transform(fdct_int32, fdct_mmx, "mmx", TEST_FDCT, flags); |
846 |
if (cpu_flags & XVID_CPU_SSE2) |
if (cpu_flags & XVID_CPU_SSE2) |
847 |
test_transform(fdct_int32, fdct_sse2, "sse2", TEST_FDCT, flags); |
test_transform(fdct_int32, fdct_sse2, "sse2", TEST_FDCT, flags); |
848 |
|
#endif |
849 |
|
|
850 |
|
/* iDCT test */ |
851 |
printf("\n--- idct ---\n"); |
printf("\n--- idct ---\n"); |
852 |
test_transform(idct_int32, idct_int32, "c", TEST_IDCT, flags); |
test_transform(idct_int32, idct_int32, "c", TEST_IDCT, flags); |
853 |
|
|
854 |
|
#if defined(ARCH_IS_IA32) |
855 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
856 |
test_transform(idct_int32, idct_mmx, "mmx", TEST_IDCT, flags); |
test_transform(idct_int32, idct_mmx, "mmx", TEST_IDCT, flags); |
857 |
if (cpu_flags & XVID_CPU_MMXEXT) |
if (cpu_flags & XVID_CPU_MMXEXT) |
860 |
test_transform(idct_int32, idct_3dne, "3dne", TEST_IDCT, flags); |
test_transform(idct_int32, idct_3dne, "3dne", TEST_IDCT, flags); |
861 |
if (cpu_flags & XVID_CPU_SSE2) |
if (cpu_flags & XVID_CPU_SSE2) |
862 |
test_transform(idct_int32, idct_sse2, "sse2", TEST_IDCT, flags); |
test_transform(idct_int32, idct_sse2, "sse2", TEST_IDCT, flags); |
863 |
|
#endif |
864 |
|
|
865 |
|
/* Intra quantization test */ |
866 |
printf("\n--- quant intra ---\n"); |
printf("\n--- quant intra ---\n"); |
867 |
test_quant(quant_intra_c, quant_intra_c, "c", TEST_QUANT_INTRA, flags); |
test_quant(quant_intra_c, quant_intra_c, "c", TEST_QUANT_INTRA, flags); |
868 |
|
|
869 |
|
#if defined(ARCH_IS_IA32) |
870 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
871 |
test_quant(quant_intra_c, quant_intra_mmx, "mmx", TEST_QUANT_INTRA, flags); |
test_quant(quant_intra_c, quant_intra_mmx, "mmx", TEST_QUANT_INTRA, flags); |
872 |
if (cpu_flags & XVID_CPU_3DNOWEXT) |
if (cpu_flags & XVID_CPU_3DNOWEXT) |
873 |
test_quant(quant_intra_c, quant_intra_3dne, "3dne", TEST_QUANT_INTRA, flags); |
test_quant(quant_intra_c, quant_intra_3dne, "3dne", TEST_QUANT_INTRA, flags); |
874 |
if (cpu_flags & XVID_CPU_SSE2) |
if (cpu_flags & XVID_CPU_SSE2) |
875 |
test_quant(quant_intra_c, quant_intra_sse2, "sse2", TEST_QUANT_INTRA, flags); |
test_quant(quant_intra_c, quant_intra_sse2, "sse2", TEST_QUANT_INTRA, flags); |
876 |
|
#endif |
877 |
|
|
878 |
|
/* Inter quantization test */ |
879 |
printf("\n--- quant inter ---\n"); |
printf("\n--- quant inter ---\n"); |
880 |
test_quant(quant_inter_c, quant_inter_c, "c", TEST_QUANT_INTER, flags); |
test_quant(quant_inter_c, quant_inter_c, "c", TEST_QUANT_INTER, flags); |
881 |
|
|
882 |
|
#if defined(ARCH_IS_IA32) |
883 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
884 |
test_quant(quant_inter_c, quant_inter_mmx, "mmx", TEST_QUANT_INTER, flags); |
test_quant(quant_inter_c, quant_inter_mmx, "mmx", TEST_QUANT_INTER, flags); |
885 |
if (cpu_flags & XVID_CPU_3DNOWEXT) |
if (cpu_flags & XVID_CPU_3DNOWEXT) |
886 |
test_quant(quant_inter_c, quant_inter_3dne, "3dne", TEST_QUANT_INTER, flags); |
test_quant(quant_inter_c, quant_inter_3dne, "3dne", TEST_QUANT_INTER, flags); |
887 |
if (cpu_flags & XVID_CPU_SSE2) |
if (cpu_flags & XVID_CPU_SSE2) |
888 |
test_quant(quant_inter_c, quant_inter_sse2, "sse2", TEST_QUANT_INTER, flags); |
test_quant(quant_inter_c, quant_inter_sse2, "sse2", TEST_QUANT_INTER, flags); |
889 |
|
#endif |
890 |
|
|
891 |
|
/* Intra dequantization test */ |
892 |
printf("\n--- dequant intra ---\n"); |
