[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, Tue Jul 23 12:59:57 2002 UTC revision 1.78, Fri Nov 28 18:16:42 2008 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 50  Line 37 
37  #include "image/colorspace.h"  #include "image/colorspace.h"
38  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
39  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
40  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
41  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
42  #include "motion/motion.h"  #include "motion/motion.h"
43    #include "motion/gmc.h"
44  #include "motion/sad.h"  #include "motion/sad.h"
45  #include "utils/emms.h"  #include "utils/emms.h"
46  #include "utils/timer.h"  #include "utils/timer.h"
47  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
48    #include "image/qpel.h"
49    #include "image/postprocessing.h"
50    
51  #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE)  #if defined(_DEBUG)
52    unsigned int xvid_debug = 0; /* xvid debug mask */
53    #endif
54    
55  #ifdef WIN32  #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
56  #include <windows.h>  #include <windows.h>
57  #else  #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
58  #include <signal.h>  #include <signal.h>
59  #include <setjmp.h>  #include <setjmp.h>
 #endif  
   
   
 #ifndef WIN32  
60    
61  static jmp_buf mark;  static jmp_buf mark;
62    
# Line 81  Line 69 
69    
70    
71  /*  /*
72  calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled   * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
73  return values:   * signalled
74  -1 : could not determine   *
75  0  : SIGILL was *not* signalled   * Return values:
76  1  : SIGILL was signalled   *  -1 : could not determine
77     *   0 : SIGILL was *not* signalled
78     *   1 : SIGILL was signalled
79  */  */
80    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
81  int  static int
82  sigill_check(void (*func)())  sigill_check(void (*func)())
83  {  {
 #ifdef WIN32  
84          _try {          _try {
85                  func();                  func();
86          }          } _except(EXCEPTION_EXECUTE_HANDLER) {
         _except(EXCEPTION_EXECUTE_HANDLER) {  
87    
88                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
89                          return 1;                          return(1);
90          }          }
91          return 0;          return(0);
92  #else  }
93    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
94    static int
95    sigill_check(void (*func)())
96    {
97      void * old_handler;      void * old_handler;
98      int jmpret;      int jmpret;
99    
100        /* Set our SIGILL handler */
101      old_handler = signal(SIGILL, sigill_handler);      old_handler = signal(SIGILL, sigill_handler);
102      if (old_handler == SIG_ERR)  
103      {      /* Check for error */
104          return -1;      if (old_handler == SIG_ERR) {
105            return(-1);
106      }      }
107    
108        /* Save stack context, so if func triggers a SIGILL, we can still roll
109             * back to a valid CPU state */
110      jmpret = setjmp(mark);      jmpret = setjmp(mark);
111      if (jmpret == 0)  
112      {          /* If setjmp returned directly, then its returned value is 0, and we still
113             * have to test the passed func. Otherwise it means the stack context has
114             * been restored by a longjmp() call, which in our case happens only in the
115             * signal handler */
116        if (jmpret == 0) {
117          func();          func();
118      }      }
119    
120        /* Restore old signal handler */
121      signal(SIGILL, old_handler);      signal(SIGILL, old_handler);
122    
123      return jmpret;      return(jmpret);
 #endif  
124  }  }
125  #endif  #endif
126    
127    
128    /* detect cpu flags  */
129    static unsigned int
130    detect_cpu_flags(void)
131    {
132            /* enable native assembly optimizations by default */
133            unsigned int cpu_flags = XVID_CPU_ASM;
134    
135    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
136            cpu_flags |= check_cpu_features();
137            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
138                    cpu_flags &= ~XVID_CPU_SSE;
139    
140            if ((cpu_flags & (XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41)) && sigill_check(sse2_os_trigger))
141                    cpu_flags &= ~(XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41);
142    #endif
143    
144    #if defined(ARCH_IS_PPC)
145            if (!sigill_check(altivec_trigger))
146                    cpu_flags |= XVID_CPU_ALTIVEC;
147    #endif
148    
149            return cpu_flags;
150    }
151    
152    
153  /*****************************************************************************  /*****************************************************************************
154   * XviD Init Entry point   * XviD Init Entry point
155   *   *
# Line 139  Line 164 
164   *   *
165   ****************************************************************************/   ****************************************************************************/
166    
 int  
 xvid_init(void *handle,  
                   int opt,  
                   void *param1,  
                   void *param2)  
 {  
         int cpu_flags;  
         XVID_INIT_PARAM *init_param;  
   
         init_param = (XVID_INIT_PARAM *) param1;  
167    
168          /* Inform the client the API version */  static
169          init_param->api_version = API_VERSION;  int xvid_gbl_init(xvid_gbl_init_t * init)
   
         /* 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))  
170          {          {
171                  init_param->cpu_flags = cpu_flags;          unsigned int cpu_flags;
                 return XVID_ERR_OK;  
         }  
172    
173          init_param->cpu_flags = cpu_flags;          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
174                    return XVID_ERR_VERSION;
175    
176            cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
177    
178          /* Initialize the function pointers */          /* Initialize the function pointers */
179          idct_int32_init();          idct_int32_init();
# Line 192  Line 184 
184          idct = idct_int32;          idct = idct_int32;
185    
