[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 1.53, Thu Apr 15 12:05:19 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            interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c;
226            interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c;
227            interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c;
228    
229            interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c;
230            interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c;
231            interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c;
232    
233            interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c;
234            interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c;
235    
236            interpolate8x8_avg2 = interpolate8x8_avg2_c;
237            interpolate8x8_avg4 = interpolate8x8_avg4_c;
238    
239            /* postprocessing */
240            image_brightness = image_brightness_c;
241    
242            /* reduced resolution */
243            copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C;
244            add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C;
245            vfilter_31 = xvid_VFilter_31_C;
246            hfilter_31 = xvid_HFilter_31_C;
247            filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C;
248            filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C;
249    
250          /* Initialize internal colorspace transformation tables */          /* Initialize internal colorspace transformation tables */
251          colorspace_init();          colorspace_init();
252    
253          /* All colorspace transformation functions User Format->YV12 */          /* All colorspace transformation functions User Format->YV12 */
254            yv12_to_yv12    = yv12_to_yv12_c;
255          rgb555_to_yv12 = rgb555_to_yv12_c;          rgb555_to_yv12 = rgb555_to_yv12_c;
256          rgb565_to_yv12 = rgb565_to_yv12_c;          rgb565_to_yv12 = rgb565_to_yv12_c;
257          rgb24_to_yv12  = rgb24_to_yv12_c;          bgr_to_yv12     = bgr_to_yv12_c;
258          rgb32_to_yv12  = rgb32_to_yv12_c;          bgra_to_yv12    = bgra_to_yv12_c;
259          yuv_to_yv12    = yuv_to_yv12_c;          abgr_to_yv12    = abgr_to_yv12_c;
260            rgba_to_yv12    = rgba_to_yv12_c;
261            argb_to_yv12    = argb_to_yv12_c;
262          yuyv_to_yv12   = yuyv_to_yv12_c;          yuyv_to_yv12   = yuyv_to_yv12_c;
263          uyvy_to_yv12   = uyvy_to_yv12_c;          uyvy_to_yv12   = uyvy_to_yv12_c;
264    
265            rgb555i_to_yv12 = rgb555i_to_yv12_c;
266            rgb565i_to_yv12 = rgb565i_to_yv12_c;
267            bgri_to_yv12    = bgri_to_yv12_c;
268            bgrai_to_yv12   = bgrai_to_yv12_c;
269            abgri_to_yv12   = abgri_to_yv12_c;
270            rgbai_to_yv12   = rgbai_to_yv12_c;
271            argbi_to_yv12   = argbi_to_yv12_c;
272            yuyvi_to_yv12   = yuyvi_to_yv12_c;
273            uyvyi_to_yv12   = uyvyi_to_yv12_c;
274    
275          /* All colorspace transformation functions YV12->User format */          /* All colorspace transformation functions YV12->User format */
276          yv12_to_rgb555 = yv12_to_rgb555_c;          yv12_to_rgb555 = yv12_to_rgb555_c;
277          yv12_to_rgb565 = yv12_to_rgb565_c;          yv12_to_rgb565 = yv12_to_rgb565_c;
278          yv12_to_rgb24  = yv12_to_rgb24_c;          yv12_to_bgr     = yv12_to_bgr_c;
279          yv12_to_rgb32  = yv12_to_rgb32_c;          yv12_to_bgra    = yv12_to_bgra_c;
280          yv12_to_yuv    = yv12_to_yuv_c;          yv12_to_abgr    = yv12_to_abgr_c;
281            yv12_to_rgba    = yv12_to_rgba_c;
282            yv12_to_argb    = yv12_to_argb_c;
283          yv12_to_yuyv   = yv12_to_yuyv_c;          yv12_to_yuyv   = yv12_to_yuyv_c;
284          yv12_to_uyvy   = yv12_to_uyvy_c;          yv12_to_uyvy   = yv12_to_uyvy_c;
285    
286            yv12_to_rgb555i = yv12_to_rgb555i_c;
287            yv12_to_rgb565i = yv12_to_rgb565i_c;
288            yv12_to_bgri    = yv12_to_bgri_c;
289            yv12_to_bgrai   = yv12_to_bgrai_c;
290            yv12_to_abgri   = yv12_to_abgri_c;
291            yv12_to_rgbai   = yv12_to_rgbai_c;
292            yv12_to_argbi   = yv12_to_argbi_c;
293            yv12_to_yuyvi   = yv12_to_yuyvi_c;
294            yv12_to_uyvyi   = yv12_to_uyvyi_c;
295    