printf("\n--- dequant intra ---\n"); |
893 |
test_quant(dequant_intra_c, dequant_intra_c, "c", TEST_DEQUANT_INTRA, flags); |
test_quant(dequant_intra_c, dequant_intra_c, "c", TEST_DEQUANT_INTRA, flags); |
894 |
|
|
895 |
|
#if defined(ARCH_IS_IA32) |
896 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
897 |
test_quant(dequant_intra_c, dequant_intra_mmx, "mmx", TEST_DEQUANT_INTRA, flags); |
test_quant(dequant_intra_c, dequant_intra_mmx, "mmx", TEST_DEQUANT_INTRA, flags); |
898 |
if (cpu_flags & XVID_CPU_MMXEXT) |
if (cpu_flags & XVID_CPU_MMXEXT) |
901 |
test_quant(dequant_intra_c, dequant_intra_3dne, "3dne", TEST_DEQUANT_INTRA, flags); |
test_quant(dequant_intra_c, dequant_intra_3dne, "3dne", TEST_DEQUANT_INTRA, flags); |
902 |
if (cpu_flags & XVID_CPU_SSE2) |
if (cpu_flags & XVID_CPU_SSE2) |
903 |
test_quant(dequant_intra_c, dequant_intra_sse2, "sse2", TEST_DEQUANT_INTRA, flags); |
test_quant(dequant_intra_c, dequant_intra_sse2, "sse2", TEST_DEQUANT_INTRA, flags); |
904 |
|
#endif |
905 |
|
|
906 |
|
/* Inter dequantization test */ |
907 |
printf("\n--- dequant inter ---\n"); |
printf("\n--- dequant inter ---\n"); |
908 |
test_quant(dequant_inter_c, dequant_inter_c, "c", TEST_DEQUANT_INTER, flags); |
test_quant(dequant_inter_c, dequant_inter_c, "c", TEST_DEQUANT_INTER, flags); |
909 |
|
|
910 |
|
#if defined(ARCH_IS_IA32) |
911 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
912 |
test_quant(dequant_inter_c, dequant_inter_mmx, "mmx", TEST_DEQUANT_INTER, flags); |
test_quant(dequant_inter_c, dequant_inter_mmx, "mmx", TEST_DEQUANT_INTER, flags); |
913 |
if (cpu_flags & XVID_CPU_MMXEXT) |
if (cpu_flags & XVID_CPU_MMXEXT) |
916 |
test_quant(dequant_inter_c, dequant_inter_3dne, "3dne", TEST_DEQUANT_INTER, flags); |
test_quant(dequant_inter_c, dequant_inter_3dne, "3dne", TEST_DEQUANT_INTER, flags); |
917 |
if (cpu_flags & XVID_CPU_SSE2) |
if (cpu_flags & XVID_CPU_SSE2) |
918 |
test_quant(dequant_inter_c, dequant_inter_sse2, "sse2", TEST_DEQUANT_INTER, flags); |
test_quant(dequant_inter_c, dequant_inter_sse2, "sse2", TEST_DEQUANT_INTER, flags); |
919 |
|
#endif |
920 |
|
|
921 |
printf("\n--- quant4_intra ---\n"); |
/* Intra quantization test */ |
922 |
|
printf("\n--- quant4 intra ---\n"); |
923 |
test_quant(quant4_intra_c, quant4_intra_c, "c", TEST_QUANT_INTRA, flags); |
test_quant(quant4_intra_c, quant4_intra_c, "c", TEST_QUANT_INTRA, flags); |
924 |
|
|
925 |
|
#if defined(ARCH_IS_IA32) |
926 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
927 |
test_quant(quant4_intra_c, quant4_intra_mmx, "mmx", TEST_QUANT_INTRA, flags); |
test_quant(quant4_intra_c, quant4_intra_mmx, "mmx", TEST_QUANT_INTRA, flags); |
928 |
if (cpu_flags & XVID_CPU_MMXEXT) |
if (cpu_flags & XVID_CPU_MMXEXT) |
929 |
test_quant(quant4_intra_c, quant4_intra_xmm, "xmm", TEST_QUANT_INTRA, flags); |
test_quant(quant4_intra_c, quant4_intra_xmm, "xmm", TEST_QUANT_INTRA, flags); |
930 |
|
#endif |
931 |
|
|
932 |
printf("\n--- quant4_inter ---\n"); |
/* Inter quantization test */ |
933 |
|
printf("\n--- quant4 inter ---\n"); |
934 |
test_quant(quant4_inter_c, quant4_inter_c, "c", TEST_QUANT_INTER, flags); |
test_quant(quant4_inter_c, quant4_inter_c, "c", TEST_QUANT_INTER, flags); |
935 |
|
|
936 |
|
#if defined(ARCH_IS_IA32) |
937 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
938 |
test_quant(quant4_inter_c, quant4_inter_mmx, "mmx", TEST_QUANT_INTER, flags); |
