Parent Directory
| 
Revision Log
Revision 698 - (view) (download)
| 1 : | edgomez | 200 | /***************************************************************************** |
| 2 : | edgomez | 201 | * |
| 3 : | * XVID MPEG-4 VIDEO CODEC | ||
| 4 : | * - Native API implementation - | ||
| 5 : | * | ||
| 6 : | * This program is an implementation of a part of one or more MPEG-4 | ||
| 7 : | * Video tools as specified in ISO/IEC 14496-2 standard. Those intending | ||
| 8 : | * to use this software module in hardware or software products are | ||
| 9 : | * advised that its use may infringe existing patents or copyrights, and | ||
| 10 : | * any such use would be at such party's own risk. The original | ||
| 11 : | * developer of this software module and his/her company, and subsequent | ||
| 12 : | * editors and their companies, will have no liability for use of this | ||
| 13 : | * software or modifications or derivatives thereof. | ||
| 14 : | * | ||
| 15 : | * This program is free software ; you can redistribute it and/or modify | ||
| 16 : | * it under the terms of the GNU General Public License as published by | ||
| 17 : | * the Free Software Foundation ; either version 2 of the License, or | ||
| 18 : | * (at your option) any later version. | ||
| 19 : | * | ||
| 20 : | * This program is distributed in the hope that it will be useful, | ||
| 21 : | * but WITHOUT ANY WARRANTY ; without even the implied warranty of | ||
| 22 : | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
| 23 : | * GNU General Public License for more details. | ||
| 24 : | * | ||
| 25 : | * You should have received a copy of the GNU General Public License | ||
| 26 : | * along with this program ; if not, write to the Free Software | ||
| 27 : | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
| 28 : | * | ||
| 29 : | ****************************************************************************/ | ||
| 30 : | chenm001 | 274 | |
| 31 : | edgomez | 200 | /***************************************************************************** |
| 32 : | edgomez | 201 | * |
| 33 : | * History | ||
| 34 : | * | ||
| 35 : | suxen_drol | 234 | * - 23.06.2002 added XVID_CPU_CHKONLY |
| 36 : | edgomez | 201 | * - 17.03.2002 Added interpolate8x8_halfpel_hv_xmm |
| 37 : | * - 22.12.2001 API change: added xvid_init() - Isibaar | ||
| 38 : | * - 16.12.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> | ||
| 39 : | * | ||
| 40 : | suxen_drol | 698 | * $Id: xvid.c,v 1.33.2.16 2002-12-08 06:43:34 suxen_drol Exp $ |
| 41 : | edgomez | 201 | * |
| 42 : | ****************************************************************************/ | ||
| 43 : | Isibaar | 3 | |
| 44 : | #include "xvid.h" | ||
| 45 : | #include "decoder.h" | ||
| 46 : | #include "encoder.h" | ||
| 47 : | #include "bitstream/cbp.h" | ||
| 48 : | #include "dct/idct.h" | ||
| 49 : | #include "dct/fdct.h" | ||
| 50 : | #include "image/colorspace.h" | ||
| 51 : | #include "image/interpolate8x8.h" | ||
| 52 : | suxen_drol | 695 | #include "image/reduced.h" |
| 53 : | Isibaar | 3 | #include "utils/mem_transfer.h" |
| 54 : | h | 540 | #include "utils/mbfunctions.h" |
| 55 : | Isibaar | 3 | #include "quant/quant_h263.h" |
| 56 : | #include "quant/quant_mpeg4.h" | ||
| 57 : | ia64p | 299 | #include "motion/motion.h" |
| 58 : | Isibaar | 3 | #include "motion/sad.h" |
| 59 : | #include "utils/emms.h" | ||
| 60 : | #include "utils/timer.h" | ||
| 61 : | Isibaar | 100 | #include "bitstream/mbcoding.h" |
