Parent Directory
| 
Revision Log
Revision 1.69 - (view) (download)
| 1 : | edgomez | 1.16 | /***************************************************************************** |
| 2 : | edgomez | 1.17 | * |
| 3 : | * XVID MPEG-4 VIDEO CODEC | ||
| 4 : | * - Native API implementation - | ||
| 5 : | * | ||
| 6 : | suxen_drol | 1.49 | * Copyright(C) 2001-2004 Peter Ross <pross@xvid.org> |
| 7 : | edgomez | 1.48 | * |
| 8 : | edgomez | 1.41 | * 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 | 1.17 | * (at your option) any later version. |
| 12 : | * | ||
| 13 : | * This program is distributed in the hope that it will be useful, | ||
| 14 : | edgomez | 1.41 | * but WITHOUT ANY WARRANTY ; without even the implied warranty of |
| 15 : | edgomez | 1.17 | * 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 | 1.41 | * along with this program ; if not, write to the Free Software |
| 20 : | edgomez | 1.17 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
| 21 : | * | ||
| 22 : | Skal | 1.69 | * $Id: xvid.c,v 1.68 2006/02/25 04:41:12 suxen_drol Exp $ |
| 23 : | edgomez | 1.36 | * |
| 24 : | edgomez | 1.17 | ****************************************************************************/ |
| 25 : | chenm001 | 1.29 | |
| 26 : | edgomez | 1.41 | #include <stdio.h> |
| 27 : | #include <stdlib.h> | ||
| 28 : | #include <string.h> | ||
| 29 : | #include <time.h> | ||
| 30 : | |||
| 31 : | Isibaar | 1.1 | #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 : | #include "utils/mem_transfer.h" | ||
| 40 : | edgomez | 1.41 | #include "utils/mbfunctions.h" |
| 41 : | edgomez | 1.48 | #include "quant/quant.h" |
| 42 : | ia64p | 1.30 | #include "motion/motion.h" |
| 43 : | Skal | 1.69 | #include "motion/gmc.h" |
| 44 : | Isibaar | 1.1 | #include "motion/sad.h" |
| 45 : | #include "utils/emms.h" | ||
| 46 : | #include "utils/timer.h" | ||
| 47 : | Isibaar | 1.9 | #include "bitstream/mbcoding.h" |
| 48 : | edgomez | 1.48 | #include "image/qpel.h" |
| 49 : | #include "image/postprocessing.h" | ||
| 50 : | |||
| 51 : | #if defined(_DEBUG) | ||
| 52 : | unsigned int xvid_debug = 0; /* xvid debug mask */ | ||
| 53 : | #endif | ||
| 54 : | Isibaar | 1.1 | |
| 55 : | edgomez | 1.60 | #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER) |
| 56 : | edgomez | 1.41 | # include <windows.h> |
| 57 : | edgomez | 1.60 | #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC) |
| 58 : | edgomez | 1.41 | # include <signal.h> |
| 59 : | # include <setjmp.h> | ||
| 60 : | suxen_drol | 1.31 | |
| 61 : | edgomez | 1.41 | static jmp_buf mark; |
| 62 : | suxen_drol | 1.31 | |
| 63 : | edgomez | 1.41 | static void |
| 64 : | sigill_handler(int signal) | ||
| 65 : | { | ||
| 66 : | longjmp(mark, 1); | ||
| 67 : | } | ||
| 68 : | suxen_drol | 1.31 | #endif |
| 69 : | |||
| 70 : | |||
| 71 : | /* | ||
| 72 : | edgomez | 1.43 | * Calls the funcptr, and returns whether SIGILL (illegal instruction) was |
| 73 : | * signalled | ||
| 74 : | * | ||
| 75 : | * Return values: | ||
| 76 : | * -1 : could not determine | ||
| 77 : | * 0 : SIGILL was *not* signalled | ||
| 78 : | * 1 : SIGILL was signalled | ||
| 79 : | */ | ||
| 80 : | edgomez | 1.60 | #if (defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64)) && defined(_MSC_VER) |
| 81 : | edgomez | 1.50 | static int |
| 82 : | suxen_drol | 1.31 | sigill_check(void (*func)()) |
| 83 : | { | ||
| 84 : | _try { | ||
| 85 : | func(); | ||
| 86 : | edgomez | 1.50 | } _except(EXCEPTION_EXECUTE_HANDLER) { |
| 87 : | suxen_drol | 1.31 | |
| 88 : | if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION) | ||
| 89 : | edgomez | 1.50 | return(1); |
| 90 : | suxen_drol | 1.31 | } |
| 91 : | edgomez | 1.50 | return(0); |
| 92 : | } | ||
| 93 : | edgomez | 1.60 | #elif defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) || defined(ARCH_IS_PPC) |
| 94 : | edgomez | 1.50 | static int |
| 95 : | sigill_check(void (*func)()) | ||
| 96 : | { | ||
| 97 : | void *old_handler; | ||
| 98 : | suxen_drol | 1.31 | int jmpret; |
