[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 1.18, Fri Jun 14 13:27:52 2002 UTC revision 1.45.2.17, Mon Oct 27 01:03:06 2003 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) 2001-2003 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  
  *  
  *  - 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 47  Line 36 
36  #include "dct/fdct.h"  #include "dct/fdct.h"
37  #include "image/colorspace.h"  #include "image/colorspace.h"
38  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
39    #include "image/reduced.h"
40  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
41  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
42  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
43    #include "motion/motion.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    
50    #if defined(_DEBUG)
51    unsigned int xvid_debug = 0; /* xvid debug mask */
52    #endif
53    
54    #if defined(ARCH_IS_IA32)
55    #if defined(_MSC_VER)
56    #       include <windows.h>
57    #else
58    #       include <signal.h>
59    #       include <setjmp.h>
60    
61            static jmp_buf mark;
62    
63            static void
64            sigill_handler(int signal)
65            {
66               longjmp(mark, 1);
67            }
68    #endif
69    
70    
71    /*
72     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
73     * signalled
74     *
75     * Return values:
76     *  -1 : could not determine
77     *   0 : SIGILL was *not* signalled
78     *   1 : SIGILL was signalled
79     */
80    
81    int
82    sigill_check(void (*func)())
83    {
84    #if defined(_MSC_VER)
85            _try {
86                    func();
87            }
88            _except(EXCEPTION_EXECUTE_HANDLER) {
89    
90                    if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
91                            return 1;
92            }
93            return 0;
94    #else
95        void * old_handler;
96        int jmpret;
97    
98    
99        old_handler = signal(SIGILL, sigill_handler);
100        if (old_handler == SIG_ERR)
101        {
102            return -1;
103        }
104    
105        jmpret = setjmp(mark);
106        if (jmpret == 0)
107        {
108            func();
109        }
110    
111        signal(SIGILL, old_handler);
112    
113        return jmpret;
114    #endif
115    }
116    #endif
117    
118    
119    /* detect cpu flags  */
120    static unsigned int
121    detect_cpu_flags()
122    {
123            /* enable native assembly optimizations by default */
124            unsigned int cpu_flags = XVID_CPU_ASM;
125    
126    #if defined(ARCH_IS_IA32)
127            cpu_flags |= check_cpu_features();
128            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
129                    cpu_flags &= ~XVID_CPU_SSE;
130    
131            if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
132                    cpu_flags &= ~XVID_CPU_SSE2;
133    #endif
134    
135    #if defined(ARCH_IS_PPC)
136    #if defined(ARCH_IS_PPC_ALTIVEC)
137            cpu_flags |= XVID_CPU_ALTIVEC;
138    #endif
139    #endif
140    
141            return cpu_flags;
142    }
143    
144    
145  /*****************************************************************************  /*****************************************************************************
146   * XviD Init Entry point   * XviD Init Entry point
# Line 69  Line 156 
156   *   *
157   ****************************************************************************/   ****************************************************************************/
158    
 int  
 xvid_init(void *handle,  
                   int opt,  
                   void *param1,  
                   void *param2)  
 {  
         int cpu_flags;  
         XVID_INIT_PARAM *init_param;  
159    
160          init_param = (XVID_INIT_PARAM *) param1;  static
161    int xvid_gbl_init(xvid_gbl_init_t * init)
162    {
163            unsigned int cpu_flags;
164    
165          /* Do we have to force CPU features  ? */          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
166          if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0) {                  return XVID_ERR_VERSION;
                 cpu_flags = init_param->cpu_flags;  
         } else {  
167    
168  #ifdef ARCH_X86          cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
                 cpu_flags = check_cpu_features();  
 #else  
                 cpu_flags = 0;  
 #endif  
                 init_param->cpu_flags = cpu_flags;  
         }  
169    
170          /* Initialize the function pointers */          /* Initialize the function pointers */
171          idct_int32_init();          idct_int32_init();
# Line 107  Line 181 
181          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
182          emms = emms_c;          emms = emms_c;
183    
184            /* Qpel stuff */
185            xvid_QP_Funcs = &xvid_QP_Funcs_C;
186            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
187            xvid_Init_QP();
188    
189          /* Quantization functions */          /* Quantization functions */
190          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
191          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
192          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
193          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
194    
195          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
196          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
197          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
198          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
199    
200          /* Block transfer related functions */          /* 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;
204            transfer_8to16subro  = transfer_8to16subro_c;
205          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
206          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
207          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
208    
209            /* Interlacing functions */