186          /* Only needed on PPC Altivec archs */          /* Only needed on PPC Altivec archs */
187          sadInit = 0;          sadInit = NULL;
188    
189          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
190          emms = emms_c;          emms = emms_c;
191    
192            /* Qpel stuff */
193            xvid_QP_Funcs = &xvid_QP_Funcs_C;
194            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
195            xvid_Init_QP();
196    
197          /* Quantization functions */          /* Quantization functions */
198          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
199          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
200          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
201          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
202    
203          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
204          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
205          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
206          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
207    
208          /* Block transfer related functions */          /* Block transfer related functions */
209          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
210          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
211          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
212            transfer_8to16subro  = transfer_8to16subro_c;
213          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
214            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
215          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
216          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
217            transfer8x4_copy   = transfer8x4_copy_c;
218    
219            /* Interlacing functions */
220            MBFieldTest = MBFieldTest_c;
221    
222          /* Image interpolation related functions */          /* Image interpolation related functions */
223          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;
224          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
225          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
226    
227            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
228            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
229            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
230    
231            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
232            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
233            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
234            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
235    
236            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
237            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
238            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
239    
240            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
241            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
242            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
243    
244            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
245            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
246    
247            interpolate8x8_avg2 = interpolate8x8_avg2_c;
248            interpolate8x8_avg4 = interpolate8x8_avg4_c;
249    
250            /* postprocessing */
251            image_brightness = image_brightness_c;
252    
253          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
254          colorspace_init();          colorspace_init();
255    
256          /* All colorspace transformation functions User Format->YV12 */          /* All colorspace transformation functions User Format->YV12 */
257            yv12_to_yv12    = yv12_to_yv12_c;
258          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
259          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
260          rgb24_to_yv12  = rgb24_to_yv12_c;          rgb_to_yv12     = rgb_to_yv12_c;
261          rgb32_to_yv12  = rgb32_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
262          yuv_to_yv12    = yuv_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
263            abgr_to_yv12    = abgr_to_yv12_c;
264            rgba_to_yv12    = rgba_to_yv12_c;
265            argb_to_yv12    = argb_to_yv12_c;
266          yuyv_to_yv12   = yuyv_to_yv12_c;          yuyv_to_yv12   = yuyv_to_yv12_c;
267          uyvy_to_yv12   = uyvy_to_yv12_c;          uyvy_to_yv12   = uyvy_to_yv12_c;
268    
269            rgb555i_to_yv12 = rgb555i_to_yv12_c;
270            rgb565i_to_yv12 = rgb565i_to_yv12_c;
271            bgri_to_yv12    = bgri_to_yv12_c;
272            bgrai_to_yv12   = bgrai_to_yv12_c;
273            abgri_to_yv12   = abgri_to_yv12_c;
274            rgbai_to_yv12   = rgbai_to_yv12_c;
275            argbi_to_yv12   = argbi_to_yv12_c;
276            yuyvi_to_yv12   = yuyvi_to_yv12_c;
277            uyvyi_to_yv12   = uyvyi_to_yv12_c;
278    
279          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
280          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
281          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
282          yv12_to_rgb24  = yv12_to_rgb24_c;          yv12_to_rgb     = yv12_to_rgb_c;
283          yv12_to_rgb32  = yv12_to_rgb32_c;          yv12_to_bgr     = yv12_to_bgr_c;
284          yv12_to_yuv    = yv12_to_yuv_c;          yv12_to_bgra    = yv12_to_bgra_c;
285            yv12_to_abgr    = yv12_to_abgr_c;
286            yv12_to_rgba    = yv12_to_rgba_c;
287            yv12_to_argb    = yv12_to_argb_c;
288          yv12_to_yuyv   = yv12_to_yuyv_c;          yv12_to_yuyv   = yv12_to_yuyv_c;
289          yv12_to_uyvy   = yv12_to_uyvy_c;          yv12_to_uyvy   = yv12_to_uyvy_c;
290    
291            yv12_to_rgb555i = yv12_to_rgb555i_c;
292            yv12_to_rgb565i = yv12_to_rgb565i_c;
293            yv12_to_bgri    = yv12_to_bgri_c;
294            yv12_to_bgrai   = yv12_to_bgrai_c;
295            yv12_to_abgri   = yv12_to_abgri_c;
296            yv12_to_rgbai   = yv12_to_rgbai_c;
297            yv12_to_argbi   = yv12_to_argbi_c;
298            yv12_to_yuyvi   = yv12_to_yuyvi_c;
299            yv12_to_uyvyi   = yv12_to_uyvyi_c;
300    
301          /* Functions used in motion estimation algorithms */          /* Functions used in motion estimation algorithms */