| 99 : | |||
| 100 : | edgomez | 1.50 | /* Set our SIGILL handler */ |
| 101 : | old_handler = signal(SIGILL, sigill_handler); | ||
| 102 : | suxen_drol | 1.31 | |
| 103 : | edgomez | 1.50 | /* Check for error */ |
| 104 : | if (old_handler == SIG_ERR) { | ||
| 105 : | return(-1); | ||
| 106 : | suxen_drol | 1.31 | } |
| 107 : | |||
| 108 : | edgomez | 1.50 | /* Save stack context, so if func triggers a SIGILL, we can still roll |
| 109 : | * back to a valid CPU state */ | ||
| 110 : | jmpret = setjmp(mark); | ||
| 111 : | |||
| 112 : | /* If setjmp returned directly, then its returned value is 0, and we still | ||
| 113 : | * have to test the passed func. Otherwise it means the stack context has | ||
| 114 : | * been restored by a longjmp() call, which in our case happens only in the | ||
| 115 : | * signal handler */ | ||
| 116 : | if (jmpret == 0) { | ||
| 117 : | suxen_drol | 1.31 | func(); |
| 118 : | } | ||
| 119 : | |||
| 120 : | edgomez | 1.50 | /* Restore old signal handler */ |
| 121 : | suxen_drol | 1.31 | signal(SIGILL, old_handler); |
| 122 : | |||
| 123 : | edgomez | 1.50 | return(jmpret); |
| 124 : | suxen_drol | 1.31 | } |
| 125 : | #endif | ||
| 126 : | |||
| 127 : | edgomez | 1.41 | |
| 128 : | /* detect cpu flags */ | ||
| 129 : | static unsigned int | ||
| 130 : | suxen_drol | 1.65 | detect_cpu_flags(void) |
| 131 : | edgomez | 1.41 | { |
| 132 : | /* enable native assembly optimizations by default */ | ||
| 133 : | unsigned int cpu_flags = XVID_CPU_ASM; | ||
| 134 : | |||
| 135 : | edgomez | 1.60 | #if defined(ARCH_IS_IA32) || defined(ARCH_IS_X86_64) |
| 136 : | edgomez | 1.41 | cpu_flags |= check_cpu_features(); |
| 137 : | if ((cpu_flags & XVID_CPU_SSE) && sigill_check(sse_os_trigger)) | ||
| 138 : | cpu_flags &= ~XVID_CPU_SSE; | ||
| 139 : | |||
| 140 : | if ((cpu_flags & XVID_CPU_SSE2) && sigill_check(sse2_os_trigger)) | ||
| 141 : | cpu_flags &= ~XVID_CPU_SSE2; | ||
| 142 : | #endif | ||
| 143 : | |||
| 144 : | #if defined(ARCH_IS_PPC) | ||
| 145 : | edgomez | 1.50 | if (!sigill_check(altivec_trigger)) |
| 146 : | cpu_flags |= XVID_CPU_ALTIVEC; | ||
| 147 : | edgomez | 1.41 | #endif |
| 148 : | |||
| 149 : | return cpu_flags; | ||
| 150 : | } | ||
| 151 : | |||
| 152 : | |||
| 153 : | edgomez | 1.16 | /***************************************************************************** |
| 154 : | * XviD Init Entry point | ||
| 155 : | * | ||
| 156 : | * Well this function initialize all internal function pointers according | ||
| 157 : | * to the CPU features forced by the library client or autodetected (depending | ||
| 158 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
| 159 : | * image colorspace transformation tables. | ||
| 160 : | edgomez | 1.48 | * |
| 161 : | edgomez | 1.16 | * Returned value : XVID_ERR_OK |
| 162 : | * + API_VERSION in the input XVID_INIT_PARAM structure | ||
| 163 : | * + core build " " " " " | ||
| 164 : | * | ||
| 165 : | ****************************************************************************/ | ||
| 166 : | |||
| 167 : | edgomez | 1.41 | |
| 168 : | edgomez | 1.48 | static |
| 169 : | int xvid_gbl_init(xvid_gbl_init_t * init) | ||
| 170 : | Isibaar | 1.1 | { |
| 171 : | edgomez | 1.48 | unsigned int cpu_flags; |
| 172 : | suxen_drol | 1.21 | |
| 173 : | edgomez | 1.48 | if (XVID_VERSION_MAJOR(init->version) != 1) /* v1.x.x */ |
| 174 : | return XVID_ERR_VERSION; | ||
| 175 : | suxen_drol | 1.31 | |
| 176 : | edgomez | 1.48 | cpu_flags = (init->cpu_flags & XVID_CPU_FORCE) ? init->cpu_flags : detect_cpu_flags(); |
| 177 : | Isibaar | 1.1 | |
| 178 : | edgomez | 1.16 | /* Initialize the function pointers */ |
| 179 : | Isibaar | 1.1 | idct_int32_init(); |
| 180 : | Isibaar | 1.9 | init_vlc_tables(); |
| 181 : | |||
| 182 : | edgomez | 1.16 | /* Fixed Point Forward/Inverse DCT transformations */ |
| 183 : | Isibaar | 1.1 | fdct = fdct_int32; |
| 184 : | idct = idct_int32; | ||
| 185 : | |||
| 186 : | edgomez | 1.16 | /* Only needed on PPC Altivec archs */ |
| 187 : | suxen