[cvs] / xvidcore / src / xvid.c Repository:
ViewVC logotype

Diff of /xvidcore/src/xvid.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1.48, Mon Mar 22 22:36:23 2004 UTC revision 1.82, Fri Jun 5 07:58:41 2009 UTC
# Line 3  Line 3 
3   *  XVID MPEG-4 VIDEO CODEC   *  XVID MPEG-4 VIDEO CODEC
4   *  - Native API implementation  -   *  - Native API implementation  -
5   *   *
6   *  Copyright(C) 2001-2003 Peter Ross <pross@xvid.org>   *  Copyright(C) 2001-2004 Peter Ross <pross@xvid.org>
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
# Line 28  Line 28 
28  #include <string.h>  #include <string.h>
29  #include <time.h>  #include <time.h>
30    
31    #if !defined(_WIN32)
32    #include <unistd.h>
33    #endif
34    
35    #if defined(__APPLE__) && defined(__MACH__) && !defined(_SC_NPROCESSORS_CONF)
36    #include <sys/types.h>
37    #include <sys/sysctl.h>
38    #ifdef MAX
39    #undef MAX
40    #endif
41    #ifdef MIN
42    #undef MIN
43    #endif
44    #endif
45    
46    #if defined(__amigaos4__)
47    #include <exec/exec.h>
48    #include <proto/exec.h>
49    #endif
50    
51  #include "xvid.h"  #include "xvid.h"
52  #include "decoder.h"  #include "decoder.h"
53  #include "encoder.h"  #include "encoder.h"
# Line 36  Line 56 
56  #include "dct/fdct.h"  #include "dct/fdct.h"
57  #include "image/colorspace.h"  #include "image/colorspace.h"
58  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
 #include "image/reduced.h"  
59  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
60  #include "utils/mbfunctions.h"  #include "utils/mbfunctions.h"
61  #include "quant/quant.h"  #include "quant/quant.h"
62  #include "motion/motion.h"  #include "motion/motion.h"
63    #include "motion/gmc.h"
64  #include "motion/sad.h"  #include "motion/sad.h"
65  #include "utils/emms.h"  #include "utils/emms.h"
66  #include "utils/timer.h"  #include "utils/timer.h"
# Line 52  Line 72 
72  unsigned int xvid_debug = 0; /* xvid debug mask */  unsigned int xvid_debug = 0; /* xvid debug mask */
73  #endif  #endif
74    
75  #if defined(ARCH_IS_IA32)  #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
 #if defined(_MSC_VER)  
76  #       include <windows.h>  #       include <windows.h>
77  #else  #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || (defined(ARCH_IS_PPC) && !defined(__amigaos4__))
78  #       include <signal.h>  #       include <signal.h>
79  #       include <setjmp.h>  #       include <setjmp.h>
80    
# Line 68  Line 87 
87          }          }
88  #endif  #endif
89    
   
90  /*  /*
91   * Calls the funcptr, and returns whether SIGILL (illegal instruction) was   * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
92   * signalled   * signalled
# Line 78  Line 96 
96   *   0 : SIGILL was *not* signalled   *   0 : SIGILL was *not* signalled
97   *   1 : SIGILL was signalled   *   1 : SIGILL was signalled
98   */   */
99    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
100  int  static int
101  sigill_check(void (*func)())  sigill_check(void (*func)())
102  {  {
 #if defined(_MSC_VER)  
103          _try {          _try {
104                  func();                  func();
105          }          } _except(EXCEPTION_EXECUTE_HANDLER) {
         _except(EXCEPTION_EXECUTE_HANDLER) {  
106    
107                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
108                          return 1;                          return(1);
109          }          }
110          return 0;          return(0);
111  #else  }
112    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || (defined(ARCH_IS_PPC) && !defined(__amigaos4__))
113    static int
114    sigill_check(void (*func)())
115    {
116      void * old_handler;      void * old_handler;
117      int jmpret;      int jmpret;
118    
119        /* Set our SIGILL handler */
120      old_handler = signal(SIGILL, sigill_handler);      old_handler = signal(SIGILL, sigill_handler);
121      if (old_handler == SIG_ERR)  
122      {      /* Check for error */
123          return -1;      if (old_handler == SIG_ERR) {
124            return(-1);
125      }      }
126    
127        /* Save stack context, so if func triggers a SIGILL, we can still roll
128             * back to a valid CPU state */
129      jmpret = setjmp(mark);      jmpret = setjmp(mark);
130      if (jmpret == 0)  
131      {          /* If setjmp returned directly, then its returned value is 0, and we still
132             * have to test the passed func. Otherwise it means the stack context has
133             * been restored by a longjmp() call, which in our case happens only in the
134             * signal handler */
135        if (jmpret == 0) {
136          func();          func();
137      }      }
138    
139        /* Restore old signal handler */
140      signal(SIGILL, old_handler);      signal(SIGILL, old_handler);
