[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 1.30, Tue Jul 16 11:15:15 2002 UTC revision 1.80, Mon Dec 1 15:06:48 2008 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  
  *  
  *      - 23.06.2002    added XVID_CPU_CHKONLY  
  *  - 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  #include "xvid.h"  #include "xvid.h"
36  #include "decoder.h"  #include "decoder.h"
37  #include "encoder.h"  #include "encoder.h"
# Line 50  Line 41 
41  #include "image/colorspace.h"  #include "image/colorspace.h"
42  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
43  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
44  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
45  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
46  #include "motion/motion.h"  #include "motion/motion.h"
47    #include "motion/gmc.h"
48  #include "motion/sad.h"  #include "motion/sad.h"
49  #include "utils/emms.h"  #include "utils/emms.h"
50  #include "utils/timer.h"  #include "utils/timer.h"
51  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
52    #include "image/qpel.h"
53    #include "image/postprocessing.h"
54    
55    #if defined(_DEBUG)
56    unsigned int xvid_debug = 0; /* xvid debug mask */
57    #endif
58    
59    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
60    #       include <windows.h>
61    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
62    #       include <signal.h>
63    #       include <setjmp.h>
64    
65            static jmp_buf mark;
66    
67            static void
68            sigill_handler(int signal)
69            {
70               longjmp(mark, 1);
71            }
72    #endif
73    
74    
75    /*
76     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
77     * signalled
78     *
79     * Return values:
80     *  -1 : could not determine
81     *   0 : SIGILL was *not* signalled
82     *   1 : SIGILL was signalled
83     */
84    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
85    static int
86    sigill_check(void (*func)())
87    {
88            _try {
89                    func();
90            } _except(EXCEPTION_EXECUTE_HANDLER) {
91    
92                    if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
93                            return(1);
94            }
95            return(0);
96    }
97    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
98    static int
99    sigill_check(void (*func)())
100    {
101        void *old_handler;
102        int jmpret;
103    
104        /* Set our SIGILL handler */
105        old_handler = signal(SIGILL, sigill_handler);
106    
107        /* Check for error */
108        if (old_handler == SIG_ERR) {
109            return(-1);
110        }
111    
112        /* Save stack context, so if func triggers a SIGILL, we can still roll
113             * back to a valid CPU state */
114            jmpret = setjmp(mark);
115    
116            /* If setjmp returned directly, then its returned value is 0, and we still
117             * have to test the passed func. Otherwise it means the stack context has
118             * been restored by a longjmp() call, which in our case happens only in the
119             * signal handler */
120        if (jmpret == 0) {
121            func();
122        }
123    
124        /* Restore old signal handler */
125        signal(SIGILL, old_handler);
126    
127        return(jmpret);
128    }
129    #endif
130    
131    
132    /* detect cpu flags  */
133    static unsigned int
134    detect_cpu_flags(void)
135    {
136            /* enable native assembly optimizations by default */
137            unsigned int cpu_flags = XVID_CPU_ASM;
138    
139    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
140            cpu_flags |= check_cpu_features();
141            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
142                    cpu_flags &= ~XVID_CPU_SSE;
143    
144            if ((cpu_flags & (XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41)) && sigill_check(sse2_os_trigger))
145                    cpu_flags &= ~(XVID_CPU_SSE2|XVID_CPU_SSE3|XVID_CPU_SSE41);
146    #endif
147    
148    #if defined(ARCH_IS_PPC)
149            if (!sigill_check(altivec_trigger))
150                    cpu_flags |= XVID_CPU_ALTIVEC;
151    #endif
152    
153            return cpu_flags;
154    }
155    
156    
157  /*****************************************************************************  /*****************************************************************************
158   * XviD Init Entry point   * XviD Init Entry point
# Line 72  Line 168 
168   *   *
169   ****************************************************************************/   ****************************************************************************/
170    
 int  
 xvid_init(void *handle,  
                   int opt,  
                   void *param1,  
                   void *param2)  
 {  
         int cpu_flags;  
         XVID_INIT_PARAM *init_param;  
   
         init_param = (XVID_INIT_PARAM *) param1;  
171    
172          /* Inform the client the API version */  static
173          init_param->api_version = API_VERSION;  int xvid_gbl_init(xvid_gbl_init_t * init)
   
         /* Inform the client the core build - unused because we're still alpha */  
         init_param->core_build = 1000;  
   
         if ((init_param->cpu_flags & XVID_CPU_CHKONLY))  
174          {          {
175                  init_param->cpu_flags = check_cpu_features();          unsigned int cpu_flags;
                 return XVID_ERR_OK;  
         }  
176    
177          /* Do we have to force CPU features  ? */          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
178          if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0) {                  return XVID_ERR_VERSION;
                 cpu_flags = init_param->cpu_flags;  
         } else {  
179    
180                  cpu_flags = check_cpu_features();          cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
                 init_param->cpu_flags = cpu_flags;  
         }  
181    
182          /* Initialize the function pointers */          /* Initialize the function pointers */
183          idct_int32_init();          idct_int32_init();
# Line 113  Line 188 
188          idct = idct_int32;          idct = idct_int32;
189    
190          /* Only needed on PPC Altivec archs */          /* Only needed on PPC Altivec archs */
191          sadInit = 0;          sadInit = NULL;
192    
193          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
194          emms = emms_c;          emms = emms_c;
195    
196            /* Qpel stuff */
197            xvid_QP_Funcs = &xvid_QP_Funcs_C;
198            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
199            xvid_Init_QP();
200    
201          /* Quantization functions */          /* Quantization functions */
202          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
203          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
204          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
205          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
206    
207          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
208          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
209          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
210          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
211    
212          /* Block transfer related functions */          /* Block transfer related functions */
213          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
214          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
215          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
216            transfer_8to16subro  = transfer_8to16subro_c;
217          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
218            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
219          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
220          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
221            transfer8x4_copy   = transfer8x4_copy_c;
222    
223            /* Interlacing functions */
224            MBFieldTest = MBFieldTest_c;
225    
226          /* Image interpolation related functions */          /* Image interpolation related functions */
227          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;
228          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
229          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
230    
231            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
232            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
233            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
234    
235            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
236            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
237            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
238            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
239    
240            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
241            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
242            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
243    
244            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
245            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
246            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
247    
248            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
249            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
250    
251            interpolate8x8_avg2 = interpolate8x8_avg2_c;
252            interpolate8x8_avg4 = interpolate8x8_avg4_c;
253    
254            /* postprocessing */
255            image_brightness = image_brightness_c;
256    
257          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
258          colorspace_init();          colorspace_init();
259    
260          /* All colorspace transformation functions User Format->YV12 */          /* All colorspace transformation functions User Format->YV12 */
261            yv12_to_yv12    = yv12_to_yv12_c;
262          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
263          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
264          rgb24_to_yv12  = rgb24_to_yv12_c;          rgb_to_yv12     = rgb_to_yv12_c;
265          rgb32_to_yv12  = rgb32_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
266          yuv_to_yv12    = yuv_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
267            abgr_to_yv12    = abgr_to_yv12_c;
268            rgba_to_yv12    = rgba_to_yv12_c;
269            argb_to_yv12    = argb_to_yv12_c;
270          yuyv_to_yv12   = yuyv_to_yv12_c;          yuyv_to_yv12   = yuyv_to_yv12_c;
271          uyvy_to_yv12   = uyvy_to_yv12_c;          uyvy_to_yv12   = uyvy_to_yv12_c;
272    
273            rgb555i_to_yv12 = rgb555i_to_yv12_c;
274            rgb565i_to_yv12 = rgb565i_to_yv12_c;
275            bgri_to_yv12    = bgri_to_yv12_c;
276            bgrai_to_yv12   = bgrai_to_yv12_c;
277            abgri_to_yv12   = abgri_to_yv12_c;
278            rgbai_to_yv12   = rgbai_to_yv12_c;
279            argbi_to_yv12   = argbi_to_yv12_c;
280            yuyvi_to_yv12   = yuyvi_to_yv12_c;
281            uyvyi_to_yv12   = uyvyi_to_yv12_c;
282    
283          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
284          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
285          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
286          yv12_to_rgb24  = yv12_to_rgb24_c;          yv12_to_rgb     = yv12_to_rgb_c;
287          yv12_to_rgb32  = yv12_to_rgb32_c;          yv12_to_bgr     = yv12_to_bgr_c;
288          yv12_to_yuv    = yv12_to_yuv_c;          yv12_to_bgra    = yv12_to_bgra_c;
289            yv12_to_abgr    = yv12_to_abgr_c;
290            yv12_to_rgba    = yv12_to_rgba_c;
291            yv12_to_argb    = yv12_to_argb_c;
292          yv12_to_yuyv   = yv12_to_yuyv_c;          yv12_to_yuyv   = yv12_to_yuyv_c;
293          yv12_to_uyvy   = yv12_to_uyvy_c;          yv12_to_uyvy   = yv12_to_uyvy_c;
294    
295            yv12_to_rgb555i = yv12_to_rgb555i_c;
296            yv12_to_rgb565i = yv12_to_rgb565i_c;
297            yv12_to_bgri    = yv12_to_bgri_c;
298            yv12_to_bgrai   = yv12_to_bgrai_c;
299            yv12_to_abgri   = yv12_to_abgri_c;
300            yv12_to_rgbai   = yv12_to_rgbai_c;
301            yv12_to_argbi   = yv12_to_argbi_c;
302            yv12_to_yuyvi   = yv12_to_yuyvi_c;
303            yv12_to_uyvyi   = yv12_to_uyvyi_c;
304    
305          /* Functions used in motion estimation algorithms */          /* Functions used in motion estimation algorithms */
306          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
307          sad16    = sad16_c;          sad16    = sad16_c;
         sad16bi  = sad16bi_c;  
308          sad8     = sad8_c;          sad8     = sad8_c;
309            sad16bi    = sad16bi_c;
310            sad8bi     = sad8bi_c;
311          dev16    = dev16_c;          dev16    = dev16_c;
312          Halfpel8_Refine = Halfpel8_Refine_c;          sad16v     = sad16v_c;
313            sse8_16bit = sse8_16bit_c;
314            sse8_8bit  = sse8_8bit_c;
315    
316            init_GMC(cpu_flags);
317    
318    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
319    
320            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
321                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
322                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2) ||
323            (cpu_flags & XVID_CPU_SSE3) || (cpu_flags & XVID_CPU_SSE41))
324            {
325                    /* Restore FPU context : emms_c is a nop functions */
326                    emms = emms_mmx;
327            }
328    
329  #ifdef ARCH_X86          if ((cpu_flags & XVID_CPU_MMX)) {
         if ((cpu_flags & XVID_CPU_MMX) > 0) {  
330    
331                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
332                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
333                  idct = idct_mmx;                  idct = idct_mmx;
334    
335                  /* To restore FPU context after mmx use */                  /* Qpel stuff */
336                  emms = emms_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
337                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
338    
339                  /* Quantization related functions */                  /* Quantization related functions */
340                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
341                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
342                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
343                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
344                    quant_mpeg_intra   = quant_mpeg_intra_mmx;
345                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
346                  dequant4_intra = dequant4_intra_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
347                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
348                  dequant4_inter = dequant4_inter_mmx;  
349    
350                  /* Block related functions */                  /* Block related functions */
351                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
352                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
353                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
354                    transfer_8to16subro  = transfer_8to16subro_mmx;
355                  transfer_8to16sub2 = transfer_8to16sub2_mmx;                  transfer_8to16sub2 = transfer_8to16sub2_mmx;
356                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
357                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
358                    transfer8x4_copy   = transfer8x4_copy_mmx;
359    
360                    /* Interlacing Functions */
361                    MBFieldTest = MBFieldTest_mmx;
362    
363                  /* Image Interpolation related functions */                  /* Image Interpolation related functions */
364                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;
365                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
366                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
367    
368                  /* Image RGB->YV12 related functions */                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
369                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
370                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
371                  yuv_to_yv12   = yuv_to_yv12_mmx;  
372                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
373                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
374                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
375                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
376    
377                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
378                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
379    
380                    interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
381                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
382    
383                    /* postprocessing */
384                    image_brightness = image_brightness_mmx;
385    
386                    /* image input xxx_to_yv12 related functions */
387    
388                    yv12_to_yv12  = yv12_to_yv12_mmx;
389    
390                    bgr_to_yv12   = bgr_to_yv12_mmx;
391                    rgb_to_yv12   = rgb_to_yv12_mmx;
392                    bgra_to_yv12  = bgra_to_yv12_mmx;
393                    rgba_to_yv12  = rgba_to_yv12_mmx;
394                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
395                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
396    
397                  /* Image YV12->RGB related functions */                  /* image output yv12_to_xxx related functions */
398                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
399                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgra  = yv12_to_bgra_mmx;
400                  yv12_to_yuyv  = yv12_to_yuyv_mmx;                  yv12_to_yuyv  = yv12_to_yuyv_mmx;
401                  yv12_to_uyvy  = yv12_to_uyvy_mmx;                  yv12_to_uyvy  = yv12_to_uyvy_mmx;
402    
403                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
404                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
405    
406                  /* Motion estimation related functions */                  /* Motion estimation related functions */
407                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
408                  sad16    = sad16_mmx;                  sad16    = sad16_mmx;
409                  sad8     = sad8_mmx;                  sad8     = sad8_mmx;
410                    sad16bi    = sad16bi_mmx;
411                    sad8bi     = sad8bi_mmx;
412                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
413                    sad16v     = sad16v_mmx;
414                    sse8_16bit = sse8_16bit_mmx;
415                    sse8_8bit  = sse8_8bit_mmx;
416            }
417    
418            /* these 3dnow functions are faster than mmx, but slower than xmm. */
419            if ((cpu_flags & XVID_CPU_3DNOW)) {
420    
421                    emms = emms_3dn;
422    
423                    /* ME functions */
424                    sad16bi = sad16bi_3dn;
425                    sad8bi  = sad8bi_3dn;
426    
427                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
428                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
429    
430          }          }
431    
         if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
432    
433                  /* Inverse DCT */          if ((cpu_flags & XVID_CPU_MMXEXT)) {
434    
435                    /* DCT */
436                    fdct = fdct_xmm_skal;
437                  idct = idct_xmm;                  idct = idct_xmm;
438    
439                  /* Interpolation */                  /* Interpolation */
# Line 237  Line 441 
441                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
442                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
443    
444                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
445                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
446                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
447    
448                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
449                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
450                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
451                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
452    
453                  /* Quantization */                  /* Quantization */
454                  dequant_intra = dequant_intra_xmm;                  quant_mpeg_inter = quant_mpeg_inter_xmm;
455                  dequant_inter = dequant_inter_xmm;  
456                    dequant_h263_intra = dequant_h263_intra_xmm;
457                    dequant_h263_inter = dequant_h263_inter_xmm;
458    
459                  /* Buffer transfer */                  /* Buffer transfer */
460                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
461                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
462    
463                  /* Colorspace transformation */                  /* Colorspace transformation */
464                  yuv_to_yv12 = yuv_to_yv12_xmm;                  /* yv12_to_yv12  = yv12_to_yv12_xmm; */ /* appears to be slow on many machines */
465                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
466                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
467    
468                  /* ME functions */                  /* ME functions */
469                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
                 sad16bi = sad16bi_xmm;  
470                  sad8  = sad8_xmm;                  sad8  = sad8_xmm;
471                    sad16bi = sad16bi_xmm;
472                    sad8bi  = sad8bi_xmm;
473                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
474                    sad16v   = sad16v_xmm;
475          }          }
476    
477          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
478    
479                  /* Interpolation */                  /* Interpolation */
480                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
481                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
482                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
483    
484                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
485                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
486                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
487          }          }
488    
489          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
490  #ifdef EXPERIMENTAL_SSE2_CODE  
491                    /* Buffer transfer */
492                    transfer_8to16copy =  transfer_8to16copy_3dne;
493                    transfer_16to8copy = transfer_16to8copy_3dne;
494                    transfer_8to16sub =  transfer_8to16sub_3dne;
495                    transfer_8to16subro =  transfer_8to16subro_3dne;
496                    transfer_16to8add = transfer_16to8add_3dne;
497                    transfer8x8_copy = transfer8x8_copy_3dne;
498                    transfer8x4_copy = transfer8x4_copy_3dne;
499    
500                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
501                            /* Inverse DCT */
502                            idct =  idct_3dne;
503    
504                            /* Buffer transfer */
505                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
506    
507                            /* Interpolation */
508                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
509                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
510                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
511    
512                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
513                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
514                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
515    
516                /* Quantization */
517                            quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
518                            quant_h263_inter = quant_h263_inter_3dne;
519                            dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
520                            dequant_mpeg_inter = dequant_mpeg_inter_3dne;
521                            dequant_h263_intra = dequant_h263_intra_3dne;
522                            dequant_h263_inter = dequant_h263_inter_3dne;
523    
524                /* ME functions */
525                            sad16 = sad16_3dne;
526                            sad8 = sad8_3dne;
527                            sad16bi = sad16bi_3dne;
528                            sad8bi = sad8bi_3dne;
529                            dev16 = dev16_3dne;
530                    }
531            }
532    
533            if ((cpu_flags & XVID_CPU_SSE2)) {
534    
535                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
536    
537                  /* Quantization */                  /* Quantization */
538                  quant_intra   = quant_intra_sse2;                  quant_h263_intra   = quant_h263_intra_sse2;
539                  dequant_intra = dequant_intra_sse2;                  quant_h263_inter   = quant_h263_inter_sse2;
540                  quant_inter   = quant_inter_sse2;                  dequant_h263_intra = dequant_h263_intra_sse2;
541                  dequant_inter = dequant_inter_sse2;                  dequant_h263_inter = dequant_h263_inter_sse2;
542    
543                  /* ME */                  /* SAD operators */
544                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
545                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
546    
547                  /* Forward and Inverse DCT */                  /* DCT operators */
548                  idct  = idct_sse2;                  fdct = fdct_sse2_skal;
549                  fdct = fdct_sse2;                  idct = idct_sse2_skal;   /* Is now IEEE1180 and Walken compliant. */
550  #endif  
551                    /* postprocessing */
552                    image_brightness = image_brightness_sse2;
553    
554          }          }
555    
556  #endif          if ((cpu_flags & XVID_CPU_SSE3)) {
557    
558                    /* SAD operators */
559                    sad16    = sad16_sse3;
560                    dev16    = dev16_sse3;
561            }
562    
563  #ifdef ARCH_IA64  #endif /* ARCH_IS_IA32 */
564          if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?  
565    #if defined(ARCH_IS_IA64)
566            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
567            idct_ia64_init();            idct_ia64_init();
568            fdct = fdct_ia64;            fdct = fdct_ia64;
569            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
570            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
571            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
572            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 298  Line 574 
574            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
575            sad8 = sad8_ia64;            sad8 = sad8_ia64;
576            dev16 = dev16_ia64;            dev16 = dev16_ia64;
577            Halfpel8_Refine = Halfpel8_Refine_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
578            quant_intra = quant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
579            dequant_intra = dequant_intra_ia64;            quant_h263_inter = quant_h263_inter_ia64;
580            quant_inter = quant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
581            dequant_inter = dequant_inter_ia64;            dequant_h263_inter = dequant_h263_inter_ia64;
582            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
583            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
584            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
585            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
586            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
587            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
588          }          }
589  #endif  #endif
590    
591  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
592  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
593          calc_cbp = calc_cbp_altivec;            /* sad operators */
594          fdct = fdct_altivec;                    sad16 = sad16_altivec_c;
595          idct = idct_altivec;                    sad16bi = sad16bi_altivec_c;
596          sadInit = sadInit_altivec;                    sad8 = sad8_altivec_c;
597          sad16 = sad16_altivec;                    dev16 = dev16_altivec_c;
598          sad8 = sad8_altivec;  
599          dev16 = dev16_altivec;            sse8_16bit = sse8_16bit_altivec_c;
600    
601              /* mem transfer */
602              transfer_8to16copy = transfer_8to16copy_altivec_c;
603              transfer_16to8copy = transfer_16to8copy_altivec_c;
604              transfer_8to16sub = transfer_8to16sub_altivec_c;
605              transfer_8to16subro = transfer_8to16subro_altivec_c;
606              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
607              transfer_16to8add = transfer_16to8add_altivec_c;
608              transfer8x8_copy = transfer8x8_copy_altivec_c;
609    
610              /* Inverse DCT */
611              idct = idct_altivec_c;
612    
613              /* Interpolation */
614              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
615              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
616              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
617    
618              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
619              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
620    
621                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
622                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
623                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
624                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
625    
626              /* Colorspace conversion */
627              bgra_to_yv12 = bgra_to_yv12_altivec_c;
628              abgr_to_yv12 = abgr_to_yv12_altivec_c;
629              rgba_to_yv12 = rgba_to_yv12_altivec_c;
630              argb_to_yv12 = argb_to_yv12_altivec_c;
631    
632              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
633              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
634    
635              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
636              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
637    
638              /* Quantization */
639              quant_h263_intra = quant_h263_intra_altivec_c;
640              quant_h263_inter = quant_h263_inter_altivec_c;
641              dequant_h263_intra = dequant_h263_intra_altivec_c;
642              dequant_h263_inter = dequant_h263_inter_altivec_c;
643    
644                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
645                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
646    
647                      /* Qpel stuff */
648                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
649                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
650            }
651    #endif
652    
653    #if defined(_DEBUG)
654        xvid_debug = init->debug;
655    #endif
656    
657        return(0);
658    }
659    
660    
661    static int
662    xvid_gbl_info(xvid_gbl_info_t * info)
663    {
664            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
665                    return XVID_ERR_VERSION;
666    
667            info->actual_version = XVID_VERSION;
668            info->build = "xvid-1.3.0-dev";
669            info->cpu_flags = detect_cpu_flags();
670            info->num_threads = 0;
671    
672    #if defined(_WIN32)
673    
674      {
675            SYSTEM_INFO siSysInfo;
676            GetSystemInfo(&siSysInfo);
677            info->num_threads = siSysInfo.dwNumberOfProcessors; /* number of _logical_ cores */
678      }
679    
680  #else  #else
681          calc_cbp = calc_cbp_ppc;  
682      info->num_threads = sysconf(_SC_NPROCESSORS_CONF);
683    
684  #endif  #endif
685    
686            return 0;
687    }
688    
689    
690    static int
691    xvid_gbl_convert(xvid_gbl_convert_t* convert)
692    {
693            int width;
694            int height;
695            int width2;
696            int height2;
697            IMAGE img;
698    
699            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
700                  return XVID_ERR_VERSION;
701    
702    #if 0
703            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
704  #endif  #endif
705            width = convert->width;
706            height = convert->height;
707            width2 = convert->width/2;
708            height2 = convert->height/2;
709    
710          return XVID_ERR_OK;          switch (convert->input.csp & ~XVID_CSP_VFLIP)
711            {
712                    case XVID_CSP_YV12 :
713                            img.y = convert->input.plane[0];
714                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
715                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
716                            image_output(&img, width, height, width,
717                                                    (uint8_t**)convert->output.plane, convert->output.stride,
718                                                    convert->output.csp, convert->interlacing);
719                            break;
720    
721                    default :
722                            return XVID_ERR_FORMAT;
723            }
724    
725    
726            emms();
727            return 0;
728    }
729    
730    /*****************************************************************************
731     * XviD Global Entry point
732     *
733     * Well this function initialize all internal function pointers according
734     * to the CPU features forced by the library client or autodetected (depending
735     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
736     * image colorspace transformation tables.
737     *
738     ****************************************************************************/
739    
740    
741    int
742    xvid_global(void *handle,
743                      int opt,
744                      void *param1,
745                      void *param2)
746    {
747            switch(opt)
748            {
749                    case XVID_GBL_INIT :
750                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
751    
752            case XVID_GBL_INFO :
753                return xvid_gbl_info((xvid_gbl_info_t*)param1);
754    
755                    case XVID_GBL_CONVERT :
756                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
757    
758                    default :
759                            return XVID_ERR_FAIL;
760            }
761  }  }
762    
763  /*****************************************************************************  /*****************************************************************************
# Line 347  Line 777 
777                          void *param2)                          void *param2)
778  {  {
779          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
780          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
781                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
782    
783          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
784                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
785    
786            case XVID_DEC_DECODE:
787                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
788    
789          default:          default:
790                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
791          }          }
# Line 380  Line 810 
810  {  {
811          switch (opt) {          switch (opt) {
812          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
813  #ifdef BFRAMES  
814                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
815                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
816                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
 #endif  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
817    
818          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
819                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
820    
821          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
822                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
823    
824          default:          default:
825                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.30  
changed lines
  Added in v.1.80

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