| 62 : | Isibaar | 3 | |
| 63 : | suxen_drol | 311 | #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE) |
| 64 : | |||
| 65 : | #ifdef WIN32 | ||
| 66 : | #include <windows.h> | ||
| 67 : | #else | ||
| 68 : | #include <signal.h> | ||
| 69 : | #include <setjmp.h> | ||
| 70 : | #endif | ||
| 71 : | |||
| 72 : | |||
| 73 : | #ifndef WIN32 | ||
| 74 : | |||
| 75 : | static jmp_buf mark; | ||
| 76 : | |||
| 77 : | static void | ||
| 78 : | sigill_handler(int signal) | ||
| 79 : | { | ||
| 80 : | longjmp(mark, 1); | ||
| 81 : | } | ||
| 82 : | #endif | ||
| 83 : | |||
| 84 : | |||
| 85 : | /* | ||
| 86 : | calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled | ||
| 87 : | return values: | ||
| 88 : | -1 : could not determine | ||
| 89 : | 0 : SIGILL was *not* signalled | ||
| 90 : | 1 : SIGILL was signalled | ||
| 91 : | */ | ||
| 92 : | |||
| 93 : | int | ||
| 94 : | sigill_check(void (*func)()) | ||
| 95 : | { | ||
| 96 : | #ifdef WIN32 | ||
| 97 : | _try { | ||
| 98 : | func(); | ||
| 99 : | } | ||
| 100 : | _except(EXCEPTION_EXECUTE_HANDLER) { | ||
| 101 : | |||
| 102 : | if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION) | ||
| 103 : | return 1; | ||
| 104 : | } | ||
| 105 : | return 0; | ||
| 106 : | #else | ||
| 107 : | void * old_handler; | ||
| 108 : | int jmpret; | ||
| 109 : | |||
| 110 : | |||
| 111 : | old_handler = signal(SIGILL, sigill_handler); | ||
| 112 : | if (old_handler == SIG_ERR) | ||
| 113 : | { | ||
| 114 : | return -1; | ||
| 115 : | } | ||
| 116 : | |||
| 117 : | jmpret = setjmp(mark); | ||
| 118 : | if (jmpret == 0) | ||
| 119 : | { | ||
| 120 : | func(); | ||
| 121 : | } | ||
| 122 : | |||
| 123 : | signal(SIGILL, old_handler); | ||
| 124 : | |||
| 125 : | return jmpret; | ||
| 126 : | #endif | ||
| 127 : | } | ||
| 128 : | #endif | ||
| 129 : | |||
| 130 : | edgomez | 200 | /***************************************************************************** |
| 131 : | * XviD Init Entry point | ||
| 132 : | * | ||
| 133 : | * Well this function initialize all internal function pointers according | ||
| 134 : | * to the CPU features forced by the library client or autodetected (depending | ||
| 135 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
| 136 : | * image colorspace transformation tables. | ||
| 137 : | * | ||
| 138 : | * Returned value : XVID_ERR_OK | ||
| 139 : | * + API_VERSION in the input XVID_INIT_PARAM structure | ||
| 140 : | * + core build " " " " " | ||
| 141 : | * | ||
| 142 : | ****************************************************************************/ | ||
| 143 : | |||
| 144 : | suxen_drol | 634 | |
| 145 : | static | ||
| 146 : | int xvid_init_init(XVID_INIT_PARAM * init_param) | ||
| 147 : | Isibaar | 3 | { |
| 148 : | int cpu_flags; | ||
| 149 : | |||
| 150 : | suxen_drol | 234 | /* Inform the client the API version */ |
| 151 : | init_param->api_version = API_VERSION; | ||
| 152 : | |||
| 153 : | /* Inform the client the core build - unused because we're still alpha */ | ||
| 154 : | init_param->core_build = 1000; | ||
| 155 : | |||
| 156 : | suxen_drol | 311 | /* Do we have to force CPU features ? */ |
| 157 : | if ((init_param->cpu_flags & XVID_CPU_FORCE)) { | ||
| 158 : | suxen_drol | 234 | |
| 159 : | Isibaar | 3 | cpu_flags = init_param->cpu_flags; |
| 160 : | suxen_drol | 311 | |
| 161 : | edgomez | 200 | } else { |
| 162 : | Isibaar | 3 | |
| 163 : | chenm001 | 274 | cpu_flags = check_cpu_features(); |