210            MBFieldTest = MBFieldTest_c;
211    
212          /* Image interpolation related functions */          /* Image interpolation related functions */
213          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;
214          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
215          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
216    
217            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
218            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
219            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
220    
221            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
222            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
223            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
224    
225            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
226            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
227    
228            interpolate8x8_avg2 = interpolate8x8_avg2_c;
229            interpolate8x8_avg4 = interpolate8x8_avg4_c;
230    
231            /* reduced resolution */
232            copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C;
233            add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C;
234            vfilter_31 = xvid_VFilter_31_C;
235            hfilter_31 = xvid_HFilter_31_C;
236            filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C;
237            filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C;
238    
239          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
240          colorspace_init();          colorspace_init();
241    
242          /* All colorspace transformation functions User Format->YV12 */          /* All colorspace transformation functions User Format->YV12 */
243            yv12_to_yv12    = yv12_to_yv12_c;
244          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
245          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
246          rgb24_to_yv12  = rgb24_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
247          rgb32_to_yv12  = rgb32_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
248          yuv_to_yv12    = yuv_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
249            rgba_to_yv12    = rgba_to_yv12_c;
250          yuyv_to_yv12   = yuyv_to_yv12_c;          yuyv_to_yv12   = yuyv_to_yv12_c;
251          uyvy_to_yv12   = uyvy_to_yv12_c;          uyvy_to_yv12   = uyvy_to_yv12_c;
252    
253            rgb555i_to_yv12 = rgb555i_to_yv12_c;
254            rgb565i_to_yv12 = rgb565i_to_yv12_c;
255            bgri_to_yv12    = bgri_to_yv12_c;
256            bgrai_to_yv12   = bgrai_to_yv12_c;
257            abgri_to_yv12   = abgri_to_yv12_c;
258            rgbai_to_yv12   = rgbai_to_yv12_c;
259            yuyvi_to_yv12   = yuyvi_to_yv12_c;
260            uyvyi_to_yv12   = uyvyi_to_yv12_c;
261    
262    
263          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
264          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
265          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
266          yv12_to_rgb24  = yv12_to_rgb24_c;          yv12_to_bgr     = yv12_to_bgr_c;
267          yv12_to_rgb32  = yv12_to_rgb32_c;          yv12_to_bgra    = yv12_to_bgra_c;
268          yv12_to_yuv    = yv12_to_yuv_c;          yv12_to_abgr    = yv12_to_abgr_c;
269            yv12_to_rgba    = yv12_to_rgba_c;
270          yv12_to_yuyv   = yv12_to_yuyv_c;          yv12_to_yuyv   = yv12_to_yuyv_c;
271          yv12_to_uyvy   = yv12_to_uyvy_c;          yv12_to_uyvy   = yv12_to_uyvy_c;
272    
273            yv12_to_rgb555i = yv12_to_rgb555i_c;
274            yv12_to_rgb565i = yv12_to_rgb565i_c;
275            yv12_to_bgri    = yv12_to_bgri_c;
276            yv12_to_bgrai   = yv12_to_bgrai_c;
277            yv12_to_abgri   = yv12_to_abgri_c;
278            yv12_to_rgbai   = yv12_to_rgbai_c;
279            yv12_to_yuyvi   = yv12_to_yuyvi_c;
280            yv12_to_uyvyi   = yv12_to_uyvyi_c;
281    
282          /* Functions used in motion estimation algorithms */          /* Functions used in motion estimation algorithms */
283          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
284          sad16    = sad16_c;          sad16    = sad16_c;
         sad16bi  = sad16bi_c;  
285          sad8     = sad8_c;          sad8     = sad8_c;
286            sad16bi  = sad16bi_c;
287            sad8bi   = sad8bi_c;
288          dev16    = dev16_c;          dev16    = dev16_c;
289            sad16v   = sad16v_c;
290    
291    /*      Halfpel8_Refine = Halfpel8_Refine_c; */
292    
293    #if defined(ARCH_IS_IA32)
294    
295            if ((cpu_flags & XVID_CPU_ASM)) {
296                    vfilter_31 = xvid_VFilter_31_x86;
297                    hfilter_31 = xvid_HFilter_31_x86;
298            }
299    
300            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
301                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
302                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2))
303            {
304                    /* Restore FPU context : emms_c is a nop functions */
305                    emms = emms_mmx;
306            }
307    
308  #ifdef ARCH_X86          if ((cpu_flags & XVID_CPU_MMX)) {
         if ((cpu_flags & XVID_CPU_MMX) > 0) {  
309    
310                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
311                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
312                  idct = idct_mmx;                  idct = idct_mmx;
313    
314                  /* To restore FPU context after mmx use */                  /* Qpel stuff */
315                  emms = emms_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
316                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
317    
318                  /* Quantization related functions */                  /* Quantization related functions */
319                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
320                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
321                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
322                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
323    
324                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_intra   = quant_mpeg_intra_mmx;
325                  dequant4_intra = dequant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
326                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
327                  dequant4_inter = dequant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
328    
329                  /* Block related functions */                  /* Block related functions */
330                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
331                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
332                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
333                    transfer_8to16subro  = transfer_8to16subro_mmx;
334                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
335                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
336                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
337    
338                    /* Interlacing Functions */
339                    MBFieldTest = MBFieldTest_mmx;
340    
341                  /* Image Interpolation related functions */                  /* Image Interpolation related functions */
342                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;
343                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
344                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
345    
346                  /* Image RGB->YV12 related functions */                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
347                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
348                  rgb32_to_yv12 = rgb32_to_yv12_mmx;  
349                  yuv_to_yv12   = yuv_to_yv12_mmx;                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
350                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
351    
352                    /* reduced resolution */
353                    copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx;
354                    add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx;
355                    hfilter_31 = xvid_HFilter_31_mmx;
356                    filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx;
357                    filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx;
358    
359                    /* image input xxx_to_yv12 related functions */
360                    yv12_to_yv12  = yv12_to_yv12_mmx;
361                    bgr_to_yv12   = bgr_to_yv12_mmx;
362                    bgra_to_yv12  = bgra_to_yv12_mmx;
363                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
364                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
365    
366                  /* Image YV12->RGB related functions */                  /* image output yv12_to_xxx related functions */
367                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
368                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgra  = yv12_to_bgra_mmx;
369                  yv12_to_yuyv  = yv12_to_yuyv_mmx;                  yv12_to_yuyv  = yv12_to_yuyv_mmx;
370                  yv12_to_uyvy  = yv12_to_uyvy_mmx;                  yv12_to_uyvy  = yv12_to_uyvy_mmx;
371    
372                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
373                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
374    
375                  /* Motion estimation related functions */                  /* Motion estimation related functions */
376                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
377                  sad16    = sad16_mmx;                  sad16    = sad16_mmx;
378                  sad8     = sad8_mmx;                  sad8     = sad8_mmx;
379                    sad16bi = sad16bi_mmx;
380                    sad8bi  = sad8bi_mmx;
381                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
382                    sad16v   = sad16v_mmx;
383            }
384    
385            /* these 3dnow functions are faster than mmx, but slower than xmm. */
386            if ((cpu_flags & XVID_CPU_3DNOW)) {
387    
388                    emms = emms_3dn;
389    
390                    /* ME functions */
391                    sad16bi = sad16bi_3dn;
392                    sad8bi  = sad8bi_3dn;
393    
394                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
395                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
396          }          }
397    
         if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
398    
399                  /* Inverse DCT */          if ((cpu_flags & XVID_CPU_MMXEXT)) {
400    
401                    /* DCT */
402                    fdct = fdct_xmm_skal;
403                  idct = idct_xmm;                  idct = idct_xmm;
404    
405                  /* Interpolation */                  /* Interpolation */
# Line 223  Line 407 
407                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
408                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
409    
410                    /* reduced resolution */
411                    copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm;
412                    add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm;
413    
414                    /* Quantization */
415                    quant_mpeg_intra = quant_mpeg_intra_xmm;
416                    quant_mpeg_inter = quant_mpeg_inter_xmm;
417    
418                    dequant_h263_intra = dequant_h263_intra_xmm;
419                    dequant_h263_inter = dequant_h263_inter_xmm;
420    
421                    /* Buffer transfer */
422                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
423    
424                  /* Colorspace transformation */                  /* Colorspace transformation */
425                  yuv_to_yv12 = yuv_to_yv12_xmm;                  yv12_to_yv12  = yv12_to_yv12_xmm;
426                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
427                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
428    
429                  /* ME functions */                  /* ME functions */
430                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
431                  sad8  = sad8_xmm;                  sad8  = sad8_xmm;
432                    sad16bi = sad16bi_xmm;
433                    sad8bi  = sad8bi_xmm;
434                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
435                    sad16v   = sad16v_xmm;
436          }          }
437    
438          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
439    
440                  /* Interpolation */                  /* Interpolation */
441                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
# Line 241  Line 443 
443                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
444          }          }
445    
446          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
447  #ifdef EXPERIMENTAL_SSE2_CODE  
448                    /* Inverse DCT */
449                    idct =  idct_3dne;
450    
451                    /* Buffer transfer */
452                    transfer_8to16copy =  transfer_8to16copy_3dne;
453                    transfer_16to8copy = transfer_16to8copy_3dne;
454                    transfer_8to16sub =  transfer_8to16sub_3dne;
455                    transfer_8to16subro =  transfer_8to16subro_3dne;
456                    transfer_8to16sub2 =  transfer_8to16sub2_3dne;
457                    transfer_16to8add = transfer_16to8add_3dne;
458                    transfer8x8_copy = transfer8x8_copy_3dne;
459    
460                  /* Quantization */                  /* Quantization */
461                  quant_intra   = quant_intra_sse2;                  quant_h263_intra = quant_h263_intra_3dne;
462                  dequant_intra = dequant_intra_sse2;                  quant_h263_inter = quant_h263_inter_3dne;
463                  quant_inter   = quant_inter_sse2;                  dequant_mpeg_intra = dequant_mpeg_intra_3dne;
464                  dequant_inter = dequant_inter_sse2;                  dequant_mpeg_inter = dequant_mpeg_inter_3dne;
465                    dequant_h263_intra = dequant_h263_intra_3dne;
466                    dequant_h263_inter = dequant_h263_inter_3dne;
467    
468                    /* ME functions */
469                    calc_cbp = calc_cbp_3dne;
470                    sad16 = sad16_3dne;
471                    sad8 = sad8_3dne;
472                    sad16bi = sad16bi_3dne;
473                    sad8bi = sad8bi_3dne;
474                    dev16 = dev16_3dne;
475    
476                    /* Interpolation */
477                    interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
478                    interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
479                    interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
480            }
481    
482    #if defined(EXPERIMENTAL_SSE2_CODE) /* mark the whole SSE2 stuff as experimental. At least on
483                                                                               my P4, it crashes... */
484            if ((cpu_flags & XVID_CPU_SSE2)) {
485    
                 /* ME */  
486                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
487    
488                    /* Quantization */
489                    quant_h263_intra   = quant_h263_intra_sse2;
490                    quant_h263_inter   = quant_h263_inter_sse2;
491                    dequant_h263_intra = dequant_h263_intra_sse2;
492                    dequant_h263_inter = dequant_h263_inter_sse2;
493    
494                    /* ME; slower than xmm */
495                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
496                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
   
                 /* Forward and Inverse DCT */  
                 idct  = idct_sse2;  
                 fdct = fdct_sse2;  
 #endif  
497          }          }
498    #endif
499  #endif  #endif
500    
501  #ifdef ARCH_IA64  #if defined(ARCH_IS_IA64)
502          if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?          if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
503            idct_ia64_init();            idct_ia64_init();
504            fdct = fdct_ia64;            fdct = fdct_ia64;
505            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
506            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
507            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
508            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 275  Line 510 
510            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
511            sad8 = sad8_ia64;            sad8 = sad8_ia64;
512            dev16 = dev16_ia64;            dev16 = dev16_ia64;
513            quant_intra = quant_intra_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
514            dequant_intra = dequant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
515            quant_inter = quant_inter_ia64;            quant_h263_inter = quant_h263_inter_ia64;
516            dequant_inter = dequant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
517              dequant_h263_inter = dequant_h263_inter_ia64;
518            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
519            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
520            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
521            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
522            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
523            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
524          }          }
525  #endif  #endif
526    
527  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
528  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ASM))
529            {
530                    calc_cbp = calc_cbp_ppc;
531            }
532    
533            if ((cpu_flags & XVID_CPU_ALTIVEC))
534            {
535          calc_cbp = calc_cbp_altivec;          calc_cbp = calc_cbp_altivec;
536          fdct = fdct_altivec;          fdct = fdct_altivec;
537          idct = idct_altivec;          idct = idct_altivec;
# Line 298  Line 539 
539          sad16 = sad16_altivec;          sad16 = sad16_altivec;
540          sad8 = sad8_altivec;          sad8 = sad8_altivec;
541          dev16 = dev16_altivec;          dev16 = dev16_altivec;
542            }
543    #endif
544    
545    #if defined(_DEBUG)
546        xvid_debug = init->debug;
547    #endif
548    
549        return 0;
550    }
551    
552    
553    static int
554    xvid_gbl_info(xvid_gbl_info_t * info)
555    {
556            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
557                    return XVID_ERR_VERSION;
558    
559            info->actual_version = XVID_VERSION;
560            info->build = "dev-api-4";
561            info->cpu_flags = detect_cpu_flags();
562    
563    #if defined(_SMP) && defined(WIN32)
564            info->num_threads = pthread_num_processors_np();;
565  #else  #else
566          calc_cbp = calc_cbp_ppc;          info->num_threads = 0;
567  #endif  #endif
568    
569            return 0;
570    }
571    
572    
573    static int
574    xvid_gbl_convert(xvid_gbl_convert_t* convert)
575    {
576            int width;
577            int height;
578            int width2;
579            int height2;
580            IMAGE img;
581    
582            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
583                  return XVID_ERR_VERSION;
584    
585    #if 0
586            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
587  #endif  #endif
588            width = convert->width;
589            height = convert->height;
590            width2 = convert->width/2;
591            height2 = convert->height/2;
592    
593          /* Inform the client the API version */          switch (convert->input.csp & ~XVID_CSP_VFLIP)
594          init_param->api_version = API_VERSION;          {
595                    case XVID_CSP_YV12 :
596                            img.y = convert->input.plane[0];
597                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
598                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
599                            image_output(&img, width, height, width,
600                                                    (uint8_t**)convert->output.plane, convert->output.stride,
601                                                    convert->output.csp, convert->interlacing);
602                            break;
603    
604          /* Inform the client the core build - unused because we're still alpha */                  default :
605          init_param->core_build = 1000;                          return XVID_ERR_FORMAT;
606            }
607    
608          return XVID_ERR_OK;  
609            emms();
610            return 0;
611    }
612    
613    /*****************************************************************************
614     * XviD Global Entry point
615     *
616     * Well this function initialize all internal function pointers according
617     * to the CPU features forced by the library client or autodetected (depending
618     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
619     * image colorspace transformation tables.
620     *
621     ****************************************************************************/
622    
623    
624    int
625    xvid_global(void *handle,
626                      int opt,
627                      void *param1,
628                      void *param2)
629    {
630            switch(opt)
631            {
632                    case XVID_GBL_INIT :
633                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
634    
635            case XVID_GBL_INFO :
636                return xvid_gbl_info((xvid_gbl_info_t*)param1);
637    
638                    case XVID_GBL_CONVERT :
639                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
640    
641                    default :
642                            return XVID_ERR_FAIL;
643            }
644  }  }
645    
646  /*****************************************************************************  /*****************************************************************************
# Line 329  Line 660 
660                          void *param2)                          void *param2)
661  {  {
662          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
663          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
664                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
665    
666          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
667                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
668    
669            case XVID_DEC_DECODE:
670                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
671    
672          default:          default:
673                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
674          }          }
# Line 362  Line 693 
693  {  {
694          switch (opt) {          switch (opt) {
695          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
696                  return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
697                                                            (XVID_ENC_STATS *) param2);                  return enc_encode((Encoder *) handle,
698                                                              (xvid_enc_frame_t *) param1,
699                                                              (xvid_enc_stats_t *) param2);
700    
701          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
702                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
703    
704          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
705                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
706    
707          default:          default:
708                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
Legend:
Removed from v.1.18  
changed lines
  Added in v.1.45.2.17
No admin address has been configured
 ViewVC Help
Powered by ViewVC 1.0.4