141    
142      return jmpret;      return(jmpret);
 #endif  
143  }  }
144  #endif  #endif
145    
146    
147  /* detect cpu flags  */  /* detect cpu flags  */
148  static unsigned int  static unsigned int
149  detect_cpu_flags()  detect_cpu_flags(void)
150  {  {
151          /* enable native assembly optimizations by default */          /* enable native assembly optimizations by default */
152          unsigned int cpu_flags = XVID_CPU_ASM;          unsigned int cpu_flags = XVID_CPU_ASM;
153    
154  #if defined(ARCH_IS_IA32)  #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
155          cpu_flags |= check_cpu_features();          cpu_flags |= check_cpu_features();
156          if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))          if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
157                  cpu_flags &= ~XVID_CPU_SSE;                  cpu_flags &= ~XVID_CPU_SSE;
158    
159          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))
160                  cpu_flags &= ~XVID_CPU_SSE2;                  cpu_flags &= ~(XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41);
161  #endif  #endif
162    
163  #if defined(ARCH_IS_PPC)  #if defined(ARCH_IS_PPC)
164  #if defined(ARCH_IS_PPC_ALTIVEC)  #if defined(__amigaos4__)
165            {
166                    uint32_t vector_unit = VECTORTYPE_NONE;
167                    IExec->GetCPUInfoTags(GCIT_VectorUnit, &vector_unit, TAG_END);
168                    if (vector_unit == VECTORTYPE_ALTIVEC) {
169                            cpu_flags |= XVID_CPU_ALTIVEC;
170                    }
171            }
172    #else
173            if (!sigill_check(altivec_trigger))
174          cpu_flags |= XVID_CPU_ALTIVEC;          cpu_flags |= XVID_CPU_ALTIVEC;
175  #endif  #endif
176  #endif  #endif
# Line 177  Line 213 
213          idct = idct_int32;          idct = idct_int32;
214    
215          /* Only needed on PPC Altivec archs */          /* Only needed on PPC Altivec archs */
216          sadInit = 0;          sadInit = NULL;
217    
218          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
219          emms = emms_c;          emms = emms_c;
# Line 204  Line 240 
240          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
241          transfer_8to16subro  = transfer_8to16subro_c;          transfer_8to16subro  = transfer_8to16subro_c;
242          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
243            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
244          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
245          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
246            transfer8x4_copy   = transfer8x4_copy_c;
247    
248          /* Interlacing functions */          /* Interlacing functions */
249          MBFieldTest = MBFieldTest_c;          MBFieldTest = MBFieldTest_c;
# Line 215  Line 253 
253          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
254          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
255    
256            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
257            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
258            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
259    
260            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
261            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
262            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
263            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
264    
265          interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;          interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
266          interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;          interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
267          interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;          interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
# Line 229  Line 276 
276          interpolate8x8_avg2 = interpolate8x8_avg2_c;          interpolate8x8_avg2 = interpolate8x8_avg2_c;
277          interpolate8x8_avg4 = interpolate8x8_avg4_c;          interpolate8x8_avg4 = interpolate8x8_avg4_c;
278    
279          /* reduced resolution */          /* postprocessing */
280          copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C;          image_brightness = image_brightness_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;  
281    
282          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
283          colorspace_init();          colorspace_init();
# Line 244  Line 286 
286          yv12_to_yv12    = yv12_to_yv12_c;          yv12_to_yv12    = yv12_to_yv12_c;
287          rgb555_to_yv12  = rgb555_to_yv12_c;          rgb555_to_yv12  = rgb555_to_yv12_c;
