[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 1.6, Wed Mar 27 12:58:29 2002 UTC revision 1.68, Sat Feb 25 04:41:12 2006 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   *  Copyright(C) 2001-2004 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 24  Line 17 
17   *   *
18   *      You should have received a copy of the GNU General Public License   *      You should have received a copy of the GNU General Public License
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., 675 Mass Ave, Cambridge, MA 02139, USA.   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
21   *   *
22   *************************************************************************/   * $Id$
   
 /**************************************************************************  
  *  
  *      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>  
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"
# Line 48  Line 37 
37  #include "image/colorspace.h"  #include "image/colorspace.h"
38  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
39  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
40  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
41  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
42    #include "motion/motion.h"
43  #include "motion/sad.h"  #include "motion/sad.h"
44  #include "utils/emms.h"  #include "utils/emms.h"
45  #include "utils/timer.h"  #include "utils/timer.h"
46    #include "bitstream/mbcoding.h"
47    #include "image/qpel.h"
48    #include "image/postprocessing.h"
49    
50    #if defined(_DEBUG)
51    unsigned int xvid_debug = 0; /* xvid debug mask */
52    #endif
53    
54    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
55    #       include <windows.h>
56    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
57    #       include <signal.h>
58    #       include <setjmp.h>
59    
60            static jmp_buf mark;
61    
62            static void
63            sigill_handler(int signal)
64            {
65               longjmp(mark, 1);
66            }
67    #endif
68    
69    
70  int xvid_init(void *handle, int opt, void *param1, void *param2)  /*
71     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
72     * signalled
73     *
74     * Return values:
75     *  -1 : could not determine
76     *   0 : SIGILL was *not* signalled
77     *   1 : SIGILL was signalled
78     */
79    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
80    static int
81    sigill_check(void (*func)())
82  {  {
83          int cpu_flags;          _try {
84          XVID_INIT_PARAM *init_param;                  func();
85            } _except(EXCEPTION_EXECUTE_HANDLER) {
86    
87          init_param = (XVID_INIT_PARAM *) param1;                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
88                            return(1);
89            }
90            return(0);
91    }
92    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
93    static int
94    sigill_check(void (*func)())
95    {
96        void *old_handler;
97        int jmpret;
98    
99          // force specific cpu settings?      /* Set our SIGILL handler */
100          if((init_param->cpu_flags & XVID_CPU_FORCE) > 0)      old_handler = signal(SIGILL, sigill_handler);
101                  cpu_flags = init_param->cpu_flags;  
102          else {      /* Check for error */
103        if (old_handler == SIG_ERR) {
104            return(-1);
105        }
106    
107  #ifdef ARCH_X86      /* Save stack context, so if func triggers a SIGILL, we can still roll
108                  cpu_flags = check_cpu_features();           * back to a valid CPU state */
109  #else          jmpret = setjmp(mark);
110                  cpu_flags = 0;  
111            /* If setjmp returned directly, then its returned value is 0, and we still
112             * have to test the passed func. Otherwise it means the stack context has
113             * been restored by a longjmp() call, which in our case happens only in the
114             * signal handler */
115        if (jmpret == 0) {
116            func();
117        }
118    
119        /* Restore old signal handler */
120        signal(SIGILL, old_handler);
121    
122        return(jmpret);
123    }
124  #endif  #endif
125                  init_param->cpu_flags = cpu_flags;  
126    
127    /* detect cpu flags  */
128    static unsigned int
129    detect_cpu_flags(void)
130    {
131            /* enable native assembly optimizations by default */
132            unsigned int cpu_flags = XVID_CPU_ASM;
133    
134    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
135            cpu_flags |= check_cpu_features();
136            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
137                    cpu_flags &= ~XVID_CPU_SSE;
138    
139            if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
140                    cpu_flags &= ~XVID_CPU_SSE2;
141    #endif
142    
143    #if defined(ARCH_IS_PPC)
144            if (!sigill_check(altivec_trigger))
145                    cpu_flags |= XVID_CPU_ALTIVEC;
146    #endif
147    
148            return cpu_flags;
149          }          }
150    
151          // initialize the function pointers  
152    /*****************************************************************************
153     * XviD Init Entry point
154     *
155     * Well this function initialize all internal function pointers according
156     * to the CPU features forced by the library client or autodetected (depending
157     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
158     * image colorspace transformation tables.