302          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
303          sad16    = sad16_c;          sad16    = sad16_c;
# Line 249  Line 305 
305          sad16bi  = sad16bi_c;          sad16bi  = sad16bi_c;
306          sad8bi   = sad8bi_c;          sad8bi   = sad8bi_c;
307          dev16    = dev16_c;          dev16    = dev16_c;
308            sad16v     = sad16v_c;
309            sse8_16bit = sse8_16bit_c;
310            sse8_8bit  = sse8_8bit_c;
311    
312            init_GMC(cpu_flags);
313    
314    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
315    
316            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
317                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
318                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2) ||
319            (cpu_flags & XVID_CPU_SSE3) || (cpu_flags & XVID_CPU_SSE41))
320            {
321                    /* Restore FPU context : emms_c is a nop functions */
322                    emms = emms_mmx;
323            }
324    
325          Halfpel8_Refine = Halfpel8_Refine_c;          if ((cpu_flags & XVID_CPU_MMX)) {
   
 #ifdef ARCH_X86  
         if ((cpu_flags & XVID_CPU_MMX) > 0) {  
326    
327                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
328                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
329                  idct = idct_mmx;                  idct = idct_mmx;
330    
331                  /* To restore FPU context after mmx use */                  /* Qpel stuff */
332                  emms = emms_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
333                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
334    
335                  /* Quantization related functions */                  /* Quantization related functions */
336                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
337                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
338                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
339                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
340                    quant_mpeg_intra   = quant_mpeg_intra_mmx;
341                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
342                  dequant4_intra = dequant4_intra_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
343                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
344                  dequant4_inter = dequant4_inter_mmx;  
345    
346                  /* Block related functions */                  /* Block related functions */
347                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
348                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
349                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
350                    transfer_8to16subro  = transfer_8to16subro_mmx;
351                  transfer_8to16sub2 = transfer_8to16sub2_mmx;                  transfer_8to16sub2 = transfer_8to16sub2_mmx;
352                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
353                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
354                    transfer8x4_copy   = transfer8x4_copy_mmx;
355    
356                    /* Interlacing Functions */
357                    MBFieldTest = MBFieldTest_mmx;
358    
359                  /* Image Interpolation related functions */                  /* Image Interpolation related functions */
360                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;
361                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
362                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
363    
364                  /* Image RGB->YV12 related functions */                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
365                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
366                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
367                  yuv_to_yv12   = yuv_to_yv12_mmx;  
368                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
369                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
370                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
371                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
372    
373                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
374                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
375    
376                    interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
377                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
378    
379                    /* postprocessing */
380                    image_brightness = image_brightness_mmx;
381    
382                    /* image input xxx_to_yv12 related functions */
383    
384                    yv12_to_yv12  = yv12_to_yv12_mmx;
385    
386                    bgr_to_yv12   = bgr_to_yv12_mmx;
387                    rgb_to_yv12   = rgb_to_yv12_mmx;
388                    bgra_to_yv12  = bgra_to_yv12_mmx;
389                    rgba_to_yv12  = rgba_to_yv12_mmx;
390                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
391                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
392    
393                  /* Image YV12->RGB related functions */                  /* image output yv12_to_xxx related functions */
394                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
395                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgra  = yv12_to_bgra_mmx;
396                  yv12_to_yuyv  = yv12_to_yuyv_mmx;                  yv12_to_yuyv  = yv12_to_yuyv_mmx;
397                  yv12_to_uyvy  = yv12_to_uyvy_mmx;                  yv12_to_uyvy  = yv12_to_uyvy_mmx;
398    
399                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
400                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
401    
402                  /* Motion estimation related functions */                  /* Motion estimation related functions */
403                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
404                  sad16    = sad16_mmx;                  sad16    = sad16_mmx;
# Line 307  Line 406 
406                  sad16bi = sad16bi_mmx;                  sad16bi = sad16bi_mmx;
407                  sad8bi  = sad8bi_mmx;                  sad8bi  = sad8bi_mmx;
