Parent Directory
| 
Revision Log
Revision 1290 - (view) (download)
| 1 : | edgomez | 200 | /***************************************************************************** |
| 2 : | edgomez | 201 | * |
| 3 : | * XVID MPEG-4 VIDEO CODEC | ||
| 4 : | * - Native API implementation - | ||
| 5 : | * | ||
| 6 : | edgomez | 1054 | * Copyright(C) 2001-2003 Peter Ross <pross@xvid.org> |
| 7 : | * | ||
| 8 : | edgomez | 851 | * 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 | ||
| 10 : | * the Free Software Foundation ; either version 2 of the License, or | ||
| 11 : | edgomez | 201 | * (at your option) any later version. |
| 12 : | * | ||
| 13 : | * This program is distributed in the hope that it will be useful, | ||
| 14 : | edgomez | 851 | * but WITHOUT ANY WARRANTY ; without even the implied warranty of |
| 15 : | edgomez | 201 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 16 : | * GNU General Public License for more details. | ||
| 17 : | * | ||
| 18 : | * You should have received a copy of the GNU General Public License | ||
| 19 : | edgomez | 851 | * along with this program ; if not, write to the Free Software |
| 20 : | edgomez | 201 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
| 21 : | * | ||
| 22 : | edgomez | 1290 | * $Id: xvid.c,v 1.45.2.26 2003-12-20 22:20:54 edgomez Exp $ |
| 23 : | edgomez | 410 | * |
| 24 : | edgomez | 201 | ****************************************************************************/ |
| 25 : | chenm001 | 274 | |
| 26 : | edgomez | 851 | #include <stdio.h> |
| 27 : | #include <stdlib.h> | ||
| 28 : | #include <string.h> | ||
| 29 : | #include <time.h> | ||
| 30 : | |||
| 31 : | Isibaar | 3 | #include "xvid.h" |
| 32 : | #include "decoder.h" | ||
| 33 : | #include "encoder.h" | ||
| 34 : | #include "bitstream/cbp.h" | ||
| 35 : | #include "dct/idct.h" | ||
| 36 : | #include "dct/fdct.h" | ||
| 37 : | #include "image/colorspace.h" | ||
| 38 : | #include "image/interpolate8x8.h" | ||
| 39 : | edgomez | 851 | #include "image/reduced.h" |
| 40 : | Isibaar | 3 | #include "utils/mem_transfer.h" |
| 41 : | edgomez | 851 | #include "utils/mbfunctions.h" |
| 42 : | edgomez | 1174 | #include "quant/quant.h" |
| 43 : | ia64p | 299 | #include "motion/motion.h" |
| 44 : | Isibaar | 3 | #include "motion/sad.h" |
| 45 : | #include "utils/emms.h" | ||
| 46 : | #include "utils/timer.h" | ||
| 47 : | Isibaar | 100 | #include "bitstream/mbcoding.h" |
| 48 : | Isibaar | 1125 | #include "image/qpel.h" |
| 49 : | Isibaar | 1256 | #include "image/postprocessing.h" |
| 50 : | Isibaar | 3 | |
| 51 : | suxen_drol | 1031 | #if defined(_DEBUG) |
| 52 : | unsigned int xvid_debug = 0; /* xvid debug mask */ | ||
| 53 : | #endif | ||
| 54 : | |||
| 55 : | edgomez | 851 | #if defined(ARCH_IS_IA32) |
| 56 : | #if defined(_MSC_VER) | ||
| 57 : | # include <windows.h> | ||
| 58 : | suxen_drol | 311 | #else |
| 59 : | edgomez | 851 | # include <signal.h> |
| 60 : | # include <setjmp.h> | ||
| 61 : | suxen_drol | 311 | |
| 62 : | edgomez | 851 | static jmp_buf mark; |
| 63 : | suxen_drol | 311 | |
| 64 : | edgomez | 851 | static void |
| 65 : | sigill_handler(int signal) | ||
| 66 : | { | ||
| 67 : | longjmp(mark, 1); | ||
| 68 : | } | ||
| 69 : | suxen_drol | 311 | #endif |
| 70 : | |||
| 71 : | |||
| 72 : | /* | ||
| 73 : | edgomez | 874 | * Calls the funcptr, and returns whether SIGILL (illegal instruction) was |
| 74 : | * signalled | ||
| 75 : | * | ||
| 76 : | * Return values: | ||
| 77 : | * -1 : could not determine | ||
| 78 : | * 0 : SIGILL was *not* signalled | ||
| 79 : | * 1 : SIGILL was signalled | ||
| 80 : | */ | ||
| 81 : | suxen_drol | 311 | |
| 82 : | int | ||
| 83 : | sigill_check(void (*func)()) | ||
| 84 : | { | ||
| 85 : | edgomez | 851 | #if defined(_MSC_VER) |
| 86 : | suxen_drol | 311 | _try { |
| 87 : | func(); | ||
| 88 : | } | ||
