[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 1.33.2.23, Wed Feb 12 11:46:18 2003 UTC revision 1.86, Fri Dec 31 11:30:38 2010 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>
7   *  Video tools as specified in ISO/IEC 14496-2 standard.  Those intending   *               2002-2010 Michael Militzer <isibaar@xvid.org>
  *  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.  
8   *   *
9   *  This program is free software ; you can redistribute it and/or modify   *  This program is free software ; you can redistribute it and/or modify
10   *  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 20 
20   *  along with this program ; if not, write to the Free Software   *  along with this program ; if not, write to the Free Software
21   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
22   *   *
  ****************************************************************************/  
   
 /*****************************************************************************  
  *  
  *  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>  
  *  
23   *  $Id$   *  $Id$
24   *   *
25   ****************************************************************************/   ****************************************************************************/
# Line 46  Line 29 
29  #include <string.h>  #include <string.h>
30  #include <time.h>  #include <time.h>
31    
32    #if !defined(_WIN32)
33    #include <unistd.h>
34    #endif
35    
36    #if defined(__APPLE__) && defined(__MACH__) && !defined(_SC_NPROCESSORS_CONF)
37    #include <sys/types.h>
38    #include <sys/sysctl.h>
39    #ifdef MAX
40    #undef MAX
41    #endif
42    #ifdef MIN
43    #undef MIN
44    #endif
45    #endif
46    
47    #if defined(__amigaos4__)
48    #include <exec/exec.h>
49    #include <proto/exec.h>
50    #endif
51    
52  #include "xvid.h"  #include "xvid.h"
53  #include "decoder.h"  #include "decoder.h"
54  #include "encoder.h"  #include "encoder.h"
# Line 54  Line 57 
57  #include "dct/fdct.h"  #include "dct/fdct.h"
58  #include "image/colorspace.h"  #include "image/colorspace.h"
59  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
 #include "image/reduced.h"  
60  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
61  #include "utils/mbfunctions.h"  #include "utils/mbfunctions.h"
62  #include "quant/quant_h263.h"  #include "quant/quant.h"
 #include "quant/quant_mpeg4.h"  
63  #include "motion/motion.h"  #include "motion/motion.h"
64    #include "motion/gmc.h"
65  #include "motion/sad.h"  #include "motion/sad.h"
66  #include "utils/emms.h"  #include "utils/emms.h"
67  #include "utils/timer.h"  #include "utils/timer.h"
68  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
69    #include "image/qpel.h"
70    #include "image/postprocessing.h"
71    
72  #if defined(ARCH_X86)  #if defined(_DEBUG)
73    unsigned int xvid_debug = 0; /* xvid debug mask */
74    #endif
75    
76  #if defined(_MSC_VER)  #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
77  #       include <windows.h>  #       include <windows.h>
78  #else  #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || (defined(ARCH_IS_PPC) && !defined(__amigaos4__))
79  #       include <signal.h>  #       include <signal.h>
80  #       include <setjmp.h>  #       include <setjmp.h>
81    
# Line 82  Line 88 
88          }          }
89  #endif  #endif
90    
   
91  /*  /*
92  calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled   * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
93  return values:   * signalled
94  -1 : could not determine   *
95  0  : SIGILL was *not* signalled   * Return values:
96  1  : SIGILL was signalled   *  -1 : could not determine
97     *   0 : SIGILL was *not* signalled
98     *   1 : SIGILL was signalled
99  */  */
100    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
101  int  static int
102  sigill_check(void (*func)())  sigill_check(void (*func)())
103  {  {
 #if defined(_MSC_VER)  
104          _try {          _try {
105                  func();                  func();
106          }          } _except(EXCEPTION_EXECUTE_HANDLER) {
         _except(EXCEPTION_EXECUTE_HANDLER) {  
107    
108                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
109                          return 1;                          return(1);
110          }          }
111          return 0;          return(0);
112  #else  }
113    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || (defined(ARCH_IS_PPC) && !defined(__amigaos4__))
114    static int
115    sigill_check(void (*func)())
116    {
117      void * old_handler;      void * old_handler;
118      int jmpret;      int jmpret;
119    
120        /* Set our SIGILL handler */
121      old_handler = signal(SIGILL, sigill_handler);      old_handler = signal(SIGILL, sigill_handler);
122      if (old_handler == SIG_ERR)  
123      {      /* Check for error */
124          return -1;      if (old_handler == SIG_ERR) {
125            return(-1);
126      }      }
127    
128        /* Save stack context, so if func triggers a SIGILL, we can still roll
129             * back to a valid CPU state */
130      jmpret = setjmp(mark);      jmpret = setjmp(mark);
131      if (jmpret == 0)  
132      {          /* If setjmp returned directly, then its returned value is 0, and we still
133             * have to test the passed func. Otherwise it means the stack context has
134             * been restored by a longjmp() call, which in our case happens only in the
135             * signal handler */
136        if (jmpret == 0) {
137          func();          func();
138      }      }
139    
140        /* Restore old signal handler */
141      signal(SIGILL, old_handler);      signal(SIGILL, old_handler);
142    
143      return jmpret;      return(jmpret);
 #endif  
144  }  }
145  #endif  #endif
146    
147    
148  /* detect cpu flags  */  /* detect cpu flags  */
149  static unsigned int  static unsigned int
150  detect_cpu_flags()  detect_cpu_flags(void)
151  {  {
152          /* enable native assembly optimizations by default */          /* enable native assembly optimizations by default */
153          unsigned int cpu_flags = XVID_CPU_ASM;          unsigned int cpu_flags = XVID_CPU_ASM;
154    
155  #if defined(ARCH_X86)  #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
156          cpu_flags |= check_cpu_features();          cpu_flags |= check_cpu_features();
157          if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))          if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
158                  cpu_flags &= ~XVID_CPU_SSE;                  cpu_flags &= ~XVID_CPU_SSE;
159    
160          if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))          if ((cpu_flags & (XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41)) && sigill_check(sse2_os_trigger))
161                  cpu_flags &= ~XVID_CPU_SSE2;                  cpu_flags &= ~(XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41);
162  #endif  #endif
163    
164  #if defined(ARCH_PPC)  #if defined(ARCH_IS_PPC)
165  #if defined(ARCH_PPC_ALTIVEC)  #if defined(__amigaos4__)
166            {
167                    uint32_t vector_unit = VECTORTYPE_NONE;
168                    IExec->GetCPUInfoTags(GCIT_VectorUnit, &vector_unit, TAG_END);
169                    if (vector_unit == VECTORTYPE_ALTIVEC) {
170                            cpu_flags |= XVID_CPU_ALTIVEC;
171                    }
172            }
173    #else
174            if (!sigill_check(altivec_trigger))
175          cpu_flags |= XVID_CPU_ALTIVEC;          cpu_flags |= XVID_CPU_ALTIVEC;
176  #endif  #endif
177  #endif  #endif
# Line 156  Line 181 
181    
182    
183  /*****************************************************************************  /*****************************************************************************
184   * XviD Init Entry point   * Xvid Init Entry point
185   *   *
186   * Well this function initialize all internal function pointers according   * Well this function initialize all internal function pointers according
187   * to the CPU features forced by the library client or autodetected (depending   * to the CPU features forced by the library client or autodetected (depending
# Line 171  Line 196 
196    
197    
198  static  static
199  int xvid_init_init(XVID_INIT_PARAM * init_param)  int xvid_gbl_init(xvid_gbl_init_t * init)
200  {  {
201          int cpu_flags;          unsigned int cpu_flags;
202    
203          /* Inform the client the API version */          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
204          init_param->api_version = API_VERSION;                  return XVID_ERR_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 = detect_cpu_flags();  
         }  
   