test_quant(quant4_inter_c, quant4_inter_mmx, "mmx", TEST_QUANT_INTER, flags); |
939 |
if (cpu_flags & XVID_CPU_MMXEXT) |
if (cpu_flags & XVID_CPU_MMXEXT) |
940 |
test_quant(quant4_inter_c, quant4_inter_xmm, "xmm", TEST_QUANT_INTER, flags); |
test_quant(quant4_inter_c, quant4_inter_xmm, "xmm", TEST_QUANT_INTER, flags); |
941 |
|
#endif |
942 |
|
|
943 |
printf("\n--- dequant4_intra ---\n"); |
/* Intra dequantization test */ |
944 |
|
printf("\n--- dequant4 intra ---\n"); |
945 |
test_quant(dequant4_intra_c, dequant4_intra_c, "c", TEST_DEQUANT_INTRA, flags); |
test_quant(dequant4_intra_c, dequant4_intra_c, "c", TEST_DEQUANT_INTRA, flags); |
946 |
|
|
947 |
|
#if defined(ARCH_IS_IA32) |
948 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
949 |
test_quant(dequant4_intra_c, dequant4_intra_mmx, "mmx", TEST_DEQUANT_INTRA, flags); |
test_quant(dequant4_intra_c, dequant4_intra_mmx, "mmx", TEST_DEQUANT_INTRA, flags); |
950 |
if (cpu_flags & XVID_CPU_3DNOWEXT) |
if (cpu_flags & XVID_CPU_3DNOWEXT) |
951 |
test_quant(dequant4_intra_c, dequant4_intra_3dne, "3dne", TEST_DEQUANT_INTRA, flags); |
test_quant(dequant4_intra_c, dequant4_intra_3dne, "3dne", TEST_DEQUANT_INTRA, flags); |
952 |
|
#endif |
953 |
|
|
954 |
printf("\n--- dequant4_inter ---\n"); |
/* Inter dequantization test */ |
955 |
|
printf("\n--- dequant4 inter ---\n"); |
956 |
test_quant(dequant4_inter_c, dequant4_inter_c, "c", TEST_DEQUANT_INTER, flags); |
test_quant(dequant4_inter_c, dequant4_inter_c, "c", TEST_DEQUANT_INTER, flags); |
957 |
|
|
958 |
|
#if defined(ARCH_IS_IA32) |
959 |
if (cpu_flags & XVID_CPU_MMX) |
if (cpu_flags & XVID_CPU_MMX) |
960 |
test_quant(dequant4_inter_c, dequant4_inter_mmx, "mmx", TEST_DEQUANT_INTER, flags); |
test_quant(dequant4_inter_c, dequant4_inter_mmx, "mmx", TEST_DEQUANT_INTER, flags); |
961 |
if (cpu_flags & XVID_CPU_3DNOWEXT) |
if (cpu_flags & XVID_CPU_3DNOWEXT) |
962 |
test_quant(dequant4_inter_c, dequant4_inter_3dne, "3dne", TEST_DEQUANT_INTER, flags); |
test_quant(dequant4_inter_c, dequant4_inter_3dne, "3dne", TEST_DEQUANT_INTER, flags); |
|
|
|
|
emms_mmx(); |
|
|
|
|
963 |
#endif |
#endif |
964 |
|
|
965 |
|
emms(); |
966 |
|
|
967 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
968 |
} |
} |
969 |
|
|
983 |
return xvid_init_convert((XVID_INIT_CONVERTINFO*)param1); |
return xvid_init_convert((XVID_INIT_CONVERTINFO*)param1); |
984 |
|
|
985 |
case XVID_INIT_TEST : |
case XVID_INIT_TEST : |
986 |
return xvid_init_test((int)param1); |
{ |
987 |
|
ptr_t flags = (ptr_t)param1; |
988 |
|
return xvid_init_test((int)flags); |
989 |
|
} |
990 |
default : |
default : |
991 |
return XVID_ERR_FAIL; |
return XVID_ERR_FAIL; |
992 |
} |
} |
1044 |
case XVID_ENC_ENCODE: |
case XVID_ENC_ENCODE: |
1045 |
|
|
1046 |
if (((Encoder *) handle)->mbParam.max_bframes >= 0) |
if (((Encoder *) handle)->mbParam.max_bframes >= 0) |
1047 |
return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
return encoder_encode_bframes((Encoder *) handle, |
1048 |
|
(XVID_ENC_FRAME *) param1, |
1049 |
(XVID_ENC_STATS *) param2); |
(XVID_ENC_STATS *) param2); |
1050 |
else |
else |
1051 |
return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
return encoder_encode((Encoder *) handle, |
1052 |
|
(XVID_ENC_FRAME *) param1, |
1053 |
(XVID_ENC_STATS *) param2); |
(XVID_ENC_STATS *) param2); |
1054 |
|
|
1055 |
case XVID_ENC_CREATE: |
case XVID_ENC_CREATE: |