| 164 : | suxen_drol | 311 | |
| 165 : | #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE) | ||
| 166 : | if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) | ||
| 167 : | cpu_flags &= ~XVID_CPU_SSE; | ||
| 168 : | |||
| 169 : | if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) | ||
| 170 : | cpu_flags &= ~XVID_CPU_SSE2; | ||
| 171 : | #endif | ||
| 172 : | } | ||
| 173 : | |||
| 174 : | if ((init_param->cpu_flags & XVID_CPU_CHKONLY)) | ||
| 175 : | { | ||
| 176 : | Isibaar | 3 | init_param->cpu_flags = cpu_flags; |
| 177 : | suxen_drol | 311 | return XVID_ERR_OK; |
| 178 : | Isibaar | 3 | } |
| 179 : | |||
| 180 : | suxen_drol | 311 | init_param->cpu_flags = cpu_flags; |
| 181 : | |||
| 182 : | |||
| 183 : | edgomez | 200 | /* Initialize the function pointers */ |
| 184 : | Isibaar | 3 | idct_int32_init(); |
| 185 : | Isibaar | 100 | init_vlc_tables(); |
| 186 : | |||
| 187 : | edgomez | 200 | /* Fixed Point Forward/Inverse DCT transformations */ |
| 188 : | Isibaar | 3 | fdct = fdct_int32; |
| 189 : | idct = idct_int32; | ||
| 190 : | |||
| 191 : | edgomez | 200 | /* Only needed on PPC Altivec archs */ |
| 192 : | canard | 115 | sadInit = 0; |
| 193 : | edgomez | 195 | |
| 194 : | edgomez | 200 | /* Restore FPU context : emms_c is a nop functions */ |
| 195 : | Isibaar | 3 | emms = emms_c; |
| 196 : | |||
| 197 : | edgomez | 200 | /* Quantization functions */ |
| 198 : | quant_intra = quant_intra_c; | ||
| 199 : | Isibaar | 3 | dequant_intra = dequant_intra_c; |
| 200 : | edgomez | 200 | quant_inter = quant_inter_c; |
| 201 : | Isibaar | 3 | dequant_inter = dequant_inter_c; |
| 202 : | |||
| 203 : | edgomez | 200 | quant4_intra = quant4_intra_c; |
| 204 : | Isibaar | 3 | dequant4_intra = dequant4_intra_c; |
| 205 : | edgomez | 200 | quant4_inter = quant4_inter_c; |
| 206 : | Isibaar | 3 | dequant4_inter = dequant4_inter_c; |
| 207 : | |||
| 208 : | edgomez | 200 | /* Block transfer related functions */ |
| 209 : | Isibaar | 3 | transfer_8to16copy = transfer_8to16copy_c; |
| 210 : | transfer_16to8copy = transfer_16to8copy_c; | ||
| 211 : | edgomez | 200 | transfer_8to16sub = transfer_8to16sub_c; |
| 212 : | suxen_drol | 118 | transfer_8to16sub2 = transfer_8to16sub2_c; |
| 213 : | edgomez | 200 | transfer_16to8add = transfer_16to8add_c; |
| 214 : | transfer8x8_copy = transfer8x8_copy_c; | ||
| 215 : | Isibaar | 3 | |
| 216 : | h | 540 | /* Interlacing functions */ |
| 217 : | MBFieldTest = MBFieldTest_c; | ||
| 218 : | |||
| 219 : | edgomez | 200 | /* Image interpolation related functions */ |
| 220 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; | ||
| 221 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; | ||
| 222 : | Isibaar | 3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
| 223 : | |||
| 224 : | Isibaar | 588 | interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; |
| 225 : | interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; | ||
| 226 : | interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; | ||
| 227 : | |||
| 228 : | Isibaar | 579 | interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c; |
| 229 : | interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c; | ||
| 230 : | interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c; | ||
| 231 : | |||
| 232 : | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c; | ||
| 233 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c; | ||
| 234 : | |||
| 235 : | interpolate8x8_avg2 = interpolate8x8_avg2_c; | ||
| 236 : | interpolate8x8_avg4 = interpolate8x8_avg4_c; | ||
| 237 : | |||
| 238 : | suxen_drol | 695 | /* reduced resoltuion */ |
| 239 : | |||
| 240 : | #ifdef ARCH_X86 | ||
| 241 : | vfilter_31 = xvid_VFilter_31_x86; | ||
| 242 : | hfilter_31 = xvid_HFilter_31_x86; | ||
| 243 : | #else | ||
| 244 : | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; | ||
| 245 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; | ||
| 246 : | vfilter_31 = xvid_VFilter_31_C; | ||
| 247 : | hfilter_31 = xvid_HFilter_31_C; | ||
| 248 : | #endif | ||
| 249 : | |||
| 250 : | edgomez | 200 | /* Initialize internal colorspace transformation tables */ |
| 251 : | Isibaar | 3 | colorspace_init(); |
| 252 : | |||
| 253 : | edgomez | 200 | /* All colorspace transformation functions User Format->YV12 */ |
| 254 : | suxen_drol | 631 | yv12_to_yv12 = yv12_to_yv12_c; |
| 255 : | rgb555_to_yv12 = rgb555_to_yv12_c; | ||
| 256 : | rgb565_to_yv12 = rgb565_to_yv12_c; | ||
| 257 : | bgr_to_yv12 = bgr_to_yv12_c; | ||
| 258 : | bgra_to_yv12 = bgra_to_yv12_c; | ||
| 259 : | abgr_to_yv12 = abgr_to_yv12_c; | ||
| 260 : | rgba_to_yv12 = rgba_to_yv12_c; | ||
| 261 : | yuyv_to_yv12 = yuyv_to_yv12_c; | ||
| 262 : | uyvy_to_yv12 = uyvy_to_yv12_c; | ||
| 263 : | Isibaar | 3 | |
| 264 : | suxen_drol | 631 | rgb555i_to_yv12 = rgb555i_to_yv12_c; |
| 265 : | rgb565i_to_yv12 = rgb565i_to_yv12_c; | ||
| 266 : | bgri_to_yv12 = bgri_to_yv12_c; | ||
| 267 : | bgrai_to_yv12 = bgrai_to_yv12_c; | ||
| 268 : | abgri_to_yv12 = abgri_to_yv12_c; | ||
| 269 : | rgbai_to_yv12 = rgbai_to_yv12_c; | ||
| 270 : | yuyvi_to_yv12 = yuyvi_to_yv12_c; | ||
| 271 : | uyvyi_to_yv12 = uyvyi_to_yv12_c; | ||
| 272 : | |||
| 273 : | |||
| 274 : | edgomez | 200 | /* All colorspace transformation functions YV12->User format */ |
| 275 : | suxen_drol | 631 | yv12_to_rgb555 = yv12_to_rgb555_c; |
| 276 : | yv12_to_rgb565 = yv12_to_rgb565_c; | ||
| 277 : | yv12_to_bgr = yv12_to_bgr_c; | ||
| 278 : | yv12_to_bgra = yv12_to_bgra_c; | ||
| 279 : | yv12_to_abgr = yv12_to_abgr_c; | ||
| 280 : | yv12_to_rgba = yv12_to_rgba_c; | ||
| 281 : | yv12_to_yuyv = yv12_to_yuyv_c; | ||
| 282 : | yv12_to_uyvy = yv12_to_uyvy_c; | ||
| 283 : | |||
| 284 : | yv12_to_rgb555i = yv12_to_rgb555i_c; | ||
| 285 : | yv12_to_rgb565i = yv12_to_rgb565i_c; | ||
| 286 : | yv12_to_bgri = yv12_to_bgri_c; | ||
| 287 : | yv12_to_bgrai = yv12_to_bgrai_c; | ||
| 288 : | yv12_to_abgri = yv12_to_abgri_c; | ||
| 289 : | yv12_to_rgbai = yv12_to_rgbai_c; | ||
| 290 : | yv12_to_yuyvi = yv12_to_yuyvi_c; | ||
| 291 : | yv12_to_uyvyi = yv12_to_uyvyi_c; | ||
| 292 : | Isibaar | 3 | |
| 293 : | edgomez | 200 | /* Functions used in motion estimation algorithms */ |
| 294 : | Isibaar | 3 | calc_cbp = calc_cbp_c; |
| 295 : | edgomez | 200 | sad16 = sad16_c; |
| 296 : | suxen_drol | 329 | sad8 = sad8_c; |
| 297 : | edgomez | 200 | sad16bi = sad16bi_c; |
| 298 : | suxen_drol | 329 | sad8bi = sad8bi_c; |
| 299 : | edgomez | 200 | dev16 = dev16_c; |
| 300 : | chl | 530 | sad16v = sad16v_c; |
| 301 : | suxen_drol | 329 | |
| 302 : | chl | 530 | // Halfpel8_Refine = Halfpel8_Refine_c; |
| 303 : | Isibaar | 3 | |
| 304 : | #ifdef ARCH_X86 | ||
| 305 : | suxen_drol | 631 | |
| 306 : | if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || | ||
| 307 : | (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || | ||
| 308 : | (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2)) | ||
