[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 1.3, Sun Mar 17 08:20:02 2002 UTC revision 1.83, Tue Mar 9 10:00:14 2010 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>
7   *      Video tools as specified in ISO/IEC 14496-2 standard.  Those intending   *               2002-2010 Michael Militzer <isibaar@xvid.org>
  *      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.  
8   *   *
9   *      This program is free software; you can redistribute it and/or modify   *      This program is free software; you can redistribute it and/or modify
10   *      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 18 
18   *   *
19   *      You should have received a copy of the GNU General Public License   *      You should have received a copy of the GNU General Public License
20   *      along with this program; if not, write to the Free Software   *      along with this program; if not, write to the Free Software
21   *      Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
22   *   *
23   *************************************************************************/   * $Id$
   
 /**************************************************************************  
  *  
  *      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>  
24   *   *
25   *************************************************************************/   ****************************************************************************/
26    
27    #include <stdio.h>
28    #include <stdlib.h>
29    #include <string.h>
30    #include <time.h>
31    
32    #if !defined(_WIN32)
33    #include <unistd.h>
34    #endif
35    
36    #if defined(__APPLE__) && defined(__MACH__) && !defined(_SC_NPROCESSORS_CONF)
37    #include <sys/types.h>
38    #include <sys/sysctl.h>
39    #ifdef MAX
40    #undef MAX
41    #endif
42    #ifdef MIN
43    #undef MIN
44    #endif
45    #endif
46    
47    #if defined(__amigaos4__)
48    #include <exec/exec.h>
49    #include <proto/exec.h>
50    #endif
51    
52  #include "xvid.h"  #include "xvid.h"
53  #include "decoder.h"  #include "decoder.h"
# Line 48  Line 58 
58  #include "image/colorspace.h"  #include "image/colorspace.h"
59  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
60  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
61  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
62  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
63    #include "motion/motion.h"
64    #include "motion/gmc.h"
65  #include "motion/sad.h"  #include "motion/sad.h"
66  #include "utils/emms.h"  #include "utils/emms.h"
67  #include "utils/timer.h"  #include "utils/timer.h"
68    #include "bitstream/mbcoding.h"
69    #include "image/qpel.h"
70    #include "image/postprocessing.h"
71    
72  int xvid_init(void *handle, int opt, void *param1, void *param2)  #if defined(_DEBUG)
73    unsigned int xvid_debug = 0; /* xvid debug mask */
74    #endif
75    
76    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
77    #       include <windows.h>
78    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || (defined(ARCH_IS_PPC) && !defined(__amigaos4__))
79    #       include <signal.h>
80    #       include <setjmp.h>
81    
82            static jmp_buf mark;
83    
84            static void
85            sigill_handler(int signal)
86  {  {
87          int cpu_flags;             longjmp(mark, 1);
88          XVID_INIT_PARAM *init_param;          }
89    #endif
90    
91          init_param = (XVID_INIT_PARAM *) param1;  /*
92     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
93     * signalled
94     *
95     * Return values:
96     *  -1 : could not determine
97     *   0 : SIGILL was *not* signalled
98     *   1 : SIGILL was signalled
99     */
100    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
101    static int
102    sigill_check(void (*func)())
103    {
104            _try {
105                    func();
106            } _except(EXCEPTION_EXECUTE_HANDLER) {
107    
108          // force specific cpu settings?                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
109          if((init_param->cpu_flags & XVID_CPU_FORCE) > 0)                          return(1);
110                  cpu_flags = init_param->cpu_flags;          }
111          else {          return(0);
112    }
113    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || (defined(ARCH_IS_PPC) && !defined(__amigaos4__))
114    static int
115    sigill_check(void (*func)())
116    {
117        void *old_handler;
118        int jmpret;
119    
120  #ifdef ARCH_X86      /* Set our SIGILL handler */
121                  cpu_flags = check_cpu_features();      old_handler = signal(SIGILL, sigill_handler);
122    
123        /* Check for error */
124        if (old_handler == SIG_ERR) {
125            return(-1);
126        }
127    
128        /* Save stack context, so if func triggers a SIGILL, we can still roll
129             * back to a valid CPU state */
130            jmpret = setjmp(mark);
131    
132            /* If setjmp returned directly, then its returned value is 0, and we still
133             * have to test the passed func. Otherwise it means the stack context has
134             * been restored by a longjmp() call, which in our case happens only in the
