[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 1.20, Fri Jun 21 16:12:47 2002 UTC revision 1.48, Mon Mar 22 22:36:23 2004 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    #include "image/postprocessing.h"
50    
51    #if defined(_DEBUG)
52    unsigned int xvid_debug = 0; /* xvid debug mask */
53    #endif
54    
55    #if defined(ARCH_IS_IA32)
56    #if defined(_MSC_VER)
57    #       include <windows.h>
58    #else
59    #       include <signal.h>
60    #       include <setjmp.h>
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
74     * signalled
75     *
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    #if defined(_MSC_VER)
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    /* detect cpu flags  */
121    static unsigned int
122    detect_cpu_flags()
123    {
124            /* enable native assembly optimizations by default */
125            unsigned int cpu_flags = XVID_CPU_ASM;
126    
127    #if defined(ARCH_IS_IA32)
128            cpu_flags |= check_cpu_features();
129            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
130                    cpu_flags &= ~XVID_CPU_SSE;
131    
132            if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
133                    cpu_flags &= ~XVID_CPU_SSE2;
134    #endif
135    
136    #if defined(ARCH_IS_PPC)
137    #if defined(ARCH_IS_PPC_ALTIVEC)
138            cpu_flags |= XVID_CPU_ALTIVEC;
139    #endif
140    #endif
141    
142            return cpu_flags;
143    }
144    
145    
146  /*****************************************************************************  /*****************************************************************************
147   * XviD Init Entry point   * XviD Init Entry point
# Line 69  Line 157 
157   *   *
158   ****************************************************************************/   ****************************************************************************/
159    
 int  
 xvid_init(void *handle,  
                   int opt,  
                   void *param1,  
                   void *param2)  
 {  
         int cpu_flags;  
         XVID_INIT_PARAM *init_param;  
160    
161          init_param = (XVID_INIT_PARAM *) param1;  static
162    int xvid_gbl_init(xvid_gbl_init_t * init)
163    {
164            unsigned int cpu_flags;
165    
166          /* Do we have to force CPU features  ? */          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
167          if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0) {                  return XVID_ERR_VERSION;
                 cpu_flags = init_param->cpu_flags;  
         } else {  
168    
169  #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;  
         }  
170    
171          /* Initialize the function pointers */          /* Initialize the function pointers */
172          idct_int32_init();          idct_int32_init();
# Line 107  Line 182 
182          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
183          emms = emms_c;          emms = emms_c;
184    
185            /* Qpel stuff */
186            xvid_QP_Funcs = &xvid_QP_Funcs_C;
187            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
188            xvid_Init_QP();
189    
190          /* Quantization functions */          /* Quantization functions */
191          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
192          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
193          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
194          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
195    
196          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
197          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
198          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
199          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
200    
201          /* Block transfer related functions */          /* Block transfer related functions */
202          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
203          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
204          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
205            transfer_8to16subro  = transfer_8to16subro_c;
206          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
207          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
208          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
209    
210            /* Interlacing functions */
211            MBFieldTest = MBFieldTest_c;
212    
213          /* Image interpolation related functions */          /* Image interpolation related functions */
214          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;
215          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
216          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
217    
218            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
219            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
220            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
221    
222            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
223            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
224            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
225    
226            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
227            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
228    
229            interpolate8x8_avg2 = interpolate8x8_avg2_c;
230            interpolate8x8_avg4 = interpolate8x8_avg4_c;
231    
232            /* reduced resolution */
233            copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C;
234            add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C;
235            vfilter_31 = xvid_VFilter_31_C;
236            hfilter_31 = xvid_HFilter_31_C;
237            filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C;
238            filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C;
239    
240          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
241          colorspace_init();          colorspace_init();
242    
243          /* All colorspace transformation functions User Format->YV12 */          /* All colorspace transformation functions User Format->YV12 */
244            yv12_to_yv12    = yv12_to_yv12_c;
245          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
246          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
247          rgb24_to_yv12  = rgb24_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
248          rgb32_to_yv12  = rgb32_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
249          yuv_to_yv12    = yuv_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
250            rgba_to_yv12    = rgba_to_yv12_c;
251            argb_to_yv12    = argb_to_yv12_c;
252          yuyv_to_yv12   = yuyv_to_yv12_c;          yuyv_to_yv12   = yuyv_to_yv12_c;
253          uyvy_to_yv12   = uyvy_to_yv12_c;          uyvy_to_yv12   = uyvy_to_yv12_c;
254    
255            rgb555i_to_yv12 = rgb555i_to_yv12_c;
256            rgb565i_to_yv12 = rgb565i_to_yv12_c;
257            bgri_to_yv12    = bgri_to_yv12_c;
258            bgrai_to_yv12   = bgrai_to_yv12_c;
259            abgri_to_yv12   = abgri_to_yv12_c;
260            rgbai_to_yv12   = rgbai_to_yv12_c;
261            argbi_to_yv12   = argbi_to_yv12_c;
262            yuyvi_to_yv12   = yuyvi_to_yv12_c;
263            uyvyi_to_yv12   = uyvyi_to_yv12_c;
264    
265          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
266          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
267          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
268          yv12_to_rgb24  = yv12_to_rgb24_c;          yv12_to_bgr     = yv12_to_bgr_c;
269          yv12_to_rgb32  = yv12_to_rgb32_c;          yv12_to_bgra    = yv12_to_bgra_c;
270          yv12_to_yuv    = yv12_to_yuv_c;          yv12_to_abgr    = yv12_to_abgr_c;
271            yv12_to_rgba    = yv12_to_rgba_c;
272            yv12_to_argb    = yv12_to_argb_c;
273          yv12_to_yuyv   = yv12_to_yuyv_c;          yv12_to_yuyv   = yv12_to_yuyv_c;
274          yv12_to_uyvy   = yv12_to_uyvy_c;          yv12_to_uyvy   = yv12_to_uyvy_c;
275    
276            yv12_to_rgb555i = yv12_to_rgb555i_c;
277            yv12_to_rgb565i = yv12_to_rgb565i_c;
278            yv12_to_bgri    = yv12_to_bgri_c;
279            yv12_to_bgrai   = yv12_to_bgrai_c;
280            yv12_to_abgri   = yv12_to_abgri_c;
281            yv12_to_rgbai   = yv12_to_rgbai_c;
282            yv12_to_argbi   = yv12_to_argbi_c;
283            yv12_to_yuyvi   = yv12_to_yuyvi_c;
284            yv12_to_uyvyi   = yv12_to_uyvyi_c;
285    
286          /* Functions used in motion estimation algorithms */          /* Functions used in motion estimation algorithms */
287          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
288          sad16    = sad16_c;          sad16    = sad16_c;
         sad16bi  = sad16bi_c;  
289          sad8     = sad8_c;          sad8     = sad8_c;
290            sad16bi    = sad16bi_c;
291            sad8bi     = sad8bi_c;
292          dev16    = dev16_c;          dev16    = dev16_c;
293            sad16v     = sad16v_c;
294            sse8_16bit = sse8_16bit_c;
295    
296    #if defined(ARCH_IS_IA32)
297    
298            if ((cpu_flags & XVID_CPU_ASM)) {
299                    vfilter_31 = xvid_VFilter_31_x86;
300                    hfilter_31 = xvid_HFilter_31_x86;
301            }
302    
303            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
304                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
305                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2))
306            {
307                    /* Restore FPU context : emms_c is a nop functions */
308                    emms = emms_mmx;
309            }
310    
311  #ifdef ARCH_X86          if ((cpu_flags & XVID_CPU_MMX)) {
         if ((cpu_flags & XVID_CPU_MMX) > 0) {  
312    
313                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
314                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
315                  idct = idct_mmx;                  idct = idct_mmx;
316    
317                  /* To restore FPU context after mmx use */                  /* Qpel stuff */
318                  emms = emms_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
319                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
320    
321                  /* Quantization related functions */                  /* Quantization related functions */
322                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
323                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
324                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
325                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
326    
327                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_intra   = quant_mpeg_intra_mmx;
328                  dequant4_intra = dequant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
329                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
330                  dequant4_inter = dequant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
331    
332                  /* Block related functions */                  /* Block related functions */
333                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
334                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
335                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
336                    transfer_8to16subro  = transfer_8to16subro_mmx;
337                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
338                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
339                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
340    
341                    /* Interlacing Functions */
342                    MBFieldTest = MBFieldTest_mmx;
343    
344                  /* Image Interpolation related functions */                  /* Image Interpolation related functions */
345                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;
346                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
347                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
348    
349                  /* Image RGB->YV12 related functions */                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
350                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
351                  rgb32_to_yv12 = rgb32_to_yv12_mmx;  
352                  yuv_to_yv12   = yuv_to_yv12_mmx;                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
353                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
354    
355                    /* reduced resolution */
356                    copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx;
357                    add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx;
358                    hfilter_31 = xvid_HFilter_31_mmx;
359                    filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx;
360                    filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx;
361    
362                    /* image input xxx_to_yv12 related functions */
363                    yv12_to_yv12  = yv12_to_yv12_mmx;
364                    bgr_to_yv12   = bgr_to_yv12_mmx;
365                    bgra_to_yv12  = bgra_to_yv12_mmx;
366                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
367                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
368    
369                  /* Image YV12->RGB related functions */                  /* image output yv12_to_xxx related functions */
370                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
371                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgra  = yv12_to_bgra_mmx;
372                  yv12_to_yuyv  = yv12_to_yuyv_mmx;                  yv12_to_yuyv  = yv12_to_yuyv_mmx;
373                  yv12_to_uyvy  = yv12_to_uyvy_mmx;                  yv12_to_uyvy  = yv12_to_uyvy_mmx;
374    
375                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
376                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
377    
378                  /* Motion estimation related functions */                  /* Motion estimation related functions */
379                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
380                  sad16    = sad16_mmx;                  sad16    = sad16_mmx;
381                  sad8     = sad8_mmx;                  sad8     = sad8_mmx;
382                    sad16bi = sad16bi_mmx;
383                    sad8bi  = sad8bi_mmx;
384                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
385                    sad16v   = sad16v_mmx;
386                    sse8_16bit = sse8_16bit_mmx;
387            }
388    
389            /* these 3dnow functions are faster than mmx, but slower than xmm. */
390            if ((cpu_flags & XVID_CPU_3DNOW)) {
391    
392                    emms = emms_3dn;
393    
394                    /* ME functions */
395                    sad16bi = sad16bi_3dn;
396                    sad8bi  = sad8bi_3dn;
397    
398                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
399                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
400          }          }
401    
         if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
402    
403                  /* Inverse DCT */          if ((cpu_flags & XVID_CPU_MMXEXT)) {
404    
405                    /* DCT */
406                    fdct = fdct_xmm_skal;
407                  idct = idct_xmm;                  idct = idct_xmm;
408    
409                  /* Interpolation */                  /* Interpolation */
# Line 223  Line 411 
411                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
412                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
413    
414                    /* reduced resolution */
415                    copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm;
416                    add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm;
417    
418                    /* Quantization */
419                    quant_mpeg_intra = quant_mpeg_intra_xmm;
420                    quant_mpeg_inter = quant_mpeg_inter_xmm;
421    
422                    dequant_h263_intra = dequant_h263_intra_xmm;
423                    dequant_h263_inter = dequant_h263_inter_xmm;
424    
425                  /* Buffer transfer */                  /* Buffer transfer */
426                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
427    
428                  /* Colorspace transformation */                  /* Colorspace transformation */
429                  yuv_to_yv12 = yuv_to_yv12_xmm;                  yv12_to_yv12  = yv12_to_yv12_xmm;
430                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
431                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
432    
433                  /* ME functions */                  /* ME functions */
434                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
435                  sad8  = sad8_xmm;                  sad8  = sad8_xmm;
436                    sad16bi = sad16bi_xmm;
437                    sad8bi  = sad8bi_xmm;
438                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
439                    sad16v   = sad16v_xmm;
440          }          }
441    
442          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
443    
444                  /* Interpolation */                  /* Interpolation */
445                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
# Line 244  Line 447 
447                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
448          }          }
449    
450          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
451  #ifdef EXPERIMENTAL_SSE2_CODE  
452                    /* Inverse DCT */
453                    idct =  idct_3dne;
454    
455                    /* Buffer transfer */
456                    transfer_8to16copy =  transfer_8to16copy_3dne;
457                    transfer_16to8copy = transfer_16to8copy_3dne;
458                    transfer_8to16sub =  transfer_8to16sub_3dne;
459                    transfer_8to16subro =  transfer_8to16subro_3dne;
460                    transfer_8to16sub2 =  transfer_8to16sub2_3dne;
461                    transfer_16to8add = transfer_16to8add_3dne;
462                    transfer8x8_copy = transfer8x8_copy_3dne;
463    
464                  /* Quantization */                  /* Quantization */
465                  quant_intra   = quant_intra_sse2;                  quant_h263_intra = quant_h263_intra_3dne;
466                  dequant_intra = dequant_intra_sse2;                  quant_h263_inter = quant_h263_inter_3dne;
467                  quant_inter   = quant_inter_sse2;                  dequant_mpeg_intra = dequant_mpeg_intra_3dne;
468                  dequant_inter = dequant_inter_sse2;                  dequant_mpeg_inter = dequant_mpeg_inter_3dne;
469                    dequant_h263_intra = dequant_h263_intra_3dne;
470                    dequant_h263_inter = dequant_h263_inter_3dne;
471    
472                    /* ME functions */
473                    calc_cbp = calc_cbp_3dne;
474                    sad16 = sad16_3dne;
475                    sad8 = sad8_3dne;
476                    sad16bi = sad16bi_3dne;
477                    sad8bi = sad8bi_3dne;
478                    dev16 = dev16_3dne;
479    
480                    /* Interpolation */
481                    interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
482                    interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
483                    interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
484            }
485    
486            if ((cpu_flags & XVID_CPU_SSE2)) {
487    
                 /* ME */  
488                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
489    
490                    /* Quantization */
491                    quant_h263_intra   = quant_h263_intra_sse2;
492                    quant_h263_inter   = quant_h263_inter_sse2;
493                    dequant_h263_intra = dequant_h263_intra_sse2;
494                    dequant_h263_inter = dequant_h263_inter_sse2;
495    
496                    /* SAD operators */
497                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
498                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
499    
500                  /* Forward and Inverse DCT */                  /* DCT operators
501                  idct  = idct_sse2;                   * no iDCT because it's not "Walken matching" */
502                  fdct = fdct_sse2;                  fdct = fdct_sse2_skal;
 #endif  
503          }          }
504    #endif /* ARCH_IS_IA32 */
505    
506  #endif  #if defined(ARCH_IS_IA64)
507            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
 #ifdef ARCH_IA64  
         if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?  
508            idct_ia64_init();            idct_ia64_init();
509            fdct = fdct_ia64;            fdct = fdct_ia64;
510            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
511            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
512            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
513            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 278  Line 515 
515            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
516            sad8 = sad8_ia64;            sad8 = sad8_ia64;
517            dev16 = dev16_ia64;            dev16 = dev16_ia64;
518            quant_intra = quant_intra_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
519            dequant_intra = dequant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
520            quant_inter = quant_inter_ia64;            quant_h263_inter = quant_h263_inter_ia64;
521            dequant_inter = dequant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
522              dequant_h263_inter = dequant_h263_inter_ia64;
523            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
524            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
525            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
526            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
527            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
528            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
529          }          }
530  #endif  #endif
531    
532  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
533  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ASM))
534            {
535                    calc_cbp = calc_cbp_ppc;
536            }
537    
538            if ((cpu_flags & XVID_CPU_ALTIVEC))
539            {
540          calc_cbp = calc_cbp_altivec;          calc_cbp = calc_cbp_altivec;
541          fdct = fdct_altivec;          fdct = fdct_altivec;
542          idct = idct_altivec;          idct = idct_altivec;
# Line 301  Line 544 
544          sad16 = sad16_altivec;          sad16 = sad16_altivec;
545          sad8 = sad8_altivec;          sad8 = sad8_altivec;
546          dev16 = dev16_altivec;          dev16 = dev16_altivec;
547            }
548    #endif
549    
550    #if defined(_DEBUG)
551        xvid_debug = init->debug;
552    #endif
553    
554        return 0;
555    }
556    
557    
558    static int
559    xvid_gbl_info(xvid_gbl_info_t * info)
560    {
561            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
562                    return XVID_ERR_VERSION;
563    
564            info->actual_version = XVID_VERSION;
565            info->build = "xvid-1.0.0";
566            info->cpu_flags = detect_cpu_flags();
567    
568    #if defined(_SMP) && defined(WIN32)
569            info->num_threads = pthread_num_processors_np();;
570  #else  #else
571          calc_cbp = calc_cbp_ppc;          info->num_threads = 0;
572  #endif  #endif
573    
574            return 0;
575    }
576    
577    
578    static int
579    xvid_gbl_convert(xvid_gbl_convert_t* convert)
580    {
581            int width;
582            int height;
583            int width2;
584            int height2;
585            IMAGE img;
586    
587            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
588                  return XVID_ERR_VERSION;
589    
590    #if 0
591            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
592  #endif  #endif
593            width = convert->width;
594            height = convert->height;
595            width2 = convert->width/2;
596            height2 = convert->height/2;
597    
598            switch (convert->input.csp & ~XVID_CSP_VFLIP)
599            {
600                    case XVID_CSP_YV12 :
601                            img.y = convert->input.plane[0];
602                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
603                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
604                            image_output(&img, width, height, width,
605                                                    (uint8_t**)convert->output.plane, convert->output.stride,
606                                                    convert->output.csp, convert->interlacing);
607                            break;
608    
609                    default :
610                            return XVID_ERR_FORMAT;
611            }
612    
613    
614            emms();
615            return 0;
616    }
617    
618    /*****************************************************************************
619     * XviD Global Entry point
620     *
621     * Well this function initialize all internal function pointers according
622     * to the CPU features forced by the library client or autodetected (depending
623     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
624     * image colorspace transformation tables.
625     *
626     ****************************************************************************/
627    
628    
629    int
630    xvid_global(void *handle,
631                      int opt,
632                      void *param1,
633                      void *param2)
634    {
635            switch(opt)
636            {
637                    case XVID_GBL_INIT :
638                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
639    
640          /* Inform the client the API version */          case XVID_GBL_INFO :
641          init_param->api_version = API_VERSION;              return xvid_gbl_info((xvid_gbl_info_t*)param1);
642    
643          /* Inform the client the core build - unused because we're still alpha */                  case XVID_GBL_CONVERT :
644          init_param->core_build = 1000;                          return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
645    
646          return XVID_ERR_OK;                  default :
647                            return XVID_ERR_FAIL;
648            }
649  }  }
650    
651  /*****************************************************************************  /*****************************************************************************
# Line 332  Line 665 
665                          void *param2)                          void *param2)
666  {  {
667          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
668          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
669                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
670    
671          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
672                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
673    
674            case XVID_DEC_DECODE:
675                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
676    
677          default:          default:
678                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
679          }          }
# Line 365  Line 698 
698  {  {
699          switch (opt) {          switch (opt) {
700          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
701  #ifdef BFRAMES  
702                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
703                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
704                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
 #endif  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
705    
706          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
707                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
708    
709          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
710                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
711    
712          default:          default:
713                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.20  
changed lines
  Added in v.1.48

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