| 309 : | { | ||
| 310 : | /* Restore FPU context : emms_c is a nop functions */ | ||
| 311 : | emms = emms_mmx; | ||
| 312 : | } | ||
| 313 : | |||
| 314 : | edgomez | 195 | if ((cpu_flags & XVID_CPU_MMX) > 0) { |
| 315 : | edgomez | 200 | |
| 316 : | /* Forward and Inverse Discrete Cosine Transformation functions */ | ||
| 317 : | Isibaar | 3 | fdct = fdct_mmx; |
| 318 : | idct = idct_mmx; | ||
| 319 : | |||
| 320 : | edgomez | 200 | /* Quantization related functions */ |
| 321 : | quant_intra = quant_intra_mmx; | ||
| 322 : | Isibaar | 3 | dequant_intra = dequant_intra_mmx; |
| 323 : | edgomez | 200 | quant_inter = quant_inter_mmx; |
| 324 : | Isibaar | 3 | dequant_inter = dequant_inter_mmx; |
| 325 : | |||
| 326 : | edgomez | 200 | quant4_intra = quant4_intra_mmx; |
| 327 : | Isibaar | 3 | dequant4_intra = dequant4_intra_mmx; |
| 328 : | edgomez | 200 | quant4_inter = quant4_inter_mmx; |
| 329 : | Isibaar | 3 | dequant4_inter = dequant4_inter_mmx; |
| 330 : | |||
| 331 : | edgomez | 200 | /* Block related functions */ |
| 332 : | Isibaar | 3 | transfer_8to16copy = transfer_8to16copy_mmx; |
| 333 : | transfer_16to8copy = transfer_16to8copy_mmx; | ||
| 334 : | edgomez | 200 | transfer_8to16sub = transfer_8to16sub_mmx; |
| 335 : | edgomez | 236 | transfer_8to16sub2 = transfer_8to16sub2_mmx; |
| 336 : | edgomez | 200 | transfer_16to8add = transfer_16to8add_mmx; |
| 337 : | transfer8x8_copy = transfer8x8_copy_mmx; | ||
| 338 : | Isibaar | 3 | |
| 339 : | h | 540 | /* Interlacing Functions */ |
| 340 : | MBFieldTest = MBFieldTest_mmx; | ||
| 341 : | edgomez | 236 | |
| 342 : | edgomez | 200 | /* Image Interpolation related functions */ |
| 343 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; | ||
| 344 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; | ||
| 345 : | Isibaar | 3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
| 346 : | |||
| 347 : | Isibaar | 579 | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; |
| 348 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; | ||
| 349 : | |||
| 350 : | Isibaar | 664 | interpolate8x8_avg2 = interpolate8x8_avg2_mmx; |
| 351 : | Isibaar | 579 | interpolate8x8_avg4 = interpolate8x8_avg4_mmx; |
| 352 : | |||
| 353 : | suxen_drol | 695 | /* reduced resolution */ |
| 354 : | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx; | ||
| 355 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx; | ||
| 356 : | hfilter_31 = xvid_HFilter_31_mmx; | ||
| 357 : | |||
| 358 : | suxen_drol | 631 | /* image input xxx_to_yv12 related functions */ |
| 359 : | yv12_to_yv12 = yv12_to_yv12_mmx; | ||
| 360 : | bgr_to_yv12 = bgr_to_yv12_mmx; | ||
| 361 : | bgra_to_yv12 = bgra_to_yv12_mmx; | ||
| 362 : | edgomez | 200 | yuyv_to_yv12 = yuyv_to_yv12_mmx; |
| 363 : | uyvy_to_yv12 = uyvy_to_yv12_mmx; | ||
| 364 : | Isibaar | 3 | |
| 365 : | suxen_drol | 631 | /* image output yv12_to_xxx related functions */ |
| 366 : | yv12_to_bgr = yv12_to_bgr_mmx; | ||
| 367 : | yv12_to_bgra = yv12_to_bgra_mmx; | ||
| 368 : | edgomez | 200 | yv12_to_yuyv = yv12_to_yuyv_mmx; |
| 369 : | yv12_to_uyvy = yv12_to_uyvy_mmx; | ||
| 370 : | suxen_drol | 631 | |
| 371 : | yv12_to_yuyvi = yv12_to_yuyvi_mmx; | ||
| 372 : | yv12_to_uyvyi = yv12_to_uyvyi_mmx; | ||
| 373 : | Isibaar | 3 | |
| 374 : | edgomez | 200 | /* Motion estimation related functions */ |
| 375 : | Isibaar | 3 | calc_cbp = calc_cbp_mmx; |
| 376 : | edgomez | 200 | sad16 = sad16_mmx; |
| 377 : | sad8 = sad8_mmx; | ||
| 378 : | suxen_drol | 329 | sad16bi = sad16bi_mmx; |
| 379 : | sad8bi = sad8bi_mmx; | ||
| 380 : | edgomez | 200 | dev16 = dev16_mmx; |
| 381 : | Isibaar | 534 | sad16v = sad16v_mmx; |
| 382 : | Isibaar | 3 | |
| 383 : | } | ||
| 384 : | |||
| 385 : | suxen_drol | 329 | /* these 3dnow functions are faster than mmx, but slower than xmm. */ |
| 386 : | if ((cpu_flags & XVID_CPU_3DNOW) > 0) { | ||
| 387 : | |||
| 388 : | /* ME functions */ | ||
| 389 : | sad16bi = sad16bi_3dn; | ||
| 390 : | sad8bi = sad8bi_3dn; | ||
| 391 : | suxen_drol | 631 | |
| 392 : | yuyv_to_yv12 = yuyv_to_yv12_3dn; | ||
| 393 : | uyvy_to_yv12 = uyvy_to_yv12_3dn; | ||
| 394 : | suxen_drol | 329 | } |
| 395 : | |||
| 396 : | |||
| 397 : | edgomez | 195 | if ((cpu_flags & XVID_CPU_MMXEXT) > 0) { |
| 398 : | edgomez | 200 | |
| 399 : | /* Inverse DCT */ | ||
| 400 : | Isibaar | 3 | idct = idct_xmm; |
| 401 : | edgomez | 200 | |
| 402 : | /* Interpolation */ | ||
| 403 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm; | ||
| 404 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; | ||
| 405 : | h | 38 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
| 406 : | edgomez | 200 | |
| 407 : | suxen_drol | 695 | /* reduced resolution */ |
| 408 : | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm; | ||
| 409 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm; | ||
| 410 : | |||
| 411 : | chenm001 | 274 | /* Quantization */ |
| 412 : | dequant_intra = dequant_intra_xmm; | ||
| 413 : | dequant_inter = dequant_inter_xmm; | ||
| 414 : | |||
| 415 : | edgomez | 218 | /* Buffer transfer */ |
| 416 : | transfer_8to16sub2 = transfer_8to16sub2_xmm; | ||
| 417 : | |||
| 418 : | edgomez | 200 | /* Colorspace transformation */ |
| 419 : | suxen_drol | 631 | yv12_to_yv12 = yv12_to_yv12_xmm; |
| 420 : | yuyv_to_yv12 = yuyv_to_yv12_xmm; | ||
| 421 : | Isibaar | 633 | uyvy_to_yv12 = uyvy_to_yv12_xmm; |
| 422 : | Isibaar | 3 | |
| 423 : | edgomez | 200 | /* ME functions */ |
| 424 : | Isibaar | 3 | sad16 = sad16_xmm; |
| 425 : | suxen_drol | 329 | sad8 = sad8_xmm; |
| 426 : | chenm001 | 274 | sad16bi = sad16bi_xmm; |
| 427 : | suxen_drol | 329 | sad8bi = sad8bi_xmm; |
| 428 : | Isibaar | 3 | dev16 = dev16_xmm; |
| 429 : | chl | 530 | sad16v = sad16v_xmm; |
| 430 : | Isibaar | 3 | } |
| 431 : | |||
| 432 : | edgomez | 195 | if ((cpu_flags & XVID_CPU_3DNOW) > 0) { |
| 433 : | edgomez | 200 | |
| 434 : | /* Interpolation */ | ||
| 435 : | Isibaar | 3 | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
| 436 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn; | ||
| 437 : | h | 40 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
| 438 : | Isibaar | 3 | } |
| 439 : | |||
| 440 : | edgomez | 195 | if ((cpu_flags & XVID_CPU_SSE2) > 0) { |
| 441 : | Isibaar | 154 | #ifdef EXPERIMENTAL_SSE2_CODE |
| 442 : | edgomez | 200 | |
| 443 : | chenm001 | 274 | calc_cbp = calc_cbp_sse2; |
| 444 : | |||
| 445 : | edgomez | 200 | /* Quantization */ |
| 446 : | quant_intra = quant_intra_sse2; | ||
| 447 : | Isibaar | 154 | dequant_intra = dequant_intra_sse2; |
| 448 : | edgomez | 200 | quant_inter = quant_inter_sse2; |
| 449 : | Isibaar | 154 | dequant_inter = dequant_inter_sse2; |
| 450 : | h | 135 | |
| 451 : | edgomez | 200 | /* ME */ |
| 452 : | sad16 = sad16_sse2; | ||
| 453 : | dev16 = dev16_sse2; | ||
| 454 : | |||
| 455 : | /* Forward and Inverse DCT */ | ||
| 456 : | idct = idct_sse2; | ||