159     *
160     * Returned value : XVID_ERR_OK
161     *                  + API_VERSION in the input XVID_INIT_PARAM structure
162     *                  + core build  "   "    "       "               "
163     *
164     ****************************************************************************/
165    
166    
167    static
168    int xvid_gbl_init(xvid_gbl_init_t * init)
169    {
170            unsigned int cpu_flags;
171    
172            if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
173                    return XVID_ERR_VERSION;
174    
175            cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
176    
177            /* Initialize the function pointers */
178          idct_int32_init();          idct_int32_init();
179            init_vlc_tables();
180    
181            /* Fixed Point Forward/Inverse DCT transformations */
182          fdct = fdct_int32;          fdct = fdct_int32;
183          idct = idct_int32;          idct = idct_int32;
184    
185            /* Only needed on PPC Altivec archs */
186            sadInit = NULL;
187    
188            /* Restore FPU context : emms_c is a nop functions */
189          emms = emms_c;          emms = emms_c;
190    
191          quant_intra = quant_intra_c;          /* Qpel stuff */
192          dequant_intra = dequant_intra_c;          xvid_QP_Funcs = &xvid_QP_Funcs_C;
193          quant_inter = quant_inter_c;          xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
194          dequant_inter = dequant_inter_c;          xvid_Init_QP();
195    
196          quant4_intra = quant4_intra_c;          /* Quantization functions */
197          dequant4_intra = dequant4_intra_c;          quant_h263_intra   = quant_h263_intra_c;
198          quant4_inter = quant4_inter_c;          quant_h263_inter   = quant_h263_inter_c;
199          dequant4_inter = dequant4_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
200            dequant_h263_inter = dequant_h263_inter_c;
201    
202            quant_mpeg_intra   = quant_mpeg_intra_c;
203            quant_mpeg_inter   = quant_mpeg_inter_c;
204            dequant_mpeg_intra = dequant_mpeg_intra_c;
205            dequant_mpeg_inter = dequant_mpeg_inter_c;
206    
207            /* Block transfer related functions */
208          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
209          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
210          transfer_8to16sub = transfer_8to16sub_c;          transfer_8to16sub = transfer_8to16sub_c;
211            transfer_8to16subro  = transfer_8to16subro_c;
212            transfer_8to16sub2 = transfer_8to16sub2_c;
213            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
214          transfer_16to8add = transfer_16to8add_c;          transfer_16to8add = transfer_16to8add_c;
215          transfer8x8_copy = transfer8x8_copy_c;          transfer8x8_copy = transfer8x8_copy_c;
216            transfer8x4_copy   = transfer8x4_copy_c;
217    
218            /* Interlacing functions */
219            MBFieldTest = MBFieldTest_c;
220    
221            /* Image interpolation related functions */
222          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;
223          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;
224          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
225    
226            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
227            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
228            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
229    
230            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
231            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
232            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
233            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
234    
235            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
236            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
237            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
238    
239            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
240            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
241            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
242    
243            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
244            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
245    
246            interpolate8x8_avg2 = interpolate8x8_avg2_c;
247            interpolate8x8_avg4 = interpolate8x8_avg4_c;
248    
249            /* postprocessing */
250            image_brightness = image_brightness_c;
251    
252            /* Initialize internal colorspace transformation tables */
253          colorspace_init();          colorspace_init();
254    
255            /* All colorspace transformation functions User Format->YV12 */
256            yv12_to_yv12    = yv12_to_yv12_c;
257          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
258          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
259          rgb24_to_yv12 = rgb24_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
260          rgb32_to_yv12 = rgb32_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
261          yuv_to_yv12 = yuv_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
262            rgba_to_yv12    = rgba_to_yv12_c;
263            argb_to_yv12    = argb_to_yv12_c;
264          yuyv_to_yv12 = yuyv_to_yv12_c;          yuyv_to_yv12 = yuyv_to_yv12_c;
265          uyvy_to_yv12 = uyvy_to_yv12_c;          uyvy_to_yv12 = uyvy_to_yv12_c;
266    
267            rgb555i_to_yv12 = rgb555i_to_yv12_c;
268            rgb565i_to_yv12 = rgb565i_to_yv12_c;
269            bgri_to_yv12    = bgri_to_yv12_c;
270            bgrai_to_yv12   = bgrai_to_yv12_c;
271            abgri_to_yv12   = abgri_to_yv12_c;
272            rgbai_to_yv12   = rgbai_to_yv12_c;
273            argbi_to_yv12   = argbi_to_yv12_c;
274            yuyvi_to_yv12   = yuyvi_to_yv12_c;
275            uyvyi_to_yv12   = uyvyi_to_yv12_c;
276    
277            /* All colorspace transformation functions YV12->User format */
278          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
279          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
280          yv12_to_rgb24 = yv12_to_rgb24_c;          yv12_to_bgr     = yv12_to_bgr_c;
281          yv12_to_rgb32 = yv12_to_rgb32_c;          yv12_to_bgra    = yv12_to_bgra_c;
282          yv12_to_yuv = yv12_to_yuv_c;          yv12_to_abgr    = yv12_to_abgr_c;
283            yv12_to_rgba    = yv12_to_rgba_c;
284            yv12_to_argb    = yv12_to_argb_c;
285          yv12_to_yuyv = yv12_to_yuyv_c;          yv12_to_yuyv = yv12_to_yuyv_c;
286          yv12_to_uyvy = yv12_to_uyvy_c;          yv12_to_uyvy = yv12_to_uyvy_c;
287    
288            yv12_to_rgb555i = yv12_to_rgb555i_c;
289            yv12_to_rgb565i = yv12_to_rgb565i_c;
290            yv12_to_bgri    = yv12_to_bgri_c;
291            yv12_to_bgrai   = yv12_to_bgrai_c;
292            yv12_to_abgri   = yv12_to_abgri_c;
293            yv12_to_rgbai   = yv12_to_rgbai_c;
294            yv12_to_argbi   = yv12_to_argbi_c;
295            yv12_to_yuyvi   = yv12_to_yuyvi_c;
296            yv12_to_uyvyi   = yv12_to_uyvyi_c;
297    
298            /* Functions used in motion estimation algorithms */
299          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
300          sad16 = sad16_c;          sad16 = sad16_c;
301          sad8 = sad8_c;          sad8 = sad8_c;
302            sad16bi    = sad16bi_c;
303            sad8bi     = sad8bi_c;
304          dev16 = dev16_c;          dev16 = dev16_c;
305            sad16v     = sad16v_c;
306            sse8_16bit = sse8_16bit_c;
307            sse8_8bit  = sse8_8bit_c;
308    
309    #if defined(ARCH_IS_IA32)
310    
311            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
312                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
313                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2))
314            {
315                    /* Restore FPU context : emms_c is a nop functions */
316                    emms = emms_mmx;
317            }
318    
319  #ifdef ARCH_X86          if ((cpu_flags & XVID_CPU_MMX)) {
         if((cpu_flags & XVID_CPU_MMX) > 0) {  
                 fdct = fdct_mmx;  
                 idct = idct_mmx;  
320    
321                  emms = emms_mmx;                  /* Forward and Inverse Discrete Cosine Transformation functions */
322                    fdct = fdct_mmx_skal;
323                    idct = idct_mmx;
324    
325                  quant_intra = quant_intra_mmx;                  /* Qpel stuff */
326                  dequant_intra = dequant_intra_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
327                  quant_inter = quant_inter_mmx;                  xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
328                  dequant_inter = dequant_inter_mmx;  
329                    /* Quantization related functions */
330                  quant4_intra = quant4_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
331                  dequant4_intra = dequant4_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
332                  quant4_inter = quant4_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
333                  dequant4_inter = dequant4_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
334    
335                    quant_mpeg_intra   = quant_mpeg_intra_mmx;
336                    quant_mpeg_inter   = quant_mpeg_inter_mmx;
337                    dequant_mpeg_intra = dequant_mpeg_intra_mmx;
338                    dequant_mpeg_inter = dequant_mpeg_inter_mmx;
339    
340                    /* Block related functions */
341                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
342                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
343                  transfer_8to16sub = transfer_8to16sub_mmx;                  transfer_8to16sub = transfer_8to16sub_mmx;
344                    transfer_8to16subro  = transfer_8to16subro_mmx;
345                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
346                  transfer_16to8add = transfer_16to8add_mmx;                  transfer_16to8add = transfer_16to8add_mmx;
347                  transfer8x8_copy = transfer8x8_copy_mmx;                  transfer8x8_copy = transfer8x8_copy_mmx;
348                    transfer8x4_copy   = transfer8x4_copy_mmx;
349    
350                    /* Interlacing Functions */
351                    MBFieldTest = MBFieldTest_mmx;
352    
353                    /* Image Interpolation related functions */
354                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;
355                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;
356                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
357    
358                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
359                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
360                  yuv_to_yv12 = yuv_to_yv12_mmx;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
361    
362                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
363                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
364                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
365                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
366    
367                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
368                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
369    
370                    interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
371                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
372    
373                    /* postprocessing */
374                    image_brightness = image_brightness_mmx;
375    
376                    /* image input xxx_to_yv12 related functions */
377                    yv12_to_yv12  = yv12_to_yv12_mmx;
378                    bgr_to_yv12   = bgr_to_yv12_mmx;
379                    bgra_to_yv12  = bgra_to_yv12_mmx;
380                  yuyv_to_yv12 = yuyv_to_yv12_mmx;                  yuyv_to_yv12 = yuyv_to_yv12_mmx;
381                  uyvy_to_yv12 = uyvy_to_yv12_mmx;                  uyvy_to_yv12 = uyvy_to_yv12_mmx;
382    
383                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  /* image output yv12_to_xxx related functions */
384                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
385                    yv12_to_bgra  = yv12_to_bgra_mmx;
386                  yv12_to_yuyv = yv12_to_yuyv_mmx;                  yv12_to_yuyv = yv12_to_yuyv_mmx;
387                  yv12_to_uyvy = yv12_to_uyvy_mmx;                  yv12_to_uyvy = yv12_to_uyvy_mmx;
388    
389                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
390                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
391    
392                    /* Motion estimation related functions */
393                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
394                  sad16 = sad16_mmx;                  sad16 = sad16_mmx;
395                  sad8 = sad8_mmx;                  sad8 = sad8_mmx;
396                    sad16bi    = sad16bi_mmx;
397                    sad8bi     = sad8bi_mmx;
398                  dev16 = dev16_mmx;                  dev16 = dev16_mmx;
399                    sad16v     = sad16v_mmx;
400                    sse8_16bit = sse8_16bit_mmx;
401                    sse8_8bit  = sse8_8bit_mmx;
402            }
403    
404            /* these 3dnow functions are faster than mmx, but slower than xmm. */
405            if ((cpu_flags & XVID_CPU_3DNOW)) {
406    
407                    emms = emms_3dn;
408    
409                    /* ME functions */
410                    sad16bi = sad16bi_3dn;
411                    sad8bi  = sad8bi_3dn;
412    
413                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
414                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
415          }          }
416    
417          if((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
418            if ((cpu_flags & XVID_CPU_MMXEXT)) {
419    
420                    /* DCT */
421                    fdct = fdct_xmm_skal;
422                  idct = idct_xmm;                  idct = idct_xmm;
423    
424                    /* Interpolation */
425                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;
426                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;
427                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
                 yuv_to_yv12 = yuv_to_yv12_xmm;  
428    
429                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
430                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
431                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
432    
433                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
434                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
435                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
436                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
437    
438                    /* Quantization */
439                    quant_mpeg_intra = quant_mpeg_intra_xmm;
440                    quant_mpeg_inter = quant_mpeg_inter_xmm;
441    
442                    dequant_h263_intra = dequant_h263_intra_xmm;
443                    dequant_h263_inter = dequant_h263_inter_xmm;
444    
445                    /* Buffer transfer */
446                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
447                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
448    
449                    /* Colorspace transformation */
450                    yv12_to_yv12  = yv12_to_yv12_xmm;
451                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
452                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
453    
454                    /* ME functions */
455                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
456                  sad8 = sad8_xmm;                  sad8 = sad8_xmm;
457                    sad16bi = sad16bi_xmm;
458                    sad8bi  = sad8bi_xmm;
459                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
460                    sad16v   = sad16v_xmm;
461          }          }
462    
463          if((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
464    
465                    /* Interpolation */
466                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
467                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
468                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
469    
470                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
471                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
472                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
473            }
474    
475            if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
476    
477                    /* Buffer transfer */
478                    transfer_8to16copy =  transfer_8to16copy_3dne;
479                    transfer_16to8copy = transfer_16to8copy_3dne;
480                    transfer_8to16sub =  transfer_8to16sub_3dne;
481                    transfer_8to16subro =  transfer_8to16subro_3dne;
482                    transfer_16to8add = transfer_16to8add_3dne;
483                    transfer8x8_copy = transfer8x8_copy_3dne;
484                    transfer8x4_copy = transfer8x4_copy_3dne;
485    
486                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
487                            /* Inverse DCT */
488                            idct =  idct_3dne;
489    
490                            /* Buffer transfer */
491                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
492    
493                            /* Interpolation */
494                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
495                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
496                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
497    
498                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
499                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
500                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
501    
502                            /* Quantization */
503                            quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
504                            quant_h263_inter = quant_h263_inter_3dne;
505                            dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
506                            dequant_mpeg_inter = dequant_mpeg_inter_3dne;
507                            dequant_h263_intra = dequant_h263_intra_3dne;
508                            dequant_h263_inter = dequant_h263_inter_3dne;
509    
510                            /* ME functions */
511                            calc_cbp = calc_cbp_3dne;
512    
513                            sad16 = sad16_3dne;
514                            sad8 = sad8_3dne;
515                            sad16bi = sad16bi_3dne;
516                            sad8bi = sad8bi_3dne;
517                            dev16 = dev16_3dne;
518                    }
519            }
520    
521            if ((cpu_flags & XVID_CPU_SSE2)) {
522    
523                    calc_cbp = calc_cbp_sse2;
524    
525                    /* Quantization */
526                    quant_h263_intra   = quant_h263_intra_sse2;
527                    quant_h263_inter   = quant_h263_inter_sse2;
528                    dequant_h263_intra = dequant_h263_intra_sse2;
529                    dequant_h263_inter = dequant_h263_inter_sse2;
530    
531                    /* SAD operators */
532                    sad16    = sad16_sse2;
533                    dev16    = dev16_sse2;
534    
535                    /* DCT operators */
536                    fdct = fdct_sse2_skal;
537        /* idct = idct_sse2_skal; */   /* Is now IEEE1180 and Walken compliant. Disabled until fully tested. */
538    
539                    /* postprocessing */
540                    image_brightness = image_brightness_sse2;
541          }          }
542    #endif /* ARCH_IS_IA32 */
543    
544    #if defined(ARCH_IS_IA64)
545            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
546              idct_ia64_init();
547              fdct = fdct_ia64;
548              idct = idct_ia64;   /*not yet working, crashes */
549              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
550              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
551              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
552              sad16 = sad16_ia64;
553              sad16bi = sad16bi_ia64;
554              sad8 = sad8_ia64;
555              dev16 = dev16_ia64;
556    /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
557              quant_h263_intra = quant_h263_intra_ia64;
558              quant_h263_inter = quant_h263_inter_ia64;
559              dequant_h263_intra = dequant_h263_intra_ia64;
560              dequant_h263_inter = dequant_h263_inter_ia64;
561              transfer_8to16copy = transfer_8to16copy_ia64;
562              transfer_16to8copy = transfer_16to8copy_ia64;
563              transfer_8to16sub = transfer_8to16sub_ia64;
564              transfer_8to16sub2 = transfer_8to16sub2_ia64;
565              transfer_16to8add = transfer_16to8add_ia64;
566              transfer8x8_copy = transfer8x8_copy_ia64;
567            }
568  #endif  #endif
569  #ifdef ARCH_PPC  
570  #ifdef ARCH_PPC_ALTIVEC  #if defined(ARCH_IS_PPC)
571          calc_cbp = calc_cbp_altivec;          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
572  #else            /* sad operators */
573          calc_cbp = calc_cbp_ppc;                    sad16 = sad16_altivec_c;
574                      sad16bi = sad16bi_altivec_c;
575                      sad8 = sad8_altivec_c;
576                      dev16 = dev16_altivec_c;
577    
578              sse8_16bit = sse8_16bit_altivec_c;
579    
580              /* mem transfer */
581              transfer_8to16copy = transfer_8to16copy_altivec_c;
582              transfer_16to8copy = transfer_16to8copy_altivec_c;
583              transfer_8to16sub = transfer_8to16sub_altivec_c;
584              transfer_8to16subro = transfer_8to16subro_altivec_c;
585              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
586              transfer_16to8add = transfer_16to8add_altivec_c;
587              transfer8x8_copy = transfer8x8_copy_altivec_c;
588    
589              /* Inverse DCT */
590              idct = idct_altivec_c;
591    
592              /* Interpolation */
593              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
594              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
595              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
596    
597              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
598              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
599    
600                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
601                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
602                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
603                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
604    
605              /* Colorspace conversion */
606              bgra_to_yv12 = bgra_to_yv12_altivec_c;
607              abgr_to_yv12 = abgr_to_yv12_altivec_c;
608              rgba_to_yv12 = rgba_to_yv12_altivec_c;
609              argb_to_yv12 = argb_to_yv12_altivec_c;
610    
611              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
612              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
613    
614              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
615              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
616    
617              /* Quantization */
618              quant_h263_intra = quant_h263_intra_altivec_c;
619              quant_h263_inter = quant_h263_inter_altivec_c;
620              dequant_h263_intra = dequant_h263_intra_altivec_c;
621              dequant_h263_inter = dequant_h263_inter_altivec_c;
622    
623                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
624                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
625    
626                      /* Qpel stuff */
627                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
628                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
629            }
630  #endif  #endif
631    
632    #if defined(ARCH_IS_X86_64)
633            /* For now, only XVID_CPU_ASM is looked for, so user can still
634             * disable asm usage the usual way. When Intel EMT64 cpus will
635             * be out, maybe we'll have to check more precisely what cpu
636             * features there really are. */
637            if (cpu_flags & XVID_CPU_ASM) {
638                    /* SIMD state flusher */
639                    emms = emms_mmx;
640    
641                    /* DCT operators */
642                    fdct = fdct_skal_x86_64;
643                    idct = idct_x86_64;
644    
645                    /* SAD operators */
646                    sad16      = sad16_x86_64;
647                    sad8       = sad8_x86_64;
648                    sad16bi    = sad16bi_x86_64;
649                    sad8bi     = sad8bi_x86_64;
650                    dev16      = dev16_x86_64;
651                    sad16v     = sad16v_x86_64;
652                    sse8_16bit = sse8_16bit_x86_64;
653                    sse8_8bit  = sse8_8bit_x86_64;
654    
655                    /* Interpolation operators */
656                    interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_x86_64;
657                    interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_x86_64;
658                    interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_x86_64;
659    
660                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_x86_64;
