[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 1.15, Wed Jun 12 20:38:40 2002 UTC revision 1.70, Mon Jul 10 08:09:59 2006 UTC
# Line 1  Line 1 
1  /**************************************************************************  /*****************************************************************************
2   *   *
3   *      XVID MPEG-4 VIDEO CODEC   *      XVID MPEG-4 VIDEO CODEC
4   *      native api   *  - Native API implementation  -
5   *   *
6   *      This program is an implementation of a part of one or more MPEG-4   *  Copyright(C) 2001-2004 Peter Ross <pross@xvid.org>
  *      Video tools as specified in ISO/IEC 14496-2 standard.  Those intending  
  *      to use this software module in hardware or software products are  
  *      advised that its use may infringe existing patents or copyrights, and  
  *      any such use would be at such party's own risk.  The original  
  *      developer of this software module and his/her company, and subsequent  
  *      editors and their companies, will have no liability for use of this  
  *      software or modifications or derivatives thereof.  
7   *   *
8   *      This program is free software; you can redistribute it and/or modify   *      This program is free software; you can redistribute it and/or modify
9   *      it under the terms of the GNU General Public License as published by   *      it under the terms of the GNU General Public License as published by
# Line 24  Line 17 
17   *   *
18   *      You should have received a copy of the GNU General Public License   *      You should have received a copy of the GNU General Public License
19   *      along with this program; if not, write to the Free Software   *      along with this program; if not, write to the Free Software
20   *      Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.   *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
21   *   *
22   *************************************************************************/   * $Id$
   
 /**************************************************************************  
  *  
  *      History:  
  *  
  *      17.03.2002      Added interpolate8x8_halfpel_hv_xmm  
  *  22.12.2001  API change: added xvid_init() - Isibaar  
  *      16.12.2001      inital version; (c)2001 peter ross <pross@cs.rmit.edu.au>  
23   *   *
24   *************************************************************************/   ****************************************************************************/
25    
26    #include <stdio.h>
27    #include <stdlib.h>
28    #include <string.h>
29    #include <time.h>
30    
31  #include "xvid.h"  #include "xvid.h"
32  #include "decoder.h"  #include "decoder.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/gmc.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  int  #if defined(_DEBUG)
52  xvid_init(void *handle,  unsigned int xvid_debug = 0; /* xvid debug mask */
53                    int opt,  #endif
54                    void *param1,  
55                    void *param2)  #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER)
56    #       include <windows.h>
57    #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || 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(ARCH_IS_X86_64)) && 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_X86_64) || defined(ARCH_IS_PPC)
94    static int
95    sigill_check(void (*func)())
96  {  {
97          int cpu_flags;      void *old_handler;
98          XVID_INIT_PARAM *init_param;      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          init_param = (XVID_INIT_PARAM *) param1;      /* 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          // force specific cpu settings?      /* Restore old signal handler */
121          if ((init_param->cpu_flags & XVID_CPU_FORCE) > 0)      signal(SIGILL, old_handler);
                 cpu_flags = init_param->cpu_flags;  
         else {  
122    
123  #ifdef ARCH_X86      return(jmpret);
124                  cpu_flags = check_cpu_features();  }
 #else  
                 cpu_flags = 0;  
125  #endif  #endif
126                  init_param->cpu_flags = cpu_flags;  
127    
128    /* detect cpu flags  */
129    static unsigned int
130    detect_cpu_flags(void)
131    {
132            /* enable native assembly optimizations by default */
133            unsigned int cpu_flags = XVID_CPU_ASM;
134    
135    #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)
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          // initialize the function pointers  
153    /*****************************************************************************
154     * XviD Init Entry point
155     *
156     * Well this function initialize all internal function pointers according
157     * to the CPU features forced by the library client or autodetected (depending
158     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
159     * image colorspace transformation tables.
160     *
161     * Returned value : XVID_ERR_OK
162     *                  + API_VERSION in the input XVID_INIT_PARAM structure
163     *                  + core build  "   "    "       "               "
164     *
165     ****************************************************************************/
166    
167    
168    static
169    int xvid_gbl_init(xvid_gbl_init_t * init)
170    {
171            unsigned int cpu_flags;
172    
173            if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */
174                    return XVID_ERR_VERSION;
175    
176            cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags();
177    
178            /* Initialize the function pointers */
179          idct_int32_init();          idct_int32_init();
180          init_vlc_tables();          init_vlc_tables();
181    
182            /* Fixed Point Forward/Inverse DCT transformations */
183          fdct = fdct_int32;          fdct = fdct_int32;
184          idct = idct_int32;          idct = idct_int32;
185    
186          sadInit = 0;          /* Only needed on PPC Altivec archs */
187            sadInit = NULL;
188    
189            /* Restore FPU context : emms_c is a nop functions */
190          emms = emms_c;          emms = emms_c;
191    
192          quant_intra = quant_intra_c;          /* Qpel stuff */
193          dequant_intra = dequant_intra_c;          xvid_QP_Funcs = &xvid_QP_Funcs_C;
194          quant_inter = quant_inter_c;          xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C;
195          dequant_inter = dequant_inter_c;          xvid_Init_QP();
196    
197          quant4_intra = quant4_intra_c;          /* Quantization functions */
198          dequant4_intra = dequant4_intra_c;          quant_h263_intra   = quant_h263_intra_c;
199          quant4_inter = quant4_inter_c;          quant_h263_inter   = quant_h263_inter_c;
200          dequant4_inter = dequant4_inter_c;          dequant_h263_intra = dequant_h263_intra_c;
201            dequant_h263_inter = dequant_h263_inter_c;
202    
203            quant_mpeg_intra   = quant_mpeg_intra_c;
204            quant_mpeg_inter   = quant_mpeg_inter_c;
205            dequant_mpeg_intra = dequant_mpeg_intra_c;
206            dequant_mpeg_inter = dequant_mpeg_inter_c;
207    
208            /* 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_8to16sub2ro = transfer_8to16sub2ro_c;
215          transfer_16to8add = transfer_16to8add_c;          transfer_16to8add = transfer_16to8add_c;
216          transfer8x8_copy = transfer8x8_copy_c;          transfer8x8_copy = transfer8x8_copy_c;
217            transfer8x4_copy   = transfer8x4_copy_c;
218    
219            /* Interlacing functions */
220            MBFieldTest = MBFieldTest_c;
221    
222            /* Image interpolation related functions */
223          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;          interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c;
224          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;          interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c;
225          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;          interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c;
226    
227            interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_c;
228            interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_c;
229            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c;
230    
231            interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c;
232            interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c;
233            interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c;
234            interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c;
235    
236            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
237            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
238            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
239    
240            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
241            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
242            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
243    
244            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
245            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
246    
247            interpolate8x8_avg2 = interpolate8x8_avg2_c;
248            interpolate8x8_avg4 = interpolate8x8_avg4_c;
249    
250            /* postprocessing */
251            image_brightness = image_brightness_c;
252    
253            /* Initialize internal colorspace transformation tables */
254          colorspace_init();          colorspace_init();
255    
256            /* All colorspace transformation functions User Format->YV12 */
257            yv12_to_yv12    = yv12_to_yv12_c;
258          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
259          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
260          rgb24_to_yv12 = rgb24_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
261          rgb32_to_yv12 = rgb32_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