| 89 : | _except(EXCEPTION_EXECUTE_HANDLER) { | ||
| 90 : | |||
| 91 : | if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION) | ||
| 92 : | return 1; | ||
| 93 : | } | ||
| 94 : | return 0; | ||
| 95 : | #else | ||
| 96 : | void * old_handler; | ||
| 97 : | int jmpret; | ||
| 98 : | |||
| 99 : | |||
| 100 : | old_handler = signal(SIGILL, sigill_handler); | ||
| 101 : | if (old_handler == SIG_ERR) | ||
| 102 : | { | ||
| 103 : | return -1; | ||
| 104 : | } | ||
| 105 : | |||
| 106 : | jmpret = setjmp(mark); | ||
| 107 : | if (jmpret == 0) | ||
| 108 : | { | ||
| 109 : | func(); | ||
| 110 : | } | ||
| 111 : | |||
| 112 : | signal(SIGILL, old_handler); | ||
| 113 : | |||
| 114 : | return jmpret; | ||
| 115 : | #endif | ||
| 116 : | } | ||
| 117 : | #endif | ||
| 118 : | |||
| 119 : | edgomez | 851 | |
| 120 : | /* detect cpu flags */ | ||
| 121 : | static unsigned int | ||
| 122 : | detect_cpu_flags() | ||
| 123 : | { | ||
| 124 : | /* enable native assembly optimizations by default */ | ||
| 125 : | unsigned int cpu_flags = XVID_CPU_ASM; | ||
| 126 : | |||
| 127 : | #if defined(ARCH_IS_IA32) | ||
| 128 : | cpu_flags |= check_cpu_features(); | ||
| 129 : | if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) | ||
| 130 : | cpu_flags &= ~XVID_CPU_SSE; | ||
| 131 : | |||
| 132 : | if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) | ||
| 133 : | cpu_flags &= ~XVID_CPU_SSE2; | ||
| 134 : | #endif | ||
| 135 : | |||
| 136 : | #if defined(ARCH_IS_PPC) | ||
| 137 : | #if defined(ARCH_IS_PPC_ALTIVEC) | ||
| 138 : | cpu_flags |= XVID_CPU_ALTIVEC; | ||
| 139 : | #endif | ||
| 140 : | #endif | ||
| 141 : | |||
| 142 : | return cpu_flags; | ||
| 143 : | } | ||
| 144 : | |||
| 145 : | |||
| 146 : | edgomez | 200 | /***************************************************************************** |
| 147 : | * XviD Init Entry point | ||
| 148 : | * | ||
| 149 : | * Well this function initialize all internal function pointers according | ||
| 150 : | * to the CPU features forced by the library client or autodetected (depending | ||
| 151 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
| 152 : | * image colorspace transformation tables. | ||
| 153 : | edgomez | 1161 | * |
| 154 : | edgomez | 200 | * Returned value : XVID_ERR_OK |
| 155 : | * + API_VERSION in the input XVID_INIT_PARAM structure | ||
| 156 : | * + core build " " " " " | ||
| 157 : | * | ||
| 158 : | ****************************************************************************/ | ||
| 159 : | |||
| 160 : | edgomez | 851 | |
| 161 : | edgomez | 1161 | static |
| 162 : | suxen_drol | 890 | int xvid_gbl_init(xvid_gbl_init_t * init) |
| 163 : | Isibaar | 3 | { |
| 164 : | suxen_drol | 890 | unsigned int cpu_flags; |
| 165 : | Isibaar | 3 | |
| 166 : | edgomez | 1107 | if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */ |
| 167 : | suxen_drol | 890 | return XVID_ERR_VERSION; |
| 168 : | suxen_drol | 234 | |
| 169 : | suxen_drol | 890 | cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags(); |
| 170 : | suxen_drol | 234 | |
| 171 : | edgomez | 200 | /* Initialize the function pointers */ |
| 172 : | Isibaar | 3 | idct_int32_init(); |
| 173 : | Isibaar | 100 | init_vlc_tables(); |
| 174 : | |||
| 175 : | edgomez | 200 | /* Fixed Point Forward/Inverse DCT transformations */ |
| 176 : | Isibaar | 3 | fdct = fdct_int32; |
| 177 : | idct = idct_int32; | ||
| 178 : | |||
| 179 : | edgomez | 200 | /* Only needed on PPC Altivec archs */ |
| 180 : | canard | 115 | sadInit = 0; |
| 181 : | edgomez | 195 | |
| 182 : | edgomez | 200 | /* Restore FPU context : emms_c is a nop functions */ |
| 183 : | Isibaar | 3 | emms = emms_c; |
| 184 : | |||
| 185 : | Isibaar | 1125 | /* Qpel stuff */ |
| 186 : | xvid_QP_Funcs = &xvid_QP_Funcs_C; | ||
| 187 : | xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C; | ||
| 188 : | edgomez | 1127 | xvid_Init_QP(); |