| 457 : | Isibaar | 154 | fdct = fdct_sse2; |
| 458 : | #endif | ||
| 459 : | h | 126 | } |
| 460 : | edgomez | 200 | |
| 461 : | Isibaar | 3 | #endif |
| 462 : | edgomez | 200 | |
| 463 : | Isibaar | 209 | #ifdef ARCH_IA64 |
| 464 : | if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines? | ||
| 465 : | idct_ia64_init(); | ||
| 466 : | fdct = fdct_ia64; | ||
| 467 : | idct = idct_ia64; //not yet working, crashes | ||
| 468 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; | ||
| 469 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; | ||
| 470 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; | ||
| 471 : | sad16 = sad16_ia64; | ||
| 472 : | sad16bi = sad16bi_ia64; | ||
| 473 : | sad8 = sad8_ia64; | ||
| 474 : | dev16 = dev16_ia64; | ||
| 475 : | chl | 530 | // Halfpel8_Refine = Halfpel8_Refine_ia64; |
| 476 : | Isibaar | 209 | quant_intra = quant_intra_ia64; |
| 477 : | dequant_intra = dequant_intra_ia64; | ||
| 478 : | quant_inter = quant_inter_ia64; | ||
| 479 : | dequant_inter = dequant_inter_ia64; | ||
| 480 : | transfer_8to16copy = transfer_8to16copy_ia64; | ||
| 481 : | transfer_16to8copy = transfer_16to8copy_ia64; | ||
| 482 : | transfer_8to16sub = transfer_8to16sub_ia64; | ||
| 483 : | transfer_8to16sub2 = transfer_8to16sub2_ia64; | ||
| 484 : | transfer_16to8add = transfer_16to8add_ia64; | ||
| 485 : | transfer8x8_copy = transfer8x8_copy_ia64; | ||
| 486 : | DEBUG("Using IA-64 assembler routines.\n"); | ||
| 487 : | } | ||
| 488 : | #endif | ||
| 489 : | |||
| 490 : | canard | 52 | #ifdef ARCH_PPC |
| 491 : | canard | 71 | #ifdef ARCH_PPC_ALTIVEC |
| 492 : | calc_cbp = calc_cbp_altivec; | ||
| 493 : | canard | 76 | fdct = fdct_altivec; |
| 494 : | idct = idct_altivec; | ||
| 495 : | canard | 115 | sadInit = sadInit_altivec; |
| 496 : | canard | 89 | sad16 = sad16_altivec; |
| 497 : | sad8 = sad8_altivec; | ||
| 498 : | dev16 = dev16_altivec; | ||
| 499 : | canard | 71 | #else |
| 500 : | canard | 52 | calc_cbp = calc_cbp_ppc; |
| 501 : | #endif | ||
| 502 : | canard | 71 | #endif |
| 503 : | edgomez | 195 | |
| 504 : | Isibaar | 3 | return XVID_ERR_OK; |
| 505 : | } | ||
| 506 : | |||
| 507 : | suxen_drol | 634 | |
| 508 : | |||
| 509 : | static int | ||
| 510 : | xvid_init_convert(XVID_INIT_CONVERTINFO* convert) | ||
| 511 : | { | ||
| 512 : | suxen_drol | 698 | // const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); |
| 513 : | suxen_drol | 634 | const int width = convert->width; |
| 514 : | const int height = convert->height; | ||
| 515 : | const int width2 = convert->width/2; | ||
| 516 : | const int height2 = convert->height/2; | ||
| 517 : | IMAGE img; | ||
| 518 : | |||
| 519 : | switch (convert->input.colorspace & ~XVID_CSP_VFLIP) | ||
| 520 : | { | ||
| 521 : | case XVID_CSP_YV12 : | ||
| 522 : | img.y = convert->input.y; | ||
| 523 : | img.v = (uint8_t*)convert->input.y + width*height; | ||
| 524 : | img.u = (uint8_t*)convert->input.y + width*height + width2*height2; | ||
| 525 : | image_output(&img, width, height, width, | ||
| 526 : | convert->output.y, convert->output.y_stride, | ||
| 527 : | convert->output.colorspace, convert->interlacing); | ||
| 528 : | break; | ||
| 529 : | |||
| 530 : | default : | ||
| 531 : | return XVID_ERR_FORMAT; | ||
| 532 : | } | ||
| 533 : | |||
| 534 : | |||
| 535 : | emms(); | ||
| 536 : | return XVID_ERR_OK; | ||
| 537 : | } | ||
| 538 : | |||
| 539 : | |||
| 540 : | int | ||
| 541 : | xvid_init(void *handle, | ||
| 542 : | int opt, | ||
| 543 : | void *param1, | ||
| 544 : | void *param2) | ||
| 545 : | { | ||
| 546 : | switch(opt) | ||
| 547 : | { | ||
| 548 : | case XVID_INIT_INIT : | ||
| 549 : | return xvid_init_init((XVID_INIT_PARAM*)param1); | ||
| 550 : | |||
| 551 : | case XVID_INIT_CONVERT : | ||
| 552 : | return xvid_init_convert((XVID_INIT_CONVERTINFO*)param1); | ||
| 553 : | |||
| 554 : | default : | ||
| 555 : | return XVID_ERR_FAIL; | ||
| 556 : | } | ||
| 557 : | } | ||
| 558 : | |||
| 559 : | edgomez | 200 | /***************************************************************************** |
| 560 : | * XviD Native decoder entry point | ||
| 561 : | * | ||
| 562 : | * This function is just a wrapper to all the option cases. | ||
| 563 : | * | ||
| 564 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
| 565 : | * else returns the wrapped function result | ||
| 566 : | * | ||
| 567 : | ****************************************************************************/ | ||
| 568 : | |||
| 569 : | edgomez | 195 | int |
| 570 : | xvid_decore(void *handle, | ||
| 571 : | int opt, | ||
| 572 : | void *param1, | ||
| 573 : | void *param2) | ||
| 574 : | Isibaar | 3 | { |
| 575 : | edgomez | 195 | switch (opt) { |
| 576 : | case XVID_DEC_DECODE: | ||
| 577 : | suxen_drol | 631 | return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1, (XVID_DEC_STATS*) param2); |
| 578 : | Isibaar | 3 | |
| 579 : | edgomez | 195 | case XVID_DEC_CREATE: |
| 580 : | chenm001 | 274 | return decoder_create((XVID_DEC_PARAM *) param1); |
| 581 : | Isibaar | 3 | |
| 582 : | edgomez | 195 | case XVID_DEC_DESTROY: |
| 583 : | return decoder_destroy((DECODER *) handle); | ||
| 584 : | |||
| 585 : | Isibaar | 3 | default: |
| 586 : | edgomez | 195 | return XVID_ERR_FAIL; |
| 587 : | } | ||
| 588 : | Isibaar | 3 | } |
| 589 : | |||
| 590 : | |||
| 591 : | edgomez | 200 | /***************************************************************************** |
| 592 : | * XviD Native encoder entry point | ||
| 593 : | * | ||
| 594 : | * This function is just a wrapper to all the option cases. | ||
| 595 : | * | ||
| 596 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
| 597 : | * else returns the wrapped function result | ||
| 598 : | * | ||
| 599 : | ****************************************************************************/ | ||
| 600 : | |||
| 601 : | edgomez | 195 | int |
| 602 : | xvid_encore(void *handle, | ||
| 603 : | int opt, | ||
| 604 : | void *param1, | ||
| 605 : | void *param2) | ||
| 606 : | Isibaar | 3 | { |
| 607 : | edgomez | 195 | switch (opt) { |
| 608 : | case XVID_ENC_ENCODE: | ||
| 609 : | chl | 620 | |
| 610 : | suxen_drol | 232 | if (((Encoder *) handle)->mbParam.max_bframes >= 0) |
| 611 : | return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1, | ||
| 612 : | (XVID_ENC_STATS *) param2); | ||
| 613 : | else | ||
| 614 : | edgomez | 195 | return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
| 615 : | (XVID_ENC_STATS *) param2); | ||
| 616 : | Isibaar | 3 | |
| 617 : | edgomez | 195 | case XVID_ENC_CREATE: |
| 618 : | return encoder_create((XVID_ENC_PARAM *) param1); | ||
| 619 : | Isibaar | 3 | |
| 620 : | edgomez | 195 | case XVID_ENC_DESTROY: |
| 621 : | return encoder_destroy((Encoder *) handle); | ||
| 622 : | |||
| 623 : | Isibaar | 3 | default: |
| 624 : | edgomez | 195 | return XVID_ERR_FAIL; |
| 625 : | } | ||
| 626 : | Isibaar | 3 | } |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |