[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 1.82, Fri Jun 5 07:58:41 2009 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    #if !defined(_WIN32)
32    #include <unistd.h>
33    #endif
34    
35    #if defined(__APPLE__) && defined(__MACH__) && !defined(_SC_NPROCESSORS_CONF)
36    #include <sys/types.h>
37    #include <sys/sysctl.h>
38    #ifdef MAX
39    #undef MAX
40    #endif
41    #ifdef MIN
42    #undef MIN
43    #endif
44    #endif
45    
46    #if defined(__amigaos4__)
47    #include <exec/exec.h>
48    #include <proto/exec.h>
49    #endif
50    
51  #include "xvid.h"  #include "xvid.h"
52  #include "decoder.h"  #include "decoder.h"
# Line 48  Line 57 
57  #include "image/colorspace.h"  #include "image/colorspace.h"
58  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
59  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
60  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
61  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
62    #include "motion/motion.h"
63    #include "motion/gmc.h"
64  #include "motion/sad.h"  #include "motion/sad.h"
65  #include "utils/emms.h"  #include "utils/emms.h"
66  #include "utils/timer.h"  #include "utils/timer.h"
67    #include "bitstream/mbcoding.h"
68    #include "image/qpel.h"
69    #include "image/postprocessing.h"
70    
71  int xvid_init(void *handle, int opt, void *param1, void *param2)  #if defined(_DEBUG)
72    unsigned int xvid_debug = 0; /* xvid debug mask */
73    #endif
74    
75    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
76    #       include <windows.h>
77    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || (defined(ARCH_IS_PPC) && !defined(__amigaos4__))
78    #       include <signal.h>
79    #       include <setjmp.h>
80    
81            static jmp_buf mark;
82    
83            static void
84            sigill_handler(int signal)
85  {  {
86          int cpu_flags;             longjmp(mark, 1);
87          XVID_INIT_PARAM *init_param;          }
88    #endif
89    
90          init_param = (XVID_INIT_PARAM *) param1;  /*
91     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
92     * signalled
93     *
94     * Return values:
95     *  -1 : could not determine
96     *   0 : SIGILL was *not* signalled
97     *   1 : SIGILL was signalled
98     */
99    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
100    static int
101    sigill_check(void (*func)())
102    {
103            _try {
104                    func();
105            } _except(EXCEPTION_EXECUTE_HANDLER) {
106    
107          // force specific cpu settings?                  if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
108          if((init_param->cpu_flags & XVID_CPU_FORCE) > 0)                          return(1);
109                  cpu_flags = init_param->cpu_flags;          }
110          else {          return(0);
111    }
112    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || (defined(ARCH_IS_PPC) && !defined(__amigaos4__))
113    static int
114    sigill_check(void (*func)())
115    {
116        void *old_handler;
117        int jmpret;
118    
119  #ifdef ARCH_X86      /* Set our SIGILL handler */
120                  cpu_flags = check_cpu_features();      old_handler = signal(SIGILL, sigill_handler);
121    
122        /* Check for error */
123        if (old_handler == SIG_ERR) {
124            return(-1);
125        }
126    
127        /* Save stack context, so if func triggers a SIGILL, we can still roll
128             * back to a valid CPU state */
129            jmpret = setjmp(mark);
130    
131            /* If setjmp returned directly, then its returned value is 0, and we still
132             * have to test the passed func. Otherwise it means the stack context has
133             * been restored by a longjmp() call, which in our case happens only in the
134             * signal handler */
135        if (jmpret == 0) {
136            func();
137        }
138    
139        /* Restore old signal handler */
140        signal(SIGILL, old_handler);
141    
142        return(jmpret);
143    }
144    #endif
145    
146    
147    /* detect cpu flags  */
148    static unsigned int
149    detect_cpu_flags(void)
150    {
151            /* enable native assembly optimizations by default */
152            unsigned int cpu_flags = XVID_CPU_ASM;
153    
154    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
155            cpu_flags |= check_cpu_features();
156            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
157                    cpu_flags &= ~XVID_CPU_SSE;
158    
159            if ((cpu_flags & (XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41)) && sigill_check(sse2_os_trigger))
160                    cpu_flags &= ~(XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41);
161    #endif
162    
163    #if defined(ARCH_IS_PPC)
164    #if defined(__amigaos4__)
165            {
166                    uint32_t vector_unit = VECTORTYPE_NONE;
167                    IExec->GetCPUInfoTags(GCIT_VectorUnit, &vector_unit, TAG_END);
168                    if (vector_unit == VECTORTYPE_ALTIVEC) {
169                            cpu_flags |= XVID_CPU_ALTIVEC;
170                    }
171            }
172  #else  #else
173                  cpu_flags = 0;          if (!sigill_check(altivec_trigger))
174                    cpu_flags |= XVID_CPU_ALTIVEC;
175    #endif
176  #endif  #endif
177                  init_param->cpu_flags = cpu_flags;  
178            return cpu_flags;
179          }          }
180    
181          // initialize the function pointers  
182    /*****************************************************************************
183     * XviD Init Entry point
184     *
185     * Well this function initialize all internal function pointers according
186     * to the CPU features forced by the library client or autodetected (depending
187     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
188     * image colorspace transformation tables.
189     *
190     * Returned value : XVID_ERR_OK
191     *                  + API_VERSION in the input XVID_INIT_PARAM structure
192     *                  + core build  "   "    "       "               "
193     *
194     ****************************************************************************/
195    
196    
197    static
198    int xvid_gbl_init(xvid_gbl_init_t * init)
199    {
200            unsigned int cpu_flags;
201    
202            if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
203                    return XVID_ERR_VERSION;
204    
205            cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
206    
207            /* Initialize the function pointers */
208          idct_int32_init();          idct_int32_init();
209            init_vlc_tables();
210    
211            /* Fixed Point Forward/Inverse DCT transformations */
212          fdct = fdct_int32;          fdct = fdct_int32;
213          idct = idct_int32;          idct = idct_int32;
214    
215            /* Only needed on PPC Altivec archs */
216            sadInit = NULL;
217    
218            /* Restore FPU context : emms_c is a nop functions */
219          emms = emms_c;          emms = emms_c;
220    
221          quant_intra = quant_intra_c;          /* Qpel stuff */
222          dequant_intra = dequant_intra_c;          xvid_QP_Funcs = &xvid_QP_Funcs_C;
223          quant_inter = quant_inter_c;          xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
224          dequant_inter = dequant_inter_c;          xvid_Init_QP();
225    
226          quant4_intra = quant4_intra_c;          /* Quantization functions */
227          dequant4_intra = dequant4_intra_c;          quant_h263_intra   = quant_h263_intra_c;
228          quant4_inter = quant4_inter_c;          quant_h263_inter   = quant_h263_inter_c;
229          dequant4_inter = dequant4_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
230            dequant_h263_inter = dequant_h263_inter_c;
231    
232            quant_mpeg_intra   = quant_mpeg_intra_c;
233            quant_mpeg_inter   = quant_mpeg_inter_c;
234            dequant_mpeg_intra = dequant_mpeg_intra_c;
235            dequant_mpeg_inter = dequant_mpeg_inter_c;
236    
237            /* Block transfer related functions */
238          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
239          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
240          transfer_8to16sub = transfer_8to16sub_c;          transfer_8to16sub = transfer_8to16sub_c;
241            transfer_8to16subro  = transfer_8to16subro_c;
242            transfer_8to16sub2 = transfer_8to16sub2_c;
243            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
244          transfer_16to8add = transfer_16to8add_c;          transfer_16to8add = transfer_16to8add_c;
245          transfer8x8_copy = transfer8x8_copy_c;          transfer8x8_copy = transfer8x8_copy_c;
246            transfer8x4_copy   = transfer8x4_copy_c;
247    
248            /* Interlacing functions */
249            MBFieldTest = MBFieldTest_c;
250    
251            /* Image interpolation related functions */
252          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;
253          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;
254          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
255    
256            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
257            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
258            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
259    
260            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
261            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
262            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
263            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
264    
265            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
266            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
267            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
268    
269            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
270            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
271            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
272    
273            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
274            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
275    
276            interpolate8x8_avg2 = interpolate8x8_avg2_c;
277            interpolate8x8_avg4 = interpolate8x8_avg4_c;
278    
279            /* postprocessing */
280            image_brightness = image_brightness_c;
281    
282            /* Initialize internal colorspace transformation tables */
283          colorspace_init();          colorspace_init();
284    
285            /* All colorspace transformation functions User Format->YV12 */
286            yv12_to_yv12    = yv12_to_yv12_c;
287          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
288          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
289          rgb24_to_yv12 = rgb24_to_yv12_c;          rgb_to_yv12     = rgb_to_yv12_c;
290          rgb32_to_yv12 = rgb32_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
291          yuv_to_yv12 = yuv_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
292            abgr_to_yv12    = abgr_to_yv12_c;
293            rgba_to_yv12    = rgba_to_yv12_c;
294            argb_to_yv12    = argb_to_yv12_c;
