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

Diff of /trunk/xvidcore/src/xvid.c

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

revision 195, Wed Jun 12 20:38:41 2002 UTC revision 408, Wed Sep 4 22:07:07 2002 UTC
# Line 1  Line 1 
1  /**************************************************************************  /*****************************************************************************
2   *   *
3   *      XVID MPEG-4 VIDEO CODEC   *      XVID MPEG-4 VIDEO CODEC
4   *      native api   *  - Native API implementation  -
5     *
6     *  Copyright(C) 2001-2002 Peter Ross <pross@cs.rmit.edu.au>
7   *   *
8   *      This program is an implementation of a part of one or more MPEG-4   *      This program is an implementation of a part of one or more MPEG-4
9   *      Video tools as specified in ISO/IEC 14496-2 standard.  Those intending   *      Video tools as specified in ISO/IEC 14496-2 standard.  Those intending
# Line 24  Line 26 
26   *   *
27   *      You should have received a copy of the GNU General Public License   *      You should have received a copy of the GNU General Public License
28   *      along with this program; if not, write to the Free Software   *      along with this program; if not, write to the Free Software
29   *      Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
30   *   *
31   *************************************************************************/   ****************************************************************************/
   
 /**************************************************************************  
  *  
  *      History:  
  *  
  *      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>  
  *  
  *************************************************************************/  
   
32    
33  #include "xvid.h"  #include "xvid.h"
34  #include "decoder.h"  #include "decoder.h"
# Line 50  Line 41 
41  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
42  #include "quant/quant_h263.h"  #include "quant/quant_h263.h"
43  #include "quant/quant_mpeg4.h"  #include "quant/quant_mpeg4.h"
44    #include "motion/motion.h"
45  #include "motion/sad.h"  #include "motion/sad.h"
46  #include "utils/emms.h"  #include "utils/emms.h"
47  #include "utils/timer.h"  #include "utils/timer.h"
48  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
49    
50    #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE)
51    
52    #ifdef WIN32
53    #include <windows.h>
54    #else
55    #include <signal.h>
56    #include <setjmp.h>
57    #endif
58    
59    
60    #ifndef WIN32
61    
62    static jmp_buf mark;
63    
64    static void
65    sigill_handler(int signal)
66    {
67       longjmp(mark, 1);
68    }
69    #endif
70    
71    
72    /*
73     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled
74     * Return values:
75     * -1 : could not determine
76     * 0  : SIGILL was *not* signalled
77     * 1  : SIGILL was signalled
78     */
79    
80    int
81    sigill_check(void (*func)())
82    {
83    #ifdef WIN32
84            _try {
85                    func();
86            }
87            _except(EXCEPTION_EXECUTE_HANDLER) {
88    
89                    if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
90                            return 1;
91            }
92            return 0;
93    #else
94        void * old_handler;
95        int jmpret;
96    
97    
98        old_handler = signal(SIGILL, sigill_handler);
99        if (old_handler == SIG_ERR)
100        {
101            return -1;
102        }
103    
104        jmpret = setjmp(mark);
105        if (jmpret == 0)
106        {
107            func();
108        }
109    
110        signal(SIGILL, old_handler);
111    
112        return jmpret;
113    #endif
114    }
115    #endif
116    
117    /*****************************************************************************
118     * XviD Init Entry point
119     *
120     * Well this function initialize all internal function pointers according
121     * to the CPU features forced by the library client or autodetected (depending
122     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
123     * image colorspace transformation tables.
124     *
125     * Returned value : XVID_ERR_OK
126     *                  + API_VERSION in the input XVID_INIT_PARAM structure
127     *                  + core build  "   "    "       "               "
128     *
129     ****************************************************************************/
130    
131  int  int
132  xvid_init(void *handle,  xvid_init(void *handle,
133                    int opt,                    int opt,
# Line 66  Line 139 
139    
140          init_param = (XVID_INIT_PARAM *) param1;          init_param = (XVID_INIT_PARAM *) param1;
141    
142          // force specific cpu settings?          /* Inform the client the API version */
143          if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0)          init_param->api_version = API_VERSION;
144    
145            /* Inform the client the core build - unused because we're still alpha */
146            init_param->core_build = 1000;
147    
148            /* Do we have to force CPU features  ? */
149            if ((init_param->cpu_flags & XVID_CPU_FORCE)) {
150    
151                  cpu_flags = init_param->cpu_flags;                  cpu_flags = init_param->cpu_flags;
         else {  
152    
153  #ifdef ARCH_X86          } else {
154    
155                  cpu_flags = check_cpu_features();                  cpu_flags = check_cpu_features();
156  #else  
157                  cpu_flags = 0;  #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE)
158                    if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
159                            cpu_flags &= ~XVID_CPU_SSE;
160    
161                    if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
162                            cpu_flags &= ~XVID_CPU_SSE2;
163  #endif  #endif
164            }
165    
166            if ((init_param->cpu_flags & XVID_CPU_CHKONLY))
167            {
168                  init_param->cpu_flags = cpu_flags;                  init_param->cpu_flags = cpu_flags;
169                    return XVID_ERR_OK;
170          }          }
171    
172          // initialize the function pointers          init_param->cpu_flags = cpu_flags;
173    
174    
175            /* Initialize the function pointers */
176          idct_int32_init();          idct_int32_init();
177          init_vlc_tables();          init_vlc_tables();
178    
179            /* Fixed Point Forward/Inverse DCT transformations */
180          fdct = fdct_int32;          fdct = fdct_int32;
181          idct = idct_int32;          idct = idct_int32;
182    
183            /* Only needed on PPC Altivec archs */
184          sadInit = 0;          sadInit = 0;
185    
186            /* Restore FPU context : emms_c is a nop functions */
187          emms = emms_c;          emms = emms_c;
188    
189            /* Quantization functions */
190          quant_intra = quant_intra_c;          quant_intra = quant_intra_c;
191          dequant_intra = dequant_intra_c;          dequant_intra = dequant_intra_c;
192          quant_inter = quant_inter_c;          quant_inter = quant_inter_c;
# Line 100  Line 197 
197          quant4_inter = quant4_inter_c;          quant4_inter = quant4_inter_c;
198          dequant4_inter = dequant4_inter_c;          dequant4_inter = dequant4_inter_c;
199    
200            /* Block transfer related functions */
201          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
202          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
203          transfer_8to16sub = transfer_8to16sub_c;          transfer_8to16sub = transfer_8to16sub_c;
# Line 107  Line 205 
205          transfer_16to8add = transfer_16to8add_c;          transfer_16to8add = transfer_16to8add_c;
206          transfer8x8_copy = transfer8x8_copy_c;          transfer8x8_copy = transfer8x8_copy_c;
207    
208            /* Image interpolation related functions */
209          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;
210          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;
211          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
212    
213            /* Initialize internal colorspace transformation tables */
214          colorspace_init();          colorspace_init();
215    
216            /* All colorspace transformation functions User Format->YV12 */
217          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
218          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
219          rgb24_to_yv12 = rgb24_to_yv12_c;          rgb24_to_yv12 = rgb24_to_yv12_c;
# Line 121  Line 222 
222          yuyv_to_yv12 = yuyv_to_yv12_c;          yuyv_to_yv12 = yuyv_to_yv12_c;
223          uyvy_to_yv12 = uyvy_to_yv12_c;          uyvy_to_yv12 = uyvy_to_yv12_c;
224    
225            /* All colorspace transformation functions YV12->User format */
226          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
227          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
228          yv12_to_rgb24 = yv12_to_rgb24_c;          yv12_to_rgb24 = yv12_to_rgb24_c;
# Line 129  Line 231 
231          yv12_to_yuyv = yv12_to_yuyv_c;          yv12_to_yuyv = yv12_to_yuyv_c;
232          yv12_to_uyvy = yv12_to_uyvy_c;          yv12_to_uyvy = yv12_to_uyvy_c;
233    
234            /* Functions used in motion estimation algorithms */
235          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
236          sad16 = sad16_c;          sad16 = sad16_c;
         sad16bi = sad16bi_c;  
237          sad8 = sad8_c;          sad8 = sad8_c;
238            sad16bi  = sad16bi_c;
239            sad8bi   = sad8bi_c;
240          dev16 = dev16_c;          dev16 = dev16_c;
241    
242            Halfpel8_Refine = Halfpel8_Refine_c;
243    
244  #ifdef ARCH_X86  #ifdef ARCH_X86
245          if ((cpu_flags & XVID_CPU_MMX) > 0) {          if ((cpu_flags & XVID_CPU_MMX) > 0) {
246    
247                    /* Forward and Inverse Discrete Cosine Transformation functions */
248                  fdct = fdct_mmx;                  fdct = fdct_mmx;
249                  idct = idct_mmx;                  idct = idct_mmx;
250    
251                    /* To restore FPU context after mmx use */
252                  emms = emms_mmx;                  emms = emms_mmx;
253    
254                    /* Quantization related functions */
255                  quant_intra = quant_intra_mmx;                  quant_intra = quant_intra_mmx;
256                  dequant_intra = dequant_intra_mmx;                  dequant_intra = dequant_intra_mmx;
257                  quant_inter = quant_inter_mmx;                  quant_inter = quant_inter_mmx;
# Line 152  Line 262 
262                  quant4_inter = quant4_inter_mmx;                  quant4_inter = quant4_inter_mmx;
263                  dequant4_inter = dequant4_inter_mmx;                  dequant4_inter = dequant4_inter_mmx;
264    
265                    /* Block related functions */
266                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
267                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
268                  transfer_8to16sub = transfer_8to16sub_mmx;                  transfer_8to16sub = transfer_8to16sub_mmx;
269                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
270                  transfer_16to8add = transfer_16to8add_mmx;                  transfer_16to8add = transfer_16to8add_mmx;
271                  transfer8x8_copy = transfer8x8_copy_mmx;                  transfer8x8_copy = transfer8x8_copy_mmx;
272    
273    
274                    /* Image Interpolation related functions */
275                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;
276                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;
277                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
278    
279                    /* Image RGB->YV12 related functions */
280                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  rgb24_to_yv12 = rgb24_to_yv12_mmx;
281                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  rgb32_to_yv12 = rgb32_to_yv12_mmx;
282                  yuv_to_yv12 = yuv_to_yv12_mmx;                  yuv_to_yv12 = yuv_to_yv12_mmx;
283                  yuyv_to_yv12 = yuyv_to_yv12_mmx;                  yuyv_to_yv12 = yuyv_to_yv12_mmx;
284                  uyvy_to_yv12 = uyvy_to_yv12_mmx;                  uyvy_to_yv12 = uyvy_to_yv12_mmx;
285    
286                    /* Image YV12->RGB related functions */
287                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_rgb24 = yv12_to_rgb24_mmx;
288                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_rgb32 = yv12_to_rgb32_mmx;
289                  yv12_to_yuyv = yv12_to_yuyv_mmx;                  yv12_to_yuyv = yv12_to_yuyv_mmx;
290                  yv12_to_uyvy = yv12_to_uyvy_mmx;                  yv12_to_uyvy = yv12_to_uyvy_mmx;
291    
292                    /* Motion estimation related functions */
293                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
294                  sad16 = sad16_mmx;                  sad16 = sad16_mmx;
295                  sad8 = sad8_mmx;                  sad8 = sad8_mmx;
296                    sad16bi = sad16bi_mmx;
297                    sad8bi  = sad8bi_mmx;
298                  dev16 = dev16_mmx;                  dev16 = dev16_mmx;
299    
300          }          }
301    
302            /* these 3dnow functions are faster than mmx, but slower than xmm. */
303            if ((cpu_flags & XVID_CPU_3DNOW) > 0) {
304    
305                    /* ME functions */
306                    sad16bi = sad16bi_3dn;
307                    sad8bi  = sad8bi_3dn;
308            }
309    
310    
311          if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {          if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {
312    
313                    /* Inverse DCT */
314                  idct = idct_xmm;                  idct = idct_xmm;
315    
316                    /* Interpolation */
317                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;
318                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;
319                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
320    
321                    /* Quantization */
322                    dequant_intra = dequant_intra_xmm;
323                    dequant_inter = dequant_inter_xmm;
324    
325                    /* Buffer transfer */
326                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
327    
328                    /* Colorspace transformation */
329                  yuv_to_yv12 = yuv_to_yv12_xmm;                  yuv_to_yv12 = yuv_to_yv12_xmm;
330    
331                    /* ME functions */
332                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
333                  sad8 = sad8_xmm;                  sad8 = sad8_xmm;
334                    sad16bi = sad16bi_xmm;
335                    sad8bi  = sad8bi_xmm;
336                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
337    
338          }          }
339    
340          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {
341    
342                    /* Interpolation */
343                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
344                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
345                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
# Line 201  Line 347 
347    
348          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_SSE2) > 0) {
349  #ifdef EXPERIMENTAL_SSE2_CODE  #ifdef EXPERIMENTAL_SSE2_CODE
350    
351                    calc_cbp = calc_cbp_sse2;
352    
353                    /* Quantization */
354                  quant_intra = quant_intra_sse2;                  quant_intra = quant_intra_sse2;
355                  dequant_intra = dequant_intra_sse2;                  dequant_intra = dequant_intra_sse2;
356                  quant_inter = quant_inter_sse2;                  quant_inter = quant_inter_sse2;
357                  dequant_inter = dequant_inter_sse2;                  dequant_inter = dequant_inter_sse2;
358    
359                  calc_cbp = calc_cbp_sse2;                  /* ME */
360                  sad16 = sad16_sse2;                  sad16 = sad16_sse2;
361                  dev16 = dev16_sse2;                  dev16 = dev16_sse2;
362    
363                    /* Forward and Inverse DCT */
364                  idct = idct_sse2;                  idct = idct_sse2;
365                  fdct = fdct_sse2;                  fdct = fdct_sse2;
366  #endif  #endif
367          }          }
368    
369    #endif
370    
371    #ifdef ARCH_IA64
372            if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?
373              idct_ia64_init();
374              fdct = fdct_ia64;
375              idct = idct_ia64;   //not yet working, crashes
376              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
377              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
378              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
379              sad16 = sad16_ia64;
380              sad16bi = sad16bi_ia64;
381              sad8 = sad8_ia64;
382              dev16 = dev16_ia64;
383              Halfpel8_Refine = Halfpel8_Refine_ia64;
384              quant_intra = quant_intra_ia64;
385              dequant_intra = dequant_intra_ia64;
386              quant_inter = quant_inter_ia64;
387              dequant_inter = dequant_inter_ia64;
388              transfer_8to16copy = transfer_8to16copy_ia64;
389              transfer_16to8copy = transfer_16to8copy_ia64;
390              transfer_8to16sub = transfer_8to16sub_ia64;
391              transfer_8to16sub2 = transfer_8to16sub2_ia64;
392              transfer_16to8add = transfer_16to8add_ia64;
393              transfer8x8_copy = transfer8x8_copy_ia64;
394              DEBUG("Using IA-64 assembler routines.\n");
395            }
396  #endif  #endif
397    
398  #ifdef ARCH_PPC  #ifdef ARCH_PPC
399  #ifdef ARCH_PPC_ALTIVEC  #ifdef ARCH_PPC_ALTIVEC
400          calc_cbp = calc_cbp_altivec;          calc_cbp = calc_cbp_altivec;
# Line 228  Line 409 
409  #endif  #endif
410  #endif  #endif
411    
         // API version  
         init_param->api_version = API_VERSION;  
   
         // something clever has to be done for this  
         init_param->core_build = 1000;  
   
412          return XVID_ERR_OK;          return XVID_ERR_OK;
413  }  }
414    
415    /*****************************************************************************
416     * XviD Native decoder entry point
417     *
418     * This function is just a wrapper to all the option cases.
419     *
420     * Returned values : XVID_ERR_FAIL when opt is invalid
421     *                   else returns the wrapped function result
422     *
423     ****************************************************************************/
424    
425  int  int
426  xvid_decore(void *handle,  xvid_decore(void *handle,
427                          int opt,                          int opt,
# Line 259  Line 444 
444  }  }
445    
446    
447    /*****************************************************************************
448     * XviD Native encoder entry point
449     *
450     * This function is just a wrapper to all the option cases.
451     *
452     * Returned values : XVID_ERR_FAIL when opt is invalid
453     *                   else returns the wrapped function result
454     *
455     ****************************************************************************/
456    
457  int  int
458  xvid_encore(void *handle,  xvid_encore(void *handle,
459                          int opt,                          int opt,
# Line 267  Line 462 
462  {  {
463          switch (opt) {          switch (opt) {
464          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
465    #ifdef BFRAMES
466                    if (((Encoder *) handle)->mbParam.max_bframes >= 0)
467                    return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,
468                                                              (XVID_ENC_STATS *) param2);
469                    else
470    #endif
471                  return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,                  return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,
472                                                            (XVID_ENC_STATS *) param2);                                                            (XVID_ENC_STATS *) param2);
473    

Legend:
Removed from v.195  
changed lines
  Added in v.408

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