135             * signal handler */
136        if (jmpret == 0) {
137            func();
138        }
139    
140        /* Restore old signal handler */
141        signal(SIGILL, old_handler);
142    
143        return(jmpret);
144    }
145    #endif
146    
147    
148    /* detect cpu flags  */
149    static unsigned int
150    detect_cpu_flags(void)
151    {
152            /* enable native assembly optimizations by default */
153            unsigned int cpu_flags = XVID_CPU_ASM;
154    
155    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
156            cpu_flags |= check_cpu_features();
157            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
158                    cpu_flags &= ~XVID_CPU_SSE;
159    
160            if ((cpu_flags & (XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41)) && sigill_check(sse2_os_trigger))
161                    cpu_flags &= ~(XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41);
162    #endif
163    
164    #if defined(ARCH_IS_PPC)
165    #if defined(__amigaos4__)
166            {
167                    uint32_t vector_unit = VECTORTYPE_NONE;
168                    IExec->GetCPUInfoTags(GCIT_VectorUnit, &vector_unit, TAG_END);
169                    if (vector_unit == VECTORTYPE_ALTIVEC) {
170                            cpu_flags |= XVID_CPU_ALTIVEC;
171                    }
172            }
173  #else  #else
174                  cpu_flags = 0;          if (!sigill_check(altivec_trigger))
175                    cpu_flags |= XVID_CPU_ALTIVEC;
176    #endif
177  #endif  #endif
178                  init_param->cpu_flags = cpu_flags;  
179            return cpu_flags;
180          }          }
181    
182          // initialize the function pointers  
183    /*****************************************************************************
184     * Xvid Init Entry point
185     *
186     * Well this function initialize all internal function pointers according
187     * to the CPU features forced by the library client or autodetected (depending
188     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
189     * image colorspace transformation tables.
190     *
191     * Returned value : XVID_ERR_OK
192     *                  + API_VERSION in the input XVID_INIT_PARAM structure
193     *                  + core build  "   "    "       "               "
194     *
195     ****************************************************************************/
196    
197    
198    static
199    int xvid_gbl_init(xvid_gbl_init_t * init)
200    {
201            unsigned int cpu_flags;
202    
203            if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
204                    return XVID_ERR_VERSION;
205    
206            cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
207    
208            /* Initialize the function pointers */
209          idct_int32_init();          idct_int32_init();
210            init_vlc_tables();
211    
212            /* Fixed Point Forward/Inverse DCT transformations */
213          fdct = fdct_int32;          fdct = fdct_int32;
214          idct = idct_int32;          idct = idct_int32;
215    
216            /* Only needed on PPC Altivec archs */
217            sadInit = NULL;
218    
219            /* Restore FPU context : emms_c is a nop functions */
220          emms = emms_c;          emms = emms_c;
221    
222          quant_intra = quant_intra_c;          /* Qpel stuff */
223          dequant_intra = dequant_intra_c;          xvid_QP_Funcs = &xvid_QP_Funcs_C;
224          quant_inter = quant_inter_c;          xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
225          dequant_inter = dequant_inter_c;          xvid_Init_QP();
226    
227          quant4_intra = quant4_intra_c;          /* Quantization functions */
228          dequant4_intra = dequant4_intra_c;          quant_h263_intra   = quant_h263_intra_c;
229          quant4_inter = quant4_inter_c;          quant_h263_inter   = quant_h263_inter_c;
230          dequant4_inter = dequant4_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
231            dequant_h263_inter = dequant_h263_inter_c;
232    
233            quant_mpeg_intra   = quant_mpeg_intra_c;
234            quant_mpeg_inter   = quant_mpeg_inter_c;
235            dequant_mpeg_intra = dequant_mpeg_intra_c;
236            dequant_mpeg_inter = dequant_mpeg_inter_c;
237    
238            /* Block transfer related functions */
239          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
240          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
241          transfer_8to16sub = transfer_8to16sub_c;          transfer_8to16sub = transfer_8to16sub_c;
242            transfer_8to16subro  = transfer_8to16subro_c;
243            transfer_8to16sub2 = transfer_8to16sub2_c;
244            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
245          transfer_16to8add = transfer_16to8add_c;          transfer_16to8add = transfer_16to8add_c;
246          transfer8x8_copy = transfer8x8_copy_c;          transfer8x8_copy = transfer8x8_copy_c;
247            transfer8x4_copy   = transfer8x4_copy_c;
248    
249            /* Interlacing functions */
250            MBFieldTest = MBFieldTest_c;
251    
252            /* Image interpolation related functions */
253          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;
254          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;
255          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
256    
257            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
258            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
259            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
260    
261            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
262            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
263            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
264            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
265    
266            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
267            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
268            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
269    
270            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
271            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
272            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
273    
274            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
275            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
276    
277            interpolate8x8_avg2 = interpolate8x8_avg2_c;
278            interpolate8x8_avg4 = interpolate8x8_avg4_c;
279    
280            /* postprocessing */
281            image_brightness = image_brightness_c;
282    
283            /* Initialize internal colorspace transformation tables */
284          colorspace_init();          colorspace_init();
285    
286            /* All colorspace transformation functions User Format->YV12 */
287            yv12_to_yv12    = yv12_to_yv12_c;
288          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
289          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
290          rgb24_to_yv12 = rgb24_to_yv12_c;          rgb_to_yv12     = rgb_to_yv12_c;
291          rgb32_to_yv12 = rgb32_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
292          yuv_to_yv12 = yuv_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
293            abgr_to_yv12    = abgr_to_yv12_c;
294            rgba_to_yv12    = rgba_to_yv12_c;
295            argb_to_yv12    = argb_to_yv12_c;
296          yuyv_to_yv12 = yuyv_to_yv12_c;          yuyv_to_yv12 = yuyv_to_yv12_c;
297          uyvy_to_yv12 = uyvy_to_yv12_c;          uyvy_to_yv12 = uyvy_to_yv12_c;
298    
299            rgb555i_to_yv12 = rgb555i_to_yv12_c;
300            rgb565i_to_yv12 = rgb565i_to_yv12_c;
301            bgri_to_yv12    = bgri_to_yv12_c;
302            bgrai_to_yv12   = bgrai_to_yv12_c;
303            abgri_to_yv12   = abgri_to_yv12_c;
304            rgbai_to_yv12   = rgbai_to_yv12_c;
305            argbi_to_yv12   = argbi_to_yv12_c;
306            yuyvi_to_yv12   = yuyvi_to_yv12_c;
307            uyvyi_to_yv12   = uyvyi_to_yv12_c;
308    
309            /* All colorspace transformation functions YV12->User format */
310          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
311          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
312          yv12_to_rgb24 = yv12_to_rgb24_c;          yv12_to_rgb     = yv12_to_rgb_c;
313          yv12_to_rgb32 = yv12_to_rgb32_c;          yv12_to_bgr     = yv12_to_bgr_c;
314          yv12_to_yuv = yv12_to_yuv_c;          yv12_to_bgra    = yv12_to_bgra_c;
315            yv12_to_abgr    = yv12_to_abgr_c;
316            yv12_to_rgba    = yv12_to_rgba_c;
317            yv12_to_argb    = yv12_to_argb_c;
318          yv12_to_yuyv = yv12_to_yuyv_c;          yv12_to_yuyv = yv12_to_yuyv_c;
319          yv12_to_uyvy = yv12_to_uyvy_c;          yv12_to_uyvy = yv12_to_uyvy_c;
320    
321            yv12_to_rgb555i = yv12_to_rgb555i_c;
322            yv12_to_rgb565i = yv12_to_rgb565i_c;
323            yv12_to_bgri    = yv12_to_bgri_c;
324            yv12_to_bgrai   = yv12_to_bgrai_c;
325            yv12_to_abgri   = yv12_to_abgri_c;
326            yv12_to_rgbai   = yv12_to_rgbai_c;
327            yv12_to_argbi   = yv12_to_argbi_c;
328            yv12_to_yuyvi   = yv12_to_yuyvi_c;
329            yv12_to_uyvyi   = yv12_to_uyvyi_c;
330    
331            /* Functions used in motion estimation algorithms */
332          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
333          sad16 = sad16_c;          sad16 = sad16_c;
334          sad8 = sad8_c;          sad8 = sad8_c;
335            sad16bi    = sad16bi_c;
336            sad8bi     = sad8bi_c;
337          dev16 = dev16_c;          dev16 = dev16_c;
338            sad16v     = sad16v_c;
339            sse8_16bit = sse8_16bit_c;
340            sse8_8bit  = sse8_8bit_c;
341    
342            init_GMC(cpu_flags);
343    
344    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
345    
346            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
347                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
348                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2) ||
349            (cpu_flags & XVID_CPU_SSE3) || (cpu_flags & XVID_CPU_SSE41))
350            {
351                    /* Restore FPU context : emms_c is a nop functions */
352                    emms = emms_mmx;
353            }
354    
355            if ((cpu_flags & XVID_CPU_MMX)) {
356    
357  #ifdef ARCH_X86                  /* Forward and Inverse Discrete Cosine Transformation functions */
358          if((cpu_flags & XVID_CPU_MMX) > 0) {                  fdct = fdct_mmx_skal;
                 fdct = fdct_mmx;  
359                  idct = idct_mmx;                  idct = idct_mmx;
360    
361                  emms = emms_mmx;                  /* Qpel stuff */
362                    xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
363                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
364    
365                    /* Quantization related functions */
366                    quant_h263_intra   = quant_h263_intra_mmx;
367                    quant_h263_inter   = quant_h263_inter_mmx;
368                    dequant_h263_intra = dequant_h263_intra_mmx;
369                    dequant_h263_inter = dequant_h263_inter_mmx;
370                    quant_mpeg_intra   = quant_mpeg_intra_mmx;
371                    quant_mpeg_inter   = quant_mpeg_inter_mmx;
372                    dequant_mpeg_intra = dequant_mpeg_intra_mmx;
373                    dequant_mpeg_inter = dequant_mpeg_inter_mmx;
374    
                 quant_intra = quant_intra_mmx;  
                 dequant_intra = dequant_intra_mmx;  
                 quant_inter = quant_inter_mmx;  
                 dequant_inter = dequant_inter_mmx;  
   
                 quant4_intra = quant4_intra_mmx;  
                 dequant4_intra = dequant4_intra_mmx;  
                 quant4_inter = quant4_inter_mmx;  
                 dequant4_inter = dequant4_inter_mmx;  
375    
376                    /* Block related functions */
377                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
378                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
379                  transfer_8to16sub = transfer_8to16sub_mmx;                  transfer_8to16sub = transfer_8to16sub_mmx;
380                    transfer_8to16subro  = transfer_8to16subro_mmx;
381                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
382                  transfer_16to8add = transfer_16to8add_mmx;                  transfer_16to8add = transfer_16to8add_mmx;
383                  transfer8x8_copy = transfer8x8_copy_mmx;                  transfer8x8_copy = transfer8x8_copy_mmx;
384                    transfer8x4_copy   = transfer8x4_copy_mmx;
385    
386                    /* Interlacing Functions */
387                    MBFieldTest = MBFieldTest_mmx;
388    
389                    /* Image Interpolation related functions */
390                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;
391                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;
392                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
393    
394                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
395                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
396                  yuv_to_yv12 = yuv_to_yv12_mmx;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
397    
398                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
399                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
400                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
401                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
402    
403                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
404                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
405    
406                    interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
407                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
408    
409                    /* postprocessing */
410                    image_brightness = image_brightness_mmx;
411    
412                    /* image input xxx_to_yv12 related functions */
413    
414                    yv12_to_yv12  = yv12_to_yv12_mmx;
415    
416                    bgr_to_yv12   = bgr_to_yv12_mmx;
417                    rgb_to_yv12   = rgb_to_yv12_mmx;
418                    bgra_to_yv12  = bgra_to_yv12_mmx;
419                    rgba_to_yv12  = rgba_to_yv12_mmx;
420                  yuyv_to_yv12 = yuyv_to_yv12_mmx;                  yuyv_to_yv12 = yuyv_to_yv12_mmx;
421                  uyvy_to_yv12 = uyvy_to_yv12_mmx;                  uyvy_to_yv12 = uyvy_to_yv12_mmx;
422    
423                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  /* image output yv12_to_xxx related functions */
424                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
425                    yv12_to_bgra  = yv12_to_bgra_mmx;
426                  yv12_to_yuyv = yv12_to_yuyv_mmx;                  yv12_to_yuyv = yv12_to_yuyv_mmx;
427                  yv12_to_uyvy = yv12_to_uyvy_mmx;                  yv12_to_uyvy = yv12_to_uyvy_mmx;
428    
429                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
430                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
431    
432                    /* Motion estimation related functions */
433                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
434                  sad16 = sad16_mmx;                  sad16 = sad16_mmx;
435                  sad8 = sad8_mmx;                  sad8 = sad8_mmx;
436                    sad16bi    = sad16bi_mmx;
437                    sad8bi     = sad8bi_mmx;
438                  dev16 = dev16_mmx;                  dev16 = dev16_mmx;
439                    sad16v     = sad16v_mmx;
440                    sse8_16bit = sse8_16bit_mmx;
441                    sse8_8bit  = sse8_8bit_mmx;
442            }
443    
444            /* these 3dnow functions are faster than mmx, but slower than xmm. */
445            if ((cpu_flags & XVID_CPU_3DNOW)) {
446    
447                    emms = emms_3dn;
448    
449                    /* ME functions */
450                    sad16bi = sad16bi_3dn;
451                    sad8bi  = sad8bi_3dn;
452    
453                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
454                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
455    
456          }          }
457    
458          if((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
459            if ((cpu_flags & XVID_CPU_MMXEXT)) {
460    
461                    /* DCT */
462                    fdct = fdct_xmm_skal;
463                  idct = idct_xmm;                  idct = idct_xmm;
464    
465                    /* Interpolation */
466                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;
467                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;
468                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
                 yuv_to_yv12 = yuv_to_yv12_xmm;  
469    
470                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
471                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
472                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
473    
474                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
475                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
476                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
477                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
478    
479            /* Quantization */
480                    quant_mpeg_inter = quant_mpeg_inter_xmm;
481    
482                    dequant_h263_intra = dequant_h263_intra_xmm;
483                    dequant_h263_inter = dequant_h263_inter_xmm;
484    
485            /* Buffer transfer */
486                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
487                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
488    
489                    /* Colorspace transformation */
490                    /* yv12_to_yv12  = yv12_to_yv12_xmm; */ /* appears to be slow on many machines */
491                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
492                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
493    
494                    /* ME functions */
495                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
496                  sad8 = sad8_xmm;                  sad8 = sad8_xmm;
497                    sad16bi = sad16bi_xmm;
498                    sad8bi  = sad8bi_xmm;
499                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
500                    sad16v   = sad16v_xmm;
501          }          }
502    
503          if((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
504    
505                    /* Interpolation */
506                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
507                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
508                    interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
509    
510                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
511                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
512                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
513            }
514    
515            if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
516    
517                    /* Buffer transfer */
518                    transfer_8to16copy =  transfer_8to16copy_3dne;
519                    transfer_16to8copy = transfer_16to8copy_3dne;
520                    transfer_8to16sub =  transfer_8to16sub_3dne;
521                    transfer_8to16subro =  transfer_8to16subro_3dne;
522                    transfer_16to8add = transfer_16to8add_3dne;
523                    transfer8x8_copy = transfer8x8_copy_3dne;
524                    transfer8x4_copy = transfer8x4_copy_3dne;
525    
526                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
527                            /* Inverse DCT */
528                            idct =  idct_3dne;
529    
530                            /* Buffer transfer */
531                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
532    
533                            /* Interpolation */
534                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
535                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