262          yuv_to_yv12 = yuv_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
263            rgba_to_yv12    = rgba_to_yv12_c;
264            argb_to_yv12    = argb_to_yv12_c;
265          yuyv_to_yv12 = yuyv_to_yv12_c;          yuyv_to_yv12 = yuyv_to_yv12_c;
266          uyvy_to_yv12 = uyvy_to_yv12_c;          uyvy_to_yv12 = uyvy_to_yv12_c;
267    
268            rgb555i_to_yv12 = rgb555i_to_yv12_c;
269            rgb565i_to_yv12 = rgb565i_to_yv12_c;
270            bgri_to_yv12    = bgri_to_yv12_c;
271            bgrai_to_yv12   = bgrai_to_yv12_c;
272            abgri_to_yv12   = abgri_to_yv12_c;
273            rgbai_to_yv12   = rgbai_to_yv12_c;
274            argbi_to_yv12   = argbi_to_yv12_c;
275            yuyvi_to_yv12   = yuyvi_to_yv12_c;
276            uyvyi_to_yv12   = uyvyi_to_yv12_c;
277    
278            /* All colorspace transformation functions YV12->User format */
279          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
280          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
281          yv12_to_rgb24 = yv12_to_rgb24_c;          yv12_to_bgr     = yv12_to_bgr_c;
282          yv12_to_rgb32 = yv12_to_rgb32_c;          yv12_to_bgra    = yv12_to_bgra_c;
283          yv12_to_yuv = yv12_to_yuv_c;          yv12_to_abgr    = yv12_to_abgr_c;
284            yv12_to_rgba    = yv12_to_rgba_c;
285            yv12_to_argb    = yv12_to_argb_c;
286          yv12_to_yuyv = yv12_to_yuyv_c;          yv12_to_yuyv = yv12_to_yuyv_c;
287          yv12_to_uyvy = yv12_to_uyvy_c;          yv12_to_uyvy = yv12_to_uyvy_c;
288    
289            yv12_to_rgb555i = yv12_to_rgb555i_c;
290            yv12_to_rgb565i = yv12_to_rgb565i_c;
291            yv12_to_bgri    = yv12_to_bgri_c;
292            yv12_to_bgrai   = yv12_to_bgrai_c;
293            yv12_to_abgri   = yv12_to_abgri_c;
294            yv12_to_rgbai   = yv12_to_rgbai_c;
295            yv12_to_argbi   = yv12_to_argbi_c;
296            yv12_to_yuyvi   = yv12_to_yuyvi_c;
297            yv12_to_uyvyi   = yv12_to_uyvyi_c;
298    
299            /* Functions used in motion estimation algorithms */
300          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
301          sad16 = sad16_c;          sad16 = sad16_c;
         sad16bi = sad16bi_c;  
302          sad8 = sad8_c;          sad8 = sad8_c;
303            sad16bi    = sad16bi_c;
304            sad8bi     = sad8bi_c;
305          dev16 = dev16_c;          dev16 = dev16_c;
306            sad16v     = sad16v_c;
307            sse8_16bit = sse8_16bit_c;
308            sse8_8bit  = sse8_8bit_c;
309    
310  #ifdef ARCH_X86          init_GMC(cpu_flags);
311          if ((cpu_flags & XVID_CPU_MMX) > 0) {  
312                  fdct = fdct_mmx;  #if defined(ARCH_IS_IA32)
                 idct = idct_mmx;  
313    
314            if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) ||
315                    (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) ||
316                    (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2))
317            {
318                    /* Restore FPU context : emms_c is a nop functions */
319                  emms = emms_mmx;                  emms = emms_mmx;
320            }
321    
322                  quant_intra = quant_intra_mmx;          if ((cpu_flags & XVID_CPU_MMX)) {
                 dequant_intra = dequant_intra_mmx;  
                 quant_inter = quant_inter_mmx;  
                 dequant_inter = dequant_inter_mmx;  
   
                 quant4_intra = quant4_intra_mmx;  
                 dequant4_intra = dequant4_intra_mmx;  
                 quant4_inter = quant4_inter_mmx;  
                 dequant4_inter = dequant4_inter_mmx;  
323    
324                    /* Forward and Inverse Discrete Cosine Transformation functions */
325                    fdct = fdct_mmx_skal;
326                    idct = idct_mmx;
327    
328                    /* Qpel stuff */
329                    xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
330                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
331    
332                    /* Quantization related functions */
333                    quant_h263_intra   = quant_h263_intra_mmx;
334                    quant_h263_inter   = quant_h263_inter_mmx;
335                    dequant_h263_intra = dequant_h263_intra_mmx;
336                    dequant_h263_inter = dequant_h263_inter_mmx;
337    
338                    quant_mpeg_intra   = quant_mpeg_intra_mmx;
339                    quant_mpeg_inter   = quant_mpeg_inter_mmx;
340                    dequant_mpeg_intra = dequant_mpeg_intra_mmx;
341                    dequant_mpeg_inter = dequant_mpeg_inter_mmx;
342    
343                    /* Block related functions */
344                  transfer_8to16copy = transfer_8to16copy_mmx;                  transfer_8to16copy = transfer_8to16copy_mmx;
345                  transfer_16to8copy = transfer_16to8copy_mmx;                  transfer_16to8copy = transfer_16to8copy_mmx;
346                  transfer_8to16sub = transfer_8to16sub_mmx;                  transfer_8to16sub = transfer_8to16sub_mmx;
347                    transfer_8to16subro  = transfer_8to16subro_mmx;
348                    transfer_8to16sub2 = transfer_8to16sub2_mmx;
349                  transfer_16to8add = transfer_16to8add_mmx;                  transfer_16to8add = transfer_16to8add_mmx;
350                  transfer8x8_copy = transfer8x8_copy_mmx;                  transfer8x8_copy = transfer8x8_copy_mmx;
351                    transfer8x4_copy   = transfer8x4_copy_mmx;
352    
353                    /* Interlacing Functions */
354                    MBFieldTest = MBFieldTest_mmx;
355    
356                    /* Image Interpolation related functions */
357                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx;
358                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx;
359                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
360    
361                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_mmx;
362                  rgb32_to_yv12 = rgb32_to_yv12_mmx;                  interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_mmx;
363                  yuv_to_yv12 = yuv_to_yv12_mmx;                  interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx;
364    
365                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx;
366                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx;
367                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx;
368                    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                    /* image input xxx_to_yv12 related functions */
380                    yv12_to_yv12  = yv12_to_yv12_mmx;
381                    bgr_to_yv12   = bgr_to_yv12_mmx;
382                    bgra_to_yv12  = bgra_to_yv12_mmx;
383                  yuyv_to_yv12 = yuyv_to_yv12_mmx;                  yuyv_to_yv12 = yuyv_to_yv12_mmx;
384                  uyvy_to_yv12 = uyvy_to_yv12_mmx;                  uyvy_to_yv12 = uyvy_to_yv12_mmx;
385    
386                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  /* image output yv12_to_xxx related functions */
387                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
388                    yv12_to_bgra  = yv12_to_bgra_mmx;
389                  yv12_to_yuyv = yv12_to_yuyv_mmx;                  yv12_to_yuyv = yv12_to_yuyv_mmx;
390                  yv12_to_uyvy = yv12_to_uyvy_mmx;                  yv12_to_uyvy = yv12_to_uyvy_mmx;
391    
392                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
393                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
394    
395                    /* Motion estimation related functions */
396                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
397                  sad16 = sad16_mmx;                  sad16 = sad16_mmx;
398                  sad8 = sad8_mmx;                  sad8 = sad8_mmx;
399                    sad16bi    = sad16bi_mmx;
400                    sad8bi     = sad8bi_mmx;
401                  dev16 = dev16_mmx;                  dev16 = dev16_mmx;
402                    sad16v     = sad16v_mmx;
403                    sse8_16bit = sse8_16bit_mmx;
404                    sse8_8bit  = sse8_8bit_mmx;
405            }
406    
407            /* these 3dnow functions are faster than mmx, but slower than xmm. */
408            if ((cpu_flags & XVID_CPU_3DNOW)) {
409    
410                    emms = emms_3dn;
411    
412                    /* ME functions */
413                    sad16bi = sad16bi_3dn;
414                    sad8bi  = sad8bi_3dn;
415    
416                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
417                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
418          }          }
419    
420          if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
421            if ((cpu_flags & XVID_CPU_MMXEXT)) {
422    
423                    /* DCT */
424                    fdct = fdct_xmm_skal;
425                  idct = idct_xmm;                  idct = idct_xmm;
426    
427                    /* Interpolation */
428                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm;
429                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm;
430                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
                 yuv_to_yv12 = yuv_to_yv12_xmm;  
431    
432                    interpolate8x4_halfpel_h  = interpolate8x4_halfpel_h_xmm;
433                    interpolate8x4_halfpel_v  = interpolate8x4_halfpel_v_xmm;
434                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm;
435    
436                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm;
437                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm;
438                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm;
439                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm;
440    
441                    /* Quantization */
442                    quant_mpeg_inter = quant_mpeg_inter_xmm;
443    
444                    dequant_h263_intra = dequant_h263_intra_xmm;
445                    dequant_h263_inter = dequant_h263_inter_xmm;
446    
447                    /* Buffer transfer */
448                    transfer_8to16sub2 = transfer_8to16sub2_xmm;
449                    transfer_8to16sub2ro = transfer_8to16sub2ro_xmm;
450    
451                    /* Colorspace transformation */
452                    yv12_to_yv12  = yv12_to_yv12_xmm;
453                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
454                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
455    
456                    /* ME functions */
457                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
458                  sad8 = sad8_xmm;                  sad8 = sad8_xmm;
459                    sad16bi = sad16bi_xmm;
460                    sad8bi  = sad8bi_xmm;
461                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
462                    sad16v   = sad16v_xmm;
463          }          }
464    
465          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
466    
467                    /* Interpolation */
468                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
469                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;                  interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn;
470                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
471    
472                    interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn;
473                    interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn;
474                    interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn;
475            }
476    
477            if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
478    
479                    /* Buffer transfer */
480                    transfer_8to16copy =  transfer_8to16copy_3dne;
481                    transfer_16to8copy = transfer_16to8copy_3dne;
482                    transfer_8to16sub =  transfer_8to16sub_3dne;
483                    transfer_8to16subro =  transfer_8to16subro_3dne;
484                    transfer_16to8add = transfer_16to8add_3dne;
485                    transfer8x8_copy = transfer8x8_copy_3dne;
486                    transfer8x4_copy = transfer8x4_copy_3dne;
487    
488                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
489                            /* Inverse DCT */
490                            idct =  idct_3dne;
491    
492                            /* Buffer transfer */
493                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
494    
495                            /* Interpolation */
496                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
497                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
498                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
499    
500                            interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne;
501                            interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne;
502                            interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne;
503    
504                            /* Quantization */
505                            quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
506                            quant_h263_inter = quant_h263_inter_3dne;
507                            dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
508                            dequant_mpeg_inter = dequant_mpeg_inter_3dne;
509                            dequant_h263_intra = dequant_h263_intra_3dne;
510                            dequant_h263_inter = dequant_h263_inter_3dne;
511    
512                            /* ME functions */
513                            calc_cbp = calc_cbp_3dne;
514    
515                            sad16 = sad16_3dne;
516                            sad8 = sad8_3dne;
517                            sad16bi = sad16bi_3dne;
518                            sad8bi = sad8bi_3dne;
519                            dev16 = dev16_3dne;
520                    }
521          }          }
522    
523          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_SSE2)) {
 #ifdef EXPERIMENTAL_SSE2_CODE  
                 quant_intra = quant_intra_sse2;  
                 dequant_intra = dequant_intra_sse2;  
                 quant_inter = quant_inter_sse2;  
                 dequant_inter = dequant_inter_sse2;  
524    
525                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
526    
527                    /* Quantization */
528                    quant_h263_intra   = quant_h263_intra_sse2;
529                    quant_h263_inter   = quant_h263_inter_sse2;
530                    dequant_h263_intra = dequant_h263_intra_sse2;
531                    dequant_h263_inter = dequant_h263_inter_sse2;
532    
533                    /* SAD operators */
534                  sad16 = sad16_sse2;                  sad16 = sad16_sse2;
535                  dev16 = dev16_sse2;                  dev16 = dev16_sse2;
536                  idct = idct_sse2;  
537                  fdct = fdct_sse2;                  /* DCT operators */
538                    fdct = fdct_sse2_skal;
539        /* idct = idct_sse2_skal; */   /* Is now IEEE1180 and Walken compliant. Disabled until fully tested. */
540    
541                    /* postprocessing */
542                    image_brightness = image_brightness_sse2;
543            }
544    #endif /* ARCH_IS_IA32 */
545    
546    #if defined(ARCH_IS_IA64)
547            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
548              idct_ia64_init();
549              fdct = fdct_ia64;
550              idct = idct_ia64;   /*not yet working, crashes */
551              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
552              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
553              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
554              sad16 = sad16_ia64;
555              sad16bi = sad16bi_ia64;
556              sad8 = sad8_ia64;
557              dev16 = dev16_ia64;
558    /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
559              quant_h263_intra = quant_h263_intra_ia64;
560              quant_h263_inter = quant_h263_inter_ia64;
561              dequant_h263_intra = dequant_h263_intra_ia64;
562              dequant_h263_inter = dequant_h263_inter_ia64;
563              transfer_8to16copy = transfer_8to16copy_ia64;
564              transfer_16to8copy = transfer_16to8copy_ia64;
565              transfer_8to16sub = transfer_8to16sub_ia64;
566              transfer_8to16sub2 = transfer_8to16sub2_ia64;
567              transfer_16to8add = transfer_16to8add_ia64;
568              transfer8x8_copy = transfer8x8_copy_ia64;
569            }
570  #endif  #endif
571    
572    #if defined(ARCH_IS_PPC)
573            if ((cpu_flags & XVID_CPU_ALTIVEC)) {
574              /* sad operators */
575                      sad16 = sad16_altivec_c;
576                      sad16bi = sad16bi_altivec_c;
577                      sad8 = sad8_altivec_c;
578                      dev16 = dev16_altivec_c;
579    
580              sse8_16bit = sse8_16bit_altivec_c;
581    
582              /* mem transfer */
583              transfer_8to16copy = transfer_8to16copy_altivec_c;
584              transfer_16to8copy = transfer_16to8copy_altivec_c;
585              transfer_8to16sub = transfer_8to16sub_altivec_c;
586              transfer_8to16subro = transfer_8to16subro_altivec_c;
587              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
588              transfer_16to8add = transfer_16to8add_altivec_c;
589              transfer8x8_copy = transfer8x8_copy_altivec_c;
590    
591              /* Inverse DCT */
592              idct = idct_altivec_c;
593    
594              /* Interpolation */
595              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
596              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
597              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
598    
599              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
600              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
601    
602                      interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c;
603                      interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c;
604                      interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c;
605                      interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c;
606    
607              /* Colorspace conversion */
608              bgra_to_yv12 = bgra_to_yv12_altivec_c;
609              abgr_to_yv12 = abgr_to_yv12_altivec_c;
610              rgba_to_yv12 = rgba_to_yv12_altivec_c;
611              argb_to_yv12 = argb_to_yv12_altivec_c;
612    
613              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
614              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
615    
616              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
617              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
618    
619              /* Quantization */
620              quant_h263_intra = quant_h263_intra_altivec_c;
621              quant_h263_inter = quant_h263_inter_altivec_c;
622              dequant_h263_intra = dequant_h263_intra_altivec_c;
623              dequant_h263_inter = dequant_h263_inter_altivec_c;
624    
625                      dequant_mpeg_intra = dequant_mpeg_intra_altivec_c;
626                      dequant_mpeg_inter = dequant_mpeg_inter_altivec_c;
627    
628                      /* Qpel stuff */
629                      xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C;
630                      xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C;
631          }          }
632  #endif  #endif
633  #ifdef ARCH_PPC  
634  #ifdef ARCH_PPC_ALTIVEC  #if defined(ARCH_IS_X86_64)
635          calc_cbp = calc_cbp_altivec;          /* For now, only XVID_CPU_ASM is looked for, so user can still
636          fdct = fdct_altivec;           * disable asm usage the usual way. When Intel EMT64 cpus will
637          idct = idct_altivec;           * be out, maybe we'll have to check more precisely what cpu
638          sadInit = sadInit_altivec;           * features there really are. */
639          sad16 = sad16_altivec;          if (cpu_flags & XVID_CPU_ASM) {
640          sad8 = sad8_altivec;                  /* SIMD state flusher */
641          dev16 = dev16_altivec;                  emms = emms_mmx;
642  #else  
643          calc_cbp = calc_cbp_ppc;                  /* DCT operators */
644                    fdct = fdct_skal_x86_64;
645                    idct = idct_x86_64;
646    
647                    /* SAD operators */
648                    sad16      = sad16_x86_64;
649                    sad8       = sad8_x86_64;
650                    sad16bi    = sad16bi_x86_64;
651                    sad8bi     = sad8bi_x86_64;
652                    dev16      = dev16_x86_64;
653                    sad16v     = sad16v_x86_64;
654                    sse8_16bit = sse8_16bit_x86_64;
655                    sse8_8bit  = sse8_8bit_x86_64;
656    
657                    /* Interpolation operators */
658                    interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_x86_64;
659                    interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_x86_64;
660                    interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_x86_64;
661    
662                    interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_x86_64;
663                    interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_x86_64;
664                    interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_x86_64;
665                    interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_x86_64;
666    
667                    interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_x86_64;
668                    interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_x86_64;
669    
670                    interpolate8x8_avg2 = interpolate8x8_avg2_x86_64;
671                    interpolate8x8_avg4 = interpolate8x8_avg4_x86_64;
672    
673                    /* Quantization related functions */
674                    quant_h263_intra   = quant_h263_intra_x86_64;
675                    quant_h263_inter   = quant_h263_inter_x86_64;
676                    dequant_h263_intra = dequant_h263_intra_x86_64;
677                    dequant_h263_inter = dequant_h263_inter_x86_64;
678                    /*quant_mpeg_intra   = quant_mpeg_intra_x86_64; fix me! */
679                    quant_mpeg_inter   = quant_mpeg_inter_x86_64;
680                    dequant_mpeg_intra   = dequant_mpeg_intra_x86_64;
681                    dequant_mpeg_inter   = dequant_mpeg_inter_x86_64;
682    
683                    /* Block related functions */
684                    transfer_8to16copy  = transfer_8to16copy_x86_64;
685                    transfer_16to8copy  = transfer_16to8copy_x86_64;
686                    transfer_8to16sub   = transfer_8to16sub_x86_64;
687                    transfer_8to16subro = transfer_8to16subro_x86_64;
688                    transfer_8to16sub2  = transfer_8to16sub2_x86_64;
689                    transfer_8to16sub2ro= transfer_8to16sub2ro_x86_64;
690                    transfer_16to8add   = transfer_16to8add_x86_64;
691                    transfer8x8_copy    = transfer8x8_copy_x86_64;
692    
693                    /* Qpel stuff */
694                    xvid_QP_Funcs = &xvid_QP_Funcs_x86_64;
695                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_x86_64;
696    
697                    /* Interlacing Functions */
698                    MBFieldTest = MBFieldTest_x86_64;
699            }
700  #endif  #endif
701    
702    #if defined(_DEBUG)
703        xvid_debug = init->debug;
704  #endif  #endif
705    
706          // API version      return(0);
707          init_param->api_version = API_VERSION;  }
708    
709    
710    static int
711    xvid_gbl_info(xvid_gbl_info_t * info)
712    {
713            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
714                    return XVID_ERR_VERSION;
715    
716          // something clever has to be done for this          info->actual_version = XVID_VERSION;
717          init_param->core_build = 1000;          info->build = "xvid-1.2.0-dev";
718            info->cpu_flags = detect_cpu_flags();
719      info->num_threads = 0;
720    
721          return XVID_ERR_OK;  #if defined(WIN32)
722      {
723        DWORD dwProcessAffinityMask, dwSystemAffinityMask;
724        if (GetProcessAffinityMask(GetCurrentProcess(), &dwProcessAffinityMask, &dwSystemAffinityMask)) {
725          int i;
726          for(i=0; i<32; i++) {
727            if ((dwProcessAffinityMask & (1<<i)))
728              info->num_threads++;
729          }
730        }
731  }  }
732    #endif
733    
734            return 0;
735    }
736    
737    
738    static int
739    xvid_gbl_convert(xvid_gbl_convert_t* convert)
740    {
741            int width;
742            int height;
743            int width2;
744            int height2;
745            IMAGE img;
746    
747            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
748                  return XVID_ERR_VERSION;
749    
750    #if 0
751            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
752    #endif
753            width = convert->width;
754            height = convert->height;
755            width2 = convert->width/2;
756            height2 = convert->height/2;
757    
758            switch (convert->input.csp & ~XVID_CSP_VFLIP)
759            {
760                    case XVID_CSP_YV12 :
761                            img.y = convert->input.plane[0];
762                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
763                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
764                            image_output(&img, width, height, width,
765                                                    (uint8_t**)convert->output.plane, convert->output.stride,
766                                                    convert->output.csp, convert->interlacing);
767                            break;
768    
769                    default :
770                            return XVID_ERR_FORMAT;
771            }
772    
773    
774            emms();
775            return 0;
776    }
777    
778    /*****************************************************************************
779     * XviD Global Entry point
780     *
781     * Well this function initialize all internal function pointers according
782     * to the CPU features forced by the library client or autodetected (depending
783     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
784     * image colorspace transformation tables.
785     *
786     ****************************************************************************/
787    
788    
789    int
790    xvid_global(void *handle,
791                      int opt,
792                      void *param1,
793                      void *param2)
794    {
795            switch(opt)
796            {
797                    case XVID_GBL_INIT :
798                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
799    
800            case XVID_GBL_INFO :
801                return xvid_gbl_info((xvid_gbl_info_t*)param1);
802    
803                    case XVID_GBL_CONVERT :
804                            return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
805    
806                    default :
807                            return XVID_ERR_FAIL;
808            }
809    }
810    
811    /*****************************************************************************
812     * XviD Native decoder entry point
813     *
814     * This function is just a wrapper to all the option cases.
815     *
816     * Returned values : XVID_ERR_FAIL when opt is invalid
817     *                   else returns the wrapped function result
818     *
819     ****************************************************************************/
820    
821  int  int
822  xvid_decore(void *handle,  xvid_decore(void *handle,
# Line 244  Line 825 
825                          void *param2)                          void *param2)
826  {  {
827          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
828          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
829                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
830    
831          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
832                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
833    
834            case XVID_DEC_DECODE:
835                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
836    
837          default:          default:
838                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
839          }          }
840  }  }
841    
842    
843    /*****************************************************************************
844     * XviD Native encoder entry point
845     *
846     * This function is just a wrapper to all the option cases.
847     *
848     * Returned values : XVID_ERR_FAIL when opt is invalid
849     *                   else returns the wrapped function result
850     *
851     ****************************************************************************/
852    
853  int  int
854  xvid_encore(void *handle,  xvid_encore(void *handle,
855                          int opt,                          int opt,
# Line 267  Line 858 
858  {  {
859          switch (opt) {          switch (opt) {
860          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
861                  return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
862                                                            (XVID_ENC_STATS *) param2);                  return enc_encode((Encoder *) handle,
863                                                              (xvid_enc_frame_t *) param1,
864                                                              (xvid_enc_stats_t *) param2);
865    
866          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
867                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
868    
869          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
870                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
871    
872          default:          default:
873                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.15  
changed lines
  Added in v.1.70

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