Parent Directory
| 
Revision Log
Revision 755 - (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 | 755 | * $Id: xvid.c,v 1.33.2.20 2003-01-02 13:58:54 suxen_drol Exp $ |
| 41 : | edgomez | 201 | * |
| 42 : | ****************************************************************************/ | ||
| 43 : | Isibaar | 3 | |
| 44 : | suxen_drol | 755 | #include <stdlib.h> |
| 45 : | #include <time.h> | ||
| 46 : | |||
| 47 : | Isibaar | 3 | #include "xvid.h" |
| 48 : | #include "decoder.h" | ||
| 49 : | #include "encoder.h" | ||
| 50 : | #include "bitstream/cbp.h" | ||
| 51 : | #include "dct/idct.h" | ||
| 52 : | #include "dct/fdct.h" | ||
| 53 : | #include "image/colorspace.h" | ||
| 54 : | #include "image/interpolate8x8.h" | ||
| 55 : | suxen_drol | 695 | #include "image/reduced.h" |
| 56 : | Isibaar | 3 | #include "utils/mem_transfer.h" |
| 57 : | h | 540 | #include "utils/mbfunctions.h" |
| 58 : | Isibaar | 3 | #include "quant/quant_h263.h" |
| 59 : | #include "quant/quant_mpeg4.h" | ||
| 60 : | ia64p | 299 | #include "motion/motion.h" |
| 61 : | Isibaar | 3 | #include "motion/sad.h" |
| 62 : | #include "utils/emms.h" | ||
| 63 : | #include "utils/timer.h" | ||
| 64 : | Isibaar | 100 | #include "bitstream/mbcoding.h" |
| 65 : | Isibaar | 3 | |
| 66 : | suxen_drol | 311 | #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE) |
| 67 : | |||
| 68 : | #ifdef WIN32 | ||
| 69 : | #include <windows.h> | ||
| 70 : | #else | ||
| 71 : | #include <signal.h> | ||
| 72 : | #include <setjmp.h> | ||
| 73 : | #endif | ||
| 74 : | |||
| 75 : | |||
| 76 : | #ifndef WIN32 | ||
| 77 : | |||
| 78 : | static jmp_buf mark; | ||
| 79 : | |||
| 80 : | static void | ||
| 81 : | sigill_handler(int signal) | ||
| 82 : | { | ||
| 83 : | longjmp(mark, 1); | ||
| 84 : | } | ||
| 85 : | #endif | ||
| 86 : | |||
| 87 : | |||
| 88 : | /* | ||
| 89 : | calls the funcptr, and returns whether SIGILL (illegal instruction) was signalled | ||
| 90 : | return values: | ||
| 91 : | -1 : could not determine | ||
| 92 : | 0 : SIGILL was *not* signalled | ||
| 93 : | 1 : SIGILL was signalled | ||
| 94 : | */ | ||
| 95 : | |||
| 96 : | int | ||
| 97 : | sigill_check(void (*func)()) | ||
| 98 : | { | ||
| 99 : | #ifdef WIN32 | ||
| 100 : | _try { | ||
| 101 : | func(); | ||
| 102 : | } | ||
| 103 : | _except(EXCEPTION_EXECUTE_HANDLER) { | ||
| 104 : | |||
| 105 : | if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION) | ||
| 106 : | return 1; | ||
| 107 : | } | ||
| 108 : | return 0; | ||
| 109 : | #else | ||
| 110 : | void * old_handler; | ||
| 111 : | int jmpret; | ||
| 112 : | |||
| 113 : | |||
| 114 : | old_handler = signal(SIGILL, sigill_handler); | ||
| 115 : | if (old_handler == SIG_ERR) | ||
| 116 : | { | ||
| 117 : | return -1; | ||
| 118 : | } | ||
| 119 : | |||
| 120 : | jmpret = setjmp(mark); | ||
| 121 : | if (jmpret == 0) | ||
| 122 : | { | ||
| 123 : | func(); | ||
| 124 : | } | ||
| 125 : | |||
| 126 : | signal(SIGILL, old_handler); | ||
| 127 : | |||
| 128 : | return jmpret; | ||
| 129 : | #endif | ||
| 130 : | } | ||
| 131 : | #endif | ||
| 132 : | |||
| 133 : | edgomez | 200 | /***************************************************************************** |
| 134 : | * XviD Init Entry point | ||
| 135 : | * | ||
| 136 : | * Well this function initialize all internal function pointers according | ||
| 137 : | * to the CPU features forced by the library client or autodetected (depending | ||
| 138 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
| 139 : | * image colorspace transformation tables. | ||
| 140 : | * | ||
| 141 : | * Returned value : XVID_ERR_OK | ||
| 142 : | * + API_VERSION in the input XVID_INIT_PARAM structure | ||
| 143 : | * + core build " " " " " | ||
| 144 : | * | ||
| 145 : | ****************************************************************************/ | ||
| 146 : | |||
| 147 : | suxen_drol | 634 | |
| 148 : | static | ||
| 149 : | int xvid_init_init(XVID_INIT_PARAM * init_param) | ||
| 150 : | Isibaar | 3 | { |
| 151 : | int cpu_flags; | ||
| 152 : | |||
| 153 : | suxen_drol | 234 | /* Inform the client the API version */ |
| 154 : | init_param->api_version = API_VERSION; | ||
| 155 : | |||
| 156 : | /* Inform the client the core build - unused because we're still alpha */ | ||
| 157 : | init_param->core_build = 1000; | ||
| 158 : | |||
| 159 : | suxen_drol | 311 | /* Do we have to force CPU features ? */ |
| 160 : | if ((init_param->cpu_flags & XVID_CPU_FORCE)) { | ||
| 161 : | suxen_drol | 234 | |
| 162 : | Isibaar | 3 | cpu_flags = init_param->cpu_flags; |
| 163 : | suxen_drol | 311 | |
| 164 : | edgomez | 200 | } else { |
| 165 : | Isibaar | 3 | |
| 166 : | chenm001 | 274 | cpu_flags = check_cpu_features(); |
| 167 : | suxen_drol | 311 | |
| 168 : | #if defined(ARCH_X86) && defined(EXPERIMENTAL_SSE2_CODE) | ||
| 169 : | if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) | ||
| 170 : | cpu_flags &= ~XVID_CPU_SSE; | ||
| 171 : | |||
| 172 : | if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) | ||
| 173 : | cpu_flags &= ~XVID_CPU_SSE2; | ||
| 174 : | #endif | ||
| 175 : | } | ||
| 176 : | |||
| 177 : | if ((init_param->cpu_flags & XVID_CPU_CHKONLY)) | ||
| 178 : | { | ||
| 179 : | Isibaar | 3 | init_param->cpu_flags = cpu_flags; |
| 180 : | suxen_drol | 311 | return XVID_ERR_OK; |
| 181 : | Isibaar | 3 | } |
| 182 : | |||
