[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 1.20, Fri Jun 21 16:12:47 2002 UTC revision 1.61, Sun Jan 16 10:34:52 2005 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  #include "xvid.h"  #include "xvid.h"
32  #include "decoder.h"  #include "decoder.h"
33  #include "encoder.h"  #include "encoder.h"
# Line 48  Line 37 
37  #include "image/colorspace.h"  #include "image/colorspace.h"
38  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
39  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
40  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
41  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
42    #include "motion/motion.h"
43  #include "motion/sad.h"  #include "motion/sad.h"
44  #include "utils/emms.h"  #include "utils/emms.h"
45  #include "utils/timer.h"  #include "utils/timer.h"
46  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
47    #include "image/qpel.h"
48    #include "image/postprocessing.h"
49    
50    #if defined(_DEBUG)
51    unsigned int xvid_debug = 0; /* xvid debug mask */
52    #endif
53    
54    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
55    #       include <windows.h>
56    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
57    #       include <signal.h>
58    #       include <setjmp.h>
59    
60            static jmp_buf mark;
61    
62            static void
63            sigill_handler(int signal)
64            {
65               longjmp(mark, 1);
66            }
67    #endif
68    
69    
70    /*
71     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
72     * signalled
73     *
74     * Return values:
75     *  -1 : could not determine
76     *   0 : SIGILL was *not* signalled
77     *   1 : SIGILL was signalled
78     */
79    #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
80    static int
81    sigill_check(void (*func)())
82    {
83            _try {
84                    func();
85            } _except(EXCEPTION_EXECUTE_HANDLER) {
86    
87                    if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
88                            return(1);
89            }
90            return(0);
91    }
92    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC)
93    static int
94    sigill_check(void (*func)())
95    {
96        void *old_handler;
97        int jmpret;
98    
99        /* Set our SIGILL handler */
100        old_handler = signal(SIGILL, sigill_handler);
101    
102        /* Check for error */
103        if (old_handler == SIG_ERR) {
104            return(-1);
105        }
106    
107        /* Save stack context, so if func triggers a SIGILL, we can still roll
108             * back to a valid CPU state */
109            jmpret = setjmp(mark);
110    
111            /* If setjmp returned directly, then its returned value is 0, and we still
112             * have to test the passed func. Otherwise it means the stack context has
113             * been restored by a longjmp() call, which in our case happens only in the
114             * signal handler */
115        if (jmpret == 0) {
116            func();
117        }
118    
119        /* Restore old signal handler */
120        signal(SIGILL, old_handler);
121    
122        return(jmpret);
123    }
124    #endif
125    
126    
127    /* detect cpu flags  */
128    static unsigned int
129    detect_cpu_flags()
130    {
131            /* enable native assembly optimizations by default */
132            unsigned int cpu_flags = XVID_CPU_ASM;
133    
134    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
135            cpu_flags |= check_cpu_features();
136            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
137                    cpu_flags &= ~XVID_CPU_SSE;
138    
139            if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
140                    cpu_flags &= ~XVID_CPU_SSE2;
141    #endif
142    
143    #if defined(ARCH_IS_PPC)
144            if (!sigill_check(altivec_trigger))
145                    cpu_flags |= XVID_CPU_ALTIVEC;
146    #endif
147    
148            return cpu_flags;
149    }
150    
151    
152  /*****************************************************************************  /*****************************************************************************
153   * XviD Init Entry point   * XviD Init Entry point
# Line 69  Line 163 
163   *   *
164   ****************************************************************************/   ****************************************************************************/
165    
 int  
 xvid_init(void *handle,  
                   int opt,  
                   void *param1,  
                   void *param2)  
 {  
         int cpu_flags;  
         XVID_INIT_PARAM *init_param;  
166    
167          init_param = (XVID_INIT_PARAM *) param1;  static
168    int xvid_gbl_init(xvid_gbl_init_t * init)
169    {
170            unsigned int cpu_flags;
171    
172          /* Do we have to force CPU features  ? */          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
173          if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0) {                  return XVID_ERR_VERSION;
                 cpu_flags = init_param->cpu_flags;  
         } else {  
174    
175  #ifdef ARCH_X86          cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
                 cpu_flags = check_cpu_features();  
 #else  
                 cpu_flags = 0;  
 #endif  
                 init_param->cpu_flags = cpu_flags;  
         }  
176    
177          /* Initialize the function pointers */          /* Initialize the function pointers */
178          idct_int32_init();          idct_int32_init();
# Line 107  Line 188 
188          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
189          emms = emms_c;          emms = emms_c;
190    
191            /* Qpel stuff */
192            xvid_QP_Funcs = &xvid_QP_Funcs_C;
193            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
194            xvid_Init_QP();
195    
196          /* Quantization functions */          /* Quantization functions */
197          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
198          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
199          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
200          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
201    
202          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