408                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
409                    sad16v     = sad16v_mmx;
410                    sse8_16bit = sse8_16bit_mmx;
411                    sse8_8bit  = sse8_8bit_mmx;
412          }          }
413    
414          /* these 3dnow functions are faster than mmx, but slower than xmm. */          /* these 3dnow functions are faster than mmx, but slower than xmm. */
415          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
416    
417                    emms = emms_3dn;
418    
419                  /* ME functions */                  /* ME functions */
420                  sad16bi = sad16bi_3dn;                  sad16bi = sad16bi_3dn;
421                  sad8bi  = sad8bi_3dn;                  sad8bi  = sad8bi_3dn;
422    
423                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
424                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
425    
426          }          }
427    
428    
429          if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {          if ((cpu_flags & XVID_CPU_MMXEXT)) {
430    
431                  /* Inverse DCT */                  /* DCT */
432                    fdct = fdct_xmm_skal;
433                  idct = idct_xmm;                  idct = idct_xmm;
434    
435                  /* Interpolation */                  /* Interpolation */
# Line 329  Line 437 
437                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
438                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
439    
440                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
441                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
442                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
443    
444                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
445                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
446                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
447                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
448    
449                  /* Quantization */                  /* Quantization */
450                  dequant_intra = dequant_intra_xmm;                  quant_mpeg_inter = quant_mpeg_inter_xmm;
451                  dequant_inter = dequant_inter_xmm;  
452                    dequant_h263_intra = dequant_h263_intra_xmm;
453                    dequant_h263_inter = dequant_h263_inter_xmm;
454    
455                  /* Buffer transfer */                  /* Buffer transfer */
456                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
457                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
458    
459                  /* Colorspace transformation */                  /* Colorspace transformation */
460                  yuv_to_yv12 = yuv_to_yv12_xmm;                  /* yv12_to_yv12  = yv12_to_yv12_xmm; */ /* appears to be slow on many machines */
461                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
462                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
463    
464                  /* ME functions */                  /* ME functions */
465                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
# Line 345  Line 467 
467                  sad16bi = sad16bi_xmm;                  sad16bi = sad16bi_xmm;
468                  sad8bi  = sad8bi_xmm;                  sad8bi  = sad8bi_xmm;
469                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
470                    sad16v   = sad16v_xmm;
471          }          }
472    
473          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
474    
475                  /* Interpolation */                  /* Interpolation */
476                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
477                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
478                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
479    
480                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
481                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
482                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
483            }
484    
485            if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
486    
487                    /* Buffer transfer */
488                    transfer_8to16copy =  transfer_8to16copy_3dne;
489                    transfer_16to8copy = transfer_16to8copy_3dne;
490                    transfer_8to16sub =  transfer_8to16sub_3dne;
491                    transfer_8to16subro =  transfer_8to16subro_3dne;
492                    transfer_16to8add = transfer_16to8add_3dne;
493                    transfer8x8_copy = transfer8x8_copy_3dne;
494                    transfer8x4_copy = transfer8x4_copy_3dne;
495    
496                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
497                            /* Inverse DCT */
498                            idct =  idct_3dne;
499    
500                            /* Buffer transfer */
501                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
502    
503                            /* Interpolation */
504                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
505                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
506                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
507    
508                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
509                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
510                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
511    
512                /* Quantization */
513                            quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
514                            quant_h263_inter = quant_h263_inter_3dne;
515                            dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
516                            dequant_mpeg_inter = dequant_mpeg_inter_3dne;
517                            dequant_h263_intra = dequant_h263_intra_3dne;
518                            dequant_h263_inter = dequant_h263_inter_3dne;
519    
520                /* ME functions */
521                            sad16 = sad16_3dne;
522                            sad8 = sad8_3dne;
523                            sad16bi = sad16bi_3dne;