| 189 : | Isibaar | 1125 | |
| 190 : | edgomez | 200 | /* Quantization functions */ |
| 191 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_c; |
| 192 : | quant_h263_inter = quant_h263_inter_c; | ||
| 193 : | dequant_h263_intra = dequant_h263_intra_c; | ||
| 194 : | dequant_h263_inter = dequant_h263_inter_c; | ||
| 195 : | Isibaar | 3 | |
| 196 : | edgomez | 1174 | quant_mpeg_intra = quant_mpeg_intra_c; |
| 197 : | quant_mpeg_inter = quant_mpeg_inter_c; | ||
| 198 : | dequant_mpeg_intra = dequant_mpeg_intra_c; | ||
| 199 : | dequant_mpeg_inter = dequant_mpeg_inter_c; | ||
| 200 : | Isibaar | 3 | |
| 201 : | edgomez | 200 | /* Block transfer related functions */ |
| 202 : | Isibaar | 3 | transfer_8to16copy = transfer_8to16copy_c; |
| 203 : | transfer_16to8copy = transfer_16to8copy_c; | ||
| 204 : | edgomez | 200 | transfer_8to16sub = transfer_8to16sub_c; |
| 205 : | edgomez | 851 | transfer_8to16subro = transfer_8to16subro_c; |
| 206 : | suxen_drol | 118 | transfer_8to16sub2 = transfer_8to16sub2_c; |
| 207 : | edgomez | 200 | transfer_16to8add = transfer_16to8add_c; |
| 208 : | transfer8x8_copy = transfer8x8_copy_c; | ||
| 209 : | Isibaar | 3 | |
| 210 : | edgomez | 851 | /* Interlacing functions */ |
| 211 : | MBFieldTest = MBFieldTest_c; | ||
| 212 : | |||
| 213 : | edgomez | 200 | /* Image interpolation related functions */ |
| 214 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; | ||
| 215 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; | ||
| 216 : | Isibaar | 3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
| 217 : | |||
| 218 : | edgomez | 851 | interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; |
| 219 : | interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; | ||
| 220 : | interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; | ||
| 221 : | |||
| 222 : | interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c; | ||
| 223 : | interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c; | ||
| 224 : | interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c; | ||
| 225 : | |||
| 226 : | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c; | ||
| 227 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c; | ||
| 228 : | |||
| 229 : | interpolate8x8_avg2 = interpolate8x8_avg2_c; | ||
| 230 : | interpolate8x8_avg4 = interpolate8x8_avg4_c; | ||
| 231 : | |||
| 232 : | edgomez | 1174 | /* reduced resolution */ |
| 233 : | edgomez | 851 | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_C; |
| 234 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_C; | ||
| 235 : | vfilter_31 = xvid_VFilter_31_C; | ||
| 236 : | hfilter_31 = xvid_HFilter_31_C; | ||
| 237 : | filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_C; | ||
| 238 : | filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_C; | ||
| 239 : | |||
| 240 : | edgomez | 200 | /* Initialize internal colorspace transformation tables */ |
| 241 : | Isibaar | 3 | colorspace_init(); |
| 242 : | |||
| 243 : | edgomez | 200 | /* All colorspace transformation functions User Format->YV12 */ |
| 244 : | edgomez | 851 | yv12_to_yv12 = yv12_to_yv12_c; |
| 245 : | rgb555_to_yv12 = rgb555_to_yv12_c; | ||
| 246 : | rgb565_to_yv12 = rgb565_to_yv12_c; | ||
| 247 : | bgr_to_yv12 = bgr_to_yv12_c; | ||
| 248 : | bgra_to_yv12 = bgra_to_yv12_c; | ||
| 249 : | abgr_to_yv12 = abgr_to_yv12_c; | ||
| 250 : | rgba_to_yv12 = rgba_to_yv12_c; | ||
| 251 : | edgomez | 1290 | argb_to_yv12 = argb_to_yv12_c; |
| 252 : | edgomez | 851 | yuyv_to_yv12 = yuyv_to_yv12_c; |
| 253 : | uyvy_to_yv12 = uyvy_to_yv12_c; | ||
| 254 : | Isibaar | 3 | |
| 255 : | edgomez | 851 | rgb555i_to_yv12 = rgb555i_to_yv12_c; |
| 256 : | rgb565i_to_yv12 = rgb565i_to_yv12_c; | ||
| 257 : | bgri_to_yv12 = bgri_to_yv12_c; | ||
| 258 : | bgrai_to_yv12 = bgrai_to_yv12_c; | ||
| 259 : | abgri_to_yv12 = abgri_to_yv12_c; | ||
