[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 1.17, Thu Jun 13 21:35:01 2002 UTC revision 1.81, Tue Jun 2 13:06:49 2009 UTC
# Line 3  Line 3 
3   *  XVID MPEG-4 VIDEO CODEC   *  XVID MPEG-4 VIDEO CODEC
4   *  - Native API implementation  -   *  - Native API implementation  -
5   *   *
6   *  This program is an implementation of a part of one or more MPEG-4   *  Copyright(C) 2001-2004 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 26  Line 19 
19   *  along with this program ; if not, write to the Free Software   *  along with this program ; if not, write to the Free Software
20   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
21   *   *
  ****************************************************************************/  
 /*****************************************************************************  
  *  
  *  History  
  *  
  *  - 17.03.2002        Added interpolate8x8_halfpel_hv_xmm  
  *  - 22.12.2001  API change: added xvid_init() - Isibaar  
  *  - 16.12.2001        inital version; (c)2001 peter ross <pross@cs.rmit.edu.au>  
  *  
22   *  $Id$   *  $Id$
23   *   *
24   ****************************************************************************/   ****************************************************************************/
25    
26    #include <stdio.h>
27    #include <stdlib.h>
28    #include <string.h>
29    #include <time.h>
30    
31    #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  #include "xvid.h"  #include "xvid.h"
47  #include "decoder.h"  #include "decoder.h"
48  #include "encoder.h"  #include "encoder.h"
# Line 48  Line 52 
52  #include "image/colorspace.h"  #include "image/colorspace.h"
53  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
54  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
55  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
56  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
57    #include "motion/motion.h"
58    #include "motion/gmc.h"
59  #include "motion/sad.h"  #include "motion/sad.h"
60  #include "utils/emms.h"  #include "utils/emms.h"
61  #include "utils/timer.h"  #include "utils/timer.h"
62  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
63    #include "image/qpel.h"
64    #include "image/postprocessing.h"
65    
66    #if defined(_DEBUG)
67    unsigned int xvid_debug = 0; /* xvid debug mask */
68    #endif
69    
70    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
71    #       include <windows.h>
72    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
73    #       include <signal.h>
74    #       include <setjmp.h>
75    
76            static jmp_buf mark;
77    
78            static void
79            sigill_handler(int signal)
80            {
81               longjmp(mark, 1);
82            }
83    #endif
84    
85    
86    /*
87     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
88     * signalled
89     *
90     * Return values:
91     *  -1 : could not determine
92     *   0 : SIGILL was *not* signalled
93     *   1 : SIGILL was signalled
94     */
95    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
96    static int
97    sigill_check(void (*func)())
98    {
99            _try {
100                    func();
101            } _except(EXCEPTION_EXECUTE_HANDLER) {
102    
103                    if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
104                            return(1);
105            }
106            return(0);
107    }
108    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
109    static int
110    sigill_check(void (*func)())
111    {
112        void *old_handler;
113        int jmpret;
114    
115        /* Set our SIGILL handler */
116        old_handler = signal(SIGILL, sigill_handler);
117    
118        /* Check for error */
119        if (old_handler == SIG_ERR) {
120            return(-1);
121        }
122    
123        /* Save stack context, so if func triggers a SIGILL, we can still roll
124             * back to a valid CPU state */
125            jmpret = setjmp(mark);
126    
127            /* If setjmp returned directly, then its returned value is 0, and we still
128             * have to test the passed func. Otherwise it means the stack context has
129             * been restored by a longjmp() call, which in our case happens only in the
130             * signal handler */
131        if (jmpret == 0) {
132            func();
133        }
134    
135        /* Restore old signal handler */
136        signal(SIGILL, old_handler);
137    
138        return(jmpret);
139    }
140    #endif
141    
142    
143    /* detect cpu flags  */
144    static unsigned int
145    detect_cpu_flags(void)
146    {
147            /* enable native assembly optimizations by default */
148            unsigned int cpu_flags = XVID_CPU_ASM;
149    
150    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
151            cpu_flags |= check_cpu_features();
152            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
153                    cpu_flags &= ~XVID_CPU_SSE;
154    
155            if ((cpu_flags & (XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41)) && sigill_check(sse2_os_trigger))
156                    cpu_flags &= ~(XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41);
157    #endif
158    
159    #if defined(ARCH_IS_PPC)
160            if (!sigill_check(altivec_trigger))
161                    cpu_flags |= XVID_CPU_ALTIVEC;
162    #endif
163    
164            return cpu_flags;
165    }
166    
167    
168  /*****************************************************************************  /*****************************************************************************
169   * XviD Init Entry point   * XviD Init Entry point
# Line 69  Line 179 
179   *   *
180   ****************************************************************************/   ****************************************************************************/
181    
182  int  
183  xvid_init(void *handle,  static
184                    int opt,  int xvid_gbl_init(xvid_gbl_init_t * init)
                   void *param1,  
                   void *param2)  
185  {  {
186          int cpu_flags;          unsigned int cpu_flags;
         XVID_INIT_PARAM *init_param;  
187    
188          init_param = (XVID_INIT_PARAM *) param1;          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
189                    return XVID_ERR_VERSION;
190    
191          /* Do we have to force CPU features  ? */          cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
         if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0) {  
                 cpu_flags = init_param->cpu_flags;  
         } else {  
   
 #ifdef ARCH_X86  
                 cpu_flags = check_cpu_features();  
 #else  
                 cpu_flags = 0;  
 #endif  
                 init_param->cpu_flags = cpu_flags;  
         }  
192    
193          /* Initialize the function pointers */          /* Initialize the function pointers */
194          idct_int32_init();          idct_int32_init();
# Line 102  Line 199 
199          idct = idct_int32;          idct = idct_int32;
200    
201          /* Only needed on PPC Altivec archs */          /* Only needed on PPC Altivec archs */
202          sadInit = 0;          sadInit = NULL;
203    
204          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
205          emms = emms_c;          emms = emms_c;
206    
207            /* Qpel stuff */
208            xvid_QP_Funcs = &xvid_QP_Funcs_C;
209            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
210            xvid_Init_QP();
211    
212          /* Quantization functions */          /* Quantization functions */
213          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
214          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
215          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
216          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
217    
218          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
219          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
220          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
221          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
222    
223          /* Block transfer related functions */          /* Block transfer related functions */
224          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
225          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
226          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
227            transfer_8to16subro  = transfer_8to16subro_c;
228          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
229            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
230          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
231          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
232            transfer8x4_copy   = transfer8x4_copy_c;
233    
234            /* Interlacing functions */
235            MBFieldTest = MBFieldTest_c;
236    
237          /* Image interpolation related functions */          /* Image interpolation related functions */
238          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;
239          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
240          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
241    
242            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
243            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
244            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
245    
246            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
247            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
248            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
249            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
250    
251            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
252            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
253            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
254    
255            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
256            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
257            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
258    
259            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
260            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
261    
262            interpolate8x8_avg2 = interpolate8x8_avg2_c;
263            interpolate8x8_avg4 = interpolate8x8_avg4_c;
264    
265            /* postprocessing */
266            image_brightness = image_brightness_c;
267    
268          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
269          colorspace_init();          colorspace_init();
270    
271          /* All colorspace transformation functions User Format->YV12 */          /* All colorspace transformation functions User Format->YV12 */
272            yv12_to_yv12    = yv12_to_yv12_c;
273          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
274          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
275          rgb24_to_yv12  = rgb24_to_yv12_c;          rgb_to_yv12     = rgb_to_yv12_c;
276          rgb32_to_yv12  = rgb32_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
277          yuv_to_yv12    = yuv_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
278            abgr_to_yv12    = abgr_to_yv12_c;
279            rgba_to_yv12    = rgba_to_yv12_c;
280            argb_to_yv12    = argb_to_yv12_c;
281          yuyv_to_yv12   = yuyv_to_yv12_c;          yuyv_to_yv12   = yuyv_to_yv12_c;
282          uyvy_to_yv12   = uyvy_to_yv12_c;          uyvy_to_yv12   = uyvy_to_yv12_c;
283    
284            rgb555i_to_yv12 = rgb555i_to_yv12_c;
285            rgb565i_to_yv12 = rgb565i_to_yv12_c;
286            bgri_to_yv12    = bgri_to_yv12_c;
287            bgrai_to_yv12   = bgrai_to_yv12_c;
288            abgri_to_yv12   = abgri_to_yv12_c;
289            rgbai_to_yv12   = rgbai_to_yv12_c;
290            argbi_to_yv12   = argbi_to_yv12_c;
291            yuyvi_to_yv12   = yuyvi_to_yv12_c;
292            uyvyi_to_yv12   = uyvyi_to_yv12_c;
293    
294          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
295          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
296          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
297          yv12_to_rgb24  = yv12_to_rgb24_c;          yv12_to_rgb     = yv12_to_rgb_c;
298          yv12_to_rgb32  = yv12_to_rgb32_c;          yv12_to_bgr     = yv12_to_bgr_c;
299          yv12_to_yuv    = yv12_to_yuv_c;          yv12_to_bgra    = yv12_to_bgra_c;
300            yv12_to_abgr    = yv12_to_abgr_c;
301            yv12_to_rgba    = yv12_to_rgba_c;
302            yv12_to_argb    = yv12_to_argb_c;
303          yv12_to_yuyv   = yv12_to_yuyv_c;          yv12_to_yuyv   = yv12_to_yuyv_c;
304          yv12_to_uyvy   = yv12_to_uyvy_c;          yv12_to_uyvy   = yv12_to_uyvy_c;
305    
306            yv12_to_rgb555i = yv12_to_rgb555i_c;
307            yv12_to_rgb565i = yv12_to_rgb565i_c;
308            yv12_to_bgri    = yv12_to_bgri_c;
309            yv12_to_bgrai   = yv12_to_bgrai_c;
310            yv12_to_abgri   = yv12_to_abgri_c;
311            yv12_to_rgbai   = yv12_to_rgbai_c;
312            yv12_to_argbi   = yv12_to_argbi_c;
313            yv12_to_yuyvi   = yv12_to_yuyvi_c;
314            yv12_to_uyvyi   = yv12_to_uyvyi_c;
315    
316          /* Functions used in motion estimation algorithms */          /* Functions used in motion estimation algorithms */
317          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
318          sad16    = sad16_c;          sad16    = sad16_c;
         sad16bi  = sad16bi_c;  
319          sad8     = sad8_c;          sad8     = sad8_c;
320            sad16bi    = sad16bi_c;
321            sad8bi     = sad8bi_c;
322          dev16    = dev16_c;          dev16    = dev16_c;
323            sad16v     = sad16v_c;
324            sse8_16bit = sse8_16bit_c;
325            sse8_8bit  = sse8_8bit_c;
326    
327            init_GMC(cpu_flags);
328    
329    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
330    
331            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
332                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
333                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2) ||
334            (cpu_flags & XVID_CPU_SSE3) || (cpu_flags & XVID_CPU_SSE41))
335            {
336                    /* Restore FPU context : emms_c is a nop functions */
337                    emms = emms_mmx;
338            }
339    
340  #ifdef ARCH_X86          if ((cpu_flags & XVID_CPU_MMX)) {
         if ((cpu_flags & XVID_CPU_MMX) > 0) {  
341    
342                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
343                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
344                  idct = idct_mmx;                  idct = idct_mmx;
345    
346                  /* To restore FPU context after mmx use */                  /* Qpel stuff */
347                  emms = emms_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
348                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
349    
350                  /* Quantization related functions */                  /* Quantization related functions */
351                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
352                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
353                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
354                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
355                    quant_mpeg_intra   = quant_mpeg_intra_mmx;
356                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
357                  dequant4_intra = dequant4_intra_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
358                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
359                  dequant4_inter = dequant4_inter_mmx;  
360    
361                  /* Block related functions */                  /* Block related functions */
362                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
363                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
364                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
365                    transfer_8to16subro  = transfer_8to16subro_mmx;
366                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
367                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
368                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
369                    transfer8x4_copy   = transfer8x4_copy_mmx;
370    
371                    /* Interlacing Functions */
372                    MBFieldTest = MBFieldTest_mmx;
373    
374                  /* Image Interpolation related functions */                  /* Image Interpolation related functions */
375                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;
376                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
377                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
378    
379                  /* Image RGB->YV12 related functions */                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
380                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
381                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
382                  yuv_to_yv12   = yuv_to_yv12_mmx;  
383                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
384                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
385                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
386                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
387    
388                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
389                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
390    
391                    interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
392                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
393    
394                    /* postprocessing */
395                    image_brightness = image_brightness_mmx;
396    
397                    /* image input xxx_to_yv12 related functions */
398    
399                    yv12_to_yv12  = yv12_to_yv12_mmx;
400    
401                    bgr_to_yv12   = bgr_to_yv12_mmx;
402                    rgb_to_yv12   = rgb_to_yv12_mmx;
403                    bgra_to_yv12  = bgra_to_yv12_mmx;
404                    rgba_to_yv12  = rgba_to_yv12_mmx;
405                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
406                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
407    
408                  /* Image YV12->RGB related functions */                  /* image output yv12_to_xxx related functions */
409                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
410                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgra  = yv12_to_bgra_mmx;
411                  yv12_to_yuyv  = yv12_to_yuyv_mmx;                  yv12_to_yuyv  = yv12_to_yuyv_mmx;
412                  yv12_to_uyvy  = yv12_to_uyvy_mmx;                  yv12_to_uyvy  = yv12_to_uyvy_mmx;
413    
414                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
415                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
416    
417                  /* Motion estimation related functions */                  /* Motion estimation related functions */
418                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
419                  sad16    = sad16_mmx;                  sad16    = sad16_mmx;
420                  sad8     = sad8_mmx;                  sad8     = sad8_mmx;
421                    sad16bi    = sad16bi_mmx;
422                    sad8bi     = sad8bi_mmx;
423                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
424                    sad16v     = sad16v_mmx;
425                    sse8_16bit = sse8_16bit_mmx;
426                    sse8_8bit  = sse8_8bit_mmx;
427            }
428    
429            /* these 3dnow functions are faster than mmx, but slower than xmm. */
430            if ((cpu_flags & XVID_CPU_3DNOW)) {
431    
432                    emms = emms_3dn;
433    
434                    /* ME functions */
435                    sad16bi = sad16bi_3dn;
436                    sad8bi  = sad8bi_3dn;
437    
438                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
439                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
440    
441          }          }
442    
         if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
443    
444                  /* Inverse DCT */          if ((cpu_flags & XVID_CPU_MMXEXT)) {
445    
446                    /* DCT */
447                    fdct = fdct_xmm_skal;
448                  idct = idct_xmm;                  idct = idct_xmm;
449    
450                  /* Interpolation */                  /* Interpolation */
# Line 223  Line 452 
452                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
453                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
454    
455                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
456                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
457                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
458    
459                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
460                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
461                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
462                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
463    
464            /* Quantization */
465                    quant_mpeg_inter = quant_mpeg_inter_xmm;
466    
467                    dequant_h263_intra = dequant_h263_intra_xmm;
468                    dequant_h263_inter = dequant_h263_inter_xmm;
469    
470            /* Buffer transfer */
471                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
472                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
473    
474                  /* Colorspace transformation */                  /* Colorspace transformation */
475                  yuv_to_yv12 = yuv_to_yv12_xmm;                  /* yv12_to_yv12  = yv12_to_yv12_xmm; */ /* appears to be slow on many machines */
476                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
477                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
478    
479                  /* ME functions */                  /* ME functions */
480                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
481                  sad8  = sad8_xmm;                  sad8  = sad8_xmm;
482                    sad16bi = sad16bi_xmm;
483                    sad8bi  = sad8bi_xmm;
484                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
485                    sad16v   = sad16v_xmm;
486          }          }
487    
488          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
489    
490                  /* Interpolation */                  /* Interpolation */
491                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
492                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
493                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
494    
495                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
496                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
497                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
498          }          }
499    
500          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
501  #ifdef EXPERIMENTAL_SSE2_CODE  
502                    /* Buffer transfer */
503                    transfer_8to16copy =  transfer_8to16copy_3dne;
504                    transfer_16to8copy = transfer_16to8copy_3dne;
505                    transfer_8to16sub =  transfer_8to16sub_3dne;
506                    transfer_8to16subro =  transfer_8to16subro_3dne;
507                    transfer_16to8add = transfer_16to8add_3dne;
508                    transfer8x8_copy = transfer8x8_copy_3dne;
509                    transfer8x4_copy = transfer8x4_copy_3dne;
510    
511                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
512                            /* Inverse DCT */
513                            idct =  idct_3dne;
514    
515                            /* Buffer transfer */
516                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
517    
518                            /* Interpolation */
519                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
520                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
521                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
522    
523                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
524                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
525                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
526    
527                  /* Quantization */                  /* Quantization */
528                  quant_intra   = quant_intra_sse2;                          quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
529                  dequant_intra = dequant_intra_sse2;                          quant_h263_inter = quant_h263_inter_3dne;
530                  quant_inter   = quant_inter_sse2;                          dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
531                  dequant_inter = dequant_inter_sse2;                          dequant_mpeg_inter = dequant_mpeg_inter_3dne;
532                            dequant_h263_intra = dequant_h263_intra_3dne;
533                            dequant_h263_inter = dequant_h263_inter_3dne;
534    
535                /* ME functions */
536                            sad16 = sad16_3dne;
537                            sad8 = sad8_3dne;
538                            sad16bi = sad16bi_3dne;
539                            sad8bi = sad8bi_3dne;
540                            dev16 = dev16_3dne;
541                    }
542            }
543    
544            if ((cpu_flags & XVID_CPU_SSE2)) {
545    
                 /* ME */  
546                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
547    
548                    /* Quantization */
549                    quant_h263_intra   = quant_h263_intra_sse2;
550                    quant_h263_inter   = quant_h263_inter_sse2;
551                    dequant_h263_intra = dequant_h263_intra_sse2;
552                    dequant_h263_inter = dequant_h263_inter_sse2;
553    
554                    /* SAD operators */
555                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
556                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
557    
558                  /* Forward and Inverse DCT */                  /* DCT operators */
559                  idct  = idct_sse2;                  fdct = fdct_sse2_skal;
560                  fdct = fdct_sse2;                  idct = idct_sse2_skal;   /* Is now IEEE1180 and Walken compliant. */
561    
562                    /* postprocessing */
563                    image_brightness = image_brightness_sse2;
564    
565            }
566    
567            if ((cpu_flags & XVID_CPU_SSE3)) {
568    
569                    /* SAD operators */
570                    sad16    = sad16_sse3;
571                    dev16    = dev16_sse3;
572            }
573    
574    #endif /* ARCH_IS_IA32 */
575    
576    #if defined(ARCH_IS_IA64)
577            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
578              idct_ia64_init();
579              fdct = fdct_ia64;
580              idct = idct_ia64;   /*not yet working, crashes */
581              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
582              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
583              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
584              sad16 = sad16_ia64;
585              sad16bi = sad16bi_ia64;
586              sad8 = sad8_ia64;
587              dev16 = dev16_ia64;
588    /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
589              quant_h263_intra = quant_h263_intra_ia64;
590              quant_h263_inter = quant_h263_inter_ia64;
591              dequant_h263_intra = dequant_h263_intra_ia64;
592              dequant_h263_inter = dequant_h263_inter_ia64;
593              transfer_8to16copy = transfer_8to16copy_ia64;
594              transfer_16to8copy = transfer_16to8copy_ia64;
595              transfer_8to16sub = transfer_8to16sub_ia64;
596              transfer_8to16sub2 = transfer_8to16sub2_ia64;
597              transfer_16to8add = transfer_16to8add_ia64;
598              transfer8x8_copy = transfer8x8_copy_ia64;
599            }
600  #endif  #endif
601    
602    #if defined(ARCH_IS_PPC)
603            if ((cpu_flags & XVID_CPU_ALTIVEC)) {
604              /* sad operators */
605                      sad16 = sad16_altivec_c;
606                      sad16bi = sad16bi_altivec_c;
607                      sad8 = sad8_altivec_c;
608                      dev16 = dev16_altivec_c;
609    
610              sse8_16bit = sse8_16bit_altivec_c;
611    
612              /* mem transfer */
613              transfer_8to16copy = transfer_8to16copy_altivec_c;
614              transfer_16to8copy = transfer_16to8copy_altivec_c;
615              transfer_8to16sub = transfer_8to16sub_altivec_c;
616              transfer_8to16subro = transfer_8to16subro_altivec_c;
617              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
618              transfer_16to8add = transfer_16to8add_altivec_c;
619              transfer8x8_copy = transfer8x8_copy_altivec_c;
620    
621              /* Inverse DCT */
622              idct = idct_altivec_c;
623    
624              /* Interpolation */
625              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
626              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
627              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
628    
629              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
630              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
631    
632                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
633                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
634                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
635                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
636    
637              /* Colorspace conversion */
638              bgra_to_yv12 = bgra_to_yv12_altivec_c;
639              abgr_to_yv12 = abgr_to_yv12_altivec_c;
640              rgba_to_yv12 = rgba_to_yv12_altivec_c;
641              argb_to_yv12 = argb_to_yv12_altivec_c;
642    
643              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
644              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
645    
646              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
647              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
648    
649              /* Quantization */
650              quant_h263_intra = quant_h263_intra_altivec_c;
651              quant_h263_inter = quant_h263_inter_altivec_c;
652              dequant_h263_intra = dequant_h263_intra_altivec_c;
653              dequant_h263_inter = dequant_h263_inter_altivec_c;
654    
655                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
656                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
657    
658                      /* Qpel stuff */
659                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
660                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
661          }          }
662    #endif
663    
664    #if defined(_DEBUG)
665        xvid_debug = init->debug;
666  #endif  #endif
667    
668  #ifdef ARCH_PPC      return(0);
669  #ifdef ARCH_PPC_ALTIVEC  }
670          calc_cbp = calc_cbp_altivec;  
671          fdct = fdct_altivec;  
672          idct = idct_altivec;  static int
673          sadInit = sadInit_altivec;  xvid_gbl_info(xvid_gbl_info_t * info)
674          sad16 = sad16_altivec;  {
675          sad8 = sad8_altivec;          if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
676          dev16 = dev16_altivec;                  return XVID_ERR_VERSION;
677  #else  
678          calc_cbp = calc_cbp_ppc;          info->actual_version = XVID_VERSION;
679            info->build = "xvid-1.3.0-dev";
680            info->cpu_flags = detect_cpu_flags();
681            info->num_threads = 0; /* single-thread */
682    
683    #if defined(_WIN32)
684    
685      {
686            SYSTEM_INFO siSysInfo;
687            GetSystemInfo(&siSysInfo);
688            info->num_threads = siSysInfo.dwNumberOfProcessors; /* number of _logical_ cores */
689      }
690    
691    #elif defined(_SC_NPROCESSORS_CONF) /* should be available on Apple too actually */
692    
693      info->num_threads = sysconf(_SC_NPROCESSORS_CONF);
694    
695    #elif defined(__APPLE__) && defined(__MACH__)
696    
697      {
698        size_t len;
699        int    mib[2], ncpu;
700    
701        mib[0] = CTL_HW;
702        mib[1] = HW_NCPU;
703        len    = sizeof(ncpu);
704        if (sysctl(mib, 2, &ncpu, &len, NULL, 0) == 0)
705          info -> num_threads = ncpu;
706        else
707          info -> num_threads = 1;
708      }
709    
710  #endif  #endif
711    
712            return 0;
713    }
714    
715    
716    static int
717    xvid_gbl_convert(xvid_gbl_convert_t* convert)
718    {
719            int width;
720            int height;
721            int width2;
722            int height2;
723            IMAGE img;
724    
725            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
726                  return XVID_ERR_VERSION;
727    
728    #if 0
729            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
730  #endif  #endif
731            width = convert->width;
732            height = convert->height;
733            width2 = convert->width/2;
734            height2 = convert->height/2;
735    
736            switch (convert->input.csp & ~XVID_CSP_VFLIP)
737            {
738                    case XVID_CSP_YV12 :
739                            img.y = convert->input.plane[0];
740                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
741                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
742                            image_output(&img, width, height, width,
743                                                    (uint8_t**)convert->output.plane, convert->output.stride,
744                                                    convert->output.csp, convert->interlacing);
745                            break;
746    
747                    default :
748                            return XVID_ERR_FORMAT;
749            }
750    
751    
752            emms();
753            return 0;
754    }
755    
756    /*****************************************************************************
757     * XviD Global Entry point
758     *
759     * Well this function initialize all internal function pointers according
760     * to the CPU features forced by the library client or autodetected (depending
761     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
762     * image colorspace transformation tables.
763     *
764     ****************************************************************************/
765    
         /* Inform the client the API version */  
         init_param->api_version = API_VERSION;  
766    
767          /* Inform the client the core build - unused because we're still alpha */  int
768          init_param->core_build = 1000;  xvid_global(void *handle,
769                      int opt,
770                      void *param1,
771                      void *param2)
772    {
773            switch(opt)
774            {
775                    case XVID_GBL_INIT :
776                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
777    
778            case XVID_GBL_INFO :
779                return xvid_gbl_info((xvid_gbl_info_t*)param1);
780    
781                    case XVID_GBL_CONVERT :
782                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
783    
784          return XVID_ERR_OK;                  default :
785                            return XVID_ERR_FAIL;
786            }
787  }  }
788    
789  /*****************************************************************************  /*****************************************************************************
# Line 303  Line 803 
803                          void *param2)                          void *param2)
804  {  {
805          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
806          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
807                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
808    
809          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
810                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
811    
812            case XVID_DEC_DECODE:
813                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
814    
815          default:          default:
816                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
817          }          }
# Line 336  Line 836 
836  {  {
837          switch (opt) {          switch (opt) {
838          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
839                  return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
840                                                            (XVID_ENC_STATS *) param2);                  return enc_encode((Encoder *) handle,
841                                                              (xvid_enc_frame_t *) param1,
842                                                              (xvid_enc_stats_t *) param2);
843    
844          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
845                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
846    
847          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
848                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
849    
850          default:          default:
851                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.17  
changed lines
  Added in v.1.81

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