         if ((init_param->cpu_flags & XVID_CPU_CHKONLY))  
         {  
                 init_param->cpu_flags = cpu_flags;  
                 return XVID_ERR_OK;  
         }  
   
         init_param->cpu_flags = cpu_flags;  
205    
206            cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
207    
208          /* Initialize the function pointers */          /* Initialize the function pointers */
209          idct_int32_init();          idct_int32_init();
# Line 209  Line 214 
214          idct = idct_int32;          idct = idct_int32;
215    
216          /* Only needed on PPC Altivec archs */          /* Only needed on PPC Altivec archs */
217          sadInit = 0;          sadInit = NULL;
218    
219          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
220          emms = emms_c;          emms = emms_c;
221    
222            /* Qpel stuff */
223            xvid_QP_Funcs = &xvid_QP_Funcs_C;
224            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
225            xvid_Init_QP();
226    
227          /* Quantization functions */          /* Quantization functions */
228          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
229          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
230          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
231          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
232    
233          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
234          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
235          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
236          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
237    
238          /* Block transfer related functions */          /* Block transfer related functions */
239          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
# Line 231  Line 241 
241          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
242          transfer_8to16subro  = transfer_8to16subro_c;          transfer_8to16subro  = transfer_8to16subro_c;
243          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
244            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
245          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
246          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
247            transfer8x4_copy   = transfer8x4_copy_c;
248    
249          /* Interlacing functions */          /* Interlacing functions */
250          MBFieldTest = MBFieldTest_c;          MBFieldTest = MBFieldTest_c;
# Line 242  Line 254 
254          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
255          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
256    
257            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
258            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
259            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
260    
261            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
262            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
263            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
264            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
265    
266          interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;          interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
267          interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;          interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
268          interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;          interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
# Line 256  Line 277 
277          interpolate8x8_avg2 = interpolate8x8_avg2_c;          interpolate8x8_avg2 = interpolate8x8_avg2_c;
278          interpolate8x8_avg4 = interpolate8x8_avg4_c;          interpolate8x8_avg4 = interpolate8x8_avg4_c;
279    
280          /* reduced resoltuion */          /* postprocessing */
281          copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C;          image_brightness = image_brightness_c;
         add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C;  
         vfilter_31 = xvid_VFilter_31_C;  
         hfilter_31 = xvid_HFilter_31_C;  
         filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C;  
         filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C;  
282    
283          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
284          colorspace_init();          colorspace_init();
# Line 271  Line 287 
287          yv12_to_yv12    = yv12_to_yv12_c;          yv12_to_yv12    = yv12_to_yv12_c;
288          rgb555_to_yv12  = rgb555_to_yv12_c;          rgb555_to_yv12  = rgb555_to_yv12_c;
289          rgb565_to_yv12  = rgb565_to_yv12_c;          rgb565_to_yv12  = rgb565_to_yv12_c;
290            rgb_to_yv12     = rgb_to_yv12_c;
291          bgr_to_yv12     = bgr_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
292          bgra_to_yv12    = bgra_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
293          abgr_to_yv12    = abgr_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
294          rgba_to_yv12    = rgba_to_yv12_c;          rgba_to_yv12    = rgba_to_yv12_c;
295            argb_to_yv12    = argb_to_yv12_c;
296          yuyv_to_yv12    = yuyv_to_yv12_c;          yuyv_to_yv12    = yuyv_to_yv12_c;
297          uyvy_to_yv12    = uyvy_to_yv12_c;          uyvy_to_yv12    = uyvy_to_yv12_c;
298    
# Line 284  Line 302 
302          bgrai_to_yv12   = bgrai_to_yv12_c;          bgrai_to_yv12   = bgrai_to_yv12_c;
303          abgri_to_yv12   = abgri_to_yv12_c;          abgri_to_yv12   = abgri_to_yv12_c;
304          rgbai_to_yv12   = rgbai_to_yv12_c;          rgbai_to_yv12   = rgbai_to_yv12_c;
305            argbi_to_yv12   = argbi_to_yv12_c;
306          yuyvi_to_yv12   = yuyvi_to_yv12_c;          yuyvi_to_yv12   = yuyvi_to_yv12_c;
307          uyvyi_to_yv12   = uyvyi_to_yv12_c;          uyvyi_to_yv12   = uyvyi_to_yv12_c;
308    
   