661                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_x86_64;
662                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_x86_64;
663                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_x86_64;
664    
665                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_x86_64;
666                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_x86_64;
667    
668                    interpolate8x8_avg2 = interpolate8x8_avg2_x86_64;
669                    interpolate8x8_avg4 = interpolate8x8_avg4_x86_64;
670    
671                    /* Quantization related functions */
672                    quant_h263_intra   = quant_h263_intra_x86_64;
673                    quant_h263_inter   = quant_h263_inter_x86_64;
674                    dequant_h263_intra = dequant_h263_intra_x86_64;
675                    dequant_h263_inter = dequant_h263_inter_x86_64;
676                    quant_mpeg_intra   = quant_mpeg_intra_x86_64;
677                    quant_mpeg_inter   = quant_mpeg_inter_x86_64;
678                    dequant_mpeg_intra   = dequant_mpeg_intra_x86_64;
679                    dequant_mpeg_inter   = dequant_mpeg_inter_x86_64;
680    
681                    /* Block related functions */
682                    transfer_8to16copy  = transfer_8to16copy_x86_64;
683                    transfer_16to8copy  = transfer_16to8copy_x86_64;
684                    transfer_8to16sub   = transfer_8to16sub_x86_64;
685                    transfer_8to16subro = transfer_8to16subro_x86_64;
686                    transfer_8to16sub2  = transfer_8to16sub2_x86_64;
687                    transfer_8to16sub2ro= transfer_8to16sub2ro_x86_64;
688                    transfer_16to8add   = transfer_16to8add_x86_64;
689                    transfer8x8_copy    = transfer8x8_copy_x86_64;
690    
691                    /* Qpel stuff */
692                    xvid_QP_Funcs = &xvid_QP_Funcs_x86_64;
693                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_x86_64;
694    
695                    /* Interlacing Functions */
696                    MBFieldTest = MBFieldTest_x86_64;
697            }
698    #endif
699    
700    #if defined(_DEBUG)
701        xvid_debug = init->debug;
702    #endif
703    
704        return(0);
705    }
706    
707    
708    static int
709    xvid_gbl_info(xvid_gbl_info_t * info)
710    {
711            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
712                    return XVID_ERR_VERSION;
713    
714            info->actual_version = XVID_VERSION;
715            info->build = "xvid-1.2.0-dev";
716            info->cpu_flags = detect_cpu_flags();
717      info->num_threads = 0;
718    
719    #if defined(WIN32)
720      {
721        DWORD dwProcessAffinityMask, dwSystemAffinityMask;
722        if (GetProcessAffinityMask(GetCurrentProcess(), &dwProcessAffinityMask, &dwSystemAffinityMask)) {
723          int i;
724          for(i=0; i<32; i++) {
725            if ((dwProcessAffinityMask & (1<<i)))
726              info->num_threads++;
727          }
728        }
729      }
730    #endif
731    
732            return 0;
733    }
734    
735    
736    static int
737    xvid_gbl_convert(xvid_gbl_convert_t* convert)
738    {
739            int width;
740            int height;
741            int width2;
742            int height2;
743            IMAGE img;
744    
745            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
746                  return XVID_ERR_VERSION;
747    
748    #if 0
749            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
750  #endif  #endif
751            width = convert->width;
752            height = convert->height;
753            width2 = convert->width/2;
754            height2 = convert->height/2;
755    
756          // API version          switch (convert->input.csp & ~XVID_CSP_VFLIP)
757          init_param->api_version = API_VERSION;          {
758                    case XVID_CSP_YV12 :
759                            img.y = convert->input.plane[0];
760                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
761                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
762                            image_output(&img, width, height, width,
763                                                    (uint8_t**)convert->output.plane, convert->output.stride,
764                                                    convert->output.csp, convert->interlacing);
765                            break;
766    
767                    default :
768                            return XVID_ERR_FORMAT;
769            }
770    
         // something clever has to be done for this  
         init_param->core_build = 1000;  
771    
772          return XVID_ERR_OK;          emms();
773            return 0;
774  }  }
775    
776  int xvid_decore(void * handle, int opt, void * param1, void * param2)  /*****************************************************************************
777     * XviD Global Entry point
778     *
779     * Well this function initialize all internal function pointers according
780     * to the CPU features forced by the library client or autodetected (depending
781     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
782     * image colorspace transformation tables.
783     *
784     ****************************************************************************/
785    
786    
787    int
788    xvid_global(void *handle,
789                      int opt,
790                      void *param1,
791                      void *param2)
792  {  {
793          switch (opt)          switch (opt)
794          {          {
795          case XVID_DEC_DECODE :                  case XVID_GBL_INIT :
796          return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);                          return xvid_gbl_init((xvid_gbl_init_t*)param1);
797    
798            case XVID_GBL_INFO :
799                return xvid_gbl_info((xvid_gbl_info_t*)param1);
800    
801                    case XVID_GBL_CONVERT :
802                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
803    
804                    default :
805                            return XVID_ERR_FAIL;
806            }
807    }
808    
809    /*****************************************************************************
810     * XviD Native decoder entry point
811     *
812     * This function is just a wrapper to all the option cases.
813     *
814     * Returned values : XVID_ERR_FAIL when opt is invalid
815     *                   else returns the wrapped function result
816     *
817     ****************************************************************************/
818    
819    int
820    xvid_decore(void *handle,
821                            int opt,
822                            void *param1,
823                            void *param2)
824    {
825            switch (opt) {
826          case XVID_DEC_CREATE :          case XVID_DEC_CREATE :
827          return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
828    
829          case XVID_DEC_DESTROY :          case XVID_DEC_DESTROY :
830          return decoder_destroy((DECODER *) handle);          return decoder_destroy((DECODER *) handle);
831    
832            case XVID_DEC_DECODE:
833                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
834    
835          default:          default:
836          return XVID_ERR_FAIL;          return XVID_ERR_FAIL;
837      }      }
838  }  }
839    
840    
841  int xvid_encore(void * handle, int opt, void * param1, void * param2)  /*****************************************************************************
842  {   * XviD Native encoder entry point
843          switch (opt)   *
844     * This function is just a wrapper to all the option cases.
845     *
846     * Returned values : XVID_ERR_FAIL when opt is invalid
847     *                   else returns the wrapped function result
848     *
849     ****************************************************************************/
850    
851    int
852    xvid_encore(void *handle,
853                            int opt,
854                            void *param1,
855                            void *param2)
856          {          {
857            switch (opt) {
858          case XVID_ENC_ENCODE :          case XVID_ENC_ENCODE :
859          return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, (XVID_ENC_STATS *) param2);  
860                    return enc_encode((Encoder *) handle,
861                                                              (xvid_enc_frame_t *) param1,
862                                                              (xvid_enc_stats_t *) param2);
863    
864          case XVID_ENC_CREATE :          case XVID_ENC_CREATE :
865          return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
866    
867          case XVID_ENC_DESTROY :          case XVID_ENC_DESTROY :
868          return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
869    
870          default:          default:
871          return XVID_ERR_FAIL;          return XVID_ERR_FAIL;

Legend:
Removed from v.1.6  
changed lines
  Added in v.1.68

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