296          /* Functions used in motion estimation algorithms */          /* Functions used in motion estimation algorithms */
297          calc_cbp = calc_cbp_c;          calc_cbp = calc_cbp_c;
298          sad16    = sad16_c;          sad16    = sad16_c;
         sad16bi  = sad16bi_c;  
299          sad8     = sad8_c;          sad8     = sad8_c;
300            sad16bi    = sad16bi_c;
301            sad8bi     = sad8bi_c;
302          dev16    = dev16_c;          dev16    = dev16_c;
303            sad16v     = sad16v_c;
304            sse8_16bit = sse8_16bit_c;
305            sse8_8bit  = sse8_8bit_c;
306    
307    #if defined(ARCH_IS_IA32)
308    
309            if ((cpu_flags & XVID_CPU_ASM)) {
310                    vfilter_31 = xvid_VFilter_31_x86;
311                    hfilter_31 = xvid_HFilter_31_x86;
312            }
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;
320            }
321    
322  #ifdef ARCH_X86          if ((cpu_flags & XVID_CPU_MMX)) {
         if ((cpu_flags & XVID_CPU_MMX) > 0) {  
323    
324                  /* Forward and Inverse Discrete Cosine Transformation functions */                  /* Forward and Inverse Discrete Cosine Transformation functions */
325                  fdct = fdct_mmx;                  fdct = fdct_mmx_skal;
326                  idct = idct_mmx;                  idct = idct_mmx;
327    
328                  /* To restore FPU context after mmx use */                  /* Qpel stuff */
329                  emms = emms_mmx;                  xvid_QP_Funcs = &xvid_QP_Funcs_mmx;
330                    xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx;
331    
332                  /* Quantization related functions */                  /* Quantization related functions */
333                  quant_intra   = quant_intra_mmx;                  quant_h263_intra   = quant_h263_intra_mmx;
334                  dequant_intra = dequant_intra_mmx;                  quant_h263_inter   = quant_h263_inter_mmx;
335                  quant_inter   = quant_inter_mmx;                  dequant_h263_intra = dequant_h263_intra_mmx;
336                  dequant_inter = dequant_inter_mmx;                  dequant_h263_inter = dequant_h263_inter_mmx;
337    
338                  quant4_intra   = quant4_intra_mmx;                  quant_mpeg_intra   = quant_mpeg_intra_mmx;
339                  dequant4_intra = dequant4_intra_mmx;                  quant_mpeg_inter   = quant_mpeg_inter_mmx;
340                  quant4_inter   = quant4_inter_mmx;                  dequant_mpeg_intra = dequant_mpeg_intra_mmx;
341                  dequant4_inter = dequant4_inter_mmx;                  dequant_mpeg_inter = dequant_mpeg_inter_mmx;
342    
343                  /* Block related functions */                  /* 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    
352                    /* Interlacing Functions */
353                    MBFieldTest = MBFieldTest_mmx;
354    
355                  /* Image Interpolation related functions */                  /* Image Interpolation related functions */
356                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;                  interpolate8x8_halfpel_h  = interpolate8x8_halfpel_h_mmx;
357                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_mmx;
358                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx;
359    
360                  /* Image RGB->YV12 related functions */                  interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx;
361                  rgb24_to_yv12 = rgb24_to_yv12_mmx;                  interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx;
362                  rgb32_to_yv12 = rgb32_to_yv12_mmx;  
363                  yuv_to_yv12   = yuv_to_yv12_mmx;                  interpolate8x8_avg2 = interpolate8x8_avg2_mmx;
364                    interpolate8x8_avg4 = interpolate8x8_avg4_mmx;
365    
366                    /* postprocessing */
367                    image_brightness = image_brightness_mmx;
368    
369                    /* reduced resolution */
370                    copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx;
371                    add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx;
372                    hfilter_31 = xvid_HFilter_31_mmx;
373                    filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx;
374                    filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx;
375    
376                    /* image input xxx_to_yv12 related functions */
377                    yv12_to_yv12  = yv12_to_yv12_mmx;
378                    bgr_to_yv12   = bgr_to_yv12_mmx;
379                    bgra_to_yv12  = bgra_to_yv12_mmx;
380                  yuyv_to_yv12  = yuyv_to_yv12_mmx;                  yuyv_to_yv12  = yuyv_to_yv12_mmx;
381                  uyvy_to_yv12  = uyvy_to_yv12_mmx;                  uyvy_to_yv12  = uyvy_to_yv12_mmx;
382    
383                  /* Image YV12->RGB related functions */                  /* image output yv12_to_xxx related functions */
384                  yv12_to_rgb24 = yv12_to_rgb24_mmx;                  yv12_to_bgr   = yv12_to_bgr_mmx;
385                  yv12_to_rgb32 = yv12_to_rgb32_mmx;                  yv12_to_bgra  = yv12_to_bgra_mmx;
386                  yv12_to_yuyv  = yv12_to_yuyv_mmx;                  yv12_to_yuyv  = yv12_to_yuyv_mmx;
387                  yv12_to_uyvy  = yv12_to_uyvy_mmx;                  yv12_to_uyvy  = yv12_to_uyvy_mmx;
388    
389                    yv12_to_yuyvi = yv12_to_yuyvi_mmx;
390                    yv12_to_uyvyi = yv12_to_uyvyi_mmx;
391    
392                  /* Motion estimation related functions */                  /* Motion estimation related functions */
393                  calc_cbp = calc_cbp_mmx;                  calc_cbp = calc_cbp_mmx;
394                  sad16    = sad16_mmx;                  sad16    = sad16_mmx;
395                  sad8     = sad8_mmx;                  sad8     = sad8_mmx;
396                    sad16bi    = sad16bi_mmx;
397                    sad8bi     = sad8bi_mmx;
398                  dev16    = dev16_mmx;                  dev16    = dev16_mmx;
399                    sad16v     = sad16v_mmx;
400                    sse8_16bit = sse8_16bit_mmx;
401                    sse8_8bit  = sse8_8bit_mmx;
402            }
403    
404            /* these 3dnow functions are faster than mmx, but slower than xmm. */
405            if ((cpu_flags & XVID_CPU_3DNOW)) {
406    
407                    emms = emms_3dn;
408    
409                    /* ME functions */
410                    sad16bi = sad16bi_3dn;
411                    sad8bi  = sad8bi_3dn;
412    
413                    yuyv_to_yv12  = yuyv_to_yv12_3dn;
414                    uyvy_to_yv12  = uyvy_to_yv12_3dn;
415          }          }
416    
         if ((cpu_flags & XVID_CPU_MMXEXT) > 0) {  
417    
418                  /* Inverse DCT */          if ((cpu_flags & XVID_CPU_MMXEXT)) {
419    
420                    /* DCT */
421                    fdct = fdct_xmm_skal;
422                  idct = idct_xmm;                  idct = idct_xmm;
423    
424                  /* Interpolation */                  /* Interpolation */
# Line 223  Line 426 
426                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;                  interpolate8x8_halfpel_v  = interpolate8x8_halfpel_v_xmm;
427                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm;
428    
429                    /* reduced resolution */
430                    copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm;
431                    add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm;
432    
433                    /* Quantization */
434                    quant_mpeg_intra = quant_mpeg_intra_xmm;
435                    quant_mpeg_inter = quant_mpeg_inter_xmm;
436    
437                    dequant_h263_intra = dequant_h263_intra_xmm;
438                    dequant_h263_inter = dequant_h263_inter_xmm;
439    
440                  /* Buffer transfer */                  /* Buffer transfer */
441                  transfer_8to16sub2 = transfer_8to16sub2_xmm;                  transfer_8to16sub2 = transfer_8to16sub2_xmm;
442    
443                  /* Colorspace transformation */                  /* Colorspace transformation */
444                  yuv_to_yv12 = yuv_to_yv12_xmm;                  yv12_to_yv12  = yv12_to_yv12_xmm;
445                    yuyv_to_yv12  = yuyv_to_yv12_xmm;
446                    uyvy_to_yv12  = uyvy_to_yv12_xmm;
447    
448                  /* ME functions */                  /* ME functions */
449                  sad16 = sad16_xmm;                  sad16 = sad16_xmm;
450                  sad8  = sad8_xmm;                  sad8  = sad8_xmm;
451                    sad16bi = sad16bi_xmm;
452                    sad8bi  = sad8bi_xmm;
453                  dev16 = dev16_xmm;                  dev16 = dev16_xmm;
454                    sad16v   = sad16v_xmm;
455          }          }
456    
457          if ((cpu_flags & XVID_CPU_3DNOW) > 0) {          if ((cpu_flags & XVID_CPU_3DNOW)) {
458    
459                  /* Interpolation */                  /* Interpolation */
460                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;                  interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn;
# Line 244  Line 462 
462                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;                  interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn;
463          }          }
464    
465          if ((cpu_flags & XVID_CPU_SSE2) > 0) {          if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
466  #ifdef EXPERIMENTAL_SSE2_CODE  
467                    /* Buffer transfer */
468                    transfer_8to16copy =  transfer_8to16copy_3dne;
469                    transfer_16to8copy = transfer_16to8copy_3dne;
470                    transfer_8to16sub =  transfer_8to16sub_3dne;
471                    transfer_8to16subro =  transfer_8to16subro_3dne;
472                    transfer_16to8add = transfer_16to8add_3dne;
473                    transfer8x8_copy = transfer8x8_copy_3dne;
474    
475                    if ((cpu_flags & XVID_CPU_MMXEXT)) {
476                            /* Inverse DCT */
477                            idct =  idct_3dne;
478    
479                            /* Buffer transfer */
480                            transfer_8to16sub2 =  transfer_8to16sub2_3dne;
481    
482                            /* Interpolation */
483                            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
484                            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
485                            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
486    
487                  /* Quantization */                  /* Quantization */
488                  quant_intra   = quant_intra_sse2;                          quant_h263_intra = quant_h263_intra_3dne;               /* cmov only */
489                  dequant_intra = dequant_intra_sse2;                          quant_h263_inter = quant_h263_inter_3dne;
490                  quant_inter   = quant_inter_sse2;                          dequant_mpeg_intra = dequant_mpeg_intra_3dne;   /* cmov only */
491                  dequant_inter = dequant_inter_sse2;                          dequant_mpeg_inter = dequant_mpeg_inter_3dne;
492                            dequant_h263_intra = dequant_h263_intra_3dne;
493                            dequant_h263_inter = dequant_h263_inter_3dne;
494    
495                            /* ME functions */
496                            calc_cbp = calc_cbp_3dne;
497    
498                            sad16 = sad16_3dne;
499                            sad8 = sad8_3dne;
500                            sad16bi = sad16bi_3dne;
501                            sad8bi = sad8bi_3dne;
502                            dev16 = dev16_3dne;
503                    }
504            }
505    
506            if ((cpu_flags & XVID_CPU_SSE2)) {
507    
                 /* ME */  
508                  calc_cbp = calc_cbp_sse2;                  calc_cbp = calc_cbp_sse2;
509    
510                    /* Quantization */
511                    quant_h263_intra   = quant_h263_intra_sse2;
512                    quant_h263_inter   = quant_h263_inter_sse2;
513                    dequant_h263_intra = dequant_h263_intra_sse2;
514                    dequant_h263_inter = dequant_h263_inter_sse2;
515    
516                    /* SAD operators */
517                  sad16    = sad16_sse2;                  sad16    = sad16_sse2;
518                  dev16    = dev16_sse2;                  dev16    = dev16_sse2;
519    
520                  /* Forward and Inverse DCT */                  /* DCT operators
521                  idct  = idct_sse2;                   * no iDCT because it's not "Walken matching" */
522                  fdct = fdct_sse2;                  fdct = fdct_sse2_skal;
 #endif  
523          }          }
524    #endif /* ARCH_IS_IA32 */
525    
526  #endif  #if defined(ARCH_IS_IA64)
527            if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */
 #ifdef ARCH_IA64  
         if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines?  
528            idct_ia64_init();            idct_ia64_init();
529            fdct = fdct_ia64;            fdct = fdct_ia64;
530            idct = idct_ia64;   //not yet working, crashes            idct = idct_ia64;   /*not yet working, crashes */
531            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;            interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64;
532            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;            interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64;
533            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;            interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64;
# Line 278  Line 535 
535            sad16bi = sad16bi_ia64;            sad16bi = sad16bi_ia64;
536            sad8 = sad8_ia64;            sad8 = sad8_ia64;
537            dev16 = dev16_ia64;            dev16 = dev16_ia64;
538            quant_intra = quant_intra_ia64;  /*        Halfpel8_Refine = Halfpel8_Refine_ia64; */
539            dequant_intra = dequant_intra_ia64;            quant_h263_intra = quant_h263_intra_ia64;
540            quant_inter = quant_inter_ia64;            quant_h263_inter = quant_h263_inter_ia64;
541            dequant_inter = dequant_inter_ia64;            dequant_h263_intra = dequant_h263_intra_ia64;
542              dequant_h263_inter = dequant_h263_inter_ia64;
543            transfer_8to16copy = transfer_8to16copy_ia64;            transfer_8to16copy = transfer_8to16copy_ia64;
544            transfer_16to8copy = transfer_16to8copy_ia64;            transfer_16to8copy = transfer_16to8copy_ia64;
545            transfer_8to16sub = transfer_8to16sub_ia64;            transfer_8to16sub = transfer_8to16sub_ia64;
546            transfer_8to16sub2 = transfer_8to16sub2_ia64;            transfer_8to16sub2 = transfer_8to16sub2_ia64;
547            transfer_16to8add = transfer_16to8add_ia64;            transfer_16to8add = transfer_16to8add_ia64;
548            transfer8x8_copy = transfer8x8_copy_ia64;            transfer8x8_copy = transfer8x8_copy_ia64;
           DEBUG("Using IA-64 assembler routines.\n");  
549          }          }
550  #endif  #endif
551    
552  #ifdef ARCH_PPC  #if defined(ARCH_IS_PPC)
553  #ifdef ARCH_PPC_ALTIVEC          if ((cpu_flags & XVID_CPU_ALTIVEC)) {
554          calc_cbp = calc_cbp_altivec;            /* sad operators */
555          fdct = fdct_altivec;            sad16 = sad16_altivec_c;
556          idct = idct_altivec;            sad16bi = sad16bi_altivec_c;
557          sadInit = sadInit_altivec;            sad8 = sad8_altivec_c;
558          sad16 = sad16_altivec;            dev16 = dev16_altivec_c;
559          sad8 = sad8_altivec;  
560          dev16 = dev16_altivec;            sse8_16bit = sse8_16bit_altivec_c;
561    
562              /* mem transfer */
563              transfer_8to16copy = transfer_8to16copy_altivec_c;
564              transfer_16to8copy = transfer_16to8copy_altivec_c;
565              transfer_8to16sub = transfer_8to16sub_altivec_c;
566              transfer_8to16subro = transfer_8to16subro_altivec_c;
567              transfer_8to16sub2 = transfer_8to16sub2_altivec_c;
568              transfer_16to8add = transfer_16to8add_altivec_c;
569              transfer8x8_copy = transfer8x8_copy_altivec_c;
570    
571              /* Inverse DCT */
572              idct = idct_altivec_c;
573    
574              /* Interpolation */
575              interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c;
576              interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c;
577              interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c;
578    
579              interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c;
580              interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c;
581    
582              interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_altivec_c;
583    
584              /* Colorspace conversion */
585              bgra_to_yv12 = bgra_to_yv12_altivec_c;
586              abgr_to_yv12 = abgr_to_yv12_altivec_c;
587              rgba_to_yv12 = rgba_to_yv12_altivec_c;
588              argb_to_yv12 = argb_to_yv12_altivec_c;
589    
590              yuyv_to_yv12 = yuyv_to_yv12_altivec_c;
591              uyvy_to_yv12 = uyvy_to_yv12_altivec_c;
592    
593              yv12_to_yuyv = yv12_to_yuyv_altivec_c;
594              yv12_to_uyvy = yv12_to_uyvy_altivec_c;
595    
596              /* Quantization */
597              quant_h263_intra = quant_h263_intra_altivec_c;
598              quant_h263_inter = quant_h263_inter_altivec_c;
599              dequant_h263_intra = dequant_h263_intra_altivec_c;
600              dequant_h263_inter = dequant_h263_inter_altivec_c;
601            }
602    #endif
603    
604    #if defined(_DEBUG)
605        xvid_debug = init->debug;
606    #endif
607    
608        return(0);
609    }
610    
611    
612    static int
613    xvid_gbl_info(xvid_gbl_info_t * info)
614    {
615            if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */
616                    return XVID_ERR_VERSION;
617    
618            info->actual_version = XVID_VERSION;
619            info->build = "xvid-1.1-cvshead";
620            info->cpu_flags = detect_cpu_flags();
621    
622    #if defined(_SMP) && defined(WIN32)
623            info->num_threads = pthread_num_processors_np();;
624  #else  #else
625          calc_cbp = calc_cbp_ppc;          info->num_threads = 0;
626  #endif  #endif
627    
628            return 0;
629    }
630    
631    
632    static int
633    xvid_gbl_convert(xvid_gbl_convert_t* convert)
634    {
635            int width;
636            int height;
637            int width2;
638            int height2;
639            IMAGE img;
640    
641            if (XVID_VERSION_MAJOR(convert->version) != 1)   /* v1.x.x */
642                  return XVID_ERR_VERSION;
643    
644    #if 0
645            const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP);
646  #endif  #endif
647            width = convert->width;
648            height = convert->height;
649            width2 = convert->width/2;
650            height2 = convert->height/2;
651    
652            switch (convert->input.csp & ~XVID_CSP_VFLIP)
653            {
654                    case XVID_CSP_YV12 :
655                            img.y = convert->input.plane[0];
656                            img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height;
657                            img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2;
658                            image_output(&img, width, height, width,
659                                                    (uint8_t**)convert->output.plane, convert->output.stride,
660                                                    convert->output.csp, convert->interlacing);
661                            break;
662    
663                    default :
664                            return XVID_ERR_FORMAT;
665            }
666    
667    
668            emms();
669            return 0;
670    }
671    
672    /*****************************************************************************
673     * XviD Global Entry point
674     *
675     * Well this function initialize all internal function pointers according
676     * to the CPU features forced by the library client or autodetected (depending
677     * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all
678     * image colorspace transformation tables.
679     *
680     ****************************************************************************/
681    
682    
683    int
684    xvid_global(void *handle,
685                      int opt,
686                      void *param1,
687                      void *param2)
688    {
689            switch(opt)
690            {
691                    case XVID_GBL_INIT :
692                            return xvid_gbl_init((xvid_gbl_init_t*)param1);
693    
694          /* Inform the client the API version */          case XVID_GBL_INFO :
695          init_param->api_version = API_VERSION;              return xvid_gbl_info((xvid_gbl_info_t*)param1);
696    
697          /* Inform the client the core build - unused because we're still alpha */                  case XVID_GBL_CONVERT :
698          init_param->core_build = 1000;                          return xvid_gbl_convert((xvid_gbl_convert_t*)param1);
699    
700          return XVID_ERR_OK;                  default :
701                            return XVID_ERR_FAIL;
702            }
703  }  }
704    
705  /*****************************************************************************  /*****************************************************************************
# Line 332  Line 719 
719                          void *param2)                          void *param2)
720  {  {
721          switch (opt) {          switch (opt) {
         case XVID_DEC_DECODE:  
                 return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1);  
   
722          case XVID_DEC_CREATE:          case XVID_DEC_CREATE:
723                  return decoder_create((XVID_DEC_PARAM *) param1);                  return decoder_create((xvid_dec_create_t *) param1);
724    
725          case XVID_DEC_DESTROY:          case XVID_DEC_DESTROY:
726                  return decoder_destroy((DECODER *) handle);                  return decoder_destroy((DECODER *) handle);
727    
728            case XVID_DEC_DECODE:
729                    return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2);
730    
731          default:          default:
732                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;
733          }          }
# Line 365  Line 752 
752  {  {
753          switch (opt) {          switch (opt) {
754          case XVID_ENC_ENCODE:          case XVID_ENC_ENCODE:
755  #ifdef BFRAMES  
756                  if (((Encoder *) handle)->mbParam.max_bframes >= 0)                  return enc_encode((Encoder *) handle,
757                  return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1,                                                            (xvid_enc_frame_t *) param1,
758                                                            (XVID_ENC_STATS *) param2);                                                            (xvid_enc_stats_t *) param2);
                 else  
 #endif  
                 return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1,  
                                                           (XVID_ENC_STATS *) param2);  
759    
760          case XVID_ENC_CREATE:          case XVID_ENC_CREATE:
761                  return encoder_create((XVID_ENC_PARAM *) param1);                  return enc_create((xvid_enc_create_t *) param1);
762    
763          case XVID_ENC_DESTROY:          case XVID_ENC_DESTROY:
764                  return encoder_destroy((Encoder *) handle);                  return enc_destroy((Encoder *) handle);
765    
766          default:          default:
767                  return XVID_ERR_FAIL;                  return XVID_ERR_FAIL;

Legend:
Removed from v.1.20  
changed lines
  Added in v.1.53

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