[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 1.33.2.17, Mon Dec 9 10:47:05 2002 UTC revision 1.58, Thu Dec 9 23:02:54 2004 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   *  This program is an implementation of a part of one or more MPEG-4   *  Copyright(C) 2001-2004 Peter Ross <pross@xvid.org>
  *  Video tools as specified in ISO/IEC 14496-2 standard.  Those intending  
  *  to use this software module in hardware or software products are  
  *  advised that its use may infringe existing patents or copyrights, and  
  *  any such use would be at such party's own risk.  The original  
  *  developer of this software module and his/her company, and subsequent  
  *  editors and their companies, will have no liability for use of this  
  *  software or modifications or derivatives thereof.  
7   *   *
8   *  This program is free software ; you can redistribute it and/or modify   *  This program is free software ; you can redistribute it and/or modify
9   *  it under the terms of the GNU General Public License as published by   *  it under the terms of the GNU General Public License as published by
# Line 26  Line 19 
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"
# Line 49  Line 36 
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_h263.h"  #include "quant/quant.h"
 #include "quant/quant_mpeg4.h"  
42  #include "motion/motion.h"  #include "motion/motion.h"
43  #include "motion/sad.h"  #include "motion/sad.h"
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    #include "image/postprocessing.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) && defined(_MSC_VER)
55  #include <windows.h>  #include <windows.h>
56  #else  #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_PPC)
57  #include <signal.h>  #include <signal.h>
58  #include <setjmp.h>  #include <setjmp.h>
 #endif  
   
   
 #ifndef WIN32  
59    
60  static jmp_buf mark;  static jmp_buf mark;
61    
# Line 83  Line 68 
68    
69    
70  /*  /*
71  calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled   * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
72  return values:   * signalled
73  -1 : could not determine   *
74  0  : SIGILL was *not* signalled   * Return values:
75  1  : SIGILL was signalled   *  -1 : could not determine
76     *   0 : SIGILL was *not* signalled
77     *   1 : SIGILL was signalled
78  */  */
79    #if defined(ARCH_IS_IA32) && defined(_MSC_VER)
80  int  static int
81  sigill_check(void (*func)())  sigill_check(void (*func)())
82  {  {
 #ifdef WIN32  
83          _try {          _try {
84                  func();                  func();
85          }          } _except(EXCEPTION_EXECUTE_HANDLER) {
         _except(EXCEPTION_EXECUTE_HANDLER) {  
86    
87                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
88                          return 1;                          return(1);
89          }          }
90          return 0;          return(0);
91  #else  }
92    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_PPC)
93    static int
94    sigill_check(void (*func)())
95    {
96      void * old_handler;      void * old_handler;
97      int jmpret;      int jmpret;
98    
99        /* Set our SIGILL handler */
100      old_handler = signal(SIGILL, sigill_handler);      old_handler = signal(SIGILL, sigill_handler);
101      if (old_handler == SIG_ERR)  
102      {      /* Check for error */
103          return -1;      if (old_handler == SIG_ERR) {
104            return(-1);
105      }      }
106    
107        /* Save stack context, so if func triggers a SIGILL, we can still roll
108             * back to a valid CPU state */
109      jmpret = setjmp(mark);      jmpret = setjmp(mark);
110      if (jmpret == 0)  
111      {          /* If setjmp returned directly, then its returned value is 0, and we still
112             * have to test the passed func. Otherwise it means the stack context has
113             * been restored by a longjmp() call, which in our case happens only in the
114             * signal handler */
115        if (jmpret == 0) {
116          func();          func();
117      }      }
118    
119        /* Restore old signal handler */
120      signal(SIGILL, old_handler);      signal(SIGILL, old_handler);
121    
122      return jmpret;      return(jmpret);
 #endif  
123  }  }
124  #endif  #endif
125    
126    
127    /* detect cpu flags  */
128    static unsigned int
129    detect_cpu_flags()
130    {
131            /* enable native assembly optimizations by default */
132            unsigned int cpu_flags = XVID_CPU_ASM;
133    
134    #if defined(ARCH_IS_IA32)
135            cpu_flags |= check_cpu_features();
136            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
137                    cpu_flags &= ~XVID_CPU_SSE;
138    
139            if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
140                    cpu_flags &= ~XVID_CPU_SSE2;
141    #endif
142    
143    #if defined(ARCH_IS_PPC)
144            if (!sigill_check(altivec_trigger))
145                    cpu_flags |= XVID_CPU_ALTIVEC;
146    #endif
147    
148            return cpu_flags;
149    }
150    
151    
152  /*****************************************************************************  /*****************************************************************************
153   * XviD Init Entry point   * XviD Init Entry point
154   *   *
# Line 143  Line 165 
165    
166    
167  static  static
168  int xvid_init_init(XVID_INIT_PARAM * init_param)  int xvid_gbl_init(xvid_gbl_init_t * init)
169  {  {
170          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;  
   
         /* Do we have to force CPU features  ? */  
         if ((init_param->cpu_flags & XVID_CPU_FORCE)) {  
171    
172                  cpu_flags = init_param->cpu_flags;          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
173                    return XVID_ERR_VERSION;
         } 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;  
174    
175            cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
176    
177          /* Initialize the function pointers */          /* Initialize the function pointers */
178          idct_int32_init();          idct_int32_init();
# Line 194  Line 188 
188          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
189          emms = emms_c;          emms = emms_c;
190    
191            /* Qpel stuff */
192            xvid_QP_Funcs = &xvid_QP_Funcs_C;
193            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
194            xvid_Init_QP();
195    
196          /* Quantization functions */          /* Quantization functions */
197          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
198          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
199          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
200          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
201    
202          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
203          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
204          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
205          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
206    
207          /* Block transfer related functions */          /* Block transfer related functions */
208          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
209          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
210          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
211            transfer_8to16subro  = transfer_8to16subro_c;
212          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
213          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
214          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
# Line 221  Line 221 
221          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
222          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
223    
224            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
225            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
226            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
227            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
228    
229          interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;          interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
230          interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;          interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
231          interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;          interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
# Line 235  Line 240 
240          interpolate8x8_avg2 = interpolate8x8_avg2_c;          interpolate8x8_avg2 = interpolate8x8_avg2_c;
241          interpolate8x8_avg4 = interpolate8x8_avg4_c;          interpolate8x8_avg4 = interpolate8x8_avg4_c;
242    
243          /* reduced resoltuion */          /* postprocessing */
244            image_brightness = image_brightness_c;
         copy_upsampled_8x8_16to8 = xvid_Copy_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  
         vfilter_31 = xvid_VFilter_31_C;  
         hfilter_31 = xvid_HFilter_31_C;  
 #endif  
         filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C;  
         filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C;  
245    
246          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
247          colorspace_init();          colorspace_init();
# Line 260  Line 254 
254          bgra_to_yv12    = bgra_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
255          abgr_to_yv12    = abgr_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
256          rgba_to_yv12    = rgba_to_yv12_c;          rgba_to_yv12    = rgba_to_yv12_c;
257            argb_to_yv12    = argb_to_yv12_c;
258          yuyv_to_yv12    = yuyv_to_yv12_c;          yuyv_to_yv12    = yuyv_to_yv12_c;
259          uyvy_to_yv12    = uyvy_to_yv12_c;          uyvy_to_yv12    = uyvy_to_yv12_c;
260    
# Line 269  Line 264 
264          bgrai_to_yv12   = bgrai_to_yv12_c;          bgrai_to_yv12   = bgrai_to_yv12_c;
265          abgri_to_yv12   = abgri_to_yv12_c;          abgri_to_yv12   = abgri_to_yv12_c;
266          rgbai_to_yv12   = rgbai_to_yv12_c;          rgbai_to_yv12   = rgbai_to_yv12_c;
267            argbi_to_yv12   = argbi_to_yv12_c;
268          yuyvi_to_yv12   = yuyvi_to_yv12_c;          yuyvi_to_yv12   = yuyvi_to_yv12_c;
269          uyvyi_to_yv12   = uyvyi_to_yv12_c;          uyvyi_to_yv12   = uyvyi_to_yv12_c;
270    
   
271          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
272          yv12_to_rgb555  = yv12_to_rgb555_c;          yv12_to_rgb555  = yv12_to_rgb555_c;
273          yv12_to_rgb565  = yv12_to_rgb565_c;          yv12_to_rgb565  = yv12_to_rgb565_c;
# Line 280  Line 275 
275          yv12_to_bgra    = yv12_to_bgra_c;          yv12_to_bgra    = yv12_to_bgra_c;
276          yv12_to_abgr    = yv12_to_abgr_c;          yv12_to_abgr    = yv12_to_abgr_c;
277          yv12_to_rgba    = yv12_to_rgba_c;          yv12_to_rgba    = yv12_to_rgba_c;
278            yv12_to_argb    = yv12_to_argb_c;
279          yv12_to_yuyv    = yv12_to_yuyv_c;          yv12_to_yuyv    = yv12_to_yuyv_c;
280          yv12_to_uyvy    = yv12_to_uyvy_c;          yv12_to_uyvy    = yv12_to_uyvy_c;
281    
# Line 289  Line 285 
285          yv12_to_bgrai   = yv12_to_bgrai_c;          yv12_to_bgrai   = yv12_to_bgrai_c;
286          yv12_to_abgri   = yv12_to_abgri_c;          yv12_to_abgri   = yv12_to_abgri_c;
287          yv12_to_rgbai   = yv12_to_rgbai_c;          yv12_to_rgbai   = yv12_to_rgbai_c;
288            yv12_to_argbi   = yv12_to_argbi_c;
289          yv12_to_yuyvi   = yv12_to_yuyvi_c;          yv12_to_yuyvi   = yv12_to_yuyvi_c;
290          yv12_to_uyvyi   = yv12_to_uyvyi_c;          yv12_to_uyvyi   = yv12_to_uyvyi_c;
291    
# Line 300  Line 297 
297          sad8bi   = sad8bi_c;          sad8bi   = sad8bi_c;
298          dev16    = dev16_c;          dev16    = dev16_c;
299          sad16v   = sad16v_c;          sad16v   = sad16v_c;
300            sse8_16bit = sse8_16bit_c;
301            sse8_8bit  = sse8_8bit_c;
302    
303  //      Halfpel8_Refine = Halfpel8_Refine_c;  #if defined(ARCH_IS_IA32)
   
 #ifdef ARCH_X86  
304    
305          if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||          if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
306                  (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||                  (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
# Line 313  Line 310 
310                  emms = emms_mmx;                  emms = emms_mmx;
311          }          }
312    
313          if ((cpu_flags & XVID_CPU_MMX) > 0) {          if ((cpu_flags & XVID_CPU_MMX)) {
314    
315                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
316                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
317                  idct = idct_mmx;                  idct = idct_mmx;
318    
319                    /* Qpel stuff */
320                    xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
321                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
322    
323                  /* Quantization related functions */                  /* Quantization related functions */
324                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
325                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
326                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
327                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
328    
329                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_intra   = quant_mpeg_intra_mmx;
330                  dequant4_intra = dequant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
331                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
332                  dequant4_inter = dequant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
333    
334                  /* Block related functions */                  /* Block related functions */
335                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
336                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
337                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
338                    transfer_8to16subro  = transfer_8to16subro_mmx;
339                  transfer_8to16sub2 = transfer_8to16sub2_mmx;                  transfer_8to16sub2 = transfer_8to16sub2_mmx;
340                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
341                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
# Line 346  Line 348 
348                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
349                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
350    
351                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
352                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
353                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
354                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
355    
356                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
357                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
358    
359                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
360                  interpolate8x8_avg4 = interpolate8x8_avg4_mmx;                  interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
361    
362                  /* reduced resolution */                  /* postprocessing */
363                  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;  
364    
365                  /* image input xxx_to_yv12 related functions */                  /* image input xxx_to_yv12 related functions */
366                  yv12_to_yv12  = yv12_to_yv12_mmx;                  yv12_to_yv12  = yv12_to_yv12_mmx;
# Line 383  Line 386 
386                  sad8bi  = sad8bi_mmx;                  sad8bi  = sad8bi_mmx;
387                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
388                  sad16v   = sad16v_mmx;                  sad16v   = sad16v_mmx;
389                    sse8_16bit = sse8_16bit_mmx;
390                    sse8_8bit  = sse8_8bit_mmx;
391          }          }
392    
393          /* these 3dnow functions are faster than mmx, but slower than xmm. */          /* these 3dnow functions are faster than mmx, but slower than xmm. */
394          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
395    
396                    emms = emms_3dn;
397    
398                  /* ME functions */                  /* ME functions */
399                  sad16bi = sad16bi_3dn;                  sad16bi = sad16bi_3dn;
# Line 398  Line 404 
404          }          }
405    
406    
407          if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {          if ((cpu_flags & XVID_CPU_MMXEXT)) {
408    
409                  /* Inverse DCT */                  /* DCT */
410                    fdct = fdct_xmm_skal;
411                  idct = idct_xmm;                  idct = idct_xmm;
412    
413                  /* Interpolation */                  /* Interpolation */
# Line 408  Line 415 
415                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
416                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
417    
418                  /* reduced resolution */                  interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
419                  copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm;                  interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
420                  add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm;                  interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
421                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
422    
423                  /* Quantization */                  /* Quantization */
424                  dequant_intra = dequant_intra_xmm;                  quant_mpeg_intra = quant_mpeg_intra_xmm;
425                  dequant_inter = dequant_inter_xmm;                  quant_mpeg_inter = quant_mpeg_inter_xmm;
426    
427                    dequant_h263_intra = dequant_h263_intra_xmm;
428                    dequant_h263_inter = dequant_h263_inter_xmm;
429    
430                  /* Buffer transfer */                  /* Buffer transfer */
431                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
# Line 433  Line 444 
444                  sad16v   = sad16v_xmm;                  sad16v   = sad16v_xmm;
445          }          }
446    
447          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
448    
449                  /* Interpolation */                  /* Interpolation */
450                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
# Line 441  Line 452 
452                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
453          }          }
454    
455          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
456  #ifdef EXPERIMENTAL_SSE2_CODE  
457                    /* Buffer transfer */
458                    transfer_8to16copy =  transfer_8to16copy_3dne;
459                    transfer_16to8copy = transfer_16to8copy_3dne;
460                    transfer_8to16sub =  transfer_8to16sub_3dne;
461                    transfer_8to16subro =  transfer_8to16subro_3dne;
462                    transfer_16to8add = transfer_16to8add_3dne;
463                    transfer8x8_copy = transfer8x8_copy_3dne;
464    
465                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
466                            /* Inverse DCT */
467                            idct =  idct_3dne;
468    
469                            /* Buffer transfer */
470                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
471    
472                            /* Interpolation */
473                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
474                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
475                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
476    
477                            /* Quantization */
478                            quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
479                            quant_h263_inter = quant_h263_inter_3dne;
480                            dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
481                            dequant_mpeg_inter = dequant_mpeg_inter_3dne;
482                            dequant_h263_intra = dequant_h263_intra_3dne;
483                            dequant_h263_inter = dequant_h263_inter_3dne;
484    
485                            /* ME functions */
486                            calc_cbp = calc_cbp_3dne;
487    
488                            sad16 = sad16_3dne;
489                            sad8 = sad8_3dne;
490                            sad16bi = sad16bi_3dne;
491                            sad8bi = sad8bi_3dne;
492                            dev16 = dev16_3dne;
493                    }
494            }
495    
496            if ((cpu_flags & XVID_CPU_SSE2)) {
497    
498                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
499    
500                  /* Quantization */                  /* Quantization */
501                  quant_intra   = quant_intra_sse2;                  quant_h263_intra   = quant_h263_intra_sse2;
502                  dequant_intra = dequant_intra_sse2;                  quant_h263_inter   = quant_h263_inter_sse2;
503                  quant_inter   = quant_inter_sse2;                  dequant_h263_intra = dequant_h263_intra_sse2;
504                  dequant_inter = dequant_inter_sse2;                  dequant_h263_inter = dequant_h263_inter_sse2;
505    
506                  /* ME */                  /* SAD operators */
507                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
508                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
509    
510                  /* Forward and Inverse DCT */                  /* DCT operators
511                  idct  = idct_sse2;                   * no iDCT because it's not "Walken matching" */
512                  fdct = fdct_sse2;                  fdct = fdct_sse2_skal;
 #endif  
         }  
513    
514  #endif                  /* postprocessing */
515                    image_brightness = image_brightness_sse2;
516            }
517    #endif /* ARCH_IS_IA32 */
518    
519  #ifdef ARCH_IA64  #if defined(ARCH_IS_IA64)
520          if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?          if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
521            idct_ia64_init();            idct_ia64_init();
522            fdct = fdct_ia64;            fdct = fdct_ia64;
523            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
524            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
525            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
526            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 476  Line 528 
528            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
529            sad8 = sad8_ia64;            sad8 = sad8_ia64;
530            dev16 = dev16_ia64;            dev16 = dev16_ia64;
531  //        Halfpel8_Refine = Halfpel8_Refine_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
532            quant_intra = quant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
533            dequant_intra = dequant_intra_ia64;            quant_h263_inter = quant_h263_inter_ia64;
534            quant_inter = quant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
535            dequant_inter = dequant_inter_ia64;            dequant_h263_inter = dequant_h263_inter_ia64;
536            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
537            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
538            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
539            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
540            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
541            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
542          }          }
543  #endif  #endif
544    
545  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
546  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
547          calc_cbp = calc_cbp_altivec;            /* sad operators */
548          fdct = fdct_altivec;                    sad16 = sad16_altivec_c;
549          idct = idct_altivec;                    sad16bi = sad16bi_altivec_c;
550          sadInit = sadInit_altivec;                    sad8 = sad8_altivec_c;
551          sad16 = sad16_altivec;                    dev16 = dev16_altivec_c;
552          sad8 = sad8_altivec;  
553          dev16 = dev16_altivec;            sse8_16bit = sse8_16bit_altivec_c;
554  #else  
555          calc_cbp = calc_cbp_ppc;            /* mem transfer */
556              transfer_8to16copy = transfer_8to16copy_altivec_c;
557              transfer_16to8copy = transfer_16to8copy_altivec_c;
558              transfer_8to16sub = transfer_8to16sub_altivec_c;
559              transfer_8to16subro = transfer_8to16subro_altivec_c;
560              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
561              transfer_16to8add = transfer_16to8add_altivec_c;
562              transfer8x8_copy = transfer8x8_copy_altivec_c;
563    
564              /* Inverse DCT */
565              idct = idct_altivec_c;
566    
567              /* Interpolation */
568              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
569              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
570              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
571    
572              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
573              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
574    
575                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
576                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
577                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
578                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
579    
580              /* Colorspace conversion */
581              bgra_to_yv12 = bgra_to_yv12_altivec_c;
582              abgr_to_yv12 = abgr_to_yv12_altivec_c;
583              rgba_to_yv12 = rgba_to_yv12_altivec_c;
584              argb_to_yv12 = argb_to_yv12_altivec_c;
585    
586              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
587              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
588    
589              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
590              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
591    
592              /* Quantization */
593              quant_h263_intra = quant_h263_intra_altivec_c;
594              quant_h263_inter = quant_h263_inter_altivec_c;
595              dequant_h263_intra = dequant_h263_intra_altivec_c;
596              dequant_h263_inter = dequant_h263_inter_altivec_c;
597    
598                      /* Qpel stuff */
599                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
600                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
601            }
602  #endif  #endif
603    
604    #if defined(_DEBUG)
605        xvid_debug = init->debug;
606  #endif  #endif
607    
608          return XVID_ERR_OK;      return(0);
609  }  }
610    
611    
612    static int
613    xvid_gbl_info(xvid_gbl_info_t * info)
614    {
615            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
616                    return XVID_ERR_VERSION;
617    
618            info->actual_version = XVID_VERSION;
619            info->build = "xvid-1.1-cvshead";
620            info->cpu_flags = detect_cpu_flags();
621    
622    #if defined(_SMP) && defined(WIN32)
623            info->num_threads = pthread_num_processors_np();;
624    #else
625            info->num_threads = 0;
626    #endif
627    
628            return 0;
629    }
630    
631    
632  static int  static int
633  xvid_init_convert(XVID_INIT_CONVERTINFO* convert)  xvid_gbl_convert(xvid_gbl_convert_t* convert)
634  {  {
635          // const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);          int width;
636          const int width = convert->width;          int height;
637          const int height = convert->height;          int width2;
638          const int width2 = convert->width/2;          int height2;
         const int height2 = convert->height/2;  
639          IMAGE img;          IMAGE img;
640    
641          switch (convert->input.colorspace & ~XVID_CSP_VFLIP)          if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
642                  return XVID_ERR_VERSION;
643    
644    #if 0
645            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
646    #endif
647            width = convert->width;
648            height = convert->height;
649            width2 = convert->width/2;
650            height2 = convert->height/2;
651    
652            switch (convert->input.csp & ~XVID_CSP_VFLIP)
653          {          {
654                  case XVID_CSP_YV12 :                  case XVID_CSP_YV12 :
655                          img.y = convert->input.y;                          img.y = convert->input.plane[0];
656                          img.v = (uint8_t*)convert->input.y + width*height;                          img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
657                          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;
658                          image_output(&img, width, height, width,                          image_output(&img, width, height, width,
659                                                  convert->output.y, convert->output.y_stride,                                                  (uint8_t**)convert->output.plane, convert->output.stride,
660                                                  convert->output.colorspace, convert->interlacing);                                                  convert->output.csp, convert->interlacing);
661                          break;                          break;
662    
663                  default :                  default :
# Line 537  Line 666 
666    
667    
668          emms();          emms();
669          return XVID_ERR_OK;          return 0;
670  }  }
671    
672    /*****************************************************************************
673     * XviD Global Entry point
674     *
675     * Well this function initialize all internal function pointers according
676     * to the CPU features forced by the library client or autodetected (depending
677     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
678     * image colorspace transformation tables.
679     *
680     ****************************************************************************/
681    
682    
683  int  int
684  xvid_init(void *handle,  xvid_global(void *handle,
685                    int opt,                    int opt,
686                    void *param1,                    void *param1,
687                    void *param2)                    void *param2)
688  {  {
689          switch(opt)          switch(opt)
690          {          {
691                  case XVID_INIT_INIT :                  case XVID_GBL_INIT :
692                          return xvid_init_init((XVID_INIT_PARAM*)param1);                          return xvid_gbl_init((xvid_gbl_init_t*)param1);
693    
694            case XVID_GBL_INFO :
695                return xvid_gbl_info((xvid_gbl_info_t*)param1);
696    
697                  case XVID_INIT_CONVERT :                  case XVID_GBL_CONVERT :
698                          return xvid_init_convert((XVID_INIT_CONVERTINFO*)param1);                          return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
699    
700                  default :                  default :
701                          return XVID_ERR_FAIL;                          return XVID_ERR_FAIL;
# Line 577  Line 719 
719                          void *param2)                          void *param2)
720  {  {
721          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1, (XVID_DEC_STATS*) param2);  
   
722          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
723                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
724    
725          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
726                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
727    
728            case XVID_DEC_DECODE:
729                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
730    
731          default:          default:
732                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
733          }          }
# Line 611  Line 753 
753          switch (opt) {          switch (opt) {
754          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
755    
756                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
757                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
758                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
759    
760          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
761                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
762    
763          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
764                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
765    
766          default:          default:
767                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
Legend:
Removed from v.1.33.2.17  
changed lines
  Added in v.1.58
No admin address has been configured
 ViewVC Help
Powered by ViewVC 1.0.4