536                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
537    
538                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
539                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
540                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
541    
542                /* Quantization */
543                            quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
544                            quant_h263_inter = quant_h263_inter_3dne;
545                            dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
546                            dequant_mpeg_inter = dequant_mpeg_inter_3dne;
547                            dequant_h263_intra = dequant_h263_intra_3dne;
548                            dequant_h263_inter = dequant_h263_inter_3dne;
549    
550                /* ME functions */
551                            sad16 = sad16_3dne;
552                            sad8 = sad8_3dne;
553                            sad16bi = sad16bi_3dne;
554                            sad8bi = sad8bi_3dne;
555                            dev16 = dev16_3dne;
556                    }
557            }
558    
559            if ((cpu_flags & XVID_CPU_SSE2)) {
560    
561                    calc_cbp = calc_cbp_sse2;
562    
563                    /* Quantization */
564                    quant_h263_intra   = quant_h263_intra_sse2;
565                    quant_h263_inter   = quant_h263_inter_sse2;
566                    dequant_h263_intra = dequant_h263_intra_sse2;
567                    dequant_h263_inter = dequant_h263_inter_sse2;
568    
569                    /* SAD operators */
570                    sad16    = sad16_sse2;
571                    dev16    = dev16_sse2;
572    
573                    /* DCT operators */
574                    fdct = fdct_sse2_skal;
575                    idct = idct_sse2_skal;   /* Is now IEEE1180 and Walken compliant. */
576    
577                    /* postprocessing */
578                    image_brightness = image_brightness_sse2;
579    
580            }
581    
582            if ((cpu_flags & XVID_CPU_SSE3)) {
583    
584                    /* SAD operators */
585                    sad16    = sad16_sse3;
586                    dev16    = dev16_sse3;
587          }          }
588    
589    #endif /* ARCH_IS_IA32 */
590    
591    #if defined(ARCH_IS_IA64)
592            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
593              idct_ia64_init();
594              fdct = fdct_ia64;
595              idct = idct_ia64;   /*not yet working, crashes */
596              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
597              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
598              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
599              sad16 = sad16_ia64;
600              sad16bi = sad16bi_ia64;
601              sad8 = sad8_ia64;
602              dev16 = dev16_ia64;
603    /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
604              quant_h263_intra = quant_h263_intra_ia64;
605              quant_h263_inter = quant_h263_inter_ia64;
606              dequant_h263_intra = dequant_h263_intra_ia64;
607              dequant_h263_inter = dequant_h263_inter_ia64;
608              transfer_8to16copy = transfer_8to16copy_ia64;
609              transfer_16to8copy = transfer_16to8copy_ia64;
610              transfer_8to16sub = transfer_8to16sub_ia64;
611              transfer_8to16sub2 = transfer_8to16sub2_ia64;
612              transfer_16to8add = transfer_16to8add_ia64;
613              transfer8x8_copy = transfer8x8_copy_ia64;
614            }
615  #endif  #endif
616    
617          // API version  #if defined(ARCH_IS_PPC)
618          init_param->api_version = API_VERSION;          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
619              /* sad operators */
620                      sad16 = sad16_altivec_c;
621                      sad16bi = sad16bi_altivec_c;
622                      sad8 = sad8_altivec_c;
623                      dev16 = dev16_altivec_c;
624    
625              sse8_16bit = sse8_16bit_altivec_c;
626    
627              /* mem transfer */
628              transfer_8to16copy = transfer_8to16copy_altivec_c;
629              transfer_16to8copy = transfer_16to8copy_altivec_c;
630              transfer_8to16sub = transfer_8to16sub_altivec_c;
631              transfer_8to16subro = transfer_8to16subro_altivec_c;
632              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
633              transfer_16to8add = transfer_16to8add_altivec_c;
634              transfer8x8_copy = transfer8x8_copy_altivec_c;
635    
636              /* Inverse DCT */
637              idct = idct_altivec_c;
638    
639              /* Interpolation */
640              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
641              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
642              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
643    
644              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
645              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
646    
647                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
648                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
649                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
650                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
651    
652              /* Colorspace conversion */
653              bgra_to_yv12 = bgra_to_yv12_altivec_c;
654              abgr_to_yv12 = abgr_to_yv12_altivec_c;
655              rgba_to_yv12 = rgba_to_yv12_altivec_c;
656              argb_to_yv12 = argb_to_yv12_altivec_c;
657    
658              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
659              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
660    
661              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
662              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
663    
664              /* Quantization */
665              quant_h263_intra = quant_h263_intra_altivec_c;
666              quant_h263_inter = quant_h263_inter_altivec_c;
667              dequant_h263_intra = dequant_h263_intra_altivec_c;
668              dequant_h263_inter = dequant_h263_inter_altivec_c;
669    
670                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
671                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
672    
673                      /* Qpel stuff */
674                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
675                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
676            }
677    #endif
678    
679          // something clever has to be done for this  #if defined(_DEBUG)
680          init_param->core_build = 1000;      xvid_debug = init->debug;
681    #endif
682    
683          return XVID_ERR_OK;      return(0);
684  }  }
685    
686  int xvid_decore(void * handle, int opt, void * param1, void * param2)  
687    static int
688    xvid_gbl_info(xvid_gbl_info_t * info)
689    {
690            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
691                    return XVID_ERR_VERSION;
692    
693            info->actual_version = XVID_VERSION;
694            info->build = "xvid-1.3.0-dev";
695            info->cpu_flags = detect_cpu_flags();
696            info->num_threads = 0; /* single-thread */
697    
698    #if defined(_WIN32)
699    
700      {
701            SYSTEM_INFO siSysInfo;
702            GetSystemInfo(&siSysInfo);
703            info->num_threads = siSysInfo.dwNumberOfProcessors; /* number of _logical_ cores */
704      }
705    
706    #elif defined(_SC_NPROCESSORS_CONF) /* should be available on Apple too actually */
707    
708      info->num_threads = sysconf(_SC_NPROCESSORS_CONF);
709    
710    #elif defined(__APPLE__) && defined(__MACH__)
711    
712      {
713        size_t len;
714        int    mib[2], ncpu;
715    
716        mib[0] = CTL_HW;
717        mib[1] = HW_NCPU;
718        len    = sizeof(ncpu);
719        if (sysctl(mib, 2, &ncpu, &len, NULL, 0) == 0)
720          info -> num_threads = ncpu;
721        else
722          info -> num_threads = 1;
723      }
724    
725    #elif defined(__amigaos4__)
726    
727      {
728        uint32_t num_threads = 1;
729        IExec->GetCPUInfoTags(GCIT_NumberOfCPUs, &num_threads, TAG_END);
730        info->num_threads = num_threads;
731      }
732    
733    #endif
734    
735            return 0;
736    }
737    
738    
739    static int
740    xvid_gbl_convert(xvid_gbl_convert_t* convert)
741    {
742            int width;
743            int height;
744            int width2;
745            int height2;
746            IMAGE img;
747    
748            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
749                  return XVID_ERR_VERSION;
750    
751    #if 0
752            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
753    #endif
754            width = convert->width;
755            height = convert->height;
756            width2 = convert->width/2;
757            height2 = convert->height/2;
758    
759            switch (convert->input.csp & ~XVID_CSP_VFLIP)
760            {
761                    case XVID_CSP_YV12 :
762                            img.y = convert->input.plane[0];
763                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
764                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
765                            image_output(&img, width, height, width,
766                                                    (uint8_t**)convert->output.plane, convert->output.stride,
767                                                    convert->output.csp, convert->interlacing);
768                            break;
769    
770                    default :
771                            return XVID_ERR_FORMAT;
772            }
773    
774    
775            emms();
776            return 0;
777    }
778    
779    /*****************************************************************************
780     * Xvid Global Entry point
781     *
782     * Well this function initialize all internal function pointers according
783     * to the CPU features forced by the library client or autodetected (depending
784     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
785     * image colorspace transformation tables.
786     *
787     ****************************************************************************/
788    
789    
790    int
791    xvid_global(void *handle,
792                      int opt,
793                      void *param1,
794                      void *param2)
795  {  {
796          switch (opt)          switch (opt)
797          {          {
798          case XVID_DEC_DECODE :                  case XVID_GBL_INIT :
799          return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);                          return xvid_gbl_init((xvid_gbl_init_t*)param1);
800    
801            case XVID_GBL_INFO :
802                return xvid_gbl_info((xvid_gbl_info_t*)param1);
803    
804                    case XVID_GBL_CONVERT :
805                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
806    
807                    default :
808                            return XVID_ERR_FAIL;
809            }
810    }
811    
812    /*****************************************************************************
813     * Xvid Native decoder entry point
814     *
815     * This function is just a wrapper to all the option cases.
816     *
817     * Returned values : XVID_ERR_FAIL when opt is invalid
818     *                   else returns the wrapped function result
819     *
820     ****************************************************************************/
821    
822    int
823    xvid_decore(void *handle,
824                            int opt,
825                            void *param1,
826                            void *param2)
827    {
828            switch (opt) {
829          case XVID_DEC_CREATE :          case XVID_DEC_CREATE :
830          return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
831    
832          case XVID_DEC_DESTROY :          case XVID_DEC_DESTROY :
833          return decoder_destroy((DECODER *) handle);          return decoder_destroy((DECODER *) handle);
834    
835            case XVID_DEC_DECODE:
836                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
837    
838          default:          default:
839          return XVID_ERR_FAIL;          return XVID_ERR_FAIL;
840      }      }
841  }  }
842    
843    
844  int xvid_encore(void * handle, int opt, void * param1, void * param2)  /*****************************************************************************
845  {   * Xvid Native encoder entry point
846          switch (opt)   *
847     * This function is just a wrapper to all the option cases.
848     *
849     * Returned values : XVID_ERR_FAIL when opt is invalid
850     *                   else returns the wrapped function result
851     *
852     ****************************************************************************/
853    
854    int
855    xvid_encore(void *handle,
856                            int opt,
857                            void *param1,
858                            void *param2)
859          {          {
860            switch (opt) {
861          case XVID_ENC_ENCODE :          case XVID_ENC_ENCODE :
862          return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, (XVID_ENC_STATS *) param2);  
863                    return enc_encode((Encoder *) handle,
864                                                              (xvid_enc_frame_t *) param1,
865                                                              (xvid_enc_stats_t *) param2);
866    
867          case XVID_ENC_CREATE :          case XVID_ENC_CREATE :
868          return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
869    
870          case XVID_ENC_DESTROY :          case XVID_ENC_DESTROY :
871          return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
872    
873          default:          default:
874          return XVID_ERR_FAIL;          return XVID_ERR_FAIL;

Legend:
Removed from v.1.3  
changed lines
  Added in v.1.83

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