524                            sad8bi = sad8bi_3dne;
525                            dev16 = dev16_3dne;
526                    }
527          }          }
528    
529          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_SSE2)) {
 #ifdef EXPERIMENTAL_SSE2_CODE  
530    
531                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
532    
533                  /* Quantization */                  /* Quantization */
534                  quant_intra   = quant_intra_sse2;                  quant_h263_intra   = quant_h263_intra_sse2;
535                  dequant_intra = dequant_intra_sse2;                  quant_h263_inter   = quant_h263_inter_sse2;
536                  quant_inter   = quant_inter_sse2;                  dequant_h263_intra = dequant_h263_intra_sse2;
537                  dequant_inter = dequant_inter_sse2;                  dequant_h263_inter = dequant_h263_inter_sse2;
538    
539                  /* ME */                  /* SAD operators */
540                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
541                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
542    
543                  /* Forward and Inverse DCT */                  /* DCT operators */
544                  idct  = idct_sse2;                  fdct = fdct_sse2_skal;
545                  fdct = fdct_sse2;                  idct = idct_sse2_skal;   /* Is now IEEE1180 and Walken compliant. */
546  #endif  
547                    /* postprocessing */
548                    image_brightness = image_brightness_sse2;
549    
550          }          }
551    
552  #endif          if ((cpu_flags & XVID_CPU_SSE3)) {
553    
554  #ifdef ARCH_IA64                  /* SAD operators */
555          if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?                  sad16    = sad16_sse3;
556                    dev16    = dev16_sse3;
557            }
558    
559    #endif /* ARCH_IS_IA32 */
560    
561    #if defined(ARCH_IS_IA64)
562            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
563            idct_ia64_init();            idct_ia64_init();
564            fdct = fdct_ia64;            fdct = fdct_ia64;
565            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
566            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
567            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
568            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 391  Line 570 
570            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
571            sad8 = sad8_ia64;            sad8 = sad8_ia64;
572            dev16 = dev16_ia64;            dev16 = dev16_ia64;
573            Halfpel8_Refine = Halfpel8_Refine_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
574            quant_intra = quant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
575            dequant_intra = dequant_intra_ia64;            quant_h263_inter = quant_h263_inter_ia64;
576            quant_inter = quant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
577            dequant_inter = dequant_inter_ia64;            dequant_h263_inter = dequant_h263_inter_ia64;
578            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
579            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
580            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
581            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
582            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
583            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
584          }          }
585  #endif  #endif
586    
587  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
588  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
589          calc_cbp = calc_cbp_altivec;            /* sad operators */
590          fdct = fdct_altivec;                    sad16 = sad16_altivec_c;
591          idct = idct_altivec;                    sad16bi = sad16bi_altivec_c;
592          sadInit = sadInit_altivec;                    sad8 = sad8_altivec_c;
593          sad16 = sad16_altivec;                    dev16 = dev16_altivec_c;
594          sad8 = sad8_altivec;  
595          dev16 = dev16_altivec;            sse8_16bit = sse8_16bit_altivec_c;
596  #else  
597          calc_cbp = calc_cbp_ppc;            /* mem transfer */
598              transfer_8to16copy = transfer_8to16copy_altivec_c;
599              transfer_16to8copy = transfer_16to8copy_altivec_c;
600              transfer_8to16sub = transfer_8to16sub_altivec_c;
601              transfer_8to16subro = transfer_8to16subro_altivec_c;
602              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
603              transfer_16to8add = transfer_16to8add_altivec_c;
604              transfer8x8_copy = transfer8x8_copy_altivec_c;
605    
606              /* Inverse DCT */
607              idct = idct_altivec_c;
608    
609              /* Interpolation */
610              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
611              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
612              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
613    
614              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
615              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
616    
617                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
618                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
619                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
620                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
621    
622              /* Colorspace conversion */
623              bgra_to_yv12 = bgra_to_yv12_altivec_c;
624              abgr_to_yv12 = abgr_to_yv12_altivec_c;
625              rgba_to_yv12 = rgba_to_yv12_altivec_c;
626              argb_to_yv12 = argb_to_yv12_altivec_c;
627    
628              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
629              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
630    
631              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
632              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
633    
634              /* Quantization */
635              quant_h263_intra = quant_h263_intra_altivec_c;
636              quant_h263_inter = quant_h263_inter_altivec_c;
637              dequant_h263_intra = dequant_h263_intra_altivec_c;