288          rgb565_to_yv12  = rgb565_to_yv12_c;          rgb565_to_yv12  = rgb565_to_yv12_c;
289            rgb_to_yv12     = rgb_to_yv12_c;
290          bgr_to_yv12     = bgr_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
291          bgra_to_yv12    = bgra_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
292          abgr_to_yv12    = abgr_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
# Line 265  Line 308 
308          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
309          yv12_to_rgb555  = yv12_to_rgb555_c;          yv12_to_rgb555  = yv12_to_rgb555_c;
310          yv12_to_rgb565  = yv12_to_rgb565_c;          yv12_to_rgb565  = yv12_to_rgb565_c;
311            yv12_to_rgb     = yv12_to_rgb_c;
312          yv12_to_bgr     = yv12_to_bgr_c;          yv12_to_bgr     = yv12_to_bgr_c;
313          yv12_to_bgra    = yv12_to_bgra_c;          yv12_to_bgra    = yv12_to_bgra_c;
314          yv12_to_abgr    = yv12_to_abgr_c;          yv12_to_abgr    = yv12_to_abgr_c;
# Line 292  Line 336 
336          dev16      = dev16_c;          dev16      = dev16_c;
337          sad16v     = sad16v_c;          sad16v     = sad16v_c;
338          sse8_16bit = sse8_16bit_c;          sse8_16bit = sse8_16bit_c;
339            sse8_8bit  = sse8_8bit_c;
340    
341  #if defined(ARCH_IS_IA32)          init_GMC(cpu_flags);
342    
343          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;  
         }  
344    
345          if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||          if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
346                  (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||                  (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
347                  (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2))                  (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2) ||
348            (cpu_flags & XVID_CPU_SSE3) || (cpu_flags & XVID_CPU_SSE41))
349          {          {
350                  /* Restore FPU context : emms_c is a nop functions */                  /* Restore FPU context : emms_c is a nop functions */
351                  emms = emms_mmx;                  emms = emms_mmx;
# Line 323  Line 366 
366                  quant_h263_inter   = quant_h263_inter_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
367                  dequant_h263_intra = dequant_h263_intra_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
368                  dequant_h263_inter = dequant_h263_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
   
369                  quant_mpeg_intra   = quant_mpeg_intra_mmx;                  quant_mpeg_intra   = quant_mpeg_intra_mmx;
370                  quant_mpeg_inter   = quant_mpeg_inter_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
371                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
372                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
373    
374    
375                  /* Block related functions */                  /* Block related functions */
376                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
377                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
# Line 337  Line 380 
380                  transfer_8to16sub2 = transfer_8to16sub2_mmx;                  transfer_8to16sub2 = transfer_8to16sub2_mmx;
381                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
382                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
383                    transfer8x4_copy   = transfer8x4_copy_mmx;
384    
385                  /* Interlacing Functions */                  /* Interlacing Functions */
386                  MBFieldTest = MBFieldTest_mmx;                  MBFieldTest = MBFieldTest_mmx;
# Line 346  Line 390 
390                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
391                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
392    
393                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
394                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
395                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
396    
397                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
398                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
399                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
400                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
401    
402                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
403                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
404    
405                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
406                  interpolate8x8_avg4 = interpolate8x8_avg4_mmx;                  interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
407    