295          yuyv_to_yv12 = yuyv_to_yv12_c;          yuyv_to_yv12 = yuyv_to_yv12_c;
296          uyvy_to_yv12 = uyvy_to_yv12_c;          uyvy_to_yv12 = uyvy_to_yv12_c;
297    
298            rgb555i_to_yv12 = rgb555i_to_yv12_c;
299            rgb565i_to_yv12 = rgb565i_to_yv12_c;
300            bgri_to_yv12    = bgri_to_yv12_c;
301            bgrai_to_yv12   = bgrai_to_yv12_c;
302            abgri_to_yv12   = abgri_to_yv12_c;
303            rgbai_to_yv12   = rgbai_to_yv12_c;
304            argbi_to_yv12   = argbi_to_yv12_c;
305            yuyvi_to_yv12   = yuyvi_to_yv12_c;
306            uyvyi_to_yv12   = uyvyi_to_yv12_c;
307    
308            /* All colorspace transformation functions YV12->User format */
309          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
310          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
311          yv12_to_rgb24 = yv12_to_rgb24_c;          yv12_to_rgb     = yv12_to_rgb_c;
312          yv12_to_rgb32 = yv12_to_rgb32_c;          yv12_to_bgr     = yv12_to_bgr_c;
313          yv12_to_yuv = yv12_to_yuv_c;          yv12_to_bgra    = yv12_to_bgra_c;
314            yv12_to_abgr    = yv12_to_abgr_c;
315            yv12_to_rgba    = yv12_to_rgba_c;
316            yv12_to_argb    = yv12_to_argb_c;
317          yv12_to_yuyv = yv12_to_yuyv_c;          yv12_to_yuyv = yv12_to_yuyv_c;
318          yv12_to_uyvy = yv12_to_uyvy_c;          yv12_to_uyvy = yv12_to_uyvy_c;
319    
320            yv12_to_rgb555i = yv12_to_rgb555i_c;
321            yv12_to_rgb565i = yv12_to_rgb565i_c;
322            yv12_to_bgri    = yv12_to_bgri_c;
323            yv12_to_bgrai   = yv12_to_bgrai_c;
324            yv12_to_abgri   = yv12_to_abgri_c;
325            yv12_to_rgbai   = yv12_to_rgbai_c;
326            yv12_to_argbi   = yv12_to_argbi_c;
327            yv12_to_yuyvi   = yv12_to_yuyvi_c;
328            yv12_to_uyvyi   = yv12_to_uyvyi_c;
329    
330            /* Functions used in motion estimation algorithms */
331          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
332          sad16 = sad16_c;          sad16 = sad16_c;
333          sad8 = sad8_c;          sad8 = sad8_c;
334            sad16bi    = sad16bi_c;
335            sad8bi     = sad8bi_c;
336          dev16 = dev16_c;          dev16 = dev16_c;
337            sad16v     = sad16v_c;
338            sse8_16bit = sse8_16bit_c;
339            sse8_8bit  = sse8_8bit_c;
340    
341            init_GMC(cpu_flags);
342    
343    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
344    
345            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
346                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
347                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2) ||
348            (cpu_flags & XVID_CPU_SSE3) || (cpu_flags & XVID_CPU_SSE41))
349            {
350                    /* Restore FPU context : emms_c is a nop functions */
351                    emms = emms_mmx;
352            }
353    
354            if ((cpu_flags & XVID_CPU_MMX)) {
355    
356  #ifdef ARCH_X86                  /* Forward and Inverse Discrete Cosine Transformation functions */
357          if((cpu_flags & XVID_CPU_MMX) > 0) {                  fdct = fdct_mmx_skal;
                 fdct = fdct_mmx;  
358                  idct = idct_mmx;                  idct = idct_mmx;
359    
360                  emms = emms_mmx;                  /* Qpel stuff */
361                    xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
362                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
363    
364                    /* Quantization related functions */
365                    quant_h263_intra   = quant_h263_intra_mmx;
366                    quant_h263_inter   = quant_h263_inter_mmx;
367                    dequant_h263_intra = dequant_h263_intra_mmx;
368                    dequant_h263_inter = dequant_h263_inter_mmx;
369                    quant_mpeg_intra   = quant_mpeg_intra_mmx;
370                    quant_mpeg_inter   = quant_mpeg_inter_mmx;
371                    dequant_mpeg_intra = dequant_mpeg_intra_mmx;
372                    dequant_mpeg_inter = dequant_mpeg_inter_mmx;
373    
                 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;  
374    
375                    /* Block related functions */
376                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
377                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
378                  transfer_8to16sub = transfer_8to16sub_mmx;                  transfer_8to16sub = transfer_8to16sub_mmx;
379                    transfer_8to16subro  = transfer_8to16subro_mmx;
380                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
381                  transfer_16to8add = transfer_16to8add_mmx;                  transfer_16to8add = transfer_16to8add_mmx;
382                  transfer8x8_copy = transfer8x8_copy_mmx;                  transfer8x8_copy = transfer8x8_copy_mmx;
383                    transfer8x4_copy   = transfer8x4_copy_mmx;
384    
385                    /* Interlacing Functions */
386                    MBFieldTest = MBFieldTest_mmx;
387    
388                    /* Image Interpolation related functions */
389                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;
390                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;
391                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
392    
393                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
394                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
395                  yuv_to_yv12 = yuv_to_yv12_mmx;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
396    
397                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
398                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
399                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
400                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
401    
402                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
403                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
404    
405                    interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
406                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
407    
408                    /* postprocessing */
409                    image_brightness = image_brightness_mmx;
410    
411                    /* image input xxx_to_yv12 related functions */
412    
413                    yv12_to_yv12  = yv12_to_yv12_mmx;
414    
415                    bgr_to_yv12   = bgr_to_yv12_mmx;
416                    rgb_to_yv12   = rgb_to_yv12_mmx;
417                    bgra_to_yv12  = bgra_to_yv12_mmx;
418                    rgba_to_yv12  = rgba_to_yv12_mmx;
419                  yuyv_to_yv12 = yuyv_to_yv12_mmx;                  yuyv_to_yv12 = yuyv_to_yv12_mmx;
420                  uyvy_to_yv12 = uyvy_to_yv12_mmx;                  uyvy_to_yv12 = uyvy_to_yv12_mmx;
421    
422                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  /* image output yv12_to_xxx related functions */
423                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
424                    yv12_to_bgra  = yv12_to_bgra_mmx;
425                  yv12_to_yuyv = yv12_to_yuyv_mmx;                  yv12_to_yuyv = yv12_to_yuyv_mmx;
426                  yv12_to_uyvy = yv12_to_uyvy_mmx;                  yv12_to_uyvy = yv12_to_uyvy_mmx;
427    
428                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
429                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
430    
431                    /* Motion estimation related functions */
432                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
433                  sad16 = sad16_mmx;                  sad16 = sad16_mmx;
434                  sad8 = sad8_mmx;                  sad8 = sad8_mmx;
435                    sad16bi    = sad16bi_mmx;
436                    sad8bi     = sad8bi_mmx;
437                  dev16 = dev16_mmx;                  dev16 = dev16_mmx;
438                    sad16v     = sad16v_mmx;
439                    sse8_16bit = sse8_16bit_mmx;
440                    sse8_8bit  = sse8_8bit_mmx;
441            }
442    
443            /* these 3dnow functions are faster than mmx, but slower than xmm. */
444            if ((cpu_flags & XVID_CPU_3DNOW)) {
445    
446                    emms = emms_3dn;
447    
448                    /* ME functions */
449                    sad16bi = sad16bi_3dn;
450                    sad8bi  = sad8bi_3dn;
451    
452                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
453                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
454    
455          }          }
456    
457          if((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
458            if ((cpu_flags & XVID_CPU_MMXEXT)) {
459    
460                    /* DCT */
461                    fdct = fdct_xmm_skal;
462                  idct = idct_xmm;                  idct = idct_xmm;
463    
464                    /* Interpolation */
465                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;
466                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;
467                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
                 yuv_to_yv12 = yuv_to_yv12_xmm;  
468    
469                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
470                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
471                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
472    
473                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
474                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
475                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
476                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
477    
478            /* Quantization */
479                    quant_mpeg_inter = quant_mpeg_inter_xmm;
480    
481                    dequant_h263_intra = dequant_h263_intra_xmm;
482                    dequant_h263_inter = dequant_h263_inter_xmm;
483    
484            /* Buffer transfer */
485                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
486                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
487    
488                    /* Colorspace transformation */
489                    /* yv12_to_yv12  = yv12_to_yv12_xmm; */ /* appears to be slow on many machines */
490                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
491                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
492    
493                    /* ME functions */
494                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
495                  sad8 = sad8_xmm;                  sad8 = sad8_xmm;
496                    sad16bi = sad16bi_xmm;
497                    sad8bi  = sad8bi_xmm;
498                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
499                    sad16v   = sad16v_xmm;
500          }          }
501    
502          if((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
503    
504                    /* Interpolation */
505                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
506                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
507                    interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
508    
509                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
510                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
511                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
512            }
513    
514            if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
515    
516                    /* Buffer transfer */
517                    transfer_8to16copy =  transfer_8to16copy_3dne;
518                    transfer_16to8copy = transfer_16to8copy_3dne;
519                    transfer_8to16sub =  transfer_8to16sub_3dne;
520                    transfer_8to16subro =  transfer_8to16subro_3dne;
521                    transfer_16to8add = transfer_16to8add_3dne;
522                    transfer8x8_copy = transfer8x8_copy_3dne;
523                    transfer8x4_copy = transfer8x4_copy_3dne;
524    
525                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
526                            /* Inverse DCT */
527                            idct =  idct_3dne;
528    
529                            /* Buffer transfer */
530                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
531    
532                            /* Interpolation */
533                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
534                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
535                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
536    
537                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
538                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
539                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
540    
541                /* Quantization */
542                            quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
543                            quant_h263_inter = quant_h263_inter_3dne;
544                            dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
545                            dequant_mpeg_inter = dequant_mpeg_inter_3dne;
546                            dequant_h263_intra = dequant_h263_intra_3dne;
547                            dequant_h263_inter = dequant_h263_inter_3dne;
548    
549                /* ME functions */
550                            sad16 = sad16_3dne;
551                            sad8 = sad8_3dne;
552                            sad16bi = sad16bi_3dne;
553                            sad8bi = sad8bi_3dne;
554                            dev16 = dev16_3dne;
555                    }
556            }
557    
558            if ((cpu_flags & XVID_CPU_SSE2)) {
559    
560                    calc_cbp = calc_cbp_sse2;
561    
562                    /* Quantization */
563                    quant_h263_intra   = quant_h263_intra_sse2;
564                    quant_h263_inter   = quant_h263_inter_sse2;
565                    dequant_h263_intra = dequant_h263_intra_sse2;
566                    dequant_h263_inter = dequant_h263_inter_sse2;
567    
568                    /* SAD operators */
569                    sad16    = sad16_sse2;
570                    dev16    = dev16_sse2;
571    
572                    /* DCT operators */
573                    fdct = fdct_sse2_skal;
574                    idct = idct_sse2_skal;   /* Is now IEEE1180 and Walken compliant. */
575    
576                    /* postprocessing */
577                    image_brightness = image_brightness_sse2;
578    
579            }
580    
581            if ((cpu_flags & XVID_CPU_SSE3)) {
582    
583                    /* SAD operators */
584                    sad16    = sad16_sse3;
585                    dev16    = dev16_sse3;
586          }          }
587    
588    #endif /* ARCH_IS_IA32 */
589    
590    #if defined(ARCH_IS_IA64)
591            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
592              idct_ia64_init();
593              fdct = fdct_ia64;
594              idct = idct_ia64;   /*not yet working, crashes */
595              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
596              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
597              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
598              sad16 = sad16_ia64;
599              sad16bi = sad16bi_ia64;
600              sad8 = sad8_ia64;
601              dev16 = dev16_ia64;
602    /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
603              quant_h263_intra = quant_h263_intra_ia64;
604              quant_h263_inter = quant_h263_inter_ia64;
605              dequant_h263_intra = dequant_h263_intra_ia64;
606              dequant_h263_inter = dequant_h263_inter_ia64;
607              transfer_8to16copy = transfer_8to16copy_ia64;
608              transfer_16to8copy = transfer_16to8copy_ia64;
609              transfer_8to16sub = transfer_8to16sub_ia64;
610              transfer_8to16sub2 = transfer_8to16sub2_ia64;
611              transfer_16to8add = transfer_16to8add_ia64;
612              transfer8x8_copy = transfer8x8_copy_ia64;
613            }
614  #endif  #endif
615    
616          // API version  #if defined(ARCH_IS_PPC)
617          init_param->api_version = API_VERSION;          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
618              /* sad operators */
619                      sad16 = sad16_altivec_c;
620                      sad16bi = sad16bi_altivec_c;
621                      sad8 = sad8_altivec_c;
622                      dev16 = dev16_altivec_c;
623    
624              sse8_16bit = sse8_16bit_altivec_c;
625    
626              /* mem transfer */
627              transfer_8to16copy = transfer_8to16copy_altivec_c;
628              transfer_16to8copy = transfer_16to8copy_altivec_c;
629              transfer_8to16sub = transfer_8to16sub_altivec_c;
630              transfer_8to16subro = transfer_8to16subro_altivec_c;
631              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
632              transfer_16to8add = transfer_16to8add_altivec_c;
633              transfer8x8_copy = transfer8x8_copy_altivec_c;
634    
635              /* Inverse DCT */
636              idct = idct_altivec_c;
637    
638              /* Interpolation */
639              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
640              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
641              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
642    
643              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
644              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
645    
646                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
647                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
648                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
649                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
650    
651              /* Colorspace conversion */
652              bgra_to_yv12 = bgra_to_yv12_altivec_c;
653              abgr_to_yv12 = abgr_to_yv12_altivec_c;
654              rgba_to_yv12 = rgba_to_yv12_altivec_c;
655              argb_to_yv12 = argb_to_yv12_altivec_c;
656    
657              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
658              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
659    
660              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
661              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
662    
663              /* Quantization */
664              quant_h263_intra = quant_h263_intra_altivec_c;
665              quant_h263_inter = quant_h263_inter_altivec_c;
666              dequant_h263_intra = dequant_h263_intra_altivec_c;
667              dequant_h263_inter = dequant_h263_inter_altivec_c;
668    
669                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
670                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
671    
672                      /* Qpel stuff */
673                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
674                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
675            }
676    #endif
677    
678          // something clever has to be done for this  #if defined(_DEBUG)
679          init_param->core_build = 1000;      xvid_debug = init->debug;
680    #endif
681    
682          return XVID_ERR_OK;      return(0);
683  }  }
684    
685  int xvid_decore(void * handle, int opt, void * param1, void * param2)  
686    static int
687    xvid_gbl_info(xvid_gbl_info_t * info)
688    {
689            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
690                    return XVID_ERR_VERSION;
691    
692            info->actual_version = XVID_VERSION;
693            info->build = "xvid-1.3.0-dev";
694            info->cpu_flags = detect_cpu_flags();
695            info->num_threads = 0; /* single-thread */
696    
697    #if defined(_WIN32)
698    
699      {
700            SYSTEM_INFO siSysInfo;
701            GetSystemInfo(&siSysInfo);
702            info->num_threads = siSysInfo.dwNumberOfProcessors; /* number of _logical_ cores */
703      }
704    
705    #elif defined(_SC_NPROCESSORS_CONF) /* should be available on Apple too actually */
706    
707      info->num_threads = sysconf(_SC_NPROCESSORS_CONF);
708    
709    #elif defined(__APPLE__) && defined(__MACH__)
710    
711      {
712        size_t len;
713        int    mib[2], ncpu;
714    
715        mib[0] = CTL_HW;
716        mib[1] = HW_NCPU;
717        len    = sizeof(ncpu);
718        if (sysctl(mib, 2, &ncpu, &len, NULL, 0) == 0)
719          info -> num_threads = ncpu;
720        else
721          info -> num_threads = 1;
722      }
723    
724    #elif defined(__amigaos4__)
725    
726      {
727        uint32_t num_threads = 1;
728        IExec->GetCPUInfoTags(GCIT_NumberOfCPUs, &num_threads, TAG_END);
729        info->num_threads = num_threads;
730      }
731    
732    #endif
733    
734            return 0;
735    }
736    
737    
738    static int
739    xvid_gbl_convert(xvid_gbl_convert_t* convert)
740    {
741            int width;
742            int height;
743            int width2;
744            int height2;
745            IMAGE img;
746    
747            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
748                  return XVID_ERR_VERSION;
749    
750    #if 0
751            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
752    #endif
753            width = convert->width;
754            height = convert->height;
755            width2 = convert->width/2;
756            height2 = convert->height/2;
757    
758            switch (convert->input.csp & ~XVID_CSP_VFLIP)
759            {
760                    case XVID_CSP_YV12 :
761                            img.y = convert->input.plane[0];
762                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
763                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
764                            image_output(&img, width, height, width,
765                                                    (uint8_t**)convert->output.plane, convert->output.stride,
766                                                    convert->output.csp, convert->interlacing);
767                            break;
768    
769                    default :
770                            return XVID_ERR_FORMAT;
771            }
772    
773    
774            emms();
775            return 0;
776    }
777    
778    /*****************************************************************************
779     * XviD Global Entry point
780     *
781     * Well this function initialize all internal function pointers according
782     * to the CPU features forced by the library client or autodetected (depending
783     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
784     * image colorspace transformation tables.
785     *
786     ****************************************************************************/
787    
788    
789    int
790    xvid_global(void *handle,
791                      int opt,
792                      void *param1,
793                      void *param2)
794  {  {
795          switch (opt)          switch (opt)
796          {          {
797          case XVID_DEC_DECODE :                  case XVID_GBL_INIT :
798          return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);                          return xvid_gbl_init((xvid_gbl_init_t*)param1);
799    
800            case XVID_GBL_INFO :
801                return xvid_gbl_info((xvid_gbl_info_t*)param1);
802    
803                    case XVID_GBL_CONVERT :
804                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
805    
806                    default :
807                            return XVID_ERR_FAIL;
808            }
809    }
810    
811    /*****************************************************************************
812     * XviD Native decoder entry point
813     *
814     * This function is just a wrapper to all the option cases.
815     *
816     * Returned values : XVID_ERR_FAIL when opt is invalid
817     *                   else returns the wrapped function result
818     *
819     ****************************************************************************/
820    
821    int
822    xvid_decore(void *handle,
823                            int opt,
824                            void *param1,
825                            void *param2)
826    {
827            switch (opt) {
828          case XVID_DEC_CREATE :          case XVID_DEC_CREATE :
829          return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
830    
831          case XVID_DEC_DESTROY :          case XVID_DEC_DESTROY :
832          return decoder_destroy((DECODER *) handle);          return decoder_destroy((DECODER *) handle);
833    
834            case XVID_DEC_DECODE:
835                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
836    
837          default:          default:
838          return XVID_ERR_FAIL;          return XVID_ERR_FAIL;
839      }      }
840  }  }
841    
842    
843  int xvid_encore(void * handle, int opt, void * param1, void * param2)  /*****************************************************************************
844  {   * XviD Native encoder entry point
845          switch (opt)   *
846     * This function is just a wrapper to all the option cases.
847     *
848     * Returned values : XVID_ERR_FAIL when opt is invalid
849     *                   else returns the wrapped function result
850     *
851     ****************************************************************************/
852    
853    int
854    xvid_encore(void *handle,
855                            int opt,
856                            void *param1,
857                            void *param2)
858          {          {
859            switch (opt) {
860          case XVID_ENC_ENCODE :          case XVID_ENC_ENCODE :
861          return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, (XVID_ENC_STATS *) param2);  
862                    return enc_encode((Encoder *) handle,
863                                                              (xvid_enc_frame_t *) param1,
864                                                              (xvid_enc_stats_t *) param2);
865    
866          case XVID_ENC_CREATE :          case XVID_ENC_CREATE :
867          return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
868    
869          case XVID_ENC_DESTROY :          case XVID_ENC_DESTROY :
870          return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
871    
872          default:          default:
873          return XVID_ERR_FAIL;          return XVID_ERR_FAIL;

Legend:
Removed from v.1.3  
changed lines
  Added in v.1.82

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