203          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
204          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
205          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
206    
207          /* Block transfer related functions */          /* Block transfer related functions */
208          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
209          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
210          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
211            transfer_8to16subro  = transfer_8to16subro_c;
212          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_c;
213            transfer_8to16sub2ro = transfer_8to16sub2ro_c;
214          transfer_16to8add  = transfer_16to8add_c;          transfer_16to8add  = transfer_16to8add_c;
215          transfer8x8_copy   = transfer8x8_copy_c;          transfer8x8_copy   = transfer8x8_copy_c;
216    
217            /* Interlacing functions */
218            MBFieldTest = MBFieldTest_c;
219    
220          /* Image interpolation related functions */          /* Image interpolation related functions */
221          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_c;
222          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_c;
223          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
224    
225            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
226            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
227            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
228            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
229    
230            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
231            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
232            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
233    
234            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
235            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
236            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
237    
238            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
239            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
240    
241            interpolate8x8_avg2 = interpolate8x8_avg2_c;
242            interpolate8x8_avg4 = interpolate8x8_avg4_c;
243    
244            /* postprocessing */
245            image_brightness = image_brightness_c;
246    
247          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
248          colorspace_init();          colorspace_init();
249    
250          /* All colorspace transformation functions User Format->YV12 */          /* All colorspace transformation functions User Format->YV12 */
251            yv12_to_yv12    = yv12_to_yv12_c;
252          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
253          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
254          rgb24_to_yv12  = rgb24_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
255          rgb32_to_yv12  = rgb32_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
256          yuv_to_yv12    = yuv_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
257            rgba_to_yv12    = rgba_to_yv12_c;
258            argb_to_yv12    = argb_to_yv12_c;
259          yuyv_to_yv12   = yuyv_to_yv12_c;          yuyv_to_yv12   = yuyv_to_yv12_c;
260          uyvy_to_yv12   = uyvy_to_yv12_c;          uyvy_to_yv12   = uyvy_to_yv12_c;
261    
262            rgb555i_to_yv12 = rgb555i_to_yv12_c;
263            rgb565i_to_yv12 = rgb565i_to_yv12_c;
264            bgri_to_yv12    = bgri_to_yv12_c;
265            bgrai_to_yv12   = bgrai_to_yv12_c;
266            abgri_to_yv12   = abgri_to_yv12_c;
267            rgbai_to_yv12   = rgbai_to_yv12_c;
268            argbi_to_yv12   = argbi_to_yv12_c;
269            yuyvi_to_yv12   = yuyvi_to_yv12_c;
270            uyvyi_to_yv12   = uyvyi_to_yv12_c;
271    
272          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
273          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
274          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
275          yv12_to_rgb24  = yv12_to_rgb24_c;          yv12_to_bgr     = yv12_to_bgr_c;
276          yv12_to_rgb32  = yv12_to_rgb32_c;          yv12_to_bgra    = yv12_to_bgra_c;
277          yv12_to_yuv    = yv12_to_yuv_c;          yv12_to_abgr    = yv12_to_abgr_c;
278            yv12_to_rgba    = yv12_to_rgba_c;
279            yv12_to_argb    = yv12_to_argb_c;
280          yv12_to_yuyv   = yv12_to_yuyv_c;          yv12_to_yuyv   = yv12_to_yuyv_c;
281          yv12_to_uyvy   = yv12_to_uyvy_c;          yv12_to_uyvy   = yv12_to_uyvy_c;
282    
283            yv12_to_rgb555i = yv12_to_rgb555i_c;
284            yv12_to_rgb565i = yv12_to_rgb565i_c;
285            yv12_to_bgri    = yv12_to_bgri_c;
286            yv12_to_bgrai   = yv12_to_bgrai_c;
287            yv12_to_abgri   = yv12_to_abgri_c;
288            yv12_to_rgbai   = yv12_to_rgbai_c;
289            yv12_to_argbi   = yv12_to_argbi_c;
290            yv12_to_yuyvi   = yv12_to_yuyvi_c;
291            yv12_to_uyvyi   = yv12_to_uyvyi_c;
292    
293          /* Functions used in motion estimation algorithms */          /* Functions used in motion estimation algorithms */
294          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
295          sad16    = sad16_c;          sad16    = sad16_c;
         sad16bi  = sad16bi_c;  
296          sad8     = sad8_c;          sad8     = sad8_c;
297            sad16bi    = sad16bi_c;
298            sad8bi     = sad8bi_c;
299          dev16    = dev16_c;          dev16    = dev16_c;
300            sad16v     = sad16v_c;
301            sse8_16bit = sse8_16bit_c;
302            sse8_8bit  = sse8_8bit_c;
303    
304    #if defined(ARCH_IS_IA32)
305    
306            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
307                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
308                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2))
309            {
310                    /* Restore FPU context : emms_c is a nop functions */
311                    emms = emms_mmx;
312            }
313    
314  #ifdef ARCH_X86          if ((cpu_flags & XVID_CPU_MMX)) {
         if ((cpu_flags & XVID_CPU_MMX) > 0) {  
315    
316                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
317                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
318                  idct = idct_mmx;                  idct = idct_mmx;
319    
320                  /* To restore FPU context after mmx use */                  /* Qpel stuff */
321                  emms = emms_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
322                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
323    
324                  /* Quantization related functions */                  /* Quantization related functions */
325                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
326                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
327                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
328                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
329    
330                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_intra   = quant_mpeg_intra_mmx;
331                  dequant4_intra = dequant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
332                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
333                  dequant4_inter = dequant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
334    
335                  /* Block related functions */                  /* Block related functions */
336                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
337                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
338                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
339                    transfer_8to16subro  = transfer_8to16subro_mmx;
340                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
341                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
342                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
343    
344                    /* Interlacing Functions */
345                    MBFieldTest = MBFieldTest_mmx;
346    
347                  /* Image Interpolation related functions */                  /* Image Interpolation related functions */
348                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;
349                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
350                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
351    
352                  /* Image RGB->YV12 related functions */                  interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
353                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
354                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
355                  yuv_to_yv12   = yuv_to_yv12_mmx;                  interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
356    
357                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
358                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
359    
360                    interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
361                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
362    
363                    /* postprocessing */
364                    image_brightness = image_brightness_mmx;
365    
366                    /* image input xxx_to_yv12 related functions */
367                    yv12_to_yv12  = yv12_to_yv12_mmx;
368                    bgr_to_yv12   = bgr_to_yv12_mmx;
369                    bgra_to_yv12  = bgra_to_yv12_mmx;
370                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
371                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
372    
373                  /* Image YV12->RGB related functions */                  /* image output yv12_to_xxx related functions */
374                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
375                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgra  = yv12_to_bgra_mmx;
376                  yv12_to_yuyv  = yv12_to_yuyv_mmx;                  yv12_to_yuyv  = yv12_to_yuyv_mmx;
377                  yv12_to_uyvy  = yv12_to_uyvy_mmx;                  yv12_to_uyvy  = yv12_to_uyvy_mmx;
378    
379                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
380                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
381    
382                  /* Motion estimation related functions */                  /* Motion estimation related functions */
383                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
384                  sad16    = sad16_mmx;                  sad16    = sad16_mmx;
385                  sad8     = sad8_mmx;                  sad8     = sad8_mmx;
386                    sad16bi    = sad16bi_mmx;
387                    sad8bi     = sad8bi_mmx;
388                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
389                    sad16v     = sad16v_mmx;
390                    sse8_16bit = sse8_16bit_mmx;
391                    sse8_8bit  = sse8_8bit_mmx;
392            }
393    
394            /* these 3dnow functions are faster than mmx, but slower than xmm. */
395            if ((cpu_flags & XVID_CPU_3DNOW)) {
396    
397                    emms = emms_3dn;
398    
399                    /* ME functions */
400                    sad16bi = sad16bi_3dn;
401                    sad8bi  = sad8bi_3dn;
402    
403                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
404                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
405          }          }
406    
         if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
407    
408                  /* Inverse DCT */          if ((cpu_flags & XVID_CPU_MMXEXT)) {
409    
410                    /* DCT */
411                    fdct = fdct_xmm_skal;
412                  idct = idct_xmm;                  idct = idct_xmm;
413    
414                  /* Interpolation */                  /* Interpolation */
# Line 223  Line 416 
416                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
417                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
418    
419                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
420                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
421                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
422                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
423    
424                    /* Quantization */
425                    quant_mpeg_intra = quant_mpeg_intra_xmm;
426                    quant_mpeg_inter = quant_mpeg_inter_xmm;
427    
428                    dequant_h263_intra = dequant_h263_intra_xmm;
429                    dequant_h263_inter = dequant_h263_inter_xmm;
430    
431                  /* Buffer transfer */                  /* Buffer transfer */
432                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
433                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
434    
435                  /* Colorspace transformation */                  /* Colorspace transformation */
436                  yuv_to_yv12 = yuv_to_yv12_xmm;                  yv12_to_yv12  = yv12_to_yv12_xmm;
437                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
438                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
439    
440                  /* ME functions */                  /* ME functions */
441                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
442                  sad8  = sad8_xmm;                  sad8  = sad8_xmm;
443                    sad16bi = sad16bi_xmm;
444                    sad8bi  = sad8bi_xmm;
445                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
446                    sad16v   = sad16v_xmm;
447          }          }
448    
449          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
450    
451                  /* Interpolation */                  /* Interpolation */
452                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
# Line 244  Line 454 
454                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
455          }          }
456    
457          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
458  #ifdef EXPERIMENTAL_SSE2_CODE  
459                    /* Buffer transfer */
460                    transfer_8to16copy =  transfer_8to16copy_3dne;
461                    transfer_16to8copy = transfer_16to8copy_3dne;
462                    transfer_8to16sub =  transfer_8to16sub_3dne;
463                    transfer_8to16subro =  transfer_8to16subro_3dne;
464                    transfer_16to8add = transfer_16to8add_3dne;
465                    transfer8x8_copy = transfer8x8_copy_3dne;
466    
467                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
468                            /* Inverse DCT */
469                            idct =  idct_3dne;
470    
471                            /* Buffer transfer */
472                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
473    
474                            /* Interpolation */
475                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
476                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
477                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
478    
479                  /* Quantization */                  /* Quantization */
480                  quant_intra   = quant_intra_sse2;                          quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
481                  dequant_intra = dequant_intra_sse2;                          quant_h263_inter = quant_h263_inter_3dne;
482                  quant_inter   = quant_inter_sse2;                          dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
483                  dequant_inter = dequant_inter_sse2;                          dequant_mpeg_inter = dequant_mpeg_inter_3dne;
484                            dequant_h263_intra = dequant_h263_intra_3dne;
485                            dequant_h263_inter = dequant_h263_inter_3dne;
486    
487                            /* ME functions */
488                            calc_cbp = calc_cbp_3dne;
489    
490                            sad16 = sad16_3dne;
491                            sad8 = sad8_3dne;
492                            sad16bi = sad16bi_3dne;
493                            sad8bi = sad8bi_3dne;
494                            dev16 = dev16_3dne;
495                    }
496            }
497    
498            if ((cpu_flags & XVID_CPU_SSE2)) {
499    
                 /* ME */  
500                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
501    
502                    /* Quantization */
503                    quant_h263_intra   = quant_h263_intra_sse2;
504                    quant_h263_inter   = quant_h263_inter_sse2;
505                    dequant_h263_intra = dequant_h263_intra_sse2;
506                    dequant_h263_inter = dequant_h263_inter_sse2;
507    
508                    /* SAD operators */
509                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
510                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
511    
512                  /* Forward and Inverse DCT */                  /* DCT operators
513                  idct  = idct_sse2;                   * no iDCT because it's not "Walken matching" */
514                  fdct = fdct_sse2;                  fdct = fdct_sse2_skal;
 #endif  
         }  
515    
516  #endif                  /* postprocessing */
517                    image_brightness = image_brightness_sse2;
518            }
519    #endif /* ARCH_IS_IA32 */
520    
521  #ifdef ARCH_IA64  #if defined(ARCH_IS_IA64)
522          if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?          if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
523            idct_ia64_init();            idct_ia64_init();
524            fdct = fdct_ia64;            fdct = fdct_ia64;
525            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
526            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
527            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
528            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 278  Line 530 
530            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
531            sad8 = sad8_ia64;            sad8 = sad8_ia64;
532            dev16 = dev16_ia64;            dev16 = dev16_ia64;
533            quant_intra = quant_intra_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
534            dequant_intra = dequant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
535            quant_inter = quant_inter_ia64;            quant_h263_inter = quant_h263_inter_ia64;
536            dequant_inter = dequant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
537              dequant_h263_inter = dequant_h263_inter_ia64;
538            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
539            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
540            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
541            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
542            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
543            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
544          }          }
545  #endif  #endif
546    
547  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
548  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
549          calc_cbp = calc_cbp_altivec;            /* sad operators */
550          fdct = fdct_altivec;                    sad16 = sad16_altivec_c;
551          idct = idct_altivec;                    sad16bi = sad16bi_altivec_c;
552          sadInit = sadInit_altivec;                    sad8 = sad8_altivec_c;
553          sad16 = sad16_altivec;                    dev16 = dev16_altivec_c;
554          sad8 = sad8_altivec;  
555          dev16 = dev16_altivec;            sse8_16bit = sse8_16bit_altivec_c;
556    
557              /* mem transfer */
558              transfer_8to16copy = transfer_8to16copy_altivec_c;
559              transfer_16to8copy = transfer_16to8copy_altivec_c;
560              transfer_8to16sub = transfer_8to16sub_altivec_c;
561              transfer_8to16subro = transfer_8to16subro_altivec_c;
562              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
563              transfer_16to8add = transfer_16to8add_altivec_c;
564              transfer8x8_copy = transfer8x8_copy_altivec_c;
565    
566              /* Inverse DCT */
567              idct = idct_altivec_c;
568    
569              /* Interpolation */
570              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
571              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
572              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
573    
574              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
575              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
576    
577                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
578                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
579                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
580                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
581    
582              /* Colorspace conversion */
583              bgra_to_yv12 = bgra_to_yv12_altivec_c;
584              abgr_to_yv12 = abgr_to_yv12_altivec_c;
585              rgba_to_yv12 = rgba_to_yv12_altivec_c;
586              argb_to_yv12 = argb_to_yv12_altivec_c;
587    
588              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
589              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
590    
591              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
592              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
593    
594              /* Quantization */
595              quant_h263_intra = quant_h263_intra_altivec_c;
596              quant_h263_inter = quant_h263_inter_altivec_c;
597              dequant_h263_intra = dequant_h263_intra_altivec_c;
598              dequant_h263_inter = dequant_h263_inter_altivec_c;
599    
600                      /* Qpel stuff */
601                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
602                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
603            }
604    #endif
605    
606    #if defined(ARCH_IS_X86_64)
607            /* For now, only XVID_CPU_ASM is looked for, so user can still
608             * disable asm usage the usual way. When Intel EMT64 cpus will
609             * be out, maybe we'll have to check more precisely what cpu
610             * features there really are. */
611            if (cpu_flags & XVID_CPU_ASM) {
612                    /* SIMD state flusher */
613                    emms = emms_3dn;
614    
615                    /* DCT operators */
616                    fdct = fdct_skal_x86_64;
617                    idct = idct_x86_64;
618    
619                    /* SAD operators */
620                    sad16      = sad16_x86_64;
621                    sad8       = sad8_x86_64;
622                    sad16bi    = sad16bi_x86_64;
623                    sad8bi     = sad8bi_x86_64;
624                    dev16      = dev16_x86_64;
625                    sad16v     = sad16v_x86_64;
626                    sse8_16bit = sse8_16bit_x86_64;
627                    sse8_8bit  = sse8_8bit_x86_64;
628    
629                    /* Interpolation operators */
630                    interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_x86_64;
631                    interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_x86_64;
632                    interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_x86_64;
633    
634                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_x86_64;
635                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_x86_64;
636                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_x86_64;
637                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_x86_64;
638    
639                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_x86_64;
640                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_x86_64;
641    
642                    interpolate8x8_avg2 = interpolate8x8_avg2_x86_64;
643                    interpolate8x8_avg4 = interpolate8x8_avg4_x86_64;
644    
645                    /* Quantization related functions */
646                    quant_h263_intra   = quant_h263_intra_x86_64;
647                    quant_h263_inter   = quant_h263_inter_x86_64;
648                    dequant_h263_intra = dequant_h263_intra_x86_64;
649                    dequant_h263_inter = dequant_h263_inter_x86_64;
650                    quant_mpeg_intra   = quant_mpeg_intra_x86_64;
651                    quant_mpeg_inter   = quant_mpeg_inter_x86_64;
652                    dequant_mpeg_intra   = dequant_mpeg_intra_x86_64;
653                    dequant_mpeg_inter   = dequant_mpeg_inter_x86_64;
654    
655                    /* Block related functions */
656                    transfer_8to16copy  = transfer_8to16copy_x86_64;
657                    transfer_16to8copy  = transfer_16to8copy_x86_64;
658                    transfer_8to16sub   = transfer_8to16sub_x86_64;
659                    transfer_8to16subro = transfer_8to16subro_x86_64;
660                    transfer_8to16sub2  = transfer_8to16sub2_x86_64;
661                    transfer_8to16sub2ro= transfer_8to16sub2ro_x86_64;
662                    transfer_16to8add   = transfer_16to8add_x86_64;
663                    transfer8x8_copy    = transfer8x8_copy_x86_64;
664    
665                    /* Qpel stuff */
666                    xvid_QP_Funcs = &xvid_QP_Funcs_x86_64;
667                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_x86_64;
668    
669                    /* Interlacing Functions */
670                    MBFieldTest = MBFieldTest_x86_64;
671            }
672    #endif
673    
674    #if defined(_DEBUG)
675        xvid_debug = init->debug;
676    #endif
677    
678        return(0);
679    }
680    
681    
682    static int
683    xvid_gbl_info(xvid_gbl_info_t * info)
684    {
685            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
686                    return XVID_ERR_VERSION;
687    
688            info->actual_version = XVID_VERSION;
689            info->build = "xvid-1.1.0-beta1";
690            info->cpu_flags = detect_cpu_flags();
691    
692    #if defined(_SMP) && defined(WIN32)
693            info->num_threads = pthread_num_processors_np();;
694  #else  #else
695          calc_cbp = calc_cbp_ppc;          info->num_threads = 0;
696  #endif  #endif
697    
698            return 0;
699    }
700    
701    
702    static int
703    xvid_gbl_convert(xvid_gbl_convert_t* convert)
704    {
705            int width;
706            int height;
707            int width2;
708            int height2;
709            IMAGE img;
710    
711            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
712                  return XVID_ERR_VERSION;
713    
714    #if 0
715            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
716  #endif  #endif
717            width = convert->width;
718            height = convert->height;
719            width2 = convert->width/2;
720            height2 = convert->height/2;
721    
722            switch (convert->input.csp & ~XVID_CSP_VFLIP)
723            {
724                    case XVID_CSP_YV12 :
725                            img.y = convert->input.plane[0];
726                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
727                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
728                            image_output(&img, width, height, width,
729                                                    (uint8_t**)convert->output.plane, convert->output.stride,
730                                                    convert->output.csp, convert->interlacing);
731                            break;
732    
733                    default :
734                            return XVID_ERR_FORMAT;
735            }
736    
737    
738            emms();
739            return 0;
740    }
741    
742    /*****************************************************************************
743     * XviD Global Entry point
744     *
745     * Well this function initialize all internal function pointers according
746     * to the CPU features forced by the library client or autodetected (depending
747     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
748     * image colorspace transformation tables.
749     *
750     ****************************************************************************/
751    
752    
753          /* Inform the client the API version */  int
754          init_param->api_version = API_VERSION;  xvid_global(void *handle,
755                      int opt,
756                      void *param1,
757                      void *param2)
758    {
759            switch(opt)
760            {
761                    case XVID_GBL_INIT :
762                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
763    
764          /* Inform the client the core build - unused because we're still alpha */          case XVID_GBL_INFO :
765          init_param->core_build = 1000;              return xvid_gbl_info((xvid_gbl_info_t*)param1);
766    
767          return XVID_ERR_OK;                  case XVID_GBL_CONVERT :
768                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
769    
770                    default :
771                            return XVID_ERR_FAIL;
772            }
773  }  }
774    
775  /*****************************************************************************  /*****************************************************************************
# Line 332  Line 789 
789                          void *param2)                          void *param2)
790  {  {
791          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
792          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
793                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
794    
795          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
796                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
797    
798            case XVID_DEC_DECODE:
799                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
800    
801          default:          default:
802                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
803          }          }
# Line 365  Line 822 
822  {  {
823          switch (opt) {          switch (opt) {
824          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
825  #ifdef BFRAMES  
826                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
827                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
828                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
 #endif  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
829    
830          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
831                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
832    
833          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
834                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
835    
836          default:          default:
837                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.20  
changed lines
  Added in v.1.61

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