[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 1.12, Wed Apr 17 10:54:19 2002 UTC revision 1.38, Wed Oct 9 14:35:57 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@xvid.org>
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:   * $Id$
32   *   *
33   *      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>  
  *  
  *************************************************************************/  
   
34    
35  #include "xvid.h"  #include "xvid.h"
36  #include "decoder.h"  #include "decoder.h"
# Line 50  Line 43 
43  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
44  #include "quant/quant_h263.h"  #include "quant/quant_h263.h"
45  #include "quant/quant_mpeg4.h"  #include "quant/quant_mpeg4.h"
46    #include "motion/motion.h"
47  #include "motion/sad.h"  #include "motion/sad.h"
48  #include "utils/emms.h"  #include "utils/emms.h"
49  #include "utils/timer.h"  #include "utils/timer.h"
50  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
51    
52  int xvid_init(void *handle, int opt, void *param1, void *param2)  #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE)
53    
54    #ifdef WIN32
55    #include <windows.h>
56    #else
57    #include <signal.h>
58    #include <setjmp.h>
59    #endif
60    
61    
62    #ifndef WIN32
63    
64    static jmp_buf mark;
65    
66    static void
67    sigill_handler(int signal)
68    {
69       longjmp(mark, 1);
70    }
71    #endif
72    
73    
74    /*
75     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled
76     * Return values:
77     * -1 : could not determine
78     * 0  : SIGILL was *not* signalled
79     * 1  : SIGILL was signalled
80     */
81    
82    int
83    sigill_check(void (*func)())
84    {
85    #ifdef WIN32
86            _try {
87                    func();
88            }
89            _except(EXCEPTION_EXECUTE_HANDLER) {
90    
91                    if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
92                            return 1;
93            }
94            return 0;
95    #else
96        void * old_handler;
97        int jmpret;
98    
99    
100        old_handler = signal(SIGILL, sigill_handler);
101        if (old_handler == SIG_ERR)
102        {
103            return -1;
104        }
105    
106        jmpret = setjmp(mark);
107        if (jmpret == 0)
108        {
109            func();
110        }
111    
112        signal(SIGILL, old_handler);
113    
114        return jmpret;
115    #endif
116    }
117    #endif
118    
119    /*****************************************************************************
120     * XviD Init Entry point
121     *
122     * Well this function initialize all internal function pointers according
123     * to the CPU features forced by the library client or autodetected (depending
124     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
125     * image colorspace transformation tables.
126     *
127     * Returned value : XVID_ERR_OK
128     *                  + API_VERSION in the input XVID_INIT_PARAM structure
129     *                  + core build  "   "    "       "               "
130     *
131     ****************************************************************************/
132    
133    int
134    xvid_init(void *handle,
135                      int opt,
136                      void *param1,
137                      void *param2)
138  {  {
139          int cpu_flags;          int cpu_flags;
140          XVID_INIT_PARAM *init_param;          XVID_INIT_PARAM *init_param;
141    
142          init_param = (XVID_INIT_PARAM *) param1;          init_param = (XVID_INIT_PARAM *) param1;
143    
144          // force specific cpu settings?          /* Inform the client the API version */
145          if((init_param->cpu_flags & XVID_CPU_FORCE) > 0)          init_param->api_version = API_VERSION;
146    
147            /* Inform the client the core build - unused because we're still alpha */
148            init_param->core_build = 1000;
149    
150            /* Do we have to force CPU features  ? */
151            if ((init_param->cpu_flags & XVID_CPU_FORCE)) {
152    
153                  cpu_flags = init_param->cpu_flags;                  cpu_flags = init_param->cpu_flags;
         else {  
154    
155  #ifdef ARCH_X86          } else {
156    
157                  cpu_flags = check_cpu_features();                  cpu_flags = check_cpu_features();
158  #else  
159                  cpu_flags = 0;  #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE)
160                    if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
161                            cpu_flags &= ~XVID_CPU_SSE;
162    
163                    if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
164                            cpu_flags &= ~XVID_CPU_SSE2;
165  #endif  #endif
166            }
167    
168            if ((init_param->cpu_flags & XVID_CPU_CHKONLY))
169            {
170                  init_param->cpu_flags = cpu_flags;                  init_param->cpu_flags = cpu_flags;
171                    return XVID_ERR_OK;
172          }          }
173    
174          // initialize the function pointers          init_param->cpu_flags = cpu_flags;
175    
176    
177            /* Initialize the function pointers */
178          idct_int32_init();          idct_int32_init();
179          init_vlc_tables();          init_vlc_tables();
180    
181            /* Fixed Point Forward/Inverse DCT transformations */
182          fdct = fdct_int32;          fdct = fdct_int32;
183          idct = idct_int32;          idct = idct_int32;
184    
185            /* Only needed on PPC Altivec archs */
186          sadInit = 0;          sadInit = 0;
187    
188            /* Restore FPU context : emms_c is a nop functions */
189          emms = emms_c;          emms = emms_c;
190    
191            /* Quantization functions */
192          quant_intra = quant_intra_c;          quant_intra = quant_intra_c;
193          dequant_intra = dequant_intra_c;          dequant_intra = dequant_intra_c;
194          quant_inter = quant_inter_c;          quant_inter = quant_inter_c;
# Line 96  Line 199 
199          quant4_inter = quant4_inter_c;          quant4_inter = quant4_inter_c;
200          dequant4_inter = dequant4_inter_c;          dequant4_inter = dequant4_inter_c;
201    
202            /* Block transfer related functions */
203          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
204          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
205          transfer_8to16sub = transfer_8to16sub_c;          transfer_8to16sub = transfer_8to16sub_c;
# Line 103  Line 207 
207          transfer_16to8add = transfer_16to8add_c;          transfer_16to8add = transfer_16to8add_c;
208          transfer8x8_copy = transfer8x8_copy_c;          transfer8x8_copy = transfer8x8_copy_c;
209    
210            /* Image interpolation related functions */
211          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;
212          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;
213          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
214    
215            /* Initialize internal colorspace transformation tables */
216          colorspace_init();          colorspace_init();
217    
218            /* All colorspace transformation functions User Format->YV12 */
219          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
220          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
221          rgb24_to_yv12 = rgb24_to_yv12_c;          rgb24_to_yv12 = rgb24_to_yv12_c;
# Line 117  Line 224 
224          yuyv_to_yv12 = yuyv_to_yv12_c;          yuyv_to_yv12 = yuyv_to_yv12_c;
225          uyvy_to_yv12 = uyvy_to_yv12_c;          uyvy_to_yv12 = uyvy_to_yv12_c;
226    
227            /* All colorspace transformation functions YV12->User format */
228          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
229          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
230          yv12_to_rgb24 = yv12_to_rgb24_c;          yv12_to_rgb24 = yv12_to_rgb24_c;
# Line 125  Line 233 
233          yv12_to_yuyv = yv12_to_yuyv_c;          yv12_to_yuyv = yv12_to_yuyv_c;
234          yv12_to_uyvy = yv12_to_uyvy_c;          yv12_to_uyvy = yv12_to_uyvy_c;
235    
236            /* Functions used in motion estimation algorithms */
237          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
238          sad16 = sad16_c;          sad16 = sad16_c;
         sad16bi = sad16bi_c;  
239          sad8 = sad8_c;          sad8 = sad8_c;
240            sad16bi  = sad16bi_c;
241            sad8bi   = sad8bi_c;
242          dev16 = dev16_c;          dev16 = dev16_c;
243    
244            Halfpel8_Refine = Halfpel8_Refine_c;
245    
246  #ifdef ARCH_X86  #ifdef ARCH_X86
247          if((cpu_flags & XVID_CPU_MMX) > 0) {          if((cpu_flags & XVID_CPU_MMX) > 0) {
248    
249                    /* Forward and Inverse Discrete Cosine Transformation functions */
250                  fdct = fdct_mmx;                  fdct = fdct_mmx;
251                  idct = idct_mmx;                  idct = idct_mmx;
252    
253                    /* To restore FPU context after mmx use */
254                  emms = emms_mmx;                  emms = emms_mmx;
255    
256                    /* Quantization related functions */
257                  quant_intra = quant_intra_mmx;                  quant_intra = quant_intra_mmx;
258                  dequant_intra = dequant_intra_mmx;                  dequant_intra = dequant_intra_mmx;
259                  quant_inter = quant_inter_mmx;                  quant_inter = quant_inter_mmx;
# Line 148  Line 264 
264                  quant4_inter = quant4_inter_mmx;                  quant4_inter = quant4_inter_mmx;
265                  dequant4_inter = dequant4_inter_mmx;                  dequant4_inter = dequant4_inter_mmx;
266    
267                    /* Block related functions */
268                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
269                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
270                  transfer_8to16sub = transfer_8to16sub_mmx;                  transfer_8to16sub = transfer_8to16sub_mmx;
271                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
272                  transfer_16to8add = transfer_16to8add_mmx;                  transfer_16to8add = transfer_16to8add_mmx;
273                  transfer8x8_copy = transfer8x8_copy_mmx;                  transfer8x8_copy = transfer8x8_copy_mmx;
274    
275    
276                    /* Image Interpolation related functions */
277                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;
278                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;
279                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
280    
281                    /* Image RGB->YV12 related functions */
282                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  rgb24_to_yv12 = rgb24_to_yv12_mmx;
283                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  rgb32_to_yv12 = rgb32_to_yv12_mmx;
284                  yuv_to_yv12 = yuv_to_yv12_mmx;                  yuv_to_yv12 = yuv_to_yv12_mmx;
285                  yuyv_to_yv12 = yuyv_to_yv12_mmx;                  yuyv_to_yv12 = yuyv_to_yv12_mmx;
286                  uyvy_to_yv12 = uyvy_to_yv12_mmx;                  uyvy_to_yv12 = uyvy_to_yv12_mmx;
287    
288                    /* Image YV12->RGB related functions */
289                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_rgb24 = yv12_to_rgb24_mmx;
290                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_rgb32 = yv12_to_rgb32_mmx;
291                  yv12_to_yuyv = yv12_to_yuyv_mmx;                  yv12_to_yuyv = yv12_to_yuyv_mmx;
292                  yv12_to_uyvy = yv12_to_uyvy_mmx;                  yv12_to_uyvy = yv12_to_uyvy_mmx;
293    
294                    /* Motion estimation related functions */
295                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
296                  sad16 = sad16_mmx;                  sad16 = sad16_mmx;
297                  sad8 = sad8_mmx;                  sad8 = sad8_mmx;
298                    sad16bi = sad16bi_mmx;
299                    sad8bi  = sad8bi_mmx;
300                  dev16 = dev16_mmx;                  dev16 = dev16_mmx;
301    
302          }          }
303    
304            /* these 3dnow functions are faster than mmx, but slower than xmm. */
305            if ((cpu_flags & XVID_CPU_3DNOW) > 0) {
306    
307                    /* ME functions */
308                    sad16bi = sad16bi_3dn;
309                    sad8bi  = sad8bi_3dn;
310            }
311    
312    
313          if((cpu_flags & XVID_CPU_MMXEXT) > 0) {          if((cpu_flags & XVID_CPU_MMXEXT) > 0) {
314    
315                    /* Inverse DCT */
316                  idct = idct_xmm;                  idct = idct_xmm;
317    
318                    /* Interpolation */
319                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;
320                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;
321                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
322    
323                    /* Quantization */
324                    dequant_intra = dequant_intra_xmm;
325                    dequant_inter = dequant_inter_xmm;
326    
327                    /* Buffer transfer */
328                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
329    
330                    /* Colorspace transformation */
331                  yuv_to_yv12 = yuv_to_yv12_xmm;                  yuv_to_yv12 = yuv_to_yv12_xmm;
332    
333                    /* ME functions */
334                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
335                  sad8 = sad8_xmm;                  sad8 = sad8_xmm;
336                    sad16bi = sad16bi_xmm;
337                    sad8bi  = sad8bi_xmm;
338                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
339    
340          }          }
341    
342          if((cpu_flags & XVID_CPU_3DNOW) > 0) {          if((cpu_flags & XVID_CPU_3DNOW) > 0) {
343    
344                    /* Interpolation */
345                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
346                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
347                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
348          }          }
349    
350          if((cpu_flags & XVID_CPU_SSE2) > 0) {          if((cpu_flags & XVID_CPU_SSE2) > 0) {
351  //              calc_cbp = calc_cbp_sse2;  #ifdef EXPERIMENTAL_SSE2_CODE
352  //              sad16 = sad16_sse2;  
353  //              quant_inter = quant_inter_sse2;                  calc_cbp = calc_cbp_sse2;
354  //              dequant_inter = dequant_inter_sse2;  
355                    /* Quantization */
356                    quant_intra   = quant_intra_sse2;
357                    dequant_intra = dequant_intra_sse2;
358                    quant_inter   = quant_inter_sse2;
359                    dequant_inter = dequant_inter_sse2;
360    
361                    /* ME */
362                    sad16    = sad16_sse2;
363                    dev16    = dev16_sse2;
364    
365                    /* Forward and Inverse DCT */
366                    idct  = idct_sse2;
367                    fdct = fdct_sse2;
368    #endif
369          }          }
370    
371  #endif  #endif
372    
373    #ifdef ARCH_IA64
374            if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?
375              idct_ia64_init();
376              fdct = fdct_ia64;
377              idct = idct_ia64;
378              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
379              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
380              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
381              sad16 = sad16_ia64;
382              sad16bi = sad16bi_ia64;
383              sad8 = sad8_ia64;
384              dev16 = dev16_ia64;
385              Halfpel8_Refine = Halfpel8_Refine_ia64;
386              quant_intra = quant_intra_ia64;
387              dequant_intra = dequant_intra_ia64;
388              quant_inter = quant_inter_ia64;
389              dequant_inter = dequant_inter_ia64;
390              transfer_8to16copy = transfer_8to16copy_ia64;
391              transfer_16to8copy = transfer_16to8copy_ia64;
392              transfer_8to16sub = transfer_8to16sub_ia64;
393              transfer_8to16sub2 = transfer_8to16sub2_ia64;
394              transfer_16to8add = transfer_16to8add_ia64;
395              transfer8x8_copy = transfer8x8_copy_ia64;
396    //        DEBUG("Using IA-64 assembler routines.\n");
397            }
398    #endif
399    
400  #ifdef ARCH_PPC  #ifdef ARCH_PPC
401  #ifdef ARCH_PPC_ALTIVEC  #ifdef ARCH_PPC_ALTIVEC
402          calc_cbp = calc_cbp_altivec;          calc_cbp = calc_cbp_altivec;
# Line 217  Line 411 
411  #endif  #endif
412  #endif  #endif
413    
         // API version  
         init_param->api_version = API_VERSION;  
   
         // something clever has to be done for this  
         init_param->core_build = 1000;  
   
414          return XVID_ERR_OK;          return XVID_ERR_OK;
415  }  }
416    
417  int xvid_decore(void * handle, int opt, void * param1, void * param2)  /*****************************************************************************
418  {   * XviD Native decoder entry point
419          switch (opt)   *
420     * This function is just a wrapper to all the option cases.
421     *
422     * Returned values : XVID_ERR_FAIL when opt is invalid
423     *                   else returns the wrapped function result
424     *
425     ****************************************************************************/
426    
427    int
428    xvid_decore(void *handle,
429                            int opt,
430                            void *param1,
431                            void *param2)
432          {          {
433            switch (opt) {
434          case XVID_DEC_DECODE :          case XVID_DEC_DECODE :
435          return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);          return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);
436    
# Line 245  Line 446 
446  }  }
447    
448    
449  int xvid_encore(void * handle, int opt, void * param1, void * param2)  /*****************************************************************************
450  {   * XviD Native encoder entry point
451          switch (opt)   *
452     * This function is just a wrapper to all the option cases.
453     *
454     * Returned values : XVID_ERR_FAIL when opt is invalid
455     *                   else returns the wrapped function result
456     *
457     ****************************************************************************/
458    
459    int
460    xvid_encore(void *handle,
461                            int opt,
462                            void *param1,
463                            void *param2)
464          {          {
465            switch (opt) {
466          case XVID_ENC_ENCODE :          case XVID_ENC_ENCODE :
467          return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, (XVID_ENC_STATS *) param2);                  return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,
468                                                              (XVID_ENC_STATS *) param2);
469    
470          case XVID_ENC_CREATE :          case XVID_ENC_CREATE :
471          return encoder_create((XVID_ENC_PARAM *) param1);          return encoder_create((XVID_ENC_PARAM *) param1);

Legend:
Removed from v.1.12  
changed lines
  Added in v.1.38

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