| 260 : | rgbai_to_yv12 = rgbai_to_yv12_c; | ||
| 261 : | edgomez | 1290 | argbi_to_yv12 = argbi_to_yv12_c; |
| 262 : | edgomez | 851 | yuyvi_to_yv12 = yuyvi_to_yv12_c; |
| 263 : | uyvyi_to_yv12 = uyvyi_to_yv12_c; | ||
| 264 : | |||
| 265 : | edgomez | 200 | /* All colorspace transformation functions YV12->User format */ |
| 266 : | edgomez | 851 | yv12_to_rgb555 = yv12_to_rgb555_c; |
| 267 : | yv12_to_rgb565 = yv12_to_rgb565_c; | ||
| 268 : | yv12_to_bgr = yv12_to_bgr_c; | ||
| 269 : | yv12_to_bgra = yv12_to_bgra_c; | ||
| 270 : | yv12_to_abgr = yv12_to_abgr_c; | ||
| 271 : | yv12_to_rgba = yv12_to_rgba_c; | ||
| 272 : | edgomez | 1290 | yv12_to_argb = yv12_to_argb_c; |
| 273 : | edgomez | 851 | yv12_to_yuyv = yv12_to_yuyv_c; |
| 274 : | yv12_to_uyvy = yv12_to_uyvy_c; | ||
| 275 : | edgomez | 1161 | |
| 276 : | edgomez | 851 | yv12_to_rgb555i = yv12_to_rgb555i_c; |
| 277 : | yv12_to_rgb565i = yv12_to_rgb565i_c; | ||
| 278 : | yv12_to_bgri = yv12_to_bgri_c; | ||
| 279 : | yv12_to_bgrai = yv12_to_bgrai_c; | ||
| 280 : | yv12_to_abgri = yv12_to_abgri_c; | ||
| 281 : | yv12_to_rgbai = yv12_to_rgbai_c; | ||
| 282 : | edgomez | 1290 | yv12_to_argbi = yv12_to_argbi_c; |
| 283 : | edgomez | 851 | yv12_to_yuyvi = yv12_to_yuyvi_c; |
| 284 : | yv12_to_uyvyi = yv12_to_uyvyi_c; | ||
| 285 : | Isibaar | 3 | |
| 286 : | edgomez | 200 | /* Functions used in motion estimation algorithms */ |
| 287 : | edgomez | 1205 | calc_cbp = calc_cbp_c; |
| 288 : | sad16 = sad16_c; | ||
| 289 : | sad8 = sad8_c; | ||
| 290 : | sad16bi = sad16bi_c; | ||
| 291 : | sad8bi = sad8bi_c; | ||
| 292 : | dev16 = dev16_c; | ||
| 293 : | sad16v = sad16v_c; | ||
| 294 : | sse8_16bit = sse8_16bit_c; | ||
| 295 : | edgomez | 1161 | |
| 296 : | edgomez | 851 | #if defined(ARCH_IS_IA32) |
| 297 : | edgomez | 200 | |
| 298 : | edgomez | 1058 | if ((cpu_flags & XVID_CPU_ASM)) { |
| 299 : | edgomez | 851 | vfilter_31 = xvid_VFilter_31_x86; |
| 300 : | hfilter_31 = xvid_HFilter_31_x86; | ||
| 301 : | } | ||
| 302 : | |||
| 303 : | if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || | ||
| 304 : | (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || | ||
| 305 : | (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2)) | ||
| 306 : | { | ||
| 307 : | /* Restore FPU context : emms_c is a nop functions */ | ||
| 308 : | emms = emms_mmx; | ||
| 309 : | } | ||
| 310 : | |||
| 311 : | if ((cpu_flags & XVID_CPU_MMX)) { | ||
| 312 : | |||
| 313 : | edgomez | 200 | /* Forward and Inverse Discrete Cosine Transformation functions */ |
| 314 : | edgomez | 1190 | fdct = fdct_mmx_skal; |
| 315 : | Isibaar | 1066 | idct = idct_mmx; |
| 316 : | Isibaar | 3 | |
| 317 : | Isibaar | 1125 | /* Qpel stuff */ |
| 318 : | xvid_QP_Funcs = &xvid_QP_Funcs_mmx; | ||
| 319 : | xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx; | ||
| 320 : | |||
| 321 : | edgomez | 200 | /* Quantization related functions */ |
| 322 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_mmx; |
| 323 : | quant_h263_inter = quant_h263_inter_mmx; | ||
| 324 : | dequant_h263_intra = dequant_h263_intra_mmx; | ||
| 325 : | dequant_h263_inter = dequant_h263_inter_mmx; | ||
| 326 : | Isibaar | 3 | |
| 327 : | edgomez | 1174 | quant_mpeg_intra = quant_mpeg_intra_mmx; |
| 328 : | quant_mpeg_inter = quant_mpeg_inter_mmx; | ||
| 329 : | dequant_mpeg_intra = dequant_mpeg_intra_mmx; | ||
| 330 : | dequant_mpeg_inter = dequant_mpeg_inter_mmx; | ||
| 331 : | Isibaar | 3 | |
| 332 : | edgomez | 200 | /* Block related functions */ |
| 333 : | Isibaar | 3 | transfer_8to16copy = transfer_8to16copy_mmx; |
| 334 : | transfer_16to8copy = transfer_16to8copy_mmx; | ||
| 335 : | edgomez | 200 | transfer_8to16sub = transfer_8to16sub_mmx; |
| 336 : | edgomez | 851 | transfer_8to16subro = transfer_8to16subro_mmx; |
| 337 : | edgomez | 236 | transfer_8to16sub2 = transfer_8to16sub2_mmx; |
| 338 : | edgomez | 200 | transfer_16to8add = transfer_16to8add_mmx; |
| 339 : | transfer8x8_copy = transfer8x8_copy_mmx; | ||
| 340 : | Isibaar | 3 | |
| 341 : | edgomez | 851 | /* Interlacing Functions */ |
| 342 : | MBFieldTest = MBFieldTest_mmx; | ||
| 343 : | edgomez | 236 | |
| 344 : | edgomez | 200 | /* Image Interpolation related functions */ |
| 345 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; | ||
| 346 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; | ||
| 347 : | Isibaar | 3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
| 348 : | |||
| 349 : | edgomez | 851 | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; |
| 350 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; | ||
| 351 : | |||
| 352 : | interpolate8x8_avg2 = interpolate8x8_avg2_mmx; | ||
| 353 : | interpolate8x8_avg4 = interpolate8x8_avg4_mmx; | ||
| 354 : | |||
| 355 : | /* reduced resolution */ | ||
| 356 : | copy_upsampled_8x8_16to8 = xvid_Copy_Upsampled_8x8_16To8_mmx; | ||
| 357 : | add_upsampled_8x8_16to8 = xvid_Add_Upsampled_8x8_16To8_mmx; | ||
| 358 : | hfilter_31 = xvid_HFilter_31_mmx; | ||
| 359 : | filter_18x18_to_8x8 = xvid_Filter_18x18_To_8x8_mmx; | ||
| 360 : | filter_diff_18x18_to_8x8 = xvid_Filter_Diff_18x18_To_8x8_mmx; | ||
| 361 : | |||
| 362 : | /* image input xxx_to_yv12 related functions */ | ||
| 363 : | yv12_to_yv12 = yv12_to_yv12_mmx; | ||
| 364 : | bgr_to_yv12 = bgr_to_yv12_mmx; | ||
| 365 : | bgra_to_yv12 = bgra_to_yv12_mmx; | ||
| 366 : | edgomez | 200 | yuyv_to_yv12 = yuyv_to_yv12_mmx; |
| 367 : | uyvy_to_yv12 = uyvy_to_yv12_mmx; | ||
| 368 : | Isibaar | 3 | |
| 369 : | edgomez | 851 | /* image output yv12_to_xxx related functions */ |
| 370 : | yv12_to_bgr = yv12_to_bgr_mmx; | ||
| 371 : | yv12_to_bgra = yv12_to_bgra_mmx; | ||
| 372 : | edgomez | 200 | yv12_to_yuyv = yv12_to_yuyv_mmx; |
| 373 : | yv12_to_uyvy = yv12_to_uyvy_mmx; | ||
| 374 : | edgomez | 1161 | |
| 375 : | edgomez | 851 | yv12_to_yuyvi = yv12_to_yuyvi_mmx; |
| 376 : | yv12_to_uyvyi = yv12_to_uyvyi_mmx; | ||
| 377 : | Isibaar | 3 | |
| 378 : | edgomez | 200 | /* Motion estimation related functions */ |
| 379 : | Isibaar | 3 | calc_cbp = calc_cbp_mmx; |
| 380 : | edgomez | 200 | sad16 = sad16_mmx; |
| 381 : | sad8 = sad8_mmx; | ||
| 382 : | suxen_drol | 329 | sad16bi = sad16bi_mmx; |
| 383 : | sad8bi = sad8bi_mmx; | ||
| 384 : | edgomez | 200 | dev16 = dev16_mmx; |
| 385 : | edgomez | 851 | sad16v = sad16v_mmx; |
| 386 : | edgomez | 1205 | sse8_16bit = sse8_16bit_mmx; |
| 387 : | Isibaar | 3 | } |
| 388 : | |||
| 389 : | suxen_drol | 329 | /* these 3dnow functions are faster than mmx, but slower than xmm. */ |
| 390 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_3DNOW)) { |
| 391 : | suxen_drol | 329 | |
| 392 : | edgomez | 851 | emms = emms_3dn; |
| 393 : | |||
| 394 : | suxen_drol | 329 | /* ME functions */ |
| 395 : | sad16bi = sad16bi_3dn; | ||
| 396 : | sad8bi = sad8bi_3dn; | ||
| 397 : | edgomez | 851 | |
| 398 : | yuyv_to_yv12 = yuyv_to_yv12_3dn; | ||
| 399 : | uyvy_to_yv12 = uyvy_to_yv12_3dn; | ||
| 400 : | suxen_drol | 329 | } |
| 401 : | |||
| 402 : | |||
| 403 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_MMXEXT)) { |
| 404 : | edgomez | 200 | |
| 405 : | edgomez | 1190 | /* DCT */ |
| 406 : | fdct = fdct_xmm_skal; | ||
| 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 : | edgomez | 851 | /* 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 : | edgomez | 1174 | quant_mpeg_intra = quant_mpeg_intra_xmm; |
| 420 : | quant_mpeg_inter = quant_mpeg_inter_xmm; | ||
| 421 : | edgomez | 851 | |
| 422 : | edgomez | 1174 | dequant_h263_intra = dequant_h263_intra_xmm; |
| 423 : | dequant_h263_inter = dequant_h263_inter_xmm; | ||
| 424 : | chenm001 | 274 | |
| 425 : | edgomez | 218 | /* Buffer transfer */ |
| 426 : | transfer_8to16sub2 = transfer_8to16sub2_xmm; | ||
| 427 : | |||
| 428 : | edgomez | 200 | /* Colorspace transformation */ |
| 429 : | edgomez | 851 | yv12_to_yv12 = yv12_to_yv12_xmm; |
| 430 : | yuyv_to_yv12 = yuyv_to_yv12_xmm; | ||
| 431 : | 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 : | edgomez | 851 | sad16v = sad16v_xmm; |
| 440 : | Isibaar | 3 | } |
| 441 : | |||
| 442 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_3DNOW)) { |
| 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 : | edgomez | 851 | if ((cpu_flags & XVID_CPU_3DNOWEXT)) { |
| 451 : | edgomez | 200 | |
| 452 : | edgomez | 851 | /* 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_8to16subro = transfer_8to16subro_3dne; | ||
| 460 : | transfer_8to16sub2 = transfer_8to16sub2_3dne; | ||
| 461 : | transfer_16to8add = transfer_16to8add_3dne; | ||
| 462 : | transfer8x8_copy = transfer8x8_copy_3dne; | ||
| 463 : | |||
| 464 : | /* Quantization */ | ||
| 465 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_3dne; |
| 466 : | quant_h263_inter = quant_h263_inter_3dne; | ||
| 467 : | dequant_mpeg_intra = dequant_mpeg_intra_3dne; | ||
| 468 : | dequant_mpeg_inter = dequant_mpeg_inter_3dne; | ||
| 469 : | dequant_h263_intra = dequant_h263_intra_3dne; | ||
| 470 : | dequant_h263_inter = dequant_h263_inter_3dne; | ||
| 471 : | edgomez | 851 | |
| 472 : | /* ME functions */ | ||
| 473 : | calc_cbp = calc_cbp_3dne; | ||
| 474 : | sad16 = sad16_3dne; | ||
| 475 : | sad8 = sad8_3dne; | ||
| 476 : | sad16bi = sad16bi_3dne; | ||
| 477 : | sad8bi = sad8bi_3dne; | ||
| 478 : | dev16 = dev16_3dne; | ||
| 479 : | |||
| 480 : | /* Interpolation */ | ||
| 481 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne; | ||
| 482 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne; | ||
| 483 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; | ||
| 484 : | } | ||
| 485 : | |||
| 486 : | if ((cpu_flags & XVID_CPU_SSE2)) { | ||
| 487 : | |||
| 488 : | chenm001 | 274 | calc_cbp = calc_cbp_sse2; |
| 489 : | |||
| 490 : | edgomez | 200 | /* Quantization */ |
| 491 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_sse2; |
| 492 : | quant_h263_inter = quant_h263_inter_sse2; | ||
| 493 : | dequant_h263_intra = dequant_h263_intra_sse2; | ||
| 494 : | dequant_h263_inter = dequant_h263_inter_sse2; | ||
| 495 : | h | 135 | |
| 496 : | edgomez | 1195 | /* SAD operators */ |
| 497 : | edgomez | 200 | sad16 = sad16_sse2; |
| 498 : | dev16 = dev16_sse2; | ||
| 499 : | edgomez | 1195 | |
| 500 : | edgomez | 1261 | /* DCT operators |
| 501 : | * no iDCT because it's not "Walken matching" */ | ||
| 502 : | edgomez | 1195 | fdct = fdct_sse2_skal; |
| 503 : | h | 126 | } |
| 504 : | edgomez | 1261 | #endif /* ARCH_IS_IA32 */ |
| 505 : | edgomez | 200 | |
| 506 : | edgomez | 851 | #if defined(ARCH_IS_IA64) |
| 507 : | edgomez | 874 | if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */ |
| 508 : | Isibaar | 209 | idct_ia64_init(); |
| 509 : | fdct = fdct_ia64; | ||
| 510 : | edgomez | 874 | idct = idct_ia64; /*not yet working, crashes */ |
| 511 : | Isibaar | 209 | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
| 512 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; | ||
| 513 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; | ||
| 514 : | sad16 = sad16_ia64; | ||
| 515 : | sad16bi = sad16bi_ia64; | ||
| 516 : | sad8 = sad8_ia64; | ||
| 517 : | dev16 = dev16_ia64; | ||
| 518 : | edgomez | 874 | /* Halfpel8_Refine = Halfpel8_Refine_ia64; */ |
| 519 : | edgomez | 1174 | quant_h263_intra = quant_h263_intra_ia64; |
| 520 : | quant_h263_inter = quant_h263_inter_ia64; | ||
| 521 : | dequant_h263_intra = dequant_h263_intra_ia64; | ||
| 522 : | dequant_h263_inter = dequant_h263_inter_ia64; | ||
| 523 : | Isibaar | 209 | transfer_8to16copy = transfer_8to16copy_ia64; |
| 524 : | transfer_16to8copy = transfer_16to8copy_ia64; | ||
| 525 : | transfer_8to16sub = transfer_8to16sub_ia64; | ||
| 526 : | transfer_8to16sub2 = transfer_8to16sub2_ia64; | ||
| 527 : | transfer_16to8add = transfer_16to8add_ia64; | ||
| 528 : | transfer8x8_copy = transfer8x8_copy_ia64; | ||
| 529 : | } | ||
| 530 : | edgomez | 1161 | #endif |
| 531 : | Isibaar | 209 | |
| 532 : | edgomez | 851 | #if defined(ARCH_IS_PPC) |
| 533 : | if ((cpu_flags & XVID_CPU_ASM)) | ||
| 534 : | { | ||
| 535 : | calc_cbp = calc_cbp_ppc; | ||
| 536 : | } | ||
| 537 : | |||
| 538 : | if ((cpu_flags & XVID_CPU_ALTIVEC)) | ||
| 539 : | { | ||
| 540 : | calc_cbp = calc_cbp_altivec; | ||
| 541 : | fdct = fdct_altivec; | ||
| 542 : | idct = idct_altivec; | ||
| 543 : | sadInit = sadInit_altivec; | ||
| 544 : | sad16 = sad16_altivec; | ||
| 545 : | sad8 = sad8_altivec; | ||
| 546 : | dev16 = dev16_altivec; | ||
| 547 : | } | ||
| 548 : | canard | 52 | #endif |
| 549 : | edgomez | 851 | |
| 550 : | suxen_drol | 1031 | #if defined(_DEBUG) |
| 551 : | xvid_debug = init->debug; | ||
| 552 : | #endif | ||
| 553 : | |||
| 554 : | return 0; | ||
| 555 : | edgomez | 851 | } |
| 556 : | |||
| 557 : | |||
| 558 : | suxen_drol | 890 | static int |
| 559 : | xvid_gbl_info(xvid_gbl_info_t * info) | ||
| 560 : | { | ||
| 561 : | edgomez | 1107 | if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */ |
| 562 : | suxen_drol | 890 | return XVID_ERR_VERSION; |
| 563 : | edgomez | 851 | |
| 564 : | suxen_drol | 890 | info->actual_version = XVID_VERSION; |
| 565 : | edgomez | 1226 | info->build = "xvid-1.0.0"; |
| 566 : | suxen_drol | 890 | info->cpu_flags = detect_cpu_flags(); |
| 567 : | |||
| 568 : | #if defined(_SMP) && defined(WIN32) | ||
| 569 : | info->num_threads = pthread_num_processors_np();; | ||
| 570 : | #else | ||
| 571 : | info->num_threads = 0; | ||
| 572 : | #endif | ||
| 573 : | |||
| 574 : | return 0; | ||
| 575 : | } | ||
| 576 : | |||
| 577 : | |||
| 578 : | edgomez | 851 | static int |
| 579 : | suxen_drol | 890 | xvid_gbl_convert(xvid_gbl_convert_t* convert) |
| 580 : | edgomez | 851 | { |
| 581 : | suxen_drol | 890 | int width; |
| 582 : | int height; | ||
| 583 : | int width2; | ||
| 584 : | int height2; | ||
| 585 : | edgomez | 851 | IMAGE img; |
| 586 : | |||
| 587 : | edgomez | 1107 | if (XVID_VERSION_MAJOR(convert->version) != 1) /* v1.x.x */ |
| 588 : | suxen_drol | 890 | return XVID_ERR_VERSION; |
| 589 : | |||
| 590 : | edgomez | 1053 | #if 0 |
| 591 : | const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); | ||
| 592 : | #endif | ||
| 593 : | suxen_drol | 890 | width = convert->width; |
| 594 : | height = convert->height; | ||
| 595 : | width2 = convert->width/2; | ||
| 596 : | height2 = convert->height/2; | ||
| 597 : | |||
| 598 : | switch (convert->input.csp & ~XVID_CSP_VFLIP) | ||
| 599 : | edgomez | 851 | { |
| 600 : | case XVID_CSP_YV12 : | ||
| 601 : | suxen_drol | 890 | img.y = convert->input.plane[0]; |
| 602 : | edgomez | 1161 | img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height; |
| 603 : | suxen_drol | 890 | img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2; |
| 604 : | edgomez | 851 | image_output(&img, width, height, width, |
| 605 : | suxen_drol | 890 | (uint8_t**)convert->output.plane, convert->output.stride, |
| 606 : | convert->output.csp, convert->interlacing); | ||
| 607 : | edgomez | 851 | break; |
| 608 : | |||
| 609 : | default : | ||
| 610 : | return XVID_ERR_FORMAT; | ||
| 611 : | } | ||
| 612 : | |||
| 613 : | |||
| 614 : | emms(); | ||
| 615 : | suxen_drol | 890 | return 0; |
| 616 : | edgomez | 851 | } |
| 617 : | |||
| 618 : | suxen_drol | 890 | /***************************************************************************** |
| 619 : | * XviD Global Entry point | ||
| 620 : | * | ||
| 621 : | * Well this function initialize all internal function pointers according | ||
| 622 : | * to the CPU features forced by the library client or autodetected (depending | ||
| 623 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
| 624 : | * image colorspace transformation tables. | ||
| 625 : | edgomez | 1161 | * |
| 626 : | suxen_drol | 890 | ****************************************************************************/ |
| 627 : | |||
| 628 : | |||
| 629 : | edgomez | 851 | int |
| 630 : | suxen_drol | 890 | xvid_global(void *handle, |
| 631 : | edgomez | 851 | int opt, |
| 632 : | void *param1, | ||
| 633 : | void *param2) | ||
| 634 : | { | ||
| 635 : | switch(opt) | ||
| 636 : | { | ||
| 637 : | suxen_drol | 890 | case XVID_GBL_INIT : |
| 638 : | return xvid_gbl_init((xvid_gbl_init_t*)param1); | ||
| 639 : | edgomez | 851 | |
| 640 : | suxen_drol | 890 | case XVID_GBL_INFO : |
| 641 : | return xvid_gbl_info((xvid_gbl_info_t*)param1); | ||
| 642 : | edgomez | 851 | |
| 643 : | suxen_drol | 890 | case XVID_GBL_CONVERT : |
| 644 : | return xvid_gbl_convert((xvid_gbl_convert_t*)param1); | ||
| 645 : | |||
| 646 : | edgomez | 851 | default : |
| 647 : | return XVID_ERR_FAIL; | ||
| 648 : | } | ||
| 649 : | } | ||
| 650 : | |||
| 651 : | edgomez | 200 | /***************************************************************************** |
| 652 : | * XviD Native decoder entry point | ||
| 653 : | * | ||
| 654 : | * This function is just a wrapper to all the option cases. | ||
| 655 : | * | ||
| 656 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
| 657 : | * else returns the wrapped function result | ||
| 658 : | * | ||
| 659 : | ****************************************************************************/ | ||
| 660 : | |||
| 661 : | edgomez | 195 | int |
| 662 : | xvid_decore(void *handle, | ||
| 663 : | int opt, | ||
| 664 : | void *param1, | ||
| 665 : | void *param2) | ||
| 666 : | Isibaar | 3 | { |
| 667 : | edgomez | 195 | switch (opt) { |
| 668 : | case XVID_DEC_CREATE: | ||
| 669 : | suxen_drol | 890 | return decoder_create((xvid_dec_create_t *) param1); |
| 670 : | Isibaar | 3 | |
| 671 : | edgomez | 195 | case XVID_DEC_DESTROY: |
| 672 : | return decoder_destroy((DECODER *) handle); | ||
| 673 : | |||
| 674 : | suxen_drol | 890 | case XVID_DEC_DECODE: |
| 675 : | return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2); | ||
| 676 : | |||
| 677 : | Isibaar | 3 | default: |
| 678 : | edgomez | 195 | return XVID_ERR_FAIL; |
| 679 : | } | ||
| 680 : | Isibaar | 3 | } |
| 681 : | |||
| 682 : | |||
| 683 : | edgomez | 200 | /***************************************************************************** |
| 684 : | * XviD Native encoder entry point | ||
| 685 : | * | ||
| 686 : | * This function is just a wrapper to all the option cases. | ||
| 687 : | * | ||
| 688 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
| 689 : | * else returns the wrapped function result | ||
| 690 : | * | ||
| 691 : | ****************************************************************************/ | ||
| 692 : | |||
| 693 : | edgomez | 195 | int |
| 694 : | xvid_encore(void *handle, | ||
| 695 : | int opt, | ||
| 696 : | void *param1, | ||
| 697 : | void *param2) | ||
| 698 : | Isibaar | 3 | { |
| 699 : | edgomez | 195 | switch (opt) { |
| 700 : | case XVID_ENC_ENCODE: | ||
| 701 : | edgomez | 851 | |
| 702 : | suxen_drol | 890 | return enc_encode((Encoder *) handle, |
| 703 : | (xvid_enc_frame_t *) param1, | ||
| 704 : | (xvid_enc_stats_t *) param2); | ||
| 705 : | Isibaar | 3 | |
| 706 : | edgomez | 195 | case XVID_ENC_CREATE: |
| 707 : | suxen_drol | 948 | return enc_create((xvid_enc_create_t *) param1); |
| 708 : | Isibaar | 3 | |
| 709 : | edgomez | 195 | case XVID_ENC_DESTROY: |
| 710 : | suxen_drol | 890 | return enc_destroy((Encoder *) handle); |
| 711 : | edgomez | 195 | |
| 712 : | Isibaar | 3 | default: |
| 713 : | edgomez | 195 | return XVID_ERR_FAIL; |
| 714 : | } | ||
| 715 : | Isibaar | 3 | } |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |