[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 1.56, Sun Oct 17 10:20:15 2004 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 47  Line 36 
36  #include "dct/fdct.h"  #include "dct/fdct.h"
37  #include "image/colorspace.h"  #include "image/colorspace.h"
38  #include "image/interpolate8x8.h"  #include "image/interpolate8x8.h"
39    #include "image/reduced.h"
40  #include "utils/mem_transfer.h"  #include "utils/mem_transfer.h"
41  #include "quant/quant_h263.h"  #include "utils/mbfunctions.h"
42  #include "quant/quant_mpeg4.h"  #include "quant/quant.h"
43    #include "motion/motion.h"
44  #include "motion/sad.h"  #include "motion/sad.h"
45  #include "utils/emms.h"  #include "utils/emms.h"
46  #include "utils/timer.h"  #include "utils/timer.h"
47  #include "bitstream/mbcoding.h"  #include "bitstream/mbcoding.h"
48    #include "image/qpel.h"
49    #include "image/postprocessing.h"
50    
51    #if defined(_DEBUG)
52    unsigned int xvid_debug = 0; /* xvid debug mask */
53    #endif
54    
55    #if defined(ARCH_IS_IA32) && defined(_MSC_VER)
56    #       include <windows.h>
57    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_PPC)
58    #       include <signal.h>
59    #       include <setjmp.h>
60    
61            static jmp_buf mark;
62    
63            static void
64            sigill_handler(int signal)
65            {
66               longjmp(mark, 1);
67            }
68    #endif
69    
70    
71    /*
72     * Calls the funcptr, and returns whether SIGILL (illegal instruction) was
73     * signalled
74     *
75     * Return values:
76     *  -1 : could not determine
77     *   0 : SIGILL was *not* signalled
78     *   1 : SIGILL was signalled
79     */
80    #if defined(ARCH_IS_IA32) && defined(_MSC_VER)
81    static int
82    sigill_check(void (*func)())
83    {
84            _try {
85                    func();
86            } _except(EXCEPTION_EXECUTE_HANDLER) {
87    
88                    if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
89                            return(1);
90            }
91            return(0);
92    }
93    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_PPC)
94    static int
95    sigill_check(void (*func)())
96    {
97        void *old_handler;
98        int jmpret;
99    
100        /* Set our SIGILL handler */
101        old_handler = signal(SIGILL, sigill_handler);
102    
103        /* Check for error */
104        if (old_handler == SIG_ERR) {
105            return(-1);
106        }
107    
108        /* Save stack context, so if func triggers a SIGILL, we can still roll
109             * back to a valid CPU state */
110            jmpret = setjmp(mark);
111    
112            /* If setjmp returned directly, then its returned value is 0, and we still
113             * have to test the passed func. Otherwise it means the stack context has
114             * been restored by a longjmp() call, which in our case happens only in the
115             * signal handler */
116        if (jmpret == 0) {
117            func();
118        }
119    
120        /* Restore old signal handler */
121        signal(SIGILL, old_handler);
122    
123        return(jmpret);
124    }
125    #endif
126    
127    
128    /* detect cpu flags  */
129    static unsigned int
130    detect_cpu_flags()
131    {
132            /* enable native assembly optimizations by default */
133            unsigned int cpu_flags = XVID_CPU_ASM;
134    
135    #if defined(ARCH_IS_IA32)
136            cpu_flags |= check_cpu_features();
137            if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger))
138                    cpu_flags &= ~XVID_CPU_SSE;
139    
140            if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger))
141                    cpu_flags &= ~XVID_CPU_SSE2;
142    #endif
143    
144    #if defined(ARCH_IS_PPC)
145            if (!sigill_check(altivec_trigger))
146                    cpu_flags |= XVID_CPU_ALTIVEC;
147    #endif
148    
149            return cpu_flags;
150    }
151    
152    
153  /*****************************************************************************  /*****************************************************************************
154   * XviD Init Entry point   * XviD Init Entry point
# Line 69  Line 164 
164   *   *
165   ****************************************************************************/   ****************************************************************************/
166    
 int  
 xvid_init(void *handle,  
                   int opt,  
                   void *param1,  
                   void *param2)  
 {  
         int cpu_flags;  
         XVID_INIT_PARAM *init_param;  
167    
168          init_param = (XVID_INIT_PARAM *) param1;  static
169    int xvid_gbl_init(xvid_gbl_init_t * init)
170    {
171            unsigned int cpu_flags;
172    
173          /* Do we have to force CPU features  ? */          if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
174          if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0) {                  return XVID_ERR_VERSION;
                 cpu_flags = init_param->cpu_flags;  
         } else {  
175    
176  #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;  
         }  
177    
178          /* Initialize the function pointers */          /* Initialize the function pointers */
179          idct_int32_init();          idct_int32_init();
# Line 107  Line 189 
189          /* Restore FPU context : emms_c is a nop functions */          /* Restore FPU context : emms_c is a nop functions */
190          emms = emms_c;          emms = emms_c;
191    
192            /* Qpel stuff */
193            xvid_QP_Funcs = &xvid_QP_Funcs_C;
194            xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
195            xvid_Init_QP();
196    
197          /* Quantization functions */          /* Quantization functions */
198          quant_intra   = quant_intra_c;          quant_h263_intra   = quant_h263_intra_c;
199          dequant_intra = dequant_intra_c;          quant_h263_inter   = quant_h263_inter_c;
200          quant_inter   = quant_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
201          dequant_inter = dequant_inter_c;          dequant_h263_inter = dequant_h263_inter_c;
202    
203          quant4_intra   = quant4_intra_c;          quant_mpeg_intra   = quant_mpeg_intra_c;
204          dequant4_intra = dequant4_intra_c;          quant_mpeg_inter   = quant_mpeg_inter_c;
205          quant4_inter   = quant4_inter_c;          dequant_mpeg_intra = dequant_mpeg_intra_c;
206          dequant4_inter = dequant4_inter_c;          dequant_mpeg_inter = dequant_mpeg_inter_c;
207    
208          /* Block transfer related functions */          /* Block transfer related functions */
209          transfer_8to16copy = transfer_8to16copy_c;          transfer_8to16copy = transfer_8to16copy_c;
210          transfer_16to8copy = transfer_16to8copy_c;          transfer_16to8copy = transfer_16to8copy_c;
211          transfer_8to16sub  = transfer_8to16sub_c;          transfer_8to16sub  = transfer_8to16sub_c;
212            transfer_8to16subro  = transfer_8to16subro_c;
213          transfer_8to16sub2 = transfer_8to16sub2_c;          transfer_8to16sub2 = transfer_8to16sub2_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            /* reduced resolution */
248            copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C;
249            add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C;
250            vfilter_31 = xvid_VFilter_31_C;
251            hfilter_31 = xvid_HFilter_31_C;
252            filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C;
253            filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C;
254    
255          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
256          colorspace_init();          colorspace_init();
257    
258          /* All colorspace transformation functions User Format->YV12 */          /* All colorspace transformation functions User Format->YV12 */
259            yv12_to_yv12    = yv12_to_yv12_c;
260          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
261          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
262          rgb24_to_yv12  = rgb24_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
263          rgb32_to_yv12  = rgb32_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
264          yuv_to_yv12    = yuv_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
265            rgba_to_yv12    = rgba_to_yv12_c;
266            argb_to_yv12    = argb_to_yv12_c;
267          yuyv_to_yv12   = yuyv_to_yv12_c;          yuyv_to_yv12   = yuyv_to_yv12_c;
268          uyvy_to_yv12   = uyvy_to_yv12_c;          uyvy_to_yv12   = uyvy_to_yv12_c;
269    
270            rgb555i_to_yv12 = rgb555i_to_yv12_c;
271            rgb565i_to_yv12 = rgb565i_to_yv12_c;
272            bgri_to_yv12    = bgri_to_yv12_c;
273            bgrai_to_yv12   = bgrai_to_yv12_c;
274            abgri_to_yv12   = abgri_to_yv12_c;
275            rgbai_to_yv12   = rgbai_to_yv12_c;
276            argbi_to_yv12   = argbi_to_yv12_c;
277            yuyvi_to_yv12   = yuyvi_to_yv12_c;
278            uyvyi_to_yv12   = uyvyi_to_yv12_c;
279    
280          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
281          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
282          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
283          yv12_to_rgb24  = yv12_to_rgb24_c;          yv12_to_bgr     = yv12_to_bgr_c;
284          yv12_to_rgb32  = yv12_to_rgb32_c;          yv12_to_bgra    = yv12_to_bgra_c;
285          yv12_to_yuv    = yv12_to_yuv_c;          yv12_to_abgr    = yv12_to_abgr_c;
286            yv12_to_rgba    = yv12_to_rgba_c;
287            yv12_to_argb    = yv12_to_argb_c;
288          yv12_to_yuyv   = yv12_to_yuyv_c;          yv12_to_yuyv   = yv12_to_yuyv_c;
289          yv12_to_uyvy   = yv12_to_uyvy_c;          yv12_to_uyvy   = yv12_to_uyvy_c;
290    
291            yv12_to_rgb555i = yv12_to_rgb555i_c;
292            yv12_to_rgb565i = yv12_to_rgb565i_c;
293            yv12_to_bgri    = yv12_to_bgri_c;
294            yv12_to_bgrai   = yv12_to_bgrai_c;
295            yv12_to_abgri   = yv12_to_abgri_c;
296            yv12_to_rgbai   = yv12_to_rgbai_c;
297            yv12_to_argbi   = yv12_to_argbi_c;
298            yv12_to_yuyvi   = yv12_to_yuyvi_c;
299            yv12_to_uyvyi   = yv12_to_uyvyi_c;
300    
301          /* Functions used in motion estimation algorithms */          /* Functions used in motion estimation algorithms */
302          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
303          sad16    = sad16_c;          sad16    = sad16_c;
         sad16bi  = sad16bi_c;  
304          sad8     = sad8_c;          sad8     = sad8_c;
305            sad16bi    = sad16bi_c;
306            sad8bi     = sad8bi_c;
307          dev16    = dev16_c;          dev16    = dev16_c;
308            sad16v     = sad16v_c;
309            sse8_16bit = sse8_16bit_c;
310            sse8_8bit  = sse8_8bit_c;
311    
312    #if defined(ARCH_IS_IA32)
313    
314            if ((cpu_flags & XVID_CPU_ASM)) {
315                    vfilter_31 = xvid_VFilter_31_x86;
316                    hfilter_31 = xvid_HFilter_31_x86;
317            }
318    
319            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
320                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
321                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2))
322            {
323                    /* Restore FPU context : emms_c is a nop functions */
324                    emms = emms_mmx;
325            }
326    
327  #ifdef ARCH_X86          if ((cpu_flags & XVID_CPU_MMX)) {
         if ((cpu_flags & XVID_CPU_MMX) > 0) {  
328    
329                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
330                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
331                  idct = idct_mmx;                  idct = idct_mmx;
332    
333                  /* To restore FPU context after mmx use */                  /* Qpel stuff */
334                  emms = emms_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
335                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
336    
337                  /* Quantization related functions */                  /* Quantization related functions */
338                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
339                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
340                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
341                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
342    
343                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_intra   = quant_mpeg_intra_mmx;
344                  dequant4_intra = dequant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
345                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
346                  dequant4_inter = dequant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
347    
348                  /* Block related functions */                  /* Block related functions */
349                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
350                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
351                  transfer_8to16sub  = transfer_8to16sub_mmx;                  transfer_8to16sub  = transfer_8to16sub_mmx;
352                    transfer_8to16subro  = transfer_8to16subro_mmx;
353                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
354                  transfer_16to8add  = transfer_16to8add_mmx;                  transfer_16to8add  = transfer_16to8add_mmx;
355                  transfer8x8_copy   = transfer8x8_copy_mmx;                  transfer8x8_copy   = transfer8x8_copy_mmx;
356    
357                    /* Interlacing Functions */
358                    MBFieldTest = MBFieldTest_mmx;
359    
360                  /* Image Interpolation related functions */                  /* Image Interpolation related functions */
361                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;
362                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
363                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
364    
365                  /* Image RGB->YV12 related functions */                  interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
366                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
367                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
368                  yuv_to_yv12   = yuv_to_yv12_mmx;                  interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx;
369    
370                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
371                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
372    
373                    interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
374                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
375    
376                    /* postprocessing */
377                    image_brightness = image_brightness_mmx;
378    
379                    /* reduced resolution */
380                    copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx;
381                    add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx;
382                    hfilter_31 = xvid_HFilter_31_mmx;
383                    filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx;
384                    filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx;
385    
386                    /* image input xxx_to_yv12 related functions */
387                    yv12_to_yv12  = yv12_to_yv12_mmx;
388                    bgr_to_yv12   = bgr_to_yv12_mmx;
389                    bgra_to_yv12  = bgra_to_yv12_mmx;
390                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
391                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
392    
393                  /* Image YV12->RGB related functions */                  /* image output yv12_to_xxx related functions */
394                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
395                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgra  = yv12_to_bgra_mmx;
396                  yv12_to_yuyv  = yv12_to_yuyv_mmx;                  yv12_to_yuyv  = yv12_to_yuyv_mmx;
397                  yv12_to_uyvy  = yv12_to_uyvy_mmx;                  yv12_to_uyvy  = yv12_to_uyvy_mmx;
398    
399                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
400                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
401    
402                  /* Motion estimation related functions */                  /* Motion estimation related functions */
403                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
404                  sad16    = sad16_mmx;                  sad16    = sad16_mmx;
405                  sad8     = sad8_mmx;                  sad8     = sad8_mmx;
406                    sad16bi    = sad16bi_mmx;
407                    sad8bi     = sad8bi_mmx;
408                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
409                    sad16v     = sad16v_mmx;
410                    sse8_16bit = sse8_16bit_mmx;
411                    sse8_8bit  = sse8_8bit_mmx;
412            }
413    
414            /* these 3dnow functions are faster than mmx, but slower than xmm. */
415            if ((cpu_flags & XVID_CPU_3DNOW)) {
416    
417                    emms = emms_3dn;
418    
419                    /* ME functions */
420                    sad16bi = sad16bi_3dn;
421                    sad8bi  = sad8bi_3dn;
422    
423                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
424                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
425          }          }
426    
         if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
427    
428                  /* Inverse DCT */          if ((cpu_flags & XVID_CPU_MMXEXT)) {
429    
430                    /* DCT */
431                    fdct = fdct_xmm_skal;
432                  idct = idct_xmm;                  idct = idct_xmm;
433    
434                  /* Interpolation */                  /* Interpolation */
# Line 223  Line 436 
436                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
437                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
438    
439                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
440                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
441                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
442                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
443    
444                    /* reduced resolution */
445                    copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm;
446                    add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm;
447    
448                    /* Quantization */
449                    quant_mpeg_intra = quant_mpeg_intra_xmm;
450                    quant_mpeg_inter = quant_mpeg_inter_xmm;
451    
452                    dequant_h263_intra = dequant_h263_intra_xmm;
453                    dequant_h263_inter = dequant_h263_inter_xmm;
454    
455                  /* Buffer transfer */                  /* Buffer transfer */
456                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
457    
458                  /* Colorspace transformation */                  /* Colorspace transformation */
459                  yuv_to_yv12 = yuv_to_yv12_xmm;                  yv12_to_yv12  = yv12_to_yv12_xmm;
460                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
461                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
462    
463                  /* ME functions */                  /* ME functions */
464                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
465                  sad8  = sad8_xmm;                  sad8  = sad8_xmm;
466                    sad16bi = sad16bi_xmm;
467                    sad8bi  = sad8bi_xmm;
468                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
469                    sad16v   = sad16v_xmm;
470          }          }
471    
472          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
473    
474                  /* Interpolation */                  /* Interpolation */
475                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
# Line 244  Line 477 
477                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
478          }          }
479    
480          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
481  #ifdef EXPERIMENTAL_SSE2_CODE  
482                    /* Buffer transfer */
483                    transfer_8to16copy =  transfer_8to16copy_3dne;
484                    transfer_16to8copy = transfer_16to8copy_3dne;
485                    transfer_8to16sub =  transfer_8to16sub_3dne;
486                    transfer_8to16subro =  transfer_8to16subro_3dne;
487                    transfer_16to8add = transfer_16to8add_3dne;
488                    transfer8x8_copy = transfer8x8_copy_3dne;
489    
490                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
491                            /* Inverse DCT */
492                            idct =  idct_3dne;
493    
494                            /* Buffer transfer */
495                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
496    
497                            /* Interpolation */
498                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
499                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
500                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
501    
502                  /* Quantization */                  /* Quantization */
503                  quant_intra   = quant_intra_sse2;                          quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
504                  dequant_intra = dequant_intra_sse2;                          quant_h263_inter = quant_h263_inter_3dne;
505                  quant_inter   = quant_inter_sse2;                          dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
506                  dequant_inter = dequant_inter_sse2;                          dequant_mpeg_inter = dequant_mpeg_inter_3dne;
507                            dequant_h263_intra = dequant_h263_intra_3dne;
508                            dequant_h263_inter = dequant_h263_inter_3dne;
509    
510                            /* ME functions */
511                            calc_cbp = calc_cbp_3dne;
512    
513                            sad16 = sad16_3dne;
514                            sad8 = sad8_3dne;
515                            sad16bi = sad16bi_3dne;
516                            sad8bi = sad8bi_3dne;
517                            dev16 = dev16_3dne;
518                    }
519            }
520    
521            if ((cpu_flags & XVID_CPU_SSE2)) {
522    
                 /* ME */  
523                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
524    
525                    /* Quantization */
526                    quant_h263_intra   = quant_h263_intra_sse2;
527                    quant_h263_inter   = quant_h263_inter_sse2;
528                    dequant_h263_intra = dequant_h263_intra_sse2;
529                    dequant_h263_inter = dequant_h263_inter_sse2;
530    
531                    /* SAD operators */
532                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
533                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
534    
535                  /* Forward and Inverse DCT */                  /* DCT operators
536                  idct  = idct_sse2;                   * no iDCT because it's not "Walken matching" */
537                  fdct = fdct_sse2;                  fdct = fdct_sse2_skal;
 #endif  
         }  
538    
539  #endif                  /* postprocessing */
540                    image_brightness = image_brightness_sse2;
541            }
542    #endif /* ARCH_IS_IA32 */
543    
544  #ifdef ARCH_IA64  #if defined(ARCH_IS_IA64)
545          if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?          if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
546            idct_ia64_init();            idct_ia64_init();
547            fdct = fdct_ia64;            fdct = fdct_ia64;
548            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
549            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
550            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
551            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 278  Line 553 
553            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
554            sad8 = sad8_ia64;            sad8 = sad8_ia64;
555            dev16 = dev16_ia64;            dev16 = dev16_ia64;
556            quant_intra = quant_intra_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
557            dequant_intra = dequant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
558            quant_inter = quant_inter_ia64;            quant_h263_inter = quant_h263_inter_ia64;
559            dequant_inter = dequant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
560              dequant_h263_inter = dequant_h263_inter_ia64;
561            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
562            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
563            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
564            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
565            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
566            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
567          }          }
568  #endif  #endif
569    
570  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
571  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
572          calc_cbp = calc_cbp_altivec;            /* sad operators */
573          fdct = fdct_altivec;                    sad16 = sad16_altivec_c;
574          idct = idct_altivec;                    sad16bi = sad16bi_altivec_c;
575          sadInit = sadInit_altivec;                    sad8 = sad8_altivec_c;
576          sad16 = sad16_altivec;                    dev16 = dev16_altivec_c;
577          sad8 = sad8_altivec;  
578          dev16 = dev16_altivec;            sse8_16bit = sse8_16bit_altivec_c;
579    
580              /* mem transfer */
581              transfer_8to16copy = transfer_8to16copy_altivec_c;
582              transfer_16to8copy = transfer_16to8copy_altivec_c;
583              transfer_8to16sub = transfer_8to16sub_altivec_c;
584              transfer_8to16subro = transfer_8to16subro_altivec_c;
585              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
586              transfer_16to8add = transfer_16to8add_altivec_c;
587              transfer8x8_copy = transfer8x8_copy_altivec_c;
588    
589              /* Inverse DCT */
590              idct = idct_altivec_c;
591    
592              /* Interpolation */
593              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
594              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
595              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
596    
597              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
598              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
599    
600                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
601                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
602                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
603                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
604    
605              /* Colorspace conversion */
606              bgra_to_yv12 = bgra_to_yv12_altivec_c;
607              abgr_to_yv12 = abgr_to_yv12_altivec_c;
608              rgba_to_yv12 = rgba_to_yv12_altivec_c;
609              argb_to_yv12 = argb_to_yv12_altivec_c;
610    
611              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
612              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
613    
614              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
615              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
616    
617              /* Quantization */
618              quant_h263_intra = quant_h263_intra_altivec_c;
619              quant_h263_inter = quant_h263_inter_altivec_c;
620              dequant_h263_intra = dequant_h263_intra_altivec_c;
621              dequant_h263_inter = dequant_h263_inter_altivec_c;
622    
623                      /* Qpel stuff */
624                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
625                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
626            }
627    #endif
628    
629    #if defined(_DEBUG)
630        xvid_debug = init->debug;
631    #endif
632    
633        return(0);
634    }
635    
636    
637    static int
638    xvid_gbl_info(xvid_gbl_info_t * info)
639    {
640            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
641                    return XVID_ERR_VERSION;
642    
643            info->actual_version = XVID_VERSION;
644            info->build = "xvid-1.1-cvshead";
645            info->cpu_flags = detect_cpu_flags();
646    
647    #if defined(_SMP) && defined(WIN32)
648            info->num_threads = pthread_num_processors_np();;
649  #else  #else
650          calc_cbp = calc_cbp_ppc;          info->num_threads = 0;
651  #endif  #endif
652    
653            return 0;
654    }
655    
656    
657    static int
658    xvid_gbl_convert(xvid_gbl_convert_t* convert)
659    {
660            int width;
661            int height;
662            int width2;
663            int height2;
664            IMAGE img;
665    
666            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
667                  return XVID_ERR_VERSION;
668    
669    #if 0
670            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
671  #endif  #endif
672            width = convert->width;
673            height = convert->height;
674            width2 = convert->width/2;
675            height2 = convert->height/2;
676    
677          /* Inform the client the API version */          switch (convert->input.csp & ~XVID_CSP_VFLIP)
678          init_param->api_version = API_VERSION;          {
679                    case XVID_CSP_YV12 :
680                            img.y = convert->input.plane[0];
681                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
682                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
683                            image_output(&img, width, height, width,
684                                                    (uint8_t**)convert->output.plane, convert->output.stride,
685                                                    convert->output.csp, convert->interlacing);
686                            break;
687    
688          /* Inform the client the core build - unused because we're still alpha */                  default :
689          init_param->core_build = 1000;                          return XVID_ERR_FORMAT;
690            }
691    
692          return XVID_ERR_OK;  
693            emms();
694            return 0;
695    }
696    
697    /*****************************************************************************
698     * XviD Global Entry point
699     *
700     * Well this function initialize all internal function pointers according
701     * to the CPU features forced by the library client or autodetected (depending
702     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
703     * image colorspace transformation tables.
704     *
705     ****************************************************************************/
706    
707    
708    int
709    xvid_global(void *handle,
710                      int opt,
711                      void *param1,
712                      void *param2)
713    {
714            switch(opt)
715            {
716                    case XVID_GBL_INIT :
717                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
718    
719            case XVID_GBL_INFO :
720                return xvid_gbl_info((xvid_gbl_info_t*)param1);
721    
722                    case XVID_GBL_CONVERT :
723                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
724    
725                    default :
726                            return XVID_ERR_FAIL;
727            }
728  }  }
729    
730  /*****************************************************************************  /*****************************************************************************
# Line 332  Line 744 
744                          void *param2)                          void *param2)
745  {  {
746          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
747          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
748                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
749    
750          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
751                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
752    
753            case XVID_DEC_DECODE:
754                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
755    
756          default:          default:
757                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
758          }          }
# Line 365  Line 777 
777  {  {
778          switch (opt) {          switch (opt) {
779          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
780  #ifdef BFRAMES  
781                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
782                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
783                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
 #endif  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
784    
785          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
786                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
787    
788          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
789                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
790    
791          default:          default:
792                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.20  
changed lines
  Added in v.1.56

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