408                  /* reduced resolution */                  /* postprocessing */
409                  copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx;                  image_brightness = image_brightness_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;  
410    
411                  /* image input xxx_to_yv12 related functions */                  /* image input xxx_to_yv12 related functions */
412    
413                  yv12_to_yv12  = yv12_to_yv12_mmx;                  yv12_to_yv12  = yv12_to_yv12_mmx;
414    
415                  bgr_to_yv12   = bgr_to_yv12_mmx;                  bgr_to_yv12   = bgr_to_yv12_mmx;
416                    rgb_to_yv12   = rgb_to_yv12_mmx;
417                  bgra_to_yv12  = bgra_to_yv12_mmx;                  bgra_to_yv12  = bgra_to_yv12_mmx;
418                    rgba_to_yv12  = rgba_to_yv12_mmx;
419                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
420                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
421    
# Line 384  Line 437 
437                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
438                  sad16v   = sad16v_mmx;                  sad16v   = sad16v_mmx;
439                  sse8_16bit = sse8_16bit_mmx;                  sse8_16bit = sse8_16bit_mmx;
440                    sse8_8bit  = sse8_8bit_mmx;
441          }          }
442    
443          /* these 3dnow functions are faster than mmx, but slower than xmm. */          /* these 3dnow functions are faster than mmx, but slower than xmm. */
# Line 397  Line 451 
451    
452                  yuyv_to_yv12  = yuyv_to_yv12_3dn;                  yuyv_to_yv12  = yuyv_to_yv12_3dn;
453                  uyvy_to_yv12  = uyvy_to_yv12_3dn;                  uyvy_to_yv12  = uyvy_to_yv12_3dn;
454    
455          }          }
456    
457    
# Line 411  Line 466 
466                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
467                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
468    
469                  /* reduced resolution */                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
470                  copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
471                  add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
472    
473                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
474                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
475                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
476                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
477    
478                  /* Quantization */                  /* Quantization */
                 quant_mpeg_intra = quant_mpeg_intra_xmm;  
479                  quant_mpeg_inter = quant_mpeg_inter_xmm;                  quant_mpeg_inter = quant_mpeg_inter_xmm;
480    
481                  dequant_h263_intra = dequant_h263_intra_xmm;                  dequant_h263_intra = dequant_h263_intra_xmm;
# Line 424  Line 483 
483    
484                  /* Buffer transfer */                  /* Buffer transfer */
485                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
486                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
487    
488                  /* Colorspace transformation */                  /* Colorspace transformation */
489                  yv12_to_yv12  = yv12_to_yv12_xmm;                  /* yv12_to_yv12  = yv12_to_yv12_xmm; */ /* appears to be slow on many machines */
490                  yuyv_to_yv12  = yuyv_to_yv12_xmm;                  yuyv_to_yv12  = yuyv_to_yv12_xmm;
491                  uyvy_to_yv12  = uyvy_to_yv12_xmm;                  uyvy_to_yv12  = uyvy_to_yv12_xmm;
492    
# Line 445  Line 505 
505                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
506                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
507                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
508    
509                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
510                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
511                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
512          }          }
513    
514          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
515    
                 /* Inverse DCT */  
                 idct =  idct_3dne;  
   
516                  /* Buffer transfer */                  /* Buffer transfer */
517                  transfer_8to16copy =  transfer_8to16copy_3dne;                  transfer_8to16copy =  transfer_8to16copy_3dne;
518                  transfer_16to8copy = transfer_16to8copy_3dne;                  transfer_16to8copy = transfer_16to8copy_3dne;
519                  transfer_8to16sub =  transfer_8to16sub_3dne;                  transfer_8to16sub =  transfer_8to16sub_3dne;
520                  transfer_8to16subro =  transfer_8to16subro_3dne;                  transfer_8to16subro =  transfer_8to16subro_3dne;
                 transfer_8to16sub2 =  transfer_8to16sub2_3dne;  
521                  transfer_16to8add = transfer_16to8add_3dne;                  transfer_16to8add = transfer_16to8add_3dne;
522                  transfer8x8_copy = transfer8x8_copy_3dne;                  transfer8x8_copy = transfer8x8_copy_3dne;
523                    transfer8x4_copy = transfer8x4_copy_3dne;
524    
525                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
526                            /* Inverse DCT */
527                            idct =  idct_3dne;
528    
529                            /* Buffer transfer */
530                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
531    
532                            /* Interpolation */
533                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
534                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
535                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
536    
537                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
538                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
539                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
540    
541                  /* Quantization */                  /* Quantization */
542                  quant_h263_intra = quant_h263_intra_3dne;                          quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
543                  quant_h263_inter = quant_h263_inter_3dne;                  quant_h263_inter = quant_h263_inter_3dne;
544                  dequant_mpeg_intra = dequant_mpeg_intra_3dne;                          dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
545                  dequant_mpeg_inter = dequant_mpeg_inter_3dne;                  dequant_mpeg_inter = dequant_mpeg_inter_3dne;
546                  dequant_h263_intra = dequant_h263_intra_3dne;                  dequant_h263_intra = dequant_h263_intra_3dne;
547                  dequant_h263_inter = dequant_h263_inter_3dne;                  dequant_h263_inter = dequant_h263_inter_3dne;
548    
549                  /* ME functions */                  /* ME functions */
                 calc_cbp = calc_cbp_3dne;  
550                  sad16 = sad16_3dne;                  sad16 = sad16_3dne;
551                  sad8 = sad8_3dne;                  sad8 = sad8_3dne;
552                  sad16bi = sad16bi_3dne;                  sad16bi = sad16bi_3dne;
553                  sad8bi = sad8bi_3dne;                  sad8bi = sad8bi_3dne;
554                  dev16 = dev16_3dne;                  dev16 = dev16_3dne;
555                    }
                 /* Interpolation */  
                 interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;  
                 interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;  
                 interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;  
556          }          }
557    
558          if ((cpu_flags & XVID_CPU_SSE2)) {          if ((cpu_flags & XVID_CPU_SSE2)) {
# Line 497  Line 569 
569                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
570                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
571    
572                  /* DCT operators                  /* DCT operators */
                  * no iDCT because it's not "Walken matching" */  
573                  fdct = fdct_sse2_skal;                  fdct = fdct_sse2_skal;
574                    idct = idct_sse2_skal;   /* Is now IEEE1180 and Walken compliant. */
575    
576                    /* postprocessing */
577                    image_brightness = image_brightness_sse2;
578    
579            }
580    
581            if ((cpu_flags & XVID_CPU_SSE3)) {
582    
583                    /* SAD operators */
584                    sad16    = sad16_sse3;
585                    dev16    = dev16_sse3;
586          }          }
587    
588  #endif /* ARCH_IS_IA32 */  #endif /* ARCH_IS_IA32 */
589    
590  #if defined(ARCH_IS_IA64)  #if defined(ARCH_IS_IA64)
# Line 530  Line 614 
614  #endif  #endif
615    
616  #if defined(ARCH_IS_PPC)  #if defined(ARCH_IS_PPC)
617          if ((cpu_flags & XVID_CPU_ASM))          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
618          {            /* sad operators */
619                  calc_cbp = calc_cbp_ppc;                    sad16 = sad16_altivec_c;
620          }                    sad16bi = sad16bi_altivec_c;
621                      sad8 = sad8_altivec_c;
622                      dev16 = dev16_altivec_c;
623    
624              sse8_16bit = sse8_16bit_altivec_c;
625    
626              /* mem transfer */
627              transfer_8to16copy = transfer_8to16copy_altivec_c;
628              transfer_16to8copy = transfer_16to8copy_altivec_c;
629              transfer_8to16sub = transfer_8to16sub_altivec_c;
630              transfer_8to16subro = transfer_8to16subro_altivec_c;
631              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
632              transfer_16to8add = transfer_16to8add_altivec_c;
633              transfer8x8_copy = transfer8x8_copy_altivec_c;
634    
635          if ((cpu_flags & XVID_CPU_ALTIVEC))            /* Inverse DCT */
636          {            idct = idct_altivec_c;
637                  calc_cbp = calc_cbp_altivec;  
638                  fdct = fdct_altivec;            /* Interpolation */
639                  idct = idct_altivec;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
640                  sadInit = sadInit_altivec;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
641                  sad16 = sad16_altivec;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
642                  sad8 = sad8_altivec;  
643                  dev16 = dev16_altivec;            interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
644              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
645    
646                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
647                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
648                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
649                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
650    
651              /* Colorspace conversion */
652              bgra_to_yv12 = bgra_to_yv12_altivec_c;
653              abgr_to_yv12 = abgr_to_yv12_altivec_c;
654              rgba_to_yv12 = rgba_to_yv12_altivec_c;
655              argb_to_yv12 = argb_to_yv12_altivec_c;
656    
657              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
658              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
659    
660              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
661              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
662    
663              /* Quantization */
664              quant_h263_intra = quant_h263_intra_altivec_c;
665              quant_h263_inter = quant_h263_inter_altivec_c;
666              dequant_h263_intra = dequant_h263_intra_altivec_c;
667              dequant_h263_inter = dequant_h263_inter_altivec_c;
668    
669                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
670                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
671    
672                      /* Qpel stuff */
673                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
674                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
675          }          }
676  #endif  #endif
677    
# Line 551  Line 679 
679      xvid_debug = init->debug;      xvid_debug = init->debug;
680  #endif  #endif
681    
682      return 0;      return(0);
683  }  }
684    
685    
# Line 562  Line 690 
690                  return XVID_ERR_VERSION;                  return XVID_ERR_VERSION;
691    
692          info->actual_version = XVID_VERSION;          info->actual_version = XVID_VERSION;
693          info->build = "xvid-1.0.0";          info->build = "xvid-1.3.0-dev";
694          info->cpu_flags = detect_cpu_flags();          info->cpu_flags = detect_cpu_flags();
695            info->num_threads = 0; /* single-thread */
696    
697    #if defined(_WIN32)
698    
699      {
700            SYSTEM_INFO siSysInfo;
701            GetSystemInfo(&siSysInfo);
702            info->num_threads = siSysInfo.dwNumberOfProcessors; /* number of _logical_ cores */
703      }
704    
705    #elif defined(_SC_NPROCESSORS_CONF) /* should be available on Apple too actually */
706    
707      info->num_threads = sysconf(_SC_NPROCESSORS_CONF);
708    
709    #elif defined(__APPLE__) && defined(__MACH__)
710    
711      {
712        size_t len;
713        int    mib[2], ncpu;
714    
715        mib[0] = CTL_HW;
716        mib[1] = HW_NCPU;
717        len    = sizeof(ncpu);
718        if (sysctl(mib, 2, &ncpu, &len, NULL, 0) == 0)
719          info -> num_threads = ncpu;
720        else
721          info -> num_threads = 1;
722      }
723    
724    #elif defined(__amigaos4__)
725    
726      {
727        uint32_t num_threads = 1;
728        IExec->GetCPUInfoTags(GCIT_NumberOfCPUs, &num_threads, TAG_END);
729        info->num_threads = num_threads;
730      }
731    
 #if defined(_SMP) && defined(WIN32)  
         info->num_threads = pthread_num_processors_np();;  
 #else  
         info->num_threads = 0;  
732  #endif  #endif
733    
734          return 0;          return 0;
Legend:
Removed from v.1.48  
changed lines
  Added in v.1.82
No admin address has been configured
 ViewVC Help
Powered by ViewVC 1.0.4