638              dequant_h263_inter = dequant_h263_inter_altivec_c;
639    
640                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
641                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
642    
643                      /* Qpel stuff */
644                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
645                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
646            }
647    #endif
648    
649    #if defined(_DEBUG)
650        xvid_debug = init->debug;
651  #endif  #endif
652    
653        return(0);
654    }
655    
656    
657    static int
658    xvid_gbl_info(xvid_gbl_info_t * info)
659    {
660            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
661                    return XVID_ERR_VERSION;
662    
663            info->actual_version = XVID_VERSION;
664            info->build = "xvid-1.3.0-dev";
665            info->cpu_flags = detect_cpu_flags();
666      info->num_threads = 0;
667    
668    #if defined(_WIN32)
669      {
670        DWORD dwProcessAffinityMask, dwSystemAffinityMask;
671        if (GetProcessAffinityMask(GetCurrentProcess(), (PDWORD_PTR) &dwProcessAffinityMask, (PDWORD_PTR) &dwSystemAffinityMask)) {
672          int i;
673          for(i=0; i<32; i++) {
674            if ((dwProcessAffinityMask & (1<<i)))
675              info->num_threads++;
676          }
677          if (info->num_threads == 0) {
678            SYSTEM_INFO siSysInfo;
679            GetSystemInfo(&siSysInfo);
680            info->num_threads = siSysInfo.dwNumberOfProcessors; /* number of _logical_ cores */
681          }
682        }
683      }
684    #else
685    
686      #include <unistd.h>
687      info->num_threads = sysconf(_SC_NPROCESSORS_CONF);
688    
689  #endif  #endif
690    
691          return XVID_ERR_OK;          return 0;
692    }
693    
694    
695    static int
696    xvid_gbl_convert(xvid_gbl_convert_t* convert)
697    {
698            int width;
699            int height;
700            int width2;
701            int height2;
702            IMAGE img;
703    
704            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
705                  return XVID_ERR_VERSION;
706    
707    #if 0
708            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
709    #endif
710            width = convert->width;
711            height = convert->height;
712            width2 = convert->width/2;
713            height2 = convert->height/2;
714    
715            switch (convert->input.csp & ~XVID_CSP_VFLIP)
716            {
717                    case XVID_CSP_YV12 :
718                            img.y = convert->input.plane[0];
719                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
720                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
721                            image_output(&img, width, height, width,
722                                                    (uint8_t**)convert->output.plane, convert->output.stride,
723                                                    convert->output.csp, convert->interlacing);
724                            break;
725    
726                    default :
727                            return XVID_ERR_FORMAT;
728            }
729    
730    
731            emms();
732            return 0;
733    }
734    
735    /*****************************************************************************
736     * XviD Global Entry point
737     *
738     * Well this function initialize all internal function pointers according
739     * to the CPU features forced by the library client or autodetected (depending
740     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
741     * image colorspace transformation tables.
742     *
743     ****************************************************************************/
744    
745    
746    int
747    xvid_global(void *handle,
748                      int opt,
749                      void *param1,
750                      void *param2)
751    {
752            switch(opt)
753            {
754                    case XVID_GBL_INIT :
755                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
756    
757            case XVID_GBL_INFO :
758                return xvid_gbl_info((xvid_gbl_info_t*)param1);
759    
760                    case XVID_GBL_CONVERT :
761                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
762    
763                    default :
764                            return XVID_ERR_FAIL;
765            }
766  }  }
767    
768  /*****************************************************************************  /*****************************************************************************
# Line 440  Line 782 
782                          void *param2)                          void *param2)
783  {  {
784          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
785          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
786                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
787    
788          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
789                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
790    
791            case XVID_DEC_DECODE:
792                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
793    
794          default:          default:
795                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
796          }          }
# Line 473  Line 815 
815  {  {
816          switch (opt) {          switch (opt) {
817          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
818  #ifdef BFRAMES  
819                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
820                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
821                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
 #endif  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
822    
823          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
824                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
825    
826          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
827                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
828    
829          default:          default:
830                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.33  
changed lines
  Added in v.1.78

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