[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 1.15, Wed Jun 12 20:38:40 2002 UTC revision 1.32, Sun Jul 21 23:34:08 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   *      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
7   *      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 24 
24   *   *
25   *      You should have received a copy of the GNU General Public License   *      You should have received a copy of the GNU General Public License
26   *      along with this program; if not, write to the Free Software   *      along with this program; if not, write to the Free Software
27   *      Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
28   *   *
29   *************************************************************************/   ****************************************************************************/
30    
31  /**************************************************************************  /*****************************************************************************
32   *   *
33   *      History:   *  History
34   *   *
35   *      17.03.2002      Added interpolate8x8_halfpel_hv_xmm   *      - 23.06.2002    added XVID_CPU_CHKONLY
36   *  22.12.2001  API change: added xvid_init() - Isibaar   *  - 17.03.2002        Added interpolate8x8_halfpel_hv_xmm
37   *      16.12.2001      inital version; (c)2001 peter ross <pross@cs.rmit.edu.au>   *  - 22.12.2001  API change: added xvid_init() - Isibaar
38     *  - 16.12.2001        inital version; (c)2001 peter ross <pross@cs.rmit.edu.au>
39   *   *
40   *************************************************************************/   *  $Id$
41     *
42     ****************************************************************************/
43    
44  #include "xvid.h"  #include "xvid.h"
45  #include "decoder.h"  #include "decoder.h"
# Line 50  Line 52 
52  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
53  #include "quant/quant_h263.h"  #include "quant/quant_h263.h"
54  #include "quant/quant_mpeg4.h"  #include "quant/quant_mpeg4.h"
55    #include "motion/motion.h"
56  #include "motion/sad.h"  #include "motion/sad.h"
57  #include "utils/emms.h"  #include "utils/emms.h"
58  #include "utils/timer.h"  #include "utils/timer.h"
59  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
60    
61    #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE)
62    
63    #ifdef WIN32
64    #include <windows.h>
65    #else
66    #include <signal.h>
67    #include <setjmp.h>
68    #endif
69    
70    
71    #ifndef WIN32
72    
73    static jmp_buf mark;
74    
75    static void
76    sigill_handler(int signal)
77    {
78       longjmp(mark, 1);
79    }
80    #endif
81    
82    
83    /*
84    calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled
85    return values:
86    -1 : could not determine
87    0  : SIGILL was *not* signalled
88    1  : SIGILL was signalled
89    */
90    
91    int
92    sigill_check(void (*func)())
93    {
94    #ifdef WIN32
95            _try {
96                    func();
97            }
98            _except(EXCEPTION_EXECUTE_HANDLER) {
99    
100                    if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
101                            return 1;
102            }
103            return 0;
104    #else
105        void * old_handler;
106        int jmpret;
107    
108    
109        old_handler = signal(SIGILL, sigill_handler);
110        if (old_handler == SIG_ERR)
111        {
112            return -1;
113        }
114    
115        jmpret = setjmp(mark);
116        if (jmpret == 0)
117        {
118            func();
119        }
120    
121        signal(SIGILL, old_handler);
122    
123        return jmpret;
124    #endif
125    }
126    #endif
127    
128    /*****************************************************************************
129     * XviD Init Entry point
130     *
131     * Well this function initialize all internal function pointers according
132     * to the CPU features forced by the library client or autodetected (depending
133     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
134     * image colorspace transformation tables.
135     *
136     * Returned value : XVID_ERR_OK
137     *                  + API_VERSION in the input XVID_INIT_PARAM structure
138     *                  + core build  "   "    "       "               "
139     *
140     ****************************************************************************/
141    
142  int  int
143  xvid_init(void *handle,  xvid_init(void *handle,
144                    int opt,                    int opt,
# Line 66  Line 150 
150    
151          init_param = (XVID_INIT_PARAM *) param1;          init_param = (XVID_INIT_PARAM *) param1;
152    
153          // force specific cpu settings?          /* Inform the client the API version */
154          if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0)          init_param->api_version = API_VERSION;
155    
156            /* Inform the client the core build - unused because we're still alpha */
157            init_param->core_build = 1000;
158    
159            /* Do we have to force CPU features  ? */
160            if ((init_param->cpu_flags & XVID_CPU_FORCE)) {
161    
162                  cpu_flags = init_param->cpu_flags;                  cpu_flags = init_param->cpu_flags;
         else {  
163    
164  #ifdef ARCH_X86          } else {
165    
166                  cpu_flags = check_cpu_features();                  cpu_flags = check_cpu_features();
167  #else  
168                  cpu_flags = 0;  #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE)
169                    if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
170                            cpu_flags &= ~XVID_CPU_SSE;
171    
172                    if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
173                            cpu_flags &= ~XVID_CPU_SSE2;
174  #endif  #endif
175            }
176    
177            if ((init_param->cpu_flags & XVID_CPU_CHKONLY))
178            {
179                  init_param->cpu_flags = cpu_flags;                  init_param->cpu_flags = cpu_flags;
180                    return XVID_ERR_OK;
181          }          }
182    
183          // initialize the function pointers          init_param->cpu_flags = cpu_flags;
184    
185    
186            /* Initialize the function pointers */
187          idct_int32_init();          idct_int32_init();
188          init_vlc_tables();          init_vlc_tables();
189    
190            /* Fixed Point Forward/Inverse DCT transformations */
191          fdct = fdct_int32;          fdct = fdct_int32;
192          idct = idct_int32;          idct = idct_int32;
193    
194            /* Only needed on PPC Altivec archs */
195          sadInit = 0;          sadInit = 0;
196    
197            /* Restore FPU context : emms_c is a nop functions */
198          emms = emms_c;          emms = emms_c;
199    
200            /* Quantization functions */
201          quant_intra = quant_intra_c;          quant_intra = quant_intra_c;
202          dequant_intra = dequant_intra_c;          dequant_intra = dequant_intra_c;
203          quant_inter = quant_inter_c;          quant_inter = quant_inter_c;
# Line 100  Line 208 
208          quant4_inter = quant4_inter_c;          quant4_inter = quant4_inter_c;
209          dequant4_inter = dequant4_inter_c;          dequant4_inter = dequant4_inter_c;
210    
211            /* Block transfer related functions */
212          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
213          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
214          transfer_8to16sub = transfer_8to16sub_c;          transfer_8to16sub = transfer_8to16sub_c;
# Line 107  Line 216 
216          transfer_16to8add = transfer_16to8add_c;          transfer_16to8add = transfer_16to8add_c;
217          transfer8x8_copy = transfer8x8_copy_c;          transfer8x8_copy = transfer8x8_copy_c;
218    
219            /* Image interpolation related functions */
220          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;
221          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;
222          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
223    
224            /* Initialize internal colorspace transformation tables */
225          colorspace_init();          colorspace_init();
226    
227            /* All colorspace transformation functions User Format->YV12 */
228          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
229          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
230          rgb24_to_yv12 = rgb24_to_yv12_c;          rgb24_to_yv12 = rgb24_to_yv12_c;
# Line 121  Line 233 
233          yuyv_to_yv12 = yuyv_to_yv12_c;          yuyv_to_yv12 = yuyv_to_yv12_c;
234          uyvy_to_yv12 = uyvy_to_yv12_c;          uyvy_to_yv12 = uyvy_to_yv12_c;
235    
236            /* All colorspace transformation functions YV12->User format */
237          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
238          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
239          yv12_to_rgb24 = yv12_to_rgb24_c;          yv12_to_rgb24 = yv12_to_rgb24_c;
# Line 129  Line 242 
242          yv12_to_yuyv = yv12_to_yuyv_c;          yv12_to_yuyv = yv12_to_yuyv_c;
243          yv12_to_uyvy = yv12_to_uyvy_c;          yv12_to_uyvy = yv12_to_uyvy_c;
244    
245            /* Functions used in motion estimation algorithms */
246          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
247          sad16 = sad16_c;          sad16 = sad16_c;
248          sad16bi = sad16bi_c;          sad16bi = sad16bi_c;
249          sad8 = sad8_c;          sad8 = sad8_c;
250          dev16 = dev16_c;          dev16 = dev16_c;
251            sad8bi   = sad8bi_c;
252            Halfpel8_Refine = Halfpel8_Refine_c;
253    
254  #ifdef ARCH_X86  #ifdef ARCH_X86
255          if ((cpu_flags & XVID_CPU_MMX) > 0) {          if ((cpu_flags & XVID_CPU_MMX) > 0) {
256    
257                    /* Forward and Inverse Discrete Cosine Transformation functions */
258                  fdct = fdct_mmx;                  fdct = fdct_mmx;
259                  idct = idct_mmx;                  idct = idct_mmx;
260    
261                    /* To restore FPU context after mmx use */
262                  emms = emms_mmx;                  emms = emms_mmx;
263    
264                    /* Quantization related functions */
265                  quant_intra = quant_intra_mmx;                  quant_intra = quant_intra_mmx;
266                  dequant_intra = dequant_intra_mmx;                  dequant_intra = dequant_intra_mmx;
267                  quant_inter = quant_inter_mmx;                  quant_inter = quant_inter_mmx;
# Line 152  Line 272 
272                  quant4_inter = quant4_inter_mmx;                  quant4_inter = quant4_inter_mmx;
273                  dequant4_inter = dequant4_inter_mmx;                  dequant4_inter = dequant4_inter_mmx;
274    
275                    /* Block related functions */
276                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
277                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
278                  transfer_8to16sub = transfer_8to16sub_mmx;                  transfer_8to16sub = transfer_8to16sub_mmx;
279                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
280                  transfer_16to8add = transfer_16to8add_mmx;                  transfer_16to8add = transfer_16to8add_mmx;
281                  transfer8x8_copy = transfer8x8_copy_mmx;                  transfer8x8_copy = transfer8x8_copy_mmx;
282    
283    
284                    /* Image Interpolation related functions */
285                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;
286                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;
287                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
288    
289                    /* Image RGB->YV12 related functions */
290                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  rgb24_to_yv12 = rgb24_to_yv12_mmx;
291                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  rgb32_to_yv12 = rgb32_to_yv12_mmx;
292                  yuv_to_yv12 = yuv_to_yv12_mmx;                  yuv_to_yv12 = yuv_to_yv12_mmx;
293                  yuyv_to_yv12 = yuyv_to_yv12_mmx;                  yuyv_to_yv12 = yuyv_to_yv12_mmx;
294                  uyvy_to_yv12 = uyvy_to_yv12_mmx;                  uyvy_to_yv12 = uyvy_to_yv12_mmx;
295    
296                    /* Image YV12->RGB related functions */
297                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_rgb24 = yv12_to_rgb24_mmx;
298                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_rgb32 = yv12_to_rgb32_mmx;
299                  yv12_to_yuyv = yv12_to_yuyv_mmx;                  yv12_to_yuyv = yv12_to_yuyv_mmx;
300                  yv12_to_uyvy = yv12_to_uyvy_mmx;                  yv12_to_uyvy = yv12_to_uyvy_mmx;
301    
302                    /* Motion estimation related functions */
303                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
304                  sad16 = sad16_mmx;                  sad16 = sad16_mmx;
305                  sad8 = sad8_mmx;                  sad8 = sad8_mmx;
# Line 181  Line 308 
308          }          }
309    
310          if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {          if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {
311    
312                    /* Inverse DCT */
313                  idct = idct_xmm;                  idct = idct_xmm;
314    
315                    /* Interpolation */
316                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;
317                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;
318                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
319    
320                    /* Quantization */
321                    dequant_intra = dequant_intra_xmm;
322                    dequant_inter = dequant_inter_xmm;
323    
324                    /* Buffer transfer */
325                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
326    
327                    /* Colorspace transformation */
328                  yuv_to_yv12 = yuv_to_yv12_xmm;                  yuv_to_yv12 = yuv_to_yv12_xmm;
329    
330                    /* ME functions */
331                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
332                    sad16bi = sad16bi_xmm;
333                  sad8 = sad8_xmm;                  sad8 = sad8_xmm;
334                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
335    
336          }          }
337    
338          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {
339    
340                    /* Interpolation */
341                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
342                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
343                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
# Line 201  Line 345 
345    
346          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_SSE2) > 0) {
347  #ifdef EXPERIMENTAL_SSE2_CODE  #ifdef EXPERIMENTAL_SSE2_CODE
348    
349                    calc_cbp = calc_cbp_sse2;
350    
351                    /* Quantization */
352                  quant_intra = quant_intra_sse2;                  quant_intra = quant_intra_sse2;
353                  dequant_intra = dequant_intra_sse2;                  dequant_intra = dequant_intra_sse2;
354                  quant_inter = quant_inter_sse2;                  quant_inter = quant_inter_sse2;
355                  dequant_inter = dequant_inter_sse2;                  dequant_inter = dequant_inter_sse2;
356    
357                  calc_cbp = calc_cbp_sse2;                  /* ME */
358                  sad16 = sad16_sse2;                  sad16 = sad16_sse2;
359                  dev16 = dev16_sse2;                  dev16 = dev16_sse2;
360    
361                    /* Forward and Inverse DCT */
362                  idct = idct_sse2;                  idct = idct_sse2;
363                  fdct = fdct_sse2;                  fdct = fdct_sse2;
364  #endif  #endif
365          }          }
366    
367    #endif
368    
369    #ifdef ARCH_IA64
370            if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?
371              idct_ia64_init();
372              fdct = fdct_ia64;
373              idct = idct_ia64;   //not yet working, crashes
374              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
375              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
376              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
377              sad16 = sad16_ia64;
378              sad16bi = sad16bi_ia64;
379              sad8 = sad8_ia64;
380              dev16 = dev16_ia64;
381              Halfpel8_Refine = Halfpel8_Refine_ia64;
382              quant_intra = quant_intra_ia64;
383              dequant_intra = dequant_intra_ia64;
384              quant_inter = quant_inter_ia64;
385              dequant_inter = dequant_inter_ia64;
386              transfer_8to16copy = transfer_8to16copy_ia64;
387              transfer_16to8copy = transfer_16to8copy_ia64;
388              transfer_8to16sub = transfer_8to16sub_ia64;
389              transfer_8to16sub2 = transfer_8to16sub2_ia64;
390              transfer_16to8add = transfer_16to8add_ia64;
391              transfer8x8_copy = transfer8x8_copy_ia64;
392              DEBUG("Using IA-64 assembler routines.\n");
393            }
394  #endif  #endif
395    
396  #ifdef ARCH_PPC  #ifdef ARCH_PPC
397  #ifdef ARCH_PPC_ALTIVEC  #ifdef ARCH_PPC_ALTIVEC
398          calc_cbp = calc_cbp_altivec;          calc_cbp = calc_cbp_altivec;
# Line 228  Line 407 
407  #endif  #endif
408  #endif  #endif
409    
         // API version  
         init_param->api_version = API_VERSION;  
   
         // something clever has to be done for this  
         init_param->core_build = 1000;  
   
410          return XVID_ERR_OK;          return XVID_ERR_OK;
411  }  }
412    
413    /*****************************************************************************
414     * XviD Native decoder entry point
415     *
416     * This function is just a wrapper to all the option cases.
417     *
418     * Returned values : XVID_ERR_FAIL when opt is invalid
419     *                   else returns the wrapped function result
420     *
421     ****************************************************************************/
422    
423  int  int
424  xvid_decore(void *handle,  xvid_decore(void *handle,
425                          int opt,                          int opt,
# Line 259  Line 442 
442  }  }
443    
444    
445    /*****************************************************************************
446     * XviD Native encoder entry point
447     *
448     * This function is just a wrapper to all the option cases.
449     *
450     * Returned values : XVID_ERR_FAIL when opt is invalid
451     *                   else returns the wrapped function result
452     *
453     ****************************************************************************/
454    
455  int  int
456  xvid_encore(void *handle,  xvid_encore(void *handle,
457                          int opt,                          int opt,
# Line 267  Line 460 
460  {  {
461          switch (opt) {          switch (opt) {
462          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
463    #ifdef BFRAMES
464                    if (((Encoder *) handle)->mbParam.max_bframes >= 0)
465                    return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,
466                                                              (XVID_ENC_STATS *) param2);
467                    else
468    #endif
469                  return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,                  return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,
470                                                            (XVID_ENC_STATS *) param2);                                                            (XVID_ENC_STATS *) param2);
471    

Legend:
Removed from v.1.15  
changed lines
  Added in v.1.32

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