309          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
310          yv12_to_rgb555  = yv12_to_rgb555_c;          yv12_to_rgb555  = yv12_to_rgb555_c;
311          yv12_to_rgb565  = yv12_to_rgb565_c;          yv12_to_rgb565  = yv12_to_rgb565_c;
312            yv12_to_rgb     = yv12_to_rgb_c;
313          yv12_to_bgr     = yv12_to_bgr_c;          yv12_to_bgr     = yv12_to_bgr_c;
314          yv12_to_bgra    = yv12_to_bgra_c;          yv12_to_bgra    = yv12_to_bgra_c;
315          yv12_to_abgr    = yv12_to_abgr_c;          yv12_to_abgr    = yv12_to_abgr_c;
316          yv12_to_rgba    = yv12_to_rgba_c;          yv12_to_rgba    = yv12_to_rgba_c;
317            yv12_to_argb    = yv12_to_argb_c;
318          yv12_to_yuyv    = yv12_to_yuyv_c;          yv12_to_yuyv    = yv12_to_yuyv_c;
319          yv12_to_uyvy    = yv12_to_uyvy_c;          yv12_to_uyvy    = yv12_to_uyvy_c;
320    
# Line 304  Line 324 
324          yv12_to_bgrai   = yv12_to_bgrai_c;          yv12_to_bgrai   = yv12_to_bgrai_c;
325          yv12_to_abgri   = yv12_to_abgri_c;          yv12_to_abgri   = yv12_to_abgri_c;
326          yv12_to_rgbai   = yv12_to_rgbai_c;          yv12_to_rgbai   = yv12_to_rgbai_c;
327            yv12_to_argbi   = yv12_to_argbi_c;
328          yv12_to_yuyvi   = yv12_to_yuyvi_c;          yv12_to_yuyvi   = yv12_to_yuyvi_c;
329          yv12_to_uyvyi   = yv12_to_uyvyi_c;          yv12_to_uyvyi   = yv12_to_uyvyi_c;
330    
# Line 315  Line 336 
336          sad8bi   = sad8bi_c;          sad8bi   = sad8bi_c;
337          dev16    = dev16_c;          dev16    = dev16_c;
338          sad16v   = sad16v_c;          sad16v   = sad16v_c;
339            sse8_16bit    = sse8_16bit_c;
340            sse8_8bit     = sse8_8bit_c;
341    
342  //      Halfpel8_Refine = Halfpel8_Refine_c;          sseh8_16bit   = sseh8_16bit_c;
343            coeff8_energy = coeff8_energy_c;
344            blocksum8     = blocksum8_c;
345    
346  #if defined(ARCH_X86)          init_GMC(cpu_flags);
347    
348          if ((cpu_flags & XVID_CPU_ASM))  #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
         {  
                 vfilter_31 = xvid_VFilter_31_x86;  
                 hfilter_31 = xvid_HFilter_31_x86;  
         }  
349    
350          if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||          if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
351                  (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||                  (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
352                  (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2))                  (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2) ||
353            (cpu_flags & XVID_CPU_SSE3) || (cpu_flags & XVID_CPU_SSE41))
354          {          {
355                  /* Restore FPU context : emms_c is a nop functions */                  /* Restore FPU context : emms_c is a nop functions */
356                  emms = emms_mmx;                  emms = emms_mmx;
# Line 337  Line 359 
359          if ((cpu_flags & XVID_CPU_MMX)) {          if ((cpu_flags & XVID_CPU_MMX)) {
360    
361                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
362                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
363                  idct = idct_mmx;                  idct = idct_mmx;
364    
365                    /* Qpel stuff */
366                    xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
367                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
368    
369                  /* Quantization related functions */                  /* Quantization related functions */
370                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
371                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
372                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
373                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
374                    quant_mpeg_intra   = quant_mpeg_intra_mmx;
375                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
376                  dequant4_intra = dequant4_intra_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
377                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
378                  dequant4_inter = dequant4_inter_mmx;  
379    
380                  /* Block related functions */                  /* Block related functions */
381                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
# Line 359  Line 385 
385                  transfer_8to16sub2 = transfer_8to16sub2_mmx;                  transfer_8to16sub2 = transfer_8to16sub2_mmx;
386                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
387                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
388                    transfer8x4_copy   = transfer8x4_copy_mmx;
389    
390                  /* Interlacing Functions */                  /* Interlacing Functions */
391                  MBFieldTest = MBFieldTest_mmx;                  MBFieldTest = MBFieldTest_mmx;
# Line 368  Line 395 
395                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
396                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
397    
398                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
399                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
400                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
401    
402                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
403                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
404                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
405                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
406    
407                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
408                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
409    
410                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
411                  interpolate8x8_avg4 = interpolate8x8_avg4_mmx;                  interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
412    
413                  /* reduced resolution */                  /* postprocessing */
414                  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;  
415    
416                  /* image input xxx_to_yv12 related functions */                  /* image input xxx_to_yv12 related functions */
417    
418                  yv12_to_yv12  = yv12_to_yv12_mmx;                  yv12_to_yv12  = yv12_to_yv12_mmx;
419    
420                  bgr_to_yv12   = bgr_to_yv12_mmx;                  bgr_to_yv12   = bgr_to_yv12_mmx;
421                    rgb_to_yv12   = rgb_to_yv12_mmx;
422                  bgra_to_yv12  = bgra_to_yv12_mmx;                  bgra_to_yv12  = bgra_to_yv12_mmx;
423                    rgba_to_yv12  = rgba_to_yv12_mmx;
424                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
425                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
426    
# Line 405  Line 441 
441                  sad8bi  = sad8bi_mmx;                  sad8bi  = sad8bi_mmx;
442                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
443                  sad16v   = sad16v_mmx;                  sad16v   = sad16v_mmx;
444                    sse8_16bit = sse8_16bit_mmx;
445                    sse8_8bit  = sse8_8bit_mmx;
446          }          }
447    
448          /* these 3dnow functions are faster than mmx, but slower than xmm. */          /* these 3dnow functions are faster than mmx, but slower than xmm. */
# Line 418  Line 456 
456    
457                  yuyv_to_yv12  = yuyv_to_yv12_3dn;                  yuyv_to_yv12  = yuyv_to_yv12_3dn;
458                  uyvy_to_yv12  = uyvy_to_yv12_3dn;                  uyvy_to_yv12  = uyvy_to_yv12_3dn;
459    
460          }          }
461    
462    
463          if ((cpu_flags & XVID_CPU_MMXEXT)) {          if ((cpu_flags & XVID_CPU_MMXEXT)) {
464    
465                  /* Inverse DCT */                  /* DCT */
466                    fdct = fdct_xmm_skal;
467                  idct = idct_xmm;                  idct = idct_xmm;
468    
469                  /* Interpolation */                  /* Interpolation */
# Line 431  Line 471 
471                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
472                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
473    
474                  /* reduced resolution */                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
475                  copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
476                  add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
477    
478                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
479                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
480                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
481                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
482    
483                  /* Quantization */                  /* Quantization */
484                  quant4_intra = quant4_intra_xmm;                  quant_mpeg_inter = quant_mpeg_inter_xmm;
                 quant4_inter = quant4_inter_xmm;  
485    
486                  dequant_intra = dequant_intra_xmm;                  dequant_h263_intra = dequant_h263_intra_xmm;
487                  dequant_inter = dequant_inter_xmm;                  dequant_h263_inter = dequant_h263_inter_xmm;
488    
489                  /* Buffer transfer */                  /* Buffer transfer */
490                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
491                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
492    
493                  /* Colorspace transformation */                  /* Colorspace transformation */
494                  yv12_to_yv12  = yv12_to_yv12_xmm;                  /* yv12_to_yv12  = yv12_to_yv12_xmm; */ /* appears to be slow on many machines */
495                  yuyv_to_yv12  = yuyv_to_yv12_xmm;                  yuyv_to_yv12  = yuyv_to_yv12_xmm;
496                  uyvy_to_yv12  = uyvy_to_yv12_xmm;                  uyvy_to_yv12  = uyvy_to_yv12_xmm;
497    
# Line 465  Line 510 
510                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
511                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
512                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
513    
514                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
515                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
516                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
517          }          }
518    
519          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
520    
                 /* Inverse DCT */  
                 idct =  idct_3dne;  
   
521                  /* Buffer transfer */                  /* Buffer transfer */
522                  transfer_8to16copy =  transfer_8to16copy_3dne;                  transfer_8to16copy =  transfer_8to16copy_3dne;
523                  transfer_16to8copy = transfer_16to8copy_3dne;                  transfer_16to8copy = transfer_16to8copy_3dne;
524                  transfer_8to16sub =  transfer_8to16sub_3dne;                  transfer_8to16sub =  transfer_8to16sub_3dne;
525                  transfer_8to16subro =  transfer_8to16subro_3dne;                  transfer_8to16subro =  transfer_8to16subro_3dne;
                 transfer_8to16sub2 =  transfer_8to16sub2_3dne;  
526                  transfer_16to8add = transfer_16to8add_3dne;                  transfer_16to8add = transfer_16to8add_3dne;
527                  transfer8x8_copy = transfer8x8_copy_3dne;                  transfer8x8_copy = transfer8x8_copy_3dne;
528                    transfer8x4_copy = transfer8x4_copy_3dne;
529    
530                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
531                            /* Inverse DCT */
532                            idct =  idct_3dne;
533    
534                            /* Buffer transfer */
535                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
536    
537                            /* Interpolation */
538                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
539                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
540                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
541    
542                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
543                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
544                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
545    
546                  /* Quantization */                  /* Quantization */
547                  dequant4_intra = dequant4_intra_3dne;                          quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
548                  dequant4_inter = dequant4_inter_3dne;                          quant_h263_inter = quant_h263_inter_3dne;
549                  quant_intra = quant_intra_3dne;                          dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
550                  quant_inter = quant_inter_3dne;                          dequant_mpeg_inter = dequant_mpeg_inter_3dne;
551                  dequant_intra = dequant_intra_3dne;                          dequant_h263_intra = dequant_h263_intra_3dne;
552                  dequant_inter = dequant_inter_3dne;                          dequant_h263_inter = dequant_h263_inter_3dne;
553    
554                  /* ME functions */                  /* ME functions */
                 calc_cbp = calc_cbp_3dne;  
555                  sad16 = sad16_3dne;                  sad16 = sad16_3dne;
556                  sad8 = sad8_3dne;                  sad8 = sad8_3dne;
557                  sad16bi = sad16bi_3dne;                  sad16bi = sad16bi_3dne;
558                  sad8bi = sad8bi_3dne;                  sad8bi = sad8bi_3dne;
559                  dev16 = dev16_3dne;                  dev16 = dev16_3dne;
   
                 /* Interpolation */  
                 interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;  
                 interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;  
                 interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;  
560          }          }
561            }
562    
563          if ((cpu_flags & XVID_CPU_SSE2)) {          if ((cpu_flags & XVID_CPU_SSE2)) {
564    
565                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
566    
567                  /* Quantization */                  /* Quantization */
568                  quant_intra   = quant_intra_sse2;                  quant_h263_intra   = quant_h263_intra_sse2;
569                  dequant_intra = dequant_intra_sse2;                  quant_h263_inter   = quant_h263_inter_sse2;
570                  quant_inter   = quant_inter_sse2;                  dequant_h263_intra = dequant_h263_intra_sse2;
571                  dequant_inter = dequant_inter_sse2;                  dequant_h263_inter = dequant_h263_inter_sse2;
572    
573  #if defined(EXPERIMENTAL_SSE2_CODE)                  /* SAD operators */
                 /* ME; slower than xmm */  
574                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
575                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
576  #endif  
577                  /* Forward and Inverse DCT */                  /* PSNR-HVS-M distortion metric */
578                  idct  = idct_sse2;                  sseh8_16bit   = sseh8_16bit_sse2;
579                  fdct = fdct_sse2;                  coeff8_energy = coeff8_energy_sse2;
580                    blocksum8     = blocksum8_sse2;
581    
582                    /* DCT operators */
583                    fdct = fdct_sse2_skal;
584                    idct = idct_sse2_skal;   /* Is now IEEE1180 and Walken compliant. */
585    
586                    /* postprocessing */
587                    image_brightness = image_brightness_sse2;
588    
589            }
590    
591            if ((cpu_flags & XVID_CPU_SSE3)) {
592    
593                    /* SAD operators */
594                    sad16    = sad16_sse3;
595                    dev16    = dev16_sse3;
596          }          }
 #endif  
597    
598  #if defined(ARCH_IA64)  #endif /* ARCH_IS_IA32 */
599          if ((cpu_flags & XVID_CPU_ASM)) { //use assembler routines?  
600    #if defined(ARCH_IS_IA64)
601            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
602            idct_ia64_init();            idct_ia64_init();
603            fdct = fdct_ia64;            fdct = fdct_ia64;
604            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
605            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
606            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
607            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 537  Line 609 
609            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
610            sad8 = sad8_ia64;            sad8 = sad8_ia64;
611            dev16 = dev16_ia64;            dev16 = dev16_ia64;
612  //        Halfpel8_Refine = Halfpel8_Refine_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
613            quant_intra = quant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
614            dequant_intra = dequant_intra_ia64;            quant_h263_inter = quant_h263_inter_ia64;
615            quant_inter = quant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
616            dequant_inter = dequant_inter_ia64;            dequant_h263_inter = dequant_h263_inter_ia64;
617            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
618            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
619            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
620            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
621            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
622            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
623          }          }
624  #endif  #endif
625    
626  #if defined(ARCH_PPC)  #if defined(ARCH_IS_PPC)
627          if ((cpu_flags & XVID_CPU_ASM))          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
628          {            /* sad operators */
629                  calc_cbp = calc_cbp_ppc;                    sad16 = sad16_altivec_c;
630          }                    sad16bi = sad16bi_altivec_c;
631                      sad8 = sad8_altivec_c;
632                      dev16 = dev16_altivec_c;
633    
634              sse8_16bit = sse8_16bit_altivec_c;
635    
636              /* mem transfer */
637              transfer_8to16copy = transfer_8to16copy_altivec_c;
638              transfer_16to8copy = transfer_16to8copy_altivec_c;
639              transfer_8to16sub = transfer_8to16sub_altivec_c;
640              transfer_8to16subro = transfer_8to16subro_altivec_c;
641              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
642              transfer_16to8add = transfer_16to8add_altivec_c;
643              transfer8x8_copy = transfer8x8_copy_altivec_c;
644    
645          if ((cpu_flags & XVID_CPU_ALTIVEC))            /* Inverse DCT */
646          {            idct = idct_altivec_c;
                 calc_cbp = calc_cbp_altivec;  
                 fdct = fdct_altivec;  
                 idct = idct_altivec;  
                 sadInit = sadInit_altivec;  
                 sad16 = sad16_altivec;  
                 sad8 = sad8_altivec;  
                 dev16 = dev16_altivec;  
         }  
 #endif  
   
         return XVID_ERR_OK;  
 }  
647    
648              /* Interpolation */
649              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
650              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
651              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
652    
653              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
654              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
655    
656                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
657                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
658                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
659                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
660    
661              /* Colorspace conversion */
662              bgra_to_yv12 = bgra_to_yv12_altivec_c;
663              abgr_to_yv12 = abgr_to_yv12_altivec_c;
664              rgba_to_yv12 = rgba_to_yv12_altivec_c;
665              argb_to_yv12 = argb_to_yv12_altivec_c;
666    
667              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
668              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
669    
670  static int            yv12_to_yuyv = yv12_to_yuyv_altivec_c;
671  xvid_init_convert(XVID_INIT_CONVERTINFO* convert)            yv12_to_uyvy = yv12_to_uyvy_altivec_c;
 {  
         // const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);  
         const int width = convert->width;  
         const int height = convert->height;  
         const int width2 = convert->width/2;  
         const int height2 = convert->height/2;  
         IMAGE img;  
672    
673          switch (convert->input.colorspace & ~XVID_CSP_VFLIP)            /* Quantization */
674          {            quant_h263_intra = quant_h263_intra_altivec_c;
675                  case XVID_CSP_YV12 :            quant_h263_inter = quant_h263_inter_altivec_c;
676                          img.y = convert->input.y;            dequant_h263_intra = dequant_h263_intra_altivec_c;
677                          img.v = (uint8_t*)convert->input.y + width*height;            dequant_h263_inter = dequant_h263_inter_altivec_c;
                         img.u = (uint8_t*)convert->input.y + width*height + width2*height2;  
                         image_output(&img, width, height, width,  
                                                 convert->output.y, convert->output.y_stride,  
                                                 convert->output.colorspace, convert->interlacing);  
                         break;  
   
                 default :  
                         return XVID_ERR_FORMAT;  
         }  
678    
679                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
680                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
681    
682          emms();                    /* Qpel stuff */
683          return XVID_ERR_OK;                    xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
684                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
685  }  }
686    #endif
687    
688    #if defined(_DEBUG)
689        xvid_debug = init->debug;
690    #endif
691    
692        return(0);
 void fill8(uint8_t * block, int size, int value)  
 {  
         int i;  
         for (i = 0; i < size; i++)  
                 block[i] = value;  
 }  
   
 void fill16(int16_t * block, int size, int value)  
 {  
         int i;  
         for (i = 0; i < size; i++)  
                 block[i] = value;  
693  }  }
694    
 #define RANDOM(min,max) min + (rand() % (max-min))  
   
 void random8(uint8_t * block, int size, int min, int max)  
 {  
         int i;  
         for (i = 0; i < size; i++)  
                 block[i] = RANDOM(min,max);  
 }  
695    
696  void random16(int16_t * block, int size, int min, int max)  static int
697    xvid_gbl_info(xvid_gbl_info_t * info)
698  {  {
699          int i;          if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
700          for (i = 0; i < size; i++)                  return XVID_ERR_VERSION;
                 block[i] = RANDOM(min,max);  
 }  
701    
702  int compare16(const int16_t * blockA, const int16_t * blockB, int size)          info->actual_version = XVID_VERSION;
703  {          info->build = "xvid-1.4.0-dev";
704          int i;          info->cpu_flags = detect_cpu_flags();
705          for (i = 0; i < size; i++)          info->num_threads = 0; /* single-thread */
                 if (blockA[i] != blockB[i])  
                         return 1;  
706    
707          return 0;  #if defined(_WIN32)
 }  
708    
 int diff16(const int16_t * blockA, const int16_t * blockB, int size)  
709  {  {
710          int i, diff = 0;          SYSTEM_INFO siSysInfo;
711          for (i = 0; i < size; i++)          GetSystemInfo(&siSysInfo);
712                  diff += ABS(blockA[i]-blockB[i]);          info->num_threads = siSysInfo.dwNumberOfProcessors; /* number of _logical_ cores */
         return diff;  
713  }  }
714    
715    #elif defined(_SC_NPROCESSORS_CONF) /* should be available on Apple too actually */
716    
717  #define XVID_TEST_RANDOM        0x00000001      /* random input data */    info->num_threads = sysconf(_SC_NPROCESSORS_CONF);
 #define XVID_TEST_VERBOSE       0x00000002      /* verbose error output */  
718    
719    #elif defined(__APPLE__) && defined(__MACH__)
720    
 #define TEST_FORWARD    0x00000001      /* intra */  
 #define TEST_FDCT  (TEST_FORWARD)  
 #define TEST_IDCT  (0)  
   
 int test_transform(void * funcA, void * funcB, const char * nameB,  
                                    int test, int flags)  
 {  
         int i;  
         int64_t timeSTART;  
         int64_t timeA = 0;  
         int64_t timeB = 0;  
         DECLARE_ALIGNED_MATRIX(arrayA, 1, 64, int16_t, CACHE_LINE);  
         DECLARE_ALIGNED_MATRIX(arrayB, 1, 64, int16_t, CACHE_LINE);  
         int min, max;  
         int count = 0;  
   
         int tmp;  
         int min_error = 0x10000*64;  
         int max_error = 0;  
   
   
         if ((test & TEST_FORWARD))      /* forward */  
721          {          {
722                  min = -256;      size_t len;
723                  max = 255;      int    mib[2], ncpu;
         }else{          /* inverse */  
                 min = -2048;  
                 max = 2047;  
         }  
724    
725          for (i = 0; i < 64*64; i++)      mib[0] = CTL_HW;
726          {      mib[1] = HW_NCPU;
727                  if ((flags & XVID_TEST_RANDOM))      len    = sizeof(ncpu);
728                  {      if (sysctl(mib, 2, &ncpu, &len, NULL, 0) == 0)
729                          random16(arrayA, 64, min, max);        info -> num_threads = ncpu;
                 }else{  
                         fill16(arrayA, 64, i);  
                 }  
                 memcpy(arrayB, arrayA, 64*sizeof(int16_t));  
   
                 if ((test & TEST_FORWARD))  
                 {  
                         timeSTART = read_counter();  
                         ((fdctFunc*)funcA)(arrayA);  
                         timeA += read_counter() - timeSTART;  
   
                         timeSTART = read_counter();  
                         ((fdctFunc*)funcB)(arrayB);  
                         timeB += read_counter() - timeSTART;  
                 }  
730                  else                  else
731                  {        info -> num_threads = 1;
                         timeSTART = read_counter();  
                         ((idctFunc*)funcA)(arrayA);  
                         timeA += read_counter() - timeSTART;  
   
                         timeSTART = read_counter();  
                         ((idctFunc*)funcB)(arrayB);  
                         timeB += read_counter() - timeSTART;  
732                  }                  }
733    
734                  tmp = diff16(arrayA, arrayB, 64) / 64;  #elif defined(__amigaos4__)
                 if (tmp > max_error)  
                         max_error = tmp;  
                 if (tmp < min_error)  
                         min_error = tmp;  
735    
736                  count++;    {
737        uint32_t num_threads = 1;
738        IExec->GetCPUInfoTags(GCIT_NumberOfCPUs, &num_threads, TAG_END);
739        info->num_threads = num_threads;
740          }          }
741    
742          /* print the "average difference" of best/worst transforms */  #endif
         printf("%s:\t%i\t(min_error:%i, max_error:%i)\n", nameB, (int)(timeB / count), min_error, max_error);  
743    
744          return 0;          return 0;
745  }  }
746    
747    
748  #define TEST_QUANT      0x00000001      /* forward quantization */  static int
749  #define TEST_INTRA      0x00000002      /* intra */  xvid_gbl_convert(xvid_gbl_convert_t* convert)
 #define TEST_QUANT_INTRA        (TEST_QUANT|TEST_INTRA)  
 #define TEST_QUANT_INTER        (TEST_QUANT)  
 #define TEST_DEQUANT_INTRA      (TEST_INTRA)  
 #define TEST_DEQUANT_INTER      (0)  
   
 int test_quant(void * funcA, void * funcB, const char * nameB,  
                            int test, int flags)  
750  {  {
751          int q,i;          int width;
752          int64_t timeSTART;          int height;
753          int64_t timeA = 0;          int width2;
754          int64_t timeB = 0;          int height2;
755          int retA, retB;          IMAGE img;
         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);  
         int min, max;  
         int count = 0;  
         int errors = 0;  
756    
757          if ((test & TEST_QUANT))        /* quant */          if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
758          {                return XVID_ERR_VERSION;
                 min = -2048;  
                 max = 2047;  
         }else{          /* dequant */  
                 min = -256;  
                 max = 255;  
         }  
759    
760          for (q = 1; q <= 31; q++)       /* quantizer */  #if 0
761          {          const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
762                  for (i = min; i < max; i++)     /* input coeff */  #endif
763                  {          width = convert->width;
764                          if ((flags & XVID_TEST_RANDOM))          height = convert->height;
765                          {          width2 = convert->width/2;
766                                  random16(arrayX, 64, min, max);          height2 = convert->height/2;
                         }else{  
                                 fill16(arrayX, 64, i);  
                         }  
   
                         if ((test & TEST_INTRA))        /* intra */  
                         {  
                                 timeSTART = read_counter();  
                                 ((quanth263_intraFunc*)funcA)(arrayA, arrayX, q, q);  
                                 timeA += read_counter() - timeSTART;  
767    
768                                  timeSTART = read_counter();          switch (convert->input.csp & ~XVID_CSP_VFLIP)
                                 ((quanth263_intraFunc*)funcB)(arrayB, arrayX, q, q);  
                                 timeB += read_counter() - timeSTART;  
                         }  
                         else    /* inter */  
769                          {                          {
770                                  timeSTART = read_counter();                  case XVID_CSP_YV12 :
771                                  retA = ((quanth263_interFunc*)funcA)(arrayA, arrayX, q);                          img.y = convert->input.plane[0];
772                                  timeA += read_counter() - timeSTART;                          img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
773                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
774                                  timeSTART = read_counter();                          image_output(&img, width, height, width,
775                                  retB = ((quanth263_interFunc*)funcB)(arrayB, arrayX, q);                                                  (uint8_t**)convert->output.plane, convert->output.stride,
776                                  timeB += read_counter() - timeSTART;                                                  convert->output.csp, convert->interlacing);
777                          }                          break;
778    
779                          /* compare return value from quant_inter, and compare (de)quantiz'd arrays */                  case XVID_CSP_INTERNAL :
780                          if ( ((test&TEST_QUANT) && !(test&TEST_INTRA) && retA != retB ) ||                          img.y = (uint8_t*)convert->input.plane[0];
781                                  compare16(arrayA, arrayB, 64))                          img.u = (uint8_t*)convert->input.plane[1];
782                          {                          img.v = (uint8_t*)convert->input.plane[2];
783                                  errors++;                          image_output(&img, width, height, convert->input.stride[0],
784                                  if ((flags & XVID_TEST_VERBOSE))                                                  (uint8_t**)convert->output.plane, convert->output.stride,
785                                          printf("%s error: q=%i, i=%i\n", nameB, q, i);                                                  convert->output.csp, convert->interlacing);
786                          }                          break;
787    
788                          count++;                  default :
789                  }                          return XVID_ERR_FORMAT;
790          }          }
791    
         printf("%s:\t%i", nameB, (int)(timeB / count));  
         if (errors>0)  
                 printf("\t(%i errors out of %i)", errors, count);  
         printf("\n");  
792    
793            emms();
794          return 0;          return 0;
795  }  }
796    
797    /*****************************************************************************
798     * Xvid Global Entry point
799  int xvid_init_test(int flags)   *
800  {   * Well this function initialize all internal function pointers according
801          int cpu_flags;   * to the CPU features forced by the library client or autodetected (depending
802     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
803          srand(time(0));   * image colorspace transformation tables.
804     *
805          printf("xvid_init_test\n");   ****************************************************************************/
   
 #if defined(ARCH_X86)  
         cpu_flags = detect_cpu_flags();  
         idct_int32_init();  
         emms_mmx();  
   
         printf("--- fdct ---\n");  
                 test_transform(fdct_int32, fdct_int32, "c", TEST_FDCT, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_transform(fdct_int32, fdct_mmx, "mmx", TEST_FDCT, flags);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_transform(fdct_int32, fdct_sse2, "sse2", TEST_FDCT, flags);  
   
         printf("\n--- idct ---\n");  
                 test_transform(idct_int32, idct_int32, "c", TEST_IDCT, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_transform(idct_int32, idct_mmx, "mmx", TEST_IDCT, flags);  
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_transform(idct_int32, idct_xmm, "xmm", TEST_IDCT, flags);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_transform(idct_int32, idct_3dne, "3dne", TEST_IDCT, flags);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_transform(idct_int32, idct_sse2, "sse2", TEST_IDCT, flags);  
   
         printf("\n--- quant intra ---\n");  
                 test_quant(quant_intra_c, quant_intra_c, "c", TEST_QUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_quant(quant_intra_c, quant_intra_mmx, "mmx", TEST_QUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_quant(quant_intra_c, quant_intra_3dne, "3dne", TEST_QUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_quant(quant_intra_c, quant_intra_sse2, "sse2", TEST_QUANT_INTRA, flags);  
   
         printf("\n--- quant inter ---\n");  
                 test_quant(quant_inter_c, quant_inter_c, "c", TEST_QUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_quant(quant_inter_c, quant_inter_mmx, "mmx", TEST_QUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_quant(quant_inter_c, quant_inter_3dne, "3dne", TEST_QUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_quant(quant_inter_c, quant_inter_sse2, "sse2", TEST_QUANT_INTER, flags);  
   
         printf("\n--- dequant intra ---\n");  
                 test_quant(dequant_intra_c, dequant_intra_c, "c", TEST_DEQUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_quant(dequant_intra_c, dequant_intra_mmx, "mmx", TEST_DEQUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_quant(dequant_intra_c, dequant_intra_xmm, "xmm", TEST_DEQUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_quant(dequant_intra_c, dequant_intra_3dne, "3dne", TEST_DEQUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_quant(dequant_intra_c, dequant_intra_sse2, "sse2", TEST_DEQUANT_INTRA, flags);  
   
         printf("\n--- dequant inter ---\n");  
                 test_quant(dequant_inter_c, dequant_inter_c, "c", TEST_DEQUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_quant(dequant_inter_c, dequant_inter_mmx, "mmx", TEST_DEQUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_quant(dequant_inter_c, dequant_inter_xmm, "xmm", TEST_DEQUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_quant(dequant_inter_c, dequant_inter_3dne, "3dne", TEST_DEQUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_SSE2)  
                 test_quant(dequant_inter_c, dequant_inter_sse2, "sse2", TEST_DEQUANT_INTER, flags);  
   
         printf("\n--- quant4_intra ---\n");  
                 test_quant(quant4_intra_c, quant4_intra_c, "c", TEST_QUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_quant(quant4_intra_c, quant4_intra_mmx, "mmx", TEST_QUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_quant(quant4_intra_c, quant4_intra_xmm, "xmm", TEST_QUANT_INTRA, flags);  
   
         printf("\n--- quant4_inter ---\n");  
                 test_quant(quant4_inter_c, quant4_inter_c, "c", TEST_QUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_quant(quant4_inter_c, quant4_inter_mmx, "mmx", TEST_QUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_MMXEXT)  
                 test_quant(quant4_inter_c, quant4_inter_xmm, "xmm", TEST_QUANT_INTER, flags);  
   
         printf("\n--- dequant4_intra ---\n");  
                 test_quant(dequant4_intra_c, dequant4_intra_c, "c", TEST_DEQUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_quant(dequant4_intra_c, dequant4_intra_mmx, "mmx", TEST_DEQUANT_INTRA, flags);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_quant(dequant4_intra_c, dequant4_intra_3dne, "3dne", TEST_DEQUANT_INTRA, flags);  
   
         printf("\n--- dequant4_inter ---\n");  
                 test_quant(dequant4_inter_c, dequant4_inter_c, "c", TEST_DEQUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_MMX)  
                 test_quant(dequant4_inter_c, dequant4_inter_mmx, "mmx", TEST_DEQUANT_INTER, flags);  
         if (cpu_flags & XVID_CPU_3DNOWEXT)  
                 test_quant(dequant4_inter_c, dequant4_inter_3dne, "3dne", TEST_DEQUANT_INTER, flags);  
   
         emms_mmx();  
   
 #endif  
   
         return XVID_ERR_OK;  
 }  
806    
807    
808  int  int
809  xvid_init(void *handle,  xvid_global(void *handle,
810                    int opt,                    int opt,
811                    void *param1,                    void *param1,
812                    void *param2)                    void *param2)
813  {  {
814          switch(opt)          switch(opt)
815          {          {
816                  case XVID_INIT_INIT :                  case XVID_GBL_INIT :
817                          return xvid_init_init((XVID_INIT_PARAM*)param1);                          return xvid_gbl_init((xvid_gbl_init_t*)param1);
818    
819                  case XVID_INIT_CONVERT :          case XVID_GBL_INFO :
820                          return xvid_init_convert((XVID_INIT_CONVERTINFO*)param1);              return xvid_gbl_info((xvid_gbl_info_t*)param1);
821    
822                  case XVID_INIT_TEST :                  case XVID_GBL_CONVERT :
823                          return xvid_init_test((int)param1);                          return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
824    
825                  default :                  default :
826                          return XVID_ERR_FAIL;                          return XVID_ERR_FAIL;
# Line 953  Line 828 
828  }  }
829    
830  /*****************************************************************************  /*****************************************************************************
831   * XviD Native decoder entry point   * Xvid Native decoder entry point
832   *   *
833   * This function is just a wrapper to all the option cases.   * This function is just a wrapper to all the option cases.
834   *   *
# Line 969  Line 844 
844                          void *param2)                          void *param2)
845  {  {
846          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1, (XVID_DEC_STATS*) param2);  
   
847          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
848                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
849    
850          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
851                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
852    
853            case XVID_DEC_DECODE:
854                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
855    
856          default:          default:
857                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
858          }          }
# Line 985  Line 860 
860    
861    
862  /*****************************************************************************  /*****************************************************************************
863   * XviD Native encoder entry point   * Xvid Native encoder entry point
864   *   *
865   * This function is just a wrapper to all the option cases.   * This function is just a wrapper to all the option cases.
866   *   *
# Line 1003  Line 878 
878          switch (opt) {          switch (opt) {
879          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
880    
881                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
882                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
883                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
884    
885          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
886                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
887    
888          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
889                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
890    
891          default:          default:
892                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.33.2.23  
changed lines
  Added in v.1.86

No admin address has been configured
ViewVC Help
Powered by ViewVC 1.0.4