| 183 : | suxen_drol | 311 | init_param->cpu_flags = cpu_flags; |
| 184 : | |||
| 185 : | |||
| 186 : | edgomez | 200 | /* Initialize the function pointers */ |
| 187 : | Isibaar | 3 | idct_int32_init(); |
| 188 : | Isibaar | 100 | init_vlc_tables(); |
| 189 : | |||
| 190 : | edgomez | 200 | /* Fixed Point Forward/Inverse DCT transformations */ |
| 191 : | Isibaar | 3 | fdct = fdct_int32; |
| 192 : | idct = idct_int32; | ||
| 193 : | |||
| 194 : | edgomez | 200 | /* Only needed on PPC Altivec archs */ |
| 195 : | canard | 115 | sadInit = 0; |
| 196 : | edgomez | 195 | |
| 197 : | edgomez | 200 | /* Restore FPU context : emms_c is a nop functions */ |
| 198 : | Isibaar | 3 | emms = emms_c; |
| 199 : | |||
| 200 : | edgomez | 200 | /* Quantization functions */ |
| 201 : | quant_intra = quant_intra_c; | ||
| 202 : | Isibaar | 3 | dequant_intra = dequant_intra_c; |
| 203 : | edgomez | 200 | quant_inter = quant_inter_c; |
| 204 : | Isibaar | 3 | dequant_inter = dequant_inter_c; |
| 205 : | |||
| 206 : | edgomez | 200 | quant4_intra = quant4_intra_c; |
| 207 : | Isibaar | 3 | dequant4_intra = dequant4_intra_c; |
| 208 : | edgomez | 200 | quant4_inter = quant4_inter_c; |
| 209 : | Isibaar | 3 | dequant4_inter = dequant4_inter_c; |
| 210 : | |||
| 211 : | edgomez | 200 | /* Block transfer related functions */ |
| 212 : | Isibaar | 3 | transfer_8to16copy = transfer_8to16copy_c; |
| 213 : | transfer_16to8copy = transfer_16to8copy_c; | ||
| 214 : | edgomez | 200 | transfer_8to16sub = transfer_8to16sub_c; |
| 215 : | suxen_drol | 118 | transfer_8to16sub2 = transfer_8to16sub2_c; |
| 216 : | edgomez | 200 | transfer_16to8add = transfer_16to8add_c; |
| 217 : | transfer8x8_copy = transfer8x8_copy_c; | ||
| 218 : | Isibaar | 3 | |
| 219 : | h | 540 | /* Interlacing functions */ |
| 220 : | MBFieldTest = MBFieldTest_c; | ||
| 221 : | |||
| 222 : | edgomez | 200 | /* Image interpolation related functions */ |
| 223 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; | ||
| 224 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; | ||
| 225 : | Isibaar | 3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
| 226 : | |||
| 227 : | Isibaar | 588 | interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; |
| 228 : | interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; | ||
| 229 : | interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; | ||
| 230 : | |||
| 231 : | Isibaar | 579 | interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c; |
| 232 : | interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c; | ||
| 233 : | interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c; | ||
| 234 : | |||
| 235 : | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c; | ||
| 236 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c; | ||
| 237 : | |||
| 238 : | interpolate8x8_avg2 = interpolate8x8_avg2_c; | ||
| 239 : | interpolate8x8_avg4 = interpolate8x8_avg4_c; | ||
| 240 : | |||
| 241 : | suxen_drol | 695 | /* reduced resoltuion */ |
| 242 : | |||
| 243 : | suxen_drol | 701 | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
| 244 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; | ||
| 245 : | suxen_drol | 695 | #ifdef ARCH_X86 |
| 246 : | vfilter_31 = xvid_VFilter_31_x86; | ||
| 247 : | hfilter_31 = xvid_HFilter_31_x86; | ||
| 248 : | #else | ||
| 249 : | vfilter_31 = xvid_VFilter_31_C; | ||
| 250 : | hfilter_31 = xvid_HFilter_31_C; | ||
| 251 : | #endif | ||
| 252 : | suxen_drol | 701 | filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; |
| 253 : | filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; | ||
| 254 : | suxen_drol | 695 | |
| 255 : | edgomez | 200 | /* Initialize internal colorspace transformation tables */ |
| 256 : | Isibaar | 3 | colorspace_init(); |
| 257 : | |||
| 258 : | edgomez | 200 | /* All colorspace transformation functions User Format->YV12 */ |
| 259 : | suxen_drol | 631 | yv12_to_yv12 = yv12_to_yv12_c; |
| 260 : | rgb555_to_yv12 = rgb555_to_yv12_c; | ||
| 261 : | rgb565_to_yv12 = rgb565_to_yv12_c; | ||
| 262 : | bgr_to_yv12 = bgr_to_yv12_c; | ||
| 263 : | bgra_to_yv12 = bgra_to_yv12_c; | ||
| 264 : | abgr_to_yv12 = abgr_to_yv12_c; | ||
| 265 : | rgba_to_yv12 = rgba_to_yv12_c; | ||
| 266 : | yuyv_to_yv12 = yuyv_to_yv12_c; | ||
| 267 : | uyvy_to_yv12 = uyvy_to_yv12_c; | ||
| 268 : | Isibaar | 3 | |
| 269 : | suxen_drol | 631 | rgb555i_to_yv12 = rgb555i_to_yv12_c; |
| 270 : | rgb565i_to_yv12 = rgb565i_to_yv12_c; | ||
| 271 : | bgri_to_yv12 = bgri_to_yv12_c; | ||
| 272 : | bgrai_to_yv12 = bgrai_to_yv12_c; | ||
| 273 : | abgri_to_yv12 = abgri_to_yv12_c; | ||
| 274 : | rgbai_to_yv12 = rgbai_to_yv12_c; | ||
| 275 : | yuyvi_to_yv12 = yuyvi_to_yv12_c; | ||
| 276 : | uyvyi_to_yv12 = uyvyi_to_yv12_c; | ||
| 277 : | |||
| 278 : | |||
| 279 : | edgomez | 200 | /* All colorspace transformation functions YV12->User format */ |
| 280 : | suxen_drol | 631 | yv12_to_rgb555 = yv12_to_rgb555_c; |
| 281 : | yv12_to_rgb565 = yv12_to_rgb565_c; | ||
| 282 : | yv12_to_bgr = yv12_to_bgr_c; | ||
| 283 : | yv12_to_bgra = yv12_to_bgra_c; | ||
| 284 : | yv12_to_abgr = yv12_to_abgr_c; | ||
| 285 : | yv12_to_rgba = yv12_to_rgba_c; | ||
| 286 : | yv12_to_yuyv = yv12_to_yuyv_c; | ||
| 287 : | 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_yuyvi = yv12_to_yuyvi_c; | ||
| 296 : | yv12_to_uyvyi = yv12_to_uyvyi_c; | ||
| 297 : | Isibaar | 3 | |
| 298 : | edgomez | 200 | /* Functions used in motion estimation algorithms */ |
| 299 : | Isibaar | 3 | calc_cbp = calc_cbp_c; |
| 300 : | edgomez | 200 | sad16 = sad16_c; |
| 301 : | suxen_drol | 329 | sad8 = sad8_c; |
| 302 : | edgomez | 200 | sad16bi = sad16bi_c; |
| 303 : | suxen_drol | 329 | sad8bi = sad8bi_c; |
| 304 : | edgomez | 200 | dev16 = dev16_c; |
| 305 : | chl | 530 | sad16v = sad16v_c; |
| 306 : | suxen_drol | 329 | |
| 307 : | chl | 530 | // Halfpel8_Refine = Halfpel8_Refine_c; |
| 308 : | Isibaar | 3 | |
| 309 : | #ifdef ARCH_X86 | ||
| 310 : | suxen_drol | 631 | |
| 311 : | if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || | ||
| 312 : | (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || | ||
| 313 : | (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2)) | ||
| 314 : | { | ||
| 315 : | /* Restore FPU context : emms_c is a nop functions */ | ||
| 316 : | emms = emms_mmx; | ||
| 317 : | } | ||
| 318 : | |||
| 319 : | edgomez | 195 | if ((cpu_flags & XVID_CPU_MMX) > 0) { |
| 320 : | edgomez | 200 | |
| 321 : | /* Forward and Inverse Discrete Cosine Transformation functions */ | ||
| 322 : | Isibaar | 3 | fdct = fdct_mmx; |
| 323 : | idct = idct_mmx; | ||
| 324 : | |||
| 325 : | edgomez | 200 | /* Quantization related functions */ |
| 326 : | quant_intra = quant_intra_mmx; | ||
| 327 : | Isibaar | 3 | dequant_intra = dequant_intra_mmx; |
| 328 : | edgomez | 200 | quant_inter = quant_inter_mmx; |
| 329 : | Isibaar | 3 | dequant_inter = dequant_inter_mmx; |
| 330 : | |||
| 331 : | edgomez | 200 | quant4_intra = quant4_intra_mmx; |
| 332 : | Isibaar | 3 | dequant4_intra = dequant4_intra_mmx; |
| 333 : | edgomez | 200 | quant4_inter = quant4_inter_mmx; |
| 334 : | Isibaar | 3 | dequant4_inter = dequant4_inter_mmx; |
| 335 : | |||
| 336 : | edgomez | 200 | /* Block related functions */ |
| 337 : | Isibaar | 3 | transfer_8to16copy = transfer_8to16copy_mmx; |
| 338 : | transfer_16to8copy = transfer_16to8copy_mmx; | ||
| 339 : | edgomez | 200 | transfer_8to16sub = transfer_8to16sub_mmx; |
| 340 : | edgomez | 236 | transfer_8to16sub2 = transfer_8to16sub2_mmx; |
| 341 : | edgomez | 200 | transfer_16to8add = transfer_16to8add_mmx; |
| 342 : | transfer8x8_copy = transfer8x8_copy_mmx; | ||
| 343 : | Isibaar | 3 | |
| 344 : | h | 540 | /* Interlacing Functions */ |
| 345 : | MBFieldTest = MBFieldTest_mmx; | ||
| 346 : | edgomez | 236 | |
| 347 : | edgomez | 200 | /* Image Interpolation related functions */ |
| 348 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; | ||
| 349 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; | ||
| 350 : | Isibaar | 3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
| 351 : | |||
| 352 : | Isibaar | 579 | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; |
| 353 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; | ||
| 354 : | |||
| 355 : | Isibaar | 664 | interpolate8x8_avg2 = interpolate8x8_avg2_mmx; |
| 356 : | Isibaar | 579 | interpolate8x8_avg4 = interpolate8x8_avg4_mmx; |
| 357 : | |||
| 358 : | suxen_drol | 695 | /* reduced resolution */ |
| 359 : | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx; | ||
| 360 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx; | ||
| 361 : | hfilter_31 = xvid_HFilter_31_mmx; | ||
| 362 : | suxen_drol | 701 | filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx; |
| 363 : | filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx; | ||
| 364 : | suxen_drol | 695 | |
| 365 : | suxen_drol | 631 | /* image input xxx_to_yv12 related functions */ |
| 366 : | yv12_to_yv12 = yv12_to_yv12_mmx; | ||
| 367 : | bgr_to_yv12 = bgr_to_yv12_mmx; | ||
| 368 : | bgra_to_yv12 = bgra_to_yv12_mmx; | ||
| 369 : | edgomez | 200 | yuyv_to_yv12 = yuyv_to_yv12_mmx; |
| 370 : | uyvy_to_yv12 = uyvy_to_yv12_mmx; | ||
| 371 : | Isibaar | 3 | |
| 372 : | suxen_drol | 631 | /* image output yv12_to_xxx related functions */ |
| 373 : | yv12_to_bgr = yv12_to_bgr_mmx; | ||
| 374 : | yv12_to_bgra = yv12_to_bgra_mmx; | ||
| 375 : | edgomez | 200 | yv12_to_yuyv = yv12_to_yuyv_mmx; |
| 376 : | yv12_to_uyvy = yv12_to_uyvy_mmx; | ||
| 377 : | suxen_drol | 631 | |
| 378 : | yv12_to_yuyvi = yv12_to_yuyvi_mmx; | ||
| 379 : | yv12_to_uyvyi = yv12_to_uyvyi_mmx; | ||
| 380 : | Isibaar | 3 | |
| 381 : | edgomez | 200 | /* Motion estimation related functions */ |
| 382 : | Isibaar | 3 | calc_cbp = calc_cbp_mmx; |
| 383 : | edgomez | 200 | sad16 = sad16_mmx; |
| 384 : | sad8 = sad8_mmx; | ||
| 385 : | suxen_drol | 329 | sad16bi = sad16bi_mmx; |
| 386 : | sad8bi = sad8bi_mmx; | ||
| 387 : | edgomez | 200 | dev16 = dev16_mmx; |
| 388 : | Isibaar | 534 | sad16v = sad16v_mmx; |
| 389 : | Isibaar | 3 | |
| 390 : | } | ||
| 391 : | |||
| 392 : | suxen_drol | 329 | /* these 3dnow functions are faster than mmx, but slower than xmm. */ |
| 393 : | if ((cpu_flags & XVID_CPU_3DNOW) > 0) { | ||
| 394 : | |||
| 395 : | /* ME functions */ | ||
| 396 : | sad16bi = sad16bi_3dn; | ||
| 397 : | sad8bi = sad8bi_3dn; | ||
| 398 : | suxen_drol | 631 | |
| 399 : | yuyv_to_yv12 = yuyv_to_yv12_3dn; | ||
| 400 : | uyvy_to_yv12 = uyvy_to_yv12_3dn; | ||
| 401 : | suxen_drol | 329 | } |
| 402 : | |||
| 403 : | |||
| 404 : | edgomez | 195 | if ((cpu_flags & XVID_CPU_MMXEXT) > 0) { |
| 405 : | edgomez | 200 | |
| 406 : | /* Inverse DCT */ | ||
| 407 : | Isibaar | 3 | idct = idct_xmm; |
| 408 : | edgomez | 200 | |
| 409 : | /* Interpolation */ | ||
| 410 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm; | ||
| 411 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; | ||
| 412 : | h | 38 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
| 413 : | edgomez | 200 | |
| 414 : | suxen_drol | 695 | /* reduced resolution */ |
| 415 : | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_xmm; | ||
| 416 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_xmm; | ||
| 417 : | |||
| 418 : | chenm001 | 274 | /* Quantization */ |
| 419 : | Isibaar | 730 | quant4_intra = quant4_intra_xmm; |
| 420 : | quant4_inter = quant4_inter_xmm; | ||
| 421 : | |||
| 422 : | chenm001 | 274 | dequant_intra = dequant_intra_xmm; |
| 423 : | dequant_inter = dequant_inter_xmm; | ||
| 424 : | |||
| 425 : | edgomez | 218 | /* Buffer transfer */ |
| 426 : | transfer_8to16sub2 = transfer_8to16sub2_xmm; | ||
| 427 : | |||
| 428 : | edgomez | 200 | /* Colorspace transformation */ |
| 429 : | suxen_drol | 631 | yv12_to_yv12 = yv12_to_yv12_xmm; |
| 430 : | yuyv_to_yv12 = yuyv_to_yv12_xmm; | ||
| 431 : | Isibaar | 633 | uyvy_to_yv12 = uyvy_to_yv12_xmm; |
| 432 : | Isibaar | 3 | |
| 433 : | edgomez | 200 | /* ME functions */ |
| 434 : | Isibaar | 3 | sad16 = sad16_xmm; |
| 435 : | suxen_drol | 329 | sad8 = sad8_xmm; |
| 436 : | chenm001 | 274 | sad16bi = sad16bi_xmm; |
| 437 : | suxen_drol | 329 | sad8bi = sad8bi_xmm; |
| 438 : | Isibaar | 3 | dev16 = dev16_xmm; |
| 439 : | chl | 530 | sad16v = sad16v_xmm; |
| 440 : | Isibaar | 3 | } |
| 441 : | |||
| 442 : | edgomez | 195 | if ((cpu_flags & XVID_CPU_3DNOW) > 0) { |
| 443 : | edgomez | 200 | |
| 444 : | /* Interpolation */ | ||
| 445 : | Isibaar | 3 | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
| 446 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn; | ||
| 447 : | h | 40 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
| 448 : | Isibaar | 3 | } |
| 449 : | |||
| 450 : | Isibaar | 730 | if ((cpu_flags & XVID_CPU_3DNOWEXT) > 0) { |
| 451 : | |||
| 452 : | /* Inverse DCT */ | ||
| 453 : | idct = idct_3dne; | ||
| 454 : | |||
| 455 : | /* Buffer transfer */ | ||
| 456 : | transfer_8to16copy = transfer_8to16copy_3dne; | ||
| 457 : | transfer_16to8copy = transfer_16to8copy_3dne; | ||
| 458 : | transfer_8to16sub = transfer_8to16sub_3dne; | ||
| 459 : | transfer_8to16sub2 = transfer_8to16sub2_3dne; | ||
| 460 : | transfer_16to8add = transfer_16to8add_3dne; | ||
| 461 : | transfer8x8_copy = transfer8x8_copy_3dne; | ||
| 462 : | |||
| 463 : | /* Quantization */ | ||
| 464 : | dequant4_intra = dequant4_intra_3dne; | ||
| 465 : | dequant4_inter = dequant4_inter_3dne; | ||
| 466 : | quant_intra = quant_intra_3dne; | ||
| 467 : | quant_inter = quant_inter_3dne; | ||
| 468 : | dequant_intra = dequant_intra_3dne; | ||
| 469 : | dequant_inter = dequant_inter_3dne; | ||
| 470 : | |||
| 471 : | /* ME functions */ | ||
| 472 : | calc_cbp = calc_cbp_3dne; | ||
| 473 : | sad16 = sad16_3dne; | ||
| 474 : | sad8 = sad8_3dne; | ||
| 475 : | sad16bi = sad16bi_3dne; | ||
| 476 : | sad8bi = sad8bi_3dne; | ||
| 477 : | dev16 = dev16_3dne; | ||
| 478 : | |||
| 479 : | /* Interpolation */ | ||
| 480 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne; | ||
| 481 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne; | ||
| 482 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; | ||
| 483 : | } | ||
| 484 : | |||
| 485 : | |||
| 486 : | edgomez | 195 | if ((cpu_flags & XVID_CPU_SSE2) > 0) { |
| 487 : | Isibaar | 154 | #ifdef EXPERIMENTAL_SSE2_CODE |
| 488 : | edgomez | 200 | |
| 489 : | chenm001 | 274 | calc_cbp = calc_cbp_sse2; |
| 490 : | |||
| 491 : | edgomez | 200 | /* Quantization */ |
| 492 : | quant_intra = quant_intra_sse2; | ||
| 493 : | Isibaar | 154 | dequant_intra = dequant_intra_sse2; |
| 494 : | edgomez | 200 | quant_inter = quant_inter_sse2; |
| 495 : | Isibaar | 154 | dequant_inter = dequant_inter_sse2; |
| 496 : | h | 135 | |
| 497 : | edgomez | 200 | /* ME */ |
| 498 : | sad16 = sad16_sse2; | ||
| 499 : | dev16 = dev16_sse2; | ||
| 500 : | |||
| 501 : | /* Forward and Inverse DCT */ | ||
| 502 : | idct = idct_sse2; | ||
| 503 : | Isibaar | 154 | fdct = fdct_sse2; |
| 504 : | #endif | ||
| 505 : | h | 126 | } |
| 506 : | edgomez | 200 | |
| 507 : | Isibaar | 3 | #endif |
| 508 : | edgomez | 200 | |
| 509 : | Isibaar | 209 | #ifdef ARCH_IA64 |
| 510 : | if ((cpu_flags & XVID_CPU_IA64) > 0) { //use assembler routines? | ||
| 511 : | idct_ia64_init(); | ||
| 512 : | fdct = fdct_ia64; | ||
| 513 : | idct = idct_ia64; //not yet working, crashes | ||
| 514 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; | ||
| 515 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; | ||
| 516 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; | ||
| 517 : | sad16 = sad16_ia64; | ||
| 518 : | sad16bi = sad16bi_ia64; | ||
| 519 : | sad8 = sad8_ia64; | ||
| 520 : | dev16 = dev16_ia64; | ||
| 521 : | chl | 530 | // Halfpel8_Refine = Halfpel8_Refine_ia64; |
| 522 : | Isibaar | 209 | quant_intra = quant_intra_ia64; |
| 523 : | dequant_intra = dequant_intra_ia64; | ||
| 524 : | quant_inter = quant_inter_ia64; | ||
| 525 : | dequant_inter = dequant_inter_ia64; | ||
| 526 : | transfer_8to16copy = transfer_8to16copy_ia64; | ||
| 527 : | transfer_16to8copy = transfer_16to8copy_ia64; | ||
| 528 : | transfer_8to16sub = transfer_8to16sub_ia64; | ||
| 529 : | transfer_8to16sub2 = transfer_8to16sub2_ia64; | ||
| 530 : | transfer_16to8add = transfer_16to8add_ia64; | ||
| 531 : | transfer8x8_copy = transfer8x8_copy_ia64; | ||
| 532 : | DEBUG("Using IA-64 assembler routines.\n"); | ||
| 533 : | } | ||
| 534 : | #endif | ||
| 535 : | |||
| 536 : | canard | 52 | #ifdef ARCH_PPC |
| 537 : | canard | 71 | #ifdef ARCH_PPC_ALTIVEC |
| 538 : | calc_cbp = calc_cbp_altivec; | ||
| 539 : | canard | 76 | fdct = fdct_altivec; |
| 540 : | idct = idct_altivec; | ||
| 541 : | canard | 115 | sadInit = sadInit_altivec; |
| 542 : | canard | 89 | sad16 = sad16_altivec; |
| 543 : | sad8 = sad8_altivec; | ||
| 544 : | dev16 = dev16_altivec; | ||
| 545 : | canard | 71 | #else |
| 546 : | canard | 52 | calc_cbp = calc_cbp_ppc; |
| 547 : | #endif | ||
| 548 : | canard | 71 | #endif |
| 549 : | edgomez | 195 | |
| 550 : | Isibaar | 3 | return XVID_ERR_OK; |
| 551 : | } | ||
| 552 : | |||
| 553 : | suxen_drol | 634 | |
| 554 : | |||
| 555 : | static int | ||
| 556 : | xvid_init_convert(XVID_INIT_CONVERTINFO* convert) | ||
| 557 : | { | ||
| 558 : | suxen_drol | 698 | // const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); |
| 559 : | suxen_drol | 634 | const int width = convert->width; |
| 560 : | const int height = convert->height; | ||
| 561 : | const int width2 = convert->width/2; | ||
| 562 : | const int height2 = convert->height/2; | ||
| 563 : | IMAGE img; | ||
| 564 : | |||
| 565 : | switch (convert->input.colorspace & ~XVID_CSP_VFLIP) | ||
| 566 : | { | ||
| 567 : | case XVID_CSP_YV12 : | ||
| 568 : | img.y = convert->input.y; | ||
| 569 : | img.v = (uint8_t*)convert->input.y + width*height; | ||
| 570 : | img.u = (uint8_t*)convert->input.y + width*height + width2*height2; | ||
| 571 : | image_output(&img, width, height, width, | ||
| 572 : | convert->output.y, convert->output.y_stride, | ||
| 573 : | convert->output.colorspace, convert->interlacing); | ||
| 574 : | break; | ||
| 575 : | |||
| 576 : | default : | ||
| 577 : | return XVID_ERR_FORMAT; | ||
| 578 : | } | ||
| 579 : | |||
| 580 : | |||
| 581 : | emms(); | ||
| 582 : | return XVID_ERR_OK; | ||
| 583 : | } | ||
| 584 : | |||
| 585 : | |||
| 586 : | suxen_drol | 752 | |
| 587 : | void fill8(uint8_t * block, int size, int value) | ||
| 588 : | { | ||
| 589 : | int i; | ||
| 590 : | for (i = 0; i < size; i++) | ||
| 591 : | block[i] = value; | ||
| 592 : | } | ||
| 593 : | |||
| 594 : | void fill16(int16_t * block, int size, int value) | ||
| 595 : | { | ||
| 596 : | int i; | ||
| 597 : | for (i = 0; i < size; i++) | ||
| 598 : | block[i] = value; | ||
| 599 : | } | ||
| 600 : | |||
| 601 : | #define RANDOM(min,max) min + (rand() % (max-min)) | ||
| 602 : | |||
| 603 : | void random8(uint8_t * block, int size, int min, int max) | ||
| 604 : | { | ||
| 605 : | int i; | ||
| 606 : | for (i = 0; i < size; i++) | ||
| 607 : | block[i] = RANDOM(min,max); | ||
| 608 : | } | ||
| 609 : | |||
| 610 : | void random16(int16_t * block, int size, int min, int max) | ||
| 611 : | { | ||
| 612 : | int i; | ||
| 613 : | for (i = 0; i < size; i++) | ||
| 614 : | block[i] = RANDOM(min,max); | ||
| 615 : | } | ||
| 616 : | |||
| 617 : | int compare16(const int16_t * blockA, const int16_t * blockB, int size) | ||
| 618 : | { | ||
| 619 : | int i; | ||
| 620 : | for (i = 0; i < size; i++) | ||
| 621 : | if (blockA[i] != blockB[i]) | ||
| 622 : | return 1; | ||
| 623 : | |||
| 624 : | return 0; | ||
| 625 : | } | ||
| 626 : | |||
| 627 : | suxen_drol | 755 | int diff16(const int16_t * blockA, const int16_t * blockB, int size) |
| 628 : | { | ||
| 629 : | int i, diff = 0; | ||
| 630 : | for (i = 0; i < size; i++) | ||
| 631 : | diff += ABS(blockA[i]-blockB[i]); | ||
| 632 : | return diff; | ||
| 633 : | } | ||
| 634 : | suxen_drol | 752 | |
| 635 : | |||
| 636 : | suxen_drol | 755 | #define XVID_TEST_RANDOM 0x00000001 /* random input data */ |
| 637 : | #define XVID_TEST_VERBOSE 0x00000002 /* verbose error output */ | ||
| 638 : | |||
| 639 : | |||
| 640 : | #define TEST_FORWARD 0x00000001 /* intra */ | ||
| 641 : | #define TEST_FDCT (TEST_FORWARD) | ||
| 642 : | #define TEST_IDCT (0) | ||
| 643 : | |||
| 644 : | int test_transform(void * funcA, void * funcB, const char * nameB, | ||
| 645 : | int test, int flags) | ||
| 646 : | suxen_drol | 752 | { |
| 647 : | suxen_drol | 755 | int i; |
| 648 : | suxen_drol | 752 | int64_t timeSTART; |
| 649 : | int64_t timeA = 0; | ||
| 650 : | int64_t timeB = 0; | ||
| 651 : | DECLARE_ALIGNED_MATRIX(arrayA, 1, 64, int16_t, CACHE_LINE); | ||
| 652 : | DECLARE_ALIGNED_MATRIX(arrayB, 1, 64, int16_t, CACHE_LINE); | ||
| 653 : | suxen_drol | 755 | int min, max; |
| 654 : | int count = 0; | ||
| 655 : | suxen_drol | 752 | |
| 656 : | suxen_drol | 755 | int tmp; |
| 657 : | int min_error = 0x10000*64; | ||
| 658 : | int max_error = 0; | ||
| 659 : | |||
| 660 : | |||
| 661 : | if ((test & TEST_FORWARD)) /* forward */ | ||
| 662 : | suxen_drol | 752 | { |
| 663 : | suxen_drol | 755 | min = -256; |
| 664 : | max = 255; | ||
| 665 : | }else{ /* inverse */ | ||
| 666 : | min = -2048; | ||
| 667 : | max = 2047; | ||
| 668 : | } | ||
| 669 : | |||
| 670 : | for (i = 0; i < 64*64; i++) | ||
| 671 : | { | ||
| 672 : | if ((flags & XVID_TEST_RANDOM)) | ||
| 673 : | suxen_drol | 752 | { |
| 674 : | suxen_drol | 755 | random16(arrayA, 64, min, max); |
| 675 : | }else{ | ||
| 676 : | fill16(arrayA, 64, i); | ||
| 677 : | } | ||
| 678 : | memcpy(arrayB, arrayA, 64*sizeof(int16_t)); | ||
| 679 : | suxen_drol | 752 | |
| 680 : | suxen_drol | 755 | if ((test & TEST_FORWARD)) |
| 681 : | { | ||
| 682 : | suxen_drol | 752 | timeSTART = read_counter(); |
| 683 : | suxen_drol | 755 | ((fdctFunc*)funcA)(arrayA); |
| 684 : | suxen_drol | 752 | timeA += read_counter() - timeSTART; |
| 685 : | |||
| 686 : | timeSTART = read_counter(); | ||
| 687 : | suxen_drol | 755 | ((fdctFunc*)funcB)(arrayB); |
| 688 : | suxen_drol | 752 | timeB += read_counter() - timeSTART; |
| 689 : | suxen_drol | 755 | } |
| 690 : | else | ||
| 691 : | { | ||
| 692 : | timeSTART = read_counter(); | ||
| 693 : | ((idctFunc*)funcA)(arrayA); | ||
| 694 : | timeA += read_counter() - timeSTART; | ||
| 695 : | suxen_drol | 752 | |
| 696 : | suxen_drol | 755 | timeSTART = read_counter(); |
| 697 : | ((idctFunc*)funcB)(arrayB); | ||
| 698 : | timeB += read_counter() - timeSTART; | ||
| 699 : | suxen_drol | 752 | } |
| 700 : | suxen_drol | 755 | |
| 701 : | tmp = diff16(arrayA, arrayB, 64) / 64; | ||
| 702 : | if (tmp > max_error) | ||
| 703 : | max_error = tmp; | ||
| 704 : | if (tmp < min_error) | ||
| 705 : | min_error = tmp; | ||
| 706 : | |||
| 707 : | count++; | ||
| 708 : | suxen_drol | 752 | } |
| 709 : | |||
| 710 : | suxen_drol | 755 | /* print the "average difference" of best/worst transforms */ |
| 711 : | printf("%s:\t%I64i\t(min_error:%i, max_error:%i)\n", nameB, timeB / count, min_error, max_error); | ||
| 712 : | suxen_drol | 752 | |
| 713 : | return 0; | ||
| 714 : | } | ||
| 715 : | |||
| 716 : | suxen_drol | 755 | |
| 717 : | #define TEST_QUANT 0x00000001 /* forward quantization */ | ||
| 718 : | #define TEST_INTRA 0x00000002 /* intra */ | ||
| 719 : | #define TEST_QUANT_INTRA (TEST_QUANT|TEST_INTRA) | ||
| 720 : | #define TEST_QUANT_INTER (TEST_QUANT) | ||
| 721 : | #define TEST_DEQUANT_INTRA (TEST_INTRA) | ||
| 722 : | #define TEST_DEQUANT_INTER (0) | ||
| 723 : | |||
| 724 : | int test_quant(void * funcA, void * funcB, const char * nameB, | ||
| 725 : | int test, int flags) | ||
| 726 : | suxen_drol | 752 | { |
| 727 : | int q,i; | ||
| 728 : | int64_t timeSTART; | ||
| 729 : | int64_t timeA = 0; | ||
| 730 : | int64_t timeB = 0; | ||
| 731 : | suxen_drol | 755 | int retA, retB; |
| 732 : | suxen_drol | 752 | DECLARE_ALIGNED_MATRIX(arrayX, 1, 64, int16_t, CACHE_LINE); |
| 733 : | DECLARE_ALIGNED_MATRIX(arrayA, 1, 64, int16_t, CACHE_LINE); | ||
| 734 : | DECLARE_ALIGNED_MATRIX(arrayB, 1, 64, int16_t, CACHE_LINE); | ||
| 735 : | suxen_drol | 755 | int min, max; |
| 736 : | int count = 0; | ||
| 737 : | int errors = 0; | ||
| 738 : | |||
| 739 : | if ((test & TEST_QUANT)) /* quant */ | ||
| 740 : | { | ||
| 741 : | min = -2048; | ||
| 742 : | max = 2047; | ||
| 743 : | }else{ /* dequant */ | ||
| 744 : | min = -256; | ||
| 745 : | max = 255; | ||
| 746 : | } | ||
| 747 : | |||
| 748 : | suxen_drol | 752 | for (q = 1; q <= 31; q++) /* quantizer */ |
| 749 : | { | ||
| 750 : | for (i = min; i < max; i++) /* input coeff */ | ||
| 751 : | { | ||
| 752 : | suxen_drol | 755 | if ((flags & XVID_TEST_RANDOM)) |
| 753 : | { | ||
| 754 : | random16(arrayX, 64, min, max); | ||
| 755 : | }else{ | ||
| 756 : | fill16(arrayX, 64, i); | ||
| 757 : | } | ||
| 758 : | suxen_drol | 752 | |
| 759 : | suxen_drol | 755 | if ((test & TEST_INTRA)) /* intra */ |
| 760 : | { | ||
| 761 : | timeSTART = read_counter(); | ||
| 762 : | ((quanth263_intraFunc*)funcA)(arrayA, arrayX, q, q); | ||
| 763 : | timeA += read_counter() - timeSTART; | ||
| 764 : | suxen_drol | 752 | |
| 765 : | suxen_drol | 755 | timeSTART = read_counter(); |
| 766 : | ((quanth263_intraFunc*)funcB)(arrayB, arrayX, q, q); | ||
| 767 : | timeB += read_counter() - timeSTART; | ||
| 768 : | } | ||
| 769 : | else /* inter */ | ||
| 770 : | { | ||
| 771 : | timeSTART = read_counter(); | ||
| 772 : | retA = ((quanth263_interFunc*)funcA)(arrayA, arrayX, q); | ||
| 773 : | timeA += read_counter() - timeSTART; | ||
| 774 : | suxen_drol | 752 | |
| 775 : | suxen_drol | 755 | timeSTART = read_counter(); |
| 776 : | retB = ((quanth263_interFunc*)funcB)(arrayB, arrayX, q); | ||
| 777 : | timeB += read_counter() - timeSTART; | ||
| 778 : | } | ||
| 779 : | |||
| 780 : | /* compare return value from quant_inter, and compare (de)quantiz'd arrays */ | ||
| 781 : | if ( ((test&TEST_QUANT) && !(test&TEST_INTRA) && retA != retB ) || | ||
| 782 : | compare16(arrayA, arrayB, 64)) | ||
| 783 : | suxen_drol | 752 | { |
| 784 : | suxen_drol | 755 | errors++; |
| 785 : | if ((flags & XVID_TEST_VERBOSE)) | ||
| 786 : | printf("%s error: q=%i, i=%i\n", nameB, q, i); | ||
| 787 : | suxen_drol | 752 | } |
| 788 : | suxen_drol | 755 | |
| 789 : | count++; | ||
| 790 : | suxen_drol | 752 | } |
| 791 : | } | ||
| 792 : | |||
| 793 : | suxen_drol | 755 | printf("%s:\t%I64i", nameB, timeB / count); |
| 794 : | if (errors>0) | ||
| 795 : | printf("\t(%i errors out of %i)", errors, count); | ||
| 796 : | printf("\n"); | ||
| 797 : | |||
| 798 : | suxen_drol | 752 | return 0; |
| 799 : | } | ||
| 800 : | |||
| 801 : | |||
| 802 : | |||
| 803 : | suxen_drol | 755 | int xvid_init_test(int flags) |
| 804 : | suxen_drol | 752 | { |
| 805 : | int cpu_flags; | ||
| 806 : | |||
| 807 : | suxen_drol | 755 | srand(time(0)); |
| 808 : | |||
| 809 : | suxen_drol | 752 | printf("xvid_init_test\n"); |
| 810 : | |||
| 811 : | #if defined(ARCH_X86) | ||
| 812 : | cpu_flags = check_cpu_features(); | ||
| 813 : | suxen_drol | 755 | idct_int32_init(); |
| 814 : | suxen_drol | 752 | emms_mmx(); |
| 815 : | |||
| 816 : | suxen_drol | 755 | printf("--- fdct ---\n"); |
| 817 : | test_transform(fdct_int32, fdct_int32, "c", TEST_FDCT, flags); | ||
| 818 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMX) |
| 819 : | suxen_drol | 755 | test_transform(fdct_int32, fdct_mmx, "mmx", TEST_FDCT, flags); |
| 820 : | if (cpu_flags & XVID_CPU_SSE2) | ||
| 821 : | test_transform(fdct_int32, fdct_sse2, "sse2", TEST_FDCT, flags); | ||
| 822 : | |||
| 823 : | printf("\n--- idct ---\n"); | ||
| 824 : | test_transform(idct_int32, idct_int32, "c", TEST_IDCT, flags); | ||
| 825 : | if (cpu_flags & XVID_CPU_MMX) | ||
| 826 : | test_transform(idct_int32, idct_mmx, "mmx", TEST_IDCT, flags); | ||
| 827 : | if (cpu_flags & XVID_CPU_MMXEXT) | ||
| 828 : | test_transform(idct_int32, idct_xmm, "xmm", TEST_IDCT, flags); | ||
| 829 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_3DNOWEXT) |
| 830 : | suxen_drol | 755 | test_transform(idct_int32, idct_3dne, "3dne", TEST_IDCT, flags); |
| 831 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_SSE2) |
| 832 : | suxen_drol | 755 | test_transform(idct_int32, idct_sse2, "sse2", TEST_IDCT, flags); |
| 833 : | suxen_drol | 752 | |
| 834 : | suxen_drol | 755 | printf("\n--- quant intra ---\n"); |
| 835 : | test_quant(quant_intra_c, quant_intra_c, "c", TEST_QUANT_INTRA, flags); | ||
| 836 : | if (cpu_flags & XVID_CPU_MMX) | ||
| 837 : | test_quant(quant_intra_c, quant_intra_mmx, "mmx", TEST_QUANT_INTRA, flags); | ||
| 838 : | if (cpu_flags & XVID_CPU_3DNOWEXT) | ||
| 839 : | test_quant(quant_intra_c, quant_intra_3dne, "3dne", TEST_QUANT_INTRA, flags); | ||
| 840 : | if (cpu_flags & XVID_CPU_SSE2) | ||
| 841 : | test_quant(quant_intra_c, quant_intra_sse2, "sse2", TEST_QUANT_INTRA, flags); | ||
| 842 : | |||
| 843 : | suxen_drol | 752 | printf("\n--- quant inter ---\n"); |
| 844 : | suxen_drol | 755 | test_quant(quant_inter_c, quant_inter_c, "c", TEST_QUANT_INTER, flags); |
| 845 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMX) |
| 846 : | suxen_drol | 755 | test_quant(quant_inter_c, quant_inter_mmx, "mmx", TEST_QUANT_INTER, flags); |
| 847 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_3DNOWEXT) |
| 848 : | suxen_drol | 755 | test_quant(quant_inter_c, quant_inter_3dne, "3dne", TEST_QUANT_INTER, flags); |
| 849 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_SSE2) |
| 850 : | suxen_drol | 755 | test_quant(quant_inter_c, quant_inter_sse2, "sse2", TEST_QUANT_INTER, flags); |
| 851 : | suxen_drol | 752 | |
| 852 : | suxen_drol | 755 | printf("\n--- dequant intra ---\n"); |
| 853 : | test_quant(dequant_intra_c, dequant_intra_c, "c", TEST_DEQUANT_INTRA, flags); | ||
| 854 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMX) |
| 855 : | suxen_drol | 755 | test_quant(dequant_intra_c, dequant_intra_mmx, "mmx", TEST_DEQUANT_INTRA, flags); |
| 856 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMXEXT) |
| 857 : | suxen_drol | 755 | test_quant(dequant_intra_c, dequant_intra_xmm, "xmm", TEST_DEQUANT_INTRA, flags); |
| 858 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_3DNOWEXT) |
| 859 : | suxen_drol | 755 | test_quant(dequant_intra_c, dequant_intra_3dne, "3dne", TEST_DEQUANT_INTRA, flags); |
| 860 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_SSE2) |
| 861 : | suxen_drol | 755 | test_quant(dequant_intra_c, dequant_intra_sse2, "sse2", TEST_DEQUANT_INTRA, flags); |
| 862 : | suxen_drol | 752 | |
| 863 : | printf("\n--- dequant inter ---\n"); | ||
| 864 : | suxen_drol | 755 | test_quant(dequant_inter_c, dequant_inter_c, "c", TEST_DEQUANT_INTER, flags); |
| 865 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMX) |
| 866 : | suxen_drol | 755 | test_quant(dequant_inter_c, dequant_inter_mmx, "mmx", TEST_DEQUANT_INTER, flags); |
| 867 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMXEXT) |
| 868 : | suxen_drol | 755 | test_quant(dequant_inter_c, dequant_inter_xmm, "xmm", TEST_DEQUANT_INTER, flags); |
| 869 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_3DNOWEXT) |
| 870 : | suxen_drol | 755 | test_quant(dequant_inter_c, dequant_inter_3dne, "3dne", TEST_DEQUANT_INTER, flags); |
| 871 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_SSE2) |
| 872 : | suxen_drol | 755 | test_quant(dequant_inter_c, dequant_inter_sse2, "sse2", TEST_DEQUANT_INTER, flags); |
| 873 : | suxen_drol | 752 | |
| 874 : | printf("\n--- quant4_intra ---\n"); | ||
| 875 : | suxen_drol | 755 | test_quant(quant4_intra_c, quant4_intra_c, "c", TEST_QUANT_INTRA, flags); |
| 876 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMX) |
| 877 : | suxen_drol | 755 | test_quant(quant4_intra_c, quant4_intra_mmx, "mmx", TEST_QUANT_INTRA, flags); |
| 878 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMXEXT) |
| 879 : | suxen_drol | 755 | test_quant(quant4_intra_c, quant4_intra_xmm, "xmm", TEST_QUANT_INTRA, flags); |
| 880 : | suxen_drol | 752 | |
| 881 : | printf("\n--- quant4_inter ---\n"); | ||
| 882 : | suxen_drol | 755 | test_quant(quant4_inter_c, quant4_inter_c, "c", TEST_QUANT_INTER, flags); |
| 883 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMX) |
| 884 : | suxen_drol | 755 | test_quant(quant4_inter_c, quant4_inter_mmx, "mmx", TEST_QUANT_INTER, flags); |
| 885 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMXEXT) |
| 886 : | suxen_drol | 755 | test_quant(quant4_inter_c, quant4_inter_xmm, "xmm", TEST_QUANT_INTER, flags); |
| 887 : | suxen_drol | 752 | |
| 888 : | printf("\n--- dequant4_intra ---\n"); | ||
| 889 : | suxen_drol | 755 | test_quant(dequant4_intra_c, dequant4_intra_c, "c", TEST_DEQUANT_INTRA, flags); |
| 890 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMX) |
| 891 : | suxen_drol | 755 | test_quant(dequant4_intra_c, dequant4_intra_mmx, "mmx", TEST_DEQUANT_INTRA, flags); |
| 892 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_3DNOWEXT) |
| 893 : | suxen_drol | 755 | test_quant(dequant4_intra_c, dequant4_intra_3dne, "3dne", TEST_DEQUANT_INTRA, flags); |
| 894 : | suxen_drol | 752 | |
| 895 : | printf("\n--- dequant4_inter ---\n"); | ||
| 896 : | suxen_drol | 755 | test_quant(dequant4_inter_c, dequant4_inter_c, "c", TEST_DEQUANT_INTER, flags); |
| 897 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_MMX) |
| 898 : | suxen_drol | 755 | test_quant(dequant4_inter_c, dequant4_inter_mmx, "mmx", TEST_DEQUANT_INTER, flags); |
| 899 : | suxen_drol | 752 | if (cpu_flags & XVID_CPU_3DNOWEXT) |
| 900 : | suxen_drol | 755 | test_quant(dequant4_inter_c, dequant4_inter_3dne, "3dne", TEST_DEQUANT_INTER, flags); |
| 901 : | suxen_drol | 752 | |
| 902 : | emms_mmx(); | ||
| 903 : | |||
| 904 : | #endif | ||
| 905 : | |||
| 906 : | return XVID_ERR_OK; | ||
| 907 : | } | ||
| 908 : | |||
| 909 : | |||
| 910 : | suxen_drol | 634 | int |
| 911 : | xvid_init(void *handle, | ||
| 912 : | int opt, | ||
| 913 : | void *param1, | ||
| 914 : | void *param2) | ||
| 915 : | { | ||
| 916 : | switch(opt) | ||
| 917 : | { | ||
| 918 : | case XVID_INIT_INIT : | ||
| 919 : | return xvid_init_init((XVID_INIT_PARAM*)param1); | ||
| 920 : | |||
| 921 : | case XVID_INIT_CONVERT : | ||
| 922 : | return xvid_init_convert((XVID_INIT_CONVERTINFO*)param1); | ||
| 923 : | |||
| 924 : | suxen_drol | 752 | case XVID_INIT_TEST : |
| 925 : | suxen_drol | 755 | return xvid_init_test((int)param1); |
| 926 : | suxen_drol | 752 | |
| 927 : | suxen_drol | 634 | default : |
| 928 : | return XVID_ERR_FAIL; | ||
| 929 : | } | ||
| 930 : | } | ||
| 931 : | |||
| 932 : | edgomez | 200 | /***************************************************************************** |
| 933 : | * XviD Native decoder entry point | ||
| 934 : | * | ||
| 935 : | * This function is just a wrapper to all the option cases. | ||
| 936 : | * | ||
| 937 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
| 938 : | * else returns the wrapped function result | ||
| 939 : | * | ||
| 940 : | ****************************************************************************/ | ||
| 941 : | |||
| 942 : | edgomez | 195 | int |
| 943 : | xvid_decore(void *handle, | ||
| 944 : | int opt, | ||
| 945 : | void *param1, | ||
| 946 : | void *param2) | ||
| 947 : | Isibaar | 3 | { |
| 948 : | edgomez | 195 | switch (opt) { |
| 949 : | case XVID_DEC_DECODE: | ||
| 950 : | suxen_drol | 631 | return decoder_decode((DECODER *) handle, (XVID_DEC_FRAME *) param1, (XVID_DEC_STATS*) param2); |
| 951 : | Isibaar | 3 | |
| 952 : | edgomez | 195 | case XVID_DEC_CREATE: |
| 953 : | chenm001 | 274 | return decoder_create((XVID_DEC_PARAM *) param1); |
| 954 : | Isibaar | 3 | |
| 955 : | edgomez | 195 | case XVID_DEC_DESTROY: |
| 956 : | return decoder_destroy((DECODER *) handle); | ||
| 957 : | |||
| 958 : | Isibaar | 3 | default: |
| 959 : | edgomez | 195 | return XVID_ERR_FAIL; |
| 960 : | } | ||
| 961 : | Isibaar | 3 | } |
| 962 : | |||
| 963 : | |||
| 964 : | edgomez | 200 | /***************************************************************************** |
| 965 : | * XviD Native encoder entry point | ||
| 966 : | * | ||
| 967 : | * This function is just a wrapper to all the option cases. | ||
| 968 : | * | ||
| 969 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
| 970 : | * else returns the wrapped function result | ||
| 971 : | * | ||
| 972 : | ****************************************************************************/ | ||
| 973 : | |||
| 974 : | edgomez | 195 | int |
| 975 : | xvid_encore(void *handle, | ||
| 976 : | int opt, | ||
| 977 : | void *param1, | ||
| 978 : | void *param2) | ||
| 979 : | Isibaar | 3 | { |
| 980 : | edgomez | 195 | switch (opt) { |
| 981 : | case XVID_ENC_ENCODE: | ||
| 982 : | chl | 620 | |
| 983 : | suxen_drol | 232 | if (((Encoder *) handle)->mbParam.max_bframes >= 0) |
| 984 : | return encoder_encode_bframes((Encoder *) handle, (XVID_ENC_FRAME *) param1, | ||
| 985 : | (XVID_ENC_STATS *) param2); | ||
| 986 : | else | ||
| 987 : | edgomez | 195 | return encoder_encode((Encoder *) handle, (XVID_ENC_FRAME *) param1, |
| 988 : | (XVID_ENC_STATS *) param2); | ||
| 989 : | Isibaar | 3 | |
| 990 : | edgomez | 195 | case XVID_ENC_CREATE: |
| 991 : | return encoder_create((XVID_ENC_PARAM *) param1); | ||
| 992 : | Isibaar | 3 | |
| 993 : | edgomez | 195 | case XVID_ENC_DESTROY: |
| 994 : | return encoder_destroy((Encoder *) handle); | ||
| 995 : | |||
| 996 : | Isibaar | 3 | default: |
| 997 : | edgomez | 195 | return XVID_ERR_FAIL; |
| 998 : | } | ||
| 999 : | Isibaar | 3 | } |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |