[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 1.33.2.19, Mon Dec 30 10:49:17 2002 UTC revision 1.65.2.3, Tue Jul 11 19:11:37 2006 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(ARCH_IS_X86_64)) && defined(_MSC_VER)
55  #include <windows.h>  #include <windows.h>
56  #else  #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || 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(ARCH_IS_X86_64)) && 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_X86_64) || 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(void)
130    {
131            /* enable native assembly optimizations by default */
132            unsigned int cpu_flags = XVID_CPU_ASM;
133    
134    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
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)
 {  
         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)) {  
   
                 cpu_flags = init_param->cpu_flags;  
   
         } else {  
   
                 cpu_flags = check_cpu_features();  
   
 #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE)  
                 if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))  
                         cpu_flags &= ~XVID_CPU_SSE;  
   
                 if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))  
                         cpu_flags &= ~XVID_CPU_SSE2;  
 #endif  
         }  
   
         if ((init_param->cpu_flags & XVID_CPU_CHKONLY))  
169          {          {
170                  init_param->cpu_flags = cpu_flags;          unsigned int cpu_flags;
                 return XVID_ERR_OK;  
         }  
171    
172          init_param->cpu_flags = cpu_flags;          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
173                    return XVID_ERR_VERSION;
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 189  Line 183 
183          idct = idct_int32;          idct = idct_int32;
184    
185          /* Only needed on PPC Altivec archs */          /* Only needed on PPC Altivec archs */
186          sadInit = 0;          sadInit = NULL;
187    
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_8to16sub2ro = transfer_8to16sub2ro_c;
214          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
215          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
216            transfer8x4_copy   = transfer8x4_copy_c;
217    
218          /* Interlacing functions */          /* Interlacing functions */
219          MBFieldTest = MBFieldTest_c;          MBFieldTest = MBFieldTest_c;
# Line 221  Line 223 
223          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
224          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
225    
226            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
227            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
228            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
229    
230            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
231            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
232            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
233            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
234    
235          interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;          interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
236          interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;          interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
237          interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;          interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
# Line 235  Line 246 
246          interpolate8x8_avg2 = interpolate8x8_avg2_c;          interpolate8x8_avg2 = interpolate8x8_avg2_c;
247          interpolate8x8_avg4 = interpolate8x8_avg4_c;          interpolate8x8_avg4 = interpolate8x8_avg4_c;
248    
249          /* reduced resoltuion */          /* postprocessing */
250            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;  
251    
252          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
253          colorspace_init();          colorspace_init();
# Line 260  Line 260 
260          bgra_to_yv12    = bgra_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
261          abgr_to_yv12    = abgr_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
262          rgba_to_yv12    = rgba_to_yv12_c;          rgba_to_yv12    = rgba_to_yv12_c;
263            argb_to_yv12    = argb_to_yv12_c;
264          yuyv_to_yv12    = yuyv_to_yv12_c;          yuyv_to_yv12    = yuyv_to_yv12_c;
265          uyvy_to_yv12    = uyvy_to_yv12_c;          uyvy_to_yv12    = uyvy_to_yv12_c;
266    
# Line 269  Line 270 
270          bgrai_to_yv12   = bgrai_to_yv12_c;          bgrai_to_yv12   = bgrai_to_yv12_c;
271          abgri_to_yv12   = abgri_to_yv12_c;          abgri_to_yv12   = abgri_to_yv12_c;
272          rgbai_to_yv12   = rgbai_to_yv12_c;          rgbai_to_yv12   = rgbai_to_yv12_c;
273            argbi_to_yv12   = argbi_to_yv12_c;
274          yuyvi_to_yv12   = yuyvi_to_yv12_c;          yuyvi_to_yv12   = yuyvi_to_yv12_c;
275          uyvyi_to_yv12   = uyvyi_to_yv12_c;          uyvyi_to_yv12   = uyvyi_to_yv12_c;
276    
   
277          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
278          yv12_to_rgb555  = yv12_to_rgb555_c;          yv12_to_rgb555  = yv12_to_rgb555_c;
279          yv12_to_rgb565  = yv12_to_rgb565_c;          yv12_to_rgb565  = yv12_to_rgb565_c;
# Line 280  Line 281 
281          yv12_to_bgra    = yv12_to_bgra_c;          yv12_to_bgra    = yv12_to_bgra_c;
282          yv12_to_abgr    = yv12_to_abgr_c;          yv12_to_abgr    = yv12_to_abgr_c;
283          yv12_to_rgba    = yv12_to_rgba_c;          yv12_to_rgba    = yv12_to_rgba_c;
284            yv12_to_argb    = yv12_to_argb_c;
285          yv12_to_yuyv    = yv12_to_yuyv_c;          yv12_to_yuyv    = yv12_to_yuyv_c;
286          yv12_to_uyvy    = yv12_to_uyvy_c;          yv12_to_uyvy    = yv12_to_uyvy_c;
287    
# Line 289  Line 291 
291          yv12_to_bgrai   = yv12_to_bgrai_c;          yv12_to_bgrai   = yv12_to_bgrai_c;
292          yv12_to_abgri   = yv12_to_abgri_c;          yv12_to_abgri   = yv12_to_abgri_c;
293          yv12_to_rgbai   = yv12_to_rgbai_c;          yv12_to_rgbai   = yv12_to_rgbai_c;
294            yv12_to_argbi   = yv12_to_argbi_c;
295          yv12_to_yuyvi   = yv12_to_yuyvi_c;          yv12_to_yuyvi   = yv12_to_yuyvi_c;
296          yv12_to_uyvyi   = yv12_to_uyvyi_c;          yv12_to_uyvyi   = yv12_to_uyvyi_c;
297    
# Line 300  Line 303 
303          sad8bi   = sad8bi_c;          sad8bi   = sad8bi_c;
304          dev16    = dev16_c;          dev16    = dev16_c;
305          sad16v   = sad16v_c;          sad16v   = sad16v_c;
306            sse8_16bit = sse8_16bit_c;
307            sse8_8bit  = sse8_8bit_c;
308    
309  //      Halfpel8_Refine = Halfpel8_Refine_c;  #if defined(ARCH_IS_IA32)
   
 #ifdef ARCH_X86  
310    
311          if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||          if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
312                  (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||                  (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
# Line 313  Line 316 
316                  emms = emms_mmx;                  emms = emms_mmx;
317          }          }
318    
319          if ((cpu_flags & XVID_CPU_MMX) > 0) {          if ((cpu_flags & XVID_CPU_MMX)) {
320    
321                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
322                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
323                  idct = idct_mmx;                  idct = idct_mmx;
324    
325                    /* Qpel stuff */
326                    xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
327                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
328    
329                  /* Quantization related functions */                  /* Quantization related functions */
330                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
331                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
332                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
333                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
334    
335                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_intra   = quant_mpeg_intra_mmx;
336                  dequant4_intra = dequant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
337                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
338                  dequant4_inter = dequant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
339    
340                  /* Block related functions */                  /* Block related functions */
341                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
342                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
343                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
344                    transfer_8to16subro  = transfer_8to16subro_mmx;
345                  transfer_8to16sub2 = transfer_8to16sub2_mmx;                  transfer_8to16sub2 = transfer_8to16sub2_mmx;
346                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
347                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
348                    transfer8x4_copy   = transfer8x4_copy_mmx;
349    
350                  /* Interlacing Functions */                  /* Interlacing Functions */
351                  MBFieldTest = MBFieldTest_mmx;                  MBFieldTest = MBFieldTest_mmx;
# Line 346  Line 355 
355                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
356                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
357    
358                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
359                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
360                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
361    
362                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
363                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
364                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
365                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
366    
367                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
368                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
369    
370                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
371                  interpolate8x8_avg4 = interpolate8x8_avg4_mmx;                  interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
372    
373                  /* reduced resolution */                  /* postprocessing */
374                  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;  
375    
376                  /* image input xxx_to_yv12 related functions */                  /* image input xxx_to_yv12 related functions */
377                  yv12_to_yv12  = yv12_to_yv12_mmx;                  yv12_to_yv12  = yv12_to_yv12_mmx;
# Line 383  Line 397 
397                  sad8bi  = sad8bi_mmx;                  sad8bi  = sad8bi_mmx;
398                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
399                  sad16v   = sad16v_mmx;                  sad16v   = sad16v_mmx;
400                    sse8_16bit = sse8_16bit_mmx;
401                    sse8_8bit  = sse8_8bit_mmx;
402          }          }
403    
404          /* these 3dnow functions are faster than mmx, but slower than xmm. */          /* these 3dnow functions are faster than mmx, but slower than xmm. */
405          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
406    
407                    emms = emms_3dn;
408    
409                  /* ME functions */                  /* ME functions */
410                  sad16bi = sad16bi_3dn;                  sad16bi = sad16bi_3dn;
# Line 398  Line 415 
415          }          }
416    
417    
418          if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {          if ((cpu_flags & XVID_CPU_MMXEXT)) {
419    
420                  /* Inverse DCT */                  /* DCT */
421                    fdct = fdct_xmm_skal;
422                  idct = idct_xmm;                  idct = idct_xmm;
423    
424                  /* Interpolation */                  /* Interpolation */
# Line 408  Line 426 
426                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
427                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
428    
429                  /* reduced resolution */                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
430                  copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
431                  add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
432    
433                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
434                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
435                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
436                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
437    
438                  /* Quantization */                  /* Quantization */
439                  quant4_intra = quant4_intra_xmm;                  quant_mpeg_intra = quant_mpeg_intra_xmm;
440                  quant4_inter = quant4_inter_xmm;                  quant_mpeg_inter = quant_mpeg_inter_xmm;
441    
442                  dequant_intra = dequant_intra_xmm;                  dequant_h263_intra = dequant_h263_intra_xmm;
443                  dequant_inter = dequant_inter_xmm;                  dequant_h263_inter = dequant_h263_inter_xmm;
444    
445                  /* Buffer transfer */                  /* Buffer transfer */
446                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
447                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
448    
449                  /* Colorspace transformation */                  /* Colorspace transformation */
450                  yv12_to_yv12  = yv12_to_yv12_xmm;                  yv12_to_yv12  = yv12_to_yv12_xmm;
# Line 436  Line 460 
460                  sad16v   = sad16v_xmm;                  sad16v   = sad16v_xmm;
461          }          }
462    
463          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
464    
465                  /* Interpolation */                  /* Interpolation */
466                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
467                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
468                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
         }  
469    
470          if ((cpu_flags & XVID_CPU_3DNOWEXT) > 0) {                  interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
471                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
472                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
473            }
474    
475                  /* Inverse DCT */          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
                 idct =  idct_3dne;  
476    
477                  /* Buffer transfer */                  /* Buffer transfer */
478                  transfer_8to16copy =  transfer_8to16copy_3dne;                  transfer_8to16copy =  transfer_8to16copy_3dne;
479                  transfer_16to8copy = transfer_16to8copy_3dne;                  transfer_16to8copy = transfer_16to8copy_3dne;
480                  transfer_8to16sub =  transfer_8to16sub_3dne;                  transfer_8to16sub =  transfer_8to16sub_3dne;
481                  transfer_8to16sub2 =  transfer_8to16sub2_3dne;                  transfer_8to16subro =  transfer_8to16subro_3dne;
482                  transfer_16to8add = transfer_16to8add_3dne;                  transfer_16to8add = transfer_16to8add_3dne;
483                  transfer8x8_copy = transfer8x8_copy_3dne;                  transfer8x8_copy = transfer8x8_copy_3dne;
484                    transfer8x4_copy = transfer8x4_copy_3dne;
485    
486                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
487                            /* Inverse DCT */
488                            idct =  idct_3dne;
489    
490                            /* Buffer transfer */
491                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
492    
493                            /* Interpolation */
494                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
495                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
496                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
497    
498                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
499                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
500                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
501    
502                  /* Quantization */                  /* Quantization */
503                  dequant4_intra = dequant4_intra_3dne;                          quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
504                  dequant4_inter = dequant4_inter_3dne;                          quant_h263_inter = quant_h263_inter_3dne;
505                  quant_intra = quant_intra_3dne;                          dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
506                  quant_inter = quant_inter_3dne;                          dequant_mpeg_inter = dequant_mpeg_inter_3dne;
507                  dequant_intra = dequant_intra_3dne;                          dequant_h263_intra = dequant_h263_intra_3dne;
508                  dequant_inter = dequant_inter_3dne;                          dequant_h263_inter = dequant_h263_inter_3dne;
509    
510                  /* ME functions */                  /* ME functions */
511                  calc_cbp = calc_cbp_3dne;                  calc_cbp = calc_cbp_3dne;
512    
513                  sad16 = sad16_3dne;                  sad16 = sad16_3dne;
514                  sad8 = sad8_3dne;                  sad8 = sad8_3dne;
515                  sad16bi = sad16bi_3dne;                  sad16bi = sad16bi_3dne;
516                  sad8bi = sad8bi_3dne;                  sad8bi = sad8bi_3dne;
517                  dev16 = dev16_3dne;                  dev16 = dev16_3dne;
518                    }
                 /* Interpolation */  
                 interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;  
                 interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;  
                 interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;  
519          }          }
520    
521            if ((cpu_flags & XVID_CPU_SSE2)) {
         if ((cpu_flags & XVID_CPU_SSE2) > 0) {  
 #ifdef EXPERIMENTAL_SSE2_CODE  
522    
523                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
524    
525                  /* Quantization */                  /* Quantization */
526                  quant_intra   = quant_intra_sse2;                  quant_h263_intra   = quant_h263_intra_sse2;
527                  dequant_intra = dequant_intra_sse2;                  quant_h263_inter   = quant_h263_inter_sse2;
528                  quant_inter   = quant_inter_sse2;                  dequant_h263_intra = dequant_h263_intra_sse2;
529                  dequant_inter = dequant_inter_sse2;                  dequant_h263_inter = dequant_h263_inter_sse2;
530    
531                  /* ME */                  /* SAD operators */
532                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
533                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
534    
535                  /* Forward and Inverse DCT */                  /* DCT operators */
536                  idct  = idct_sse2;                  fdct = fdct_sse2_skal;
537                  fdct = fdct_sse2;      /* idct = idct_sse2_skal; */   /* Is now IEEE1180 and Walken compliant. Disabled until fully tested. */
 #endif  
         }  
538    
539  #endif                  /* postprocessing */
540                    image_brightness = image_brightness_sse2;
541            }
542    #endif /* ARCH_IS_IA32 */
543    
544  #ifdef ARCH_IA64  #if defined(ARCH_IS_IA64)
545          if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?          if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
546            idct_ia64_init();            idct_ia64_init();
547            fdct = fdct_ia64;            fdct = fdct_ia64;
548            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
549            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
550            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
551            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 515  Line 553 
553            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
554            sad8 = sad8_ia64;            sad8 = sad8_ia64;
555            dev16 = dev16_ia64;            dev16 = dev16_ia64;
556  //        Halfpel8_Refine = Halfpel8_Refine_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
557            quant_intra = quant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
558            dequant_intra = dequant_intra_ia64;            quant_h263_inter = quant_h263_inter_ia64;
559            quant_inter = quant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
560            dequant_inter = dequant_inter_ia64;            dequant_h263_inter = dequant_h263_inter_ia64;
561            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
562            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
563            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
564            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
565            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
566            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
567          }          }
568  #endif  #endif
569    
570  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
571  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
572          calc_cbp = calc_cbp_altivec;            /* sad operators */
573          fdct = fdct_altivec;                    sad16 = sad16_altivec_c;
574          idct = idct_altivec;                    sad16bi = sad16bi_altivec_c;
575          sadInit = sadInit_altivec;                    sad8 = sad8_altivec_c;
576          sad16 = sad16_altivec;                    dev16 = dev16_altivec_c;
577          sad8 = sad8_altivec;  
578          dev16 = dev16_altivec;            sse8_16bit = sse8_16bit_altivec_c;
579  #else  
580          calc_cbp = calc_cbp_ppc;            /* mem transfer */
581  #endif            transfer_8to16copy = transfer_8to16copy_altivec_c;
582  #endif            transfer_16to8copy = transfer_16to8copy_altivec_c;
583              transfer_8to16sub = transfer_8to16sub_altivec_c;
584          return XVID_ERR_OK;            transfer_8to16subro = transfer_8to16subro_altivec_c;
585  }            transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
586              transfer_16to8add = transfer_16to8add_altivec_c;
587              transfer8x8_copy = transfer8x8_copy_altivec_c;
588    
589              /* Inverse DCT */
590              idct = idct_altivec_c;
591    
592  static int            /* Interpolation */
593  xvid_init_convert(XVID_INIT_CONVERTINFO* convert)            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
594  {            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
595          // const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
596          const int width = convert->width;  
597          const int height = convert->height;            interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
598          const int width2 = convert->width/2;            interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
599          const int height2 = convert->height/2;  
600          IMAGE img;                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
601                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
602          switch (convert->input.colorspace & ~XVID_CSP_VFLIP)                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
603          {                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
604                  case XVID_CSP_YV12 :  
605                          img.y = convert->input.y;            /* Colorspace conversion */
606                          img.v = (uint8_t*)convert->input.y + width*height;            bgra_to_yv12 = bgra_to_yv12_altivec_c;
607                          img.u = (uint8_t*)convert->input.y + width*height + width2*height2;            abgr_to_yv12 = abgr_to_yv12_altivec_c;
608                          image_output(&img, width, height, width,            rgba_to_yv12 = rgba_to_yv12_altivec_c;
609                                                  convert->output.y, convert->output.y_stride,            argb_to_yv12 = argb_to_yv12_altivec_c;
                                                 convert->output.colorspace, convert->interlacing);  
                         break;  
610    
611                  default :            yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
612                          return XVID_ERR_FORMAT;            uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
         }  
613    
614              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
615              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
616    
617          emms();            /* Quantization */
618          return XVID_ERR_OK;            quant_h263_intra = quant_h263_intra_altivec_c;
619              quant_h263_inter = quant_h263_inter_altivec_c;
620              dequant_h263_intra = dequant_h263_intra_altivec_c;
621              dequant_h263_inter = dequant_h263_inter_altivec_c;
622    
623                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
624                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
625    
626                      /* Qpel stuff */
627                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
628                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
629  }  }
630    #endif
631    
632    #if defined(ARCH_IS_X86_64)
633            /* For now, only XVID_CPU_ASM is looked for, so user can still
634             * disable asm usage the usual way. When Intel EMT64 cpus will
635             * be out, maybe we'll have to check more precisely what cpu
636             * features there really are. */
637            if (cpu_flags & XVID_CPU_ASM) {
638                    /* SIMD state flusher */
639                    emms = emms_mmx;
640    
641                    /* DCT operators */
642                    fdct = fdct_skal_x86_64;
643                    idct = idct_x86_64;
644    
645                    /* SAD operators */
646                    sad16      = sad16_x86_64;
647                    sad8       = sad8_x86_64;
648                    sad16bi    = sad16bi_x86_64;
649                    sad8bi     = sad8bi_x86_64;
650                    dev16      = dev16_x86_64;
651                    sad16v     = sad16v_x86_64;
652                    sse8_16bit = sse8_16bit_x86_64;
653                    sse8_8bit  = sse8_8bit_x86_64;
654    
655                    /* Interpolation operators */
656                    interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_x86_64;
657                    interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_x86_64;
658                    interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_x86_64;
659    
660                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_x86_64;
661                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_x86_64;
662                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_x86_64;
663                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_x86_64;
664    
665  void fill8(uint8_t * block, int size, int value)                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_x86_64;
666  {                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_x86_64;
         int i;  
         for (i = 0; i < size; i++)  
                 block[i] = value;  
 }  
667    
668  void fill16(int16_t * block, int size, int value)                  interpolate8x8_avg2 = interpolate8x8_avg2_x86_64;
669  {                  interpolate8x8_avg4 = interpolate8x8_avg4_x86_64;
         int i;  
         for (i = 0; i < size; i++)  
                 block[i] = value;  
 }  
670    
671  #define RANDOM(min,max) min + (rand() % (max-min))                  /* Quantization related functions */
672                    quant_h263_intra   = quant_h263_intra_x86_64;
673                    quant_h263_inter   = quant_h263_inter_x86_64;
674                    dequant_h263_intra = dequant_h263_intra_x86_64;
675                    dequant_h263_inter = dequant_h263_inter_x86_64;
676                    quant_mpeg_intra   = quant_mpeg_intra_x86_64;
677                    quant_mpeg_inter   = quant_mpeg_inter_x86_64;
678                    dequant_mpeg_intra   = dequant_mpeg_intra_x86_64;
679                    dequant_mpeg_inter   = dequant_mpeg_inter_x86_64;
680    
681  void random8(uint8_t * block, int size, int min, int max)                  /* Block related functions */
682  {                  transfer_8to16copy  = transfer_8to16copy_x86_64;
683          int i;                  transfer_16to8copy  = transfer_16to8copy_x86_64;
684          for (i = 0; i < size; i++)                  transfer_8to16sub   = transfer_8to16sub_x86_64;
685                  block[i] = RANDOM(min,max);                  transfer_8to16subro = transfer_8to16subro_x86_64;
686  }                  transfer_8to16sub2  = transfer_8to16sub2_x86_64;
687                    transfer_8to16sub2ro= transfer_8to16sub2ro_x86_64;
688                    transfer_16to8add   = transfer_16to8add_x86_64;
689                    transfer8x8_copy    = transfer8x8_copy_x86_64;
690    
691                    /* Qpel stuff */
692                    xvid_QP_Funcs = &xvid_QP_Funcs_x86_64;
693                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_x86_64;
694    
695  void random16(int16_t * block, int size, int min, int max)                  /* Interlacing Functions */
696  {                  MBFieldTest = MBFieldTest_x86_64;
         int i;  
         for (i = 0; i < size; i++)  
                 block[i] = RANDOM(min,max);  
697  }  }
698    #endif
699    
700  int compare16(const int16_t * blockA, const int16_t * blockB, int size)  #if defined(_DEBUG)
701  {      xvid_debug = init->debug;
702          int i;  #endif
         for (i = 0; i < size; i++)  
                 if (blockA[i] != blockB[i])  
                         return 1;  
703    
704          return 0;      return(0);
705  }  }
706    
707    
708    static int
709  int test_h263_intra(quanth263_intraFunc * funcA, quanth263_intraFunc * funcB,  xvid_gbl_info(xvid_gbl_info_t * info)
                                                 const char * nameA, const char * nameB,  
                                                 int min, int max)  
 {  
         int q,i;  
         int64_t timeSTART;  
         int64_t timeA = 0;  
         int64_t timeB = 0;  
         DECLARE_ALIGNED_MATRIX(arrayX, 1, 64, int16_t, CACHE_LINE);  
         DECLARE_ALIGNED_MATRIX(arrayA, 1, 64, int16_t, CACHE_LINE);  
         DECLARE_ALIGNED_MATRIX(arrayB, 1, 64, int16_t, CACHE_LINE);  
   
         for (q = 1; q <= 31; q++)       /* quantizer */  
         {  
                 for (i = min; i < max; i++)     /* input coeff */  
710                  {                  {
711                          fill16(arrayX, 64, i);          if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
712                    return XVID_ERR_VERSION;
713    
714                          timeSTART = read_counter();          info->actual_version = XVID_VERSION;
715                          funcA(arrayA, arrayX, q, q);          info->build = "xvid-1.1.1;
716                          timeA += read_counter() - timeSTART;          info->cpu_flags = detect_cpu_flags();
717    
718                          timeSTART = read_counter();  #if defined(_SMP) && defined(WIN32)
719                          funcB(arrayB, arrayX, q, q);          info->num_threads = pthread_num_processors_np();;
720                          timeB += read_counter() - timeSTART;  #else
721            info->num_threads = 0;
722                          if (compare16(arrayA, arrayB, 64))  #endif
                         {  
                                 printf("%s/%s error: q=%i, i=%i\n", nameA?nameA:"?", nameB?nameB:"?", q, i);  
                                 return 0;  
                         }  
                 }  
         }  
   
         if (nameA) printf("%s:\t%I64i\n", nameA, timeA);  
         if (nameB) printf("%s:\t%I64i\n", nameB, timeB);  
723    
724          return 0;          return 0;
725  }  }
726    
 int test_h263_inter(quanth263_interFunc * funcA, quanth263_interFunc * funcB,  
                                                 const char * nameA, const char * nameB,  
                                                 int min, int max)  
 {  
         int q,i;  
         int64_t timeSTART;  
         int64_t timeA = 0;  
         int64_t timeB = 0;  
         DECLARE_ALIGNED_MATRIX(arrayX, 1, 64, int16_t, CACHE_LINE);  
         DECLARE_ALIGNED_MATRIX(arrayA, 1, 64, int16_t, CACHE_LINE);  
         DECLARE_ALIGNED_MATRIX(arrayB, 1, 64, int16_t, CACHE_LINE);  
727    
728          for (q = 1; q <= 31; q++)       /* quantizer */  static int
729          {  xvid_gbl_convert(xvid_gbl_convert_t* convert)
                 for (i = min; i < max; i++)     /* input coeff */  
730                  {                  {
731                          fill16(arrayX, 64, i);          int width;
732            int height;
733            int width2;
734            int height2;
735            IMAGE img;
736    
737                          timeSTART = read_counter();          if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
738                          funcA(arrayA, arrayX, q);                return XVID_ERR_VERSION;
                         timeA += read_counter() - timeSTART;  
739    
740                          timeSTART = read_counter();  #if 0
741                          funcB(arrayB, arrayX, q);          const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
742                          timeB += read_counter() - timeSTART;  #endif
743            width = convert->width;
744            height = convert->height;
745            width2 = convert->width/2;
746            height2 = convert->height/2;
747    
748                          if (compare16(arrayA, arrayB, 64))          switch (convert->input.csp & ~XVID_CSP_VFLIP)
749                          {                          {
750                                  printf("%s/%s error: q=%i, i=%i\n", nameA?nameA:"?", nameB?nameB:"?", q, i);                  case XVID_CSP_YV12 :
751                                  return 0;                          img.y = convert->input.plane[0];
752                          }                          img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
753                  }                          img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
754                            image_output(&img, width, height, width,
755                                                    (uint8_t**)convert->output.plane, convert->output.stride,
756                                                    convert->output.csp, convert->interlacing);
757                            break;
758    
759                    default :
760                            return XVID_ERR_FORMAT;
761          }          }
762    
         if (nameA) printf("%s:\t%I64i\n", nameA, timeA);  
         if (nameB) printf("%s:\t%I64i\n", nameB, timeB);  
763    
764            emms();
765          return 0;          return 0;
766  }  }
767    
768    /*****************************************************************************
769     * XviD Global Entry point
770  int xvid_init_test()   *
771  {   * Well this function initialize all internal function pointers according
772          int cpu_flags;   * to the CPU features forced by the library client or autodetected (depending
773     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
774          printf("xvid_init_test\n");   * image colorspace transformation tables.
775     *
776  #if defined(ARCH_X86)   ****************************************************************************/
         cpu_flags = check_cpu_features();  
   
         emms_mmx();  
   
         printf("--- quant intra ---\n");  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_h263_intra(quant_intra_c, quant_intra_mmx, "c", "mmx", -2048, 2047);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_h263_intra(quant_intra_c, quant_intra_3dne, NULL, "3dne", -2048, 2047);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_h263_intra(quant_intra_c, quant_intra_sse2, NULL, "sse2", -2048, 2047);  
   
         printf("\n--- quant inter ---\n");  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_h263_inter(quant_inter_c, quant_inter_mmx, "c", "mmx", -2048, 2047);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_h263_inter(quant_inter_c, quant_inter_3dne, NULL, "3dne", -2048, 2047);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_h263_inter(quant_inter_c, quant_inter_sse2, NULL, "sse2", -2048, 2047);  
   
         printf("\n--- dequan intra ---\n");  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_h263_intra(dequant_intra_c, dequant_intra_mmx, "c", "mmx", -256, 255);  
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_h263_intra(dequant_intra_c, dequant_intra_xmm, NULL, "xmm", -256, 255);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_h263_intra(dequant_intra_c, dequant_intra_3dne, NULL, "3dne", -256, 255);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_h263_intra(dequant_intra_c, dequant_intra_sse2, NULL, "sse2", -256, 255);  
   
         printf("\n--- dequant inter ---\n");  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_h263_inter((quanth263_interFunc*)dequant_inter_c,  
                                                 (quanth263_interFunc*)dequant_inter_mmx, "c", "mmx", -256, 255);  
   
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_h263_inter((quanth263_interFunc*)dequant_inter_c,  
                                                 (quanth263_interFunc*)dequant_inter_xmm, NULL, "xmm", -256, 255);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_h263_inter((quanth263_interFunc*)dequant_inter_c,  
                                                 (quanth263_interFunc*)dequant_inter_3dne, NULL, "3dne", -256, 255);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_h263_inter((quanth263_interFunc*)dequant_inter_c,  
                                                 (quanth263_interFunc*)dequant_inter_sse2, NULL, "sse2", -256, 255);  
   
         printf("\n--- quant4_intra ---\n");  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_h263_intra((quanth263_intraFunc*)quant4_intra_c,  
                                                 (quanth263_intraFunc*)quant4_intra_mmx, "c", "mmx", -2048, 2047);  
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_h263_intra((quanth263_intraFunc*)quant4_intra_c,  
                                                 (quanth263_intraFunc*)quant4_intra_xmm, NULL, "xmm", -2048, 2047);  
   
         printf("\n--- quant4_inter ---\n");  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_h263_inter((quanth263_interFunc*)quant4_inter_c,  
                                                 (quanth263_interFunc*)quant4_inter_mmx, "c", "mmx", -2048, 2047);  
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_h263_inter((quanth263_interFunc*)quant4_inter_c,  
                                                 (quanth263_interFunc*)quant4_inter_xmm, NULL, "xmm", -2048, 2047);  
   
   
         printf("\n--- dequant4_intra ---\n");  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_h263_intra((quanth263_intraFunc*)dequant4_intra_c,  
                                                 (quanth263_intraFunc*)dequant4_intra_mmx, "c", "mmx", -256, 255);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_h263_intra((quanth263_intraFunc*)dequant4_intra_c,  
                                                 (quanth263_intraFunc*)dequant4_intra_3dne, NULL, "sse2", -256, 255);  
   
         printf("\n--- dequant4_inter ---\n");  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_h263_inter((quanth263_interFunc*)dequant4_inter_c,  
                                                 (quanth263_interFunc*)dequant4_inter_mmx, "c", "mmx", -256, 255);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_h263_inter((quanth263_interFunc*)dequant4_inter_c,  
                                                 (quanth263_interFunc*)dequant4_inter_3dne, NULL, "sse2", -256, 255);  
   
         emms_mmx();  
   
 #endif  
   
         return XVID_ERR_OK;  
 }  
777    
778    
779  int  int
780  xvid_init(void *handle,  xvid_global(void *handle,
781                    int opt,                    int opt,
782                    void *param1,                    void *param1,
783                    void *param2)                    void *param2)
784  {  {
785          switch(opt)          switch(opt)
786          {          {
787                  case XVID_INIT_INIT :                  case XVID_GBL_INIT :
788                          return xvid_init_init((XVID_INIT_PARAM*)param1);                          return xvid_gbl_init((xvid_gbl_init_t*)param1);
789    
790                  case XVID_INIT_CONVERT :          case XVID_GBL_INFO :
791                          return xvid_init_convert((XVID_INIT_CONVERTINFO*)param1);              return xvid_gbl_info((xvid_gbl_info_t*)param1);
792    
793                  case XVID_INIT_TEST :                  case XVID_GBL_CONVERT :
794                          return xvid_init_test();                          return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
795    
796                  default :                  default :
797                          return XVID_ERR_FAIL;                          return XVID_ERR_FAIL;
# Line 837  Line 815 
815                          void *param2)                          void *param2)
816  {  {
817          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1, (XVID_DEC_STATS*) param2);  
   
818          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
819                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
820    
821          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
822                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
823    
824            case XVID_DEC_DECODE:
825                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
826    
827          default:          default:
828                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
829          }          }
# Line 871  Line 849 
849          switch (opt) {          switch (opt) {
850          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
851    
852                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
853                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
854                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
855    
856          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
857                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
858    
859          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
860                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
861    
862          default:          default:
863                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
Legend:
Removed from v.1.33.2.19  
changed lines
  Added in v.1.65.2.3
No admin address has been configured
 ViewVC Help
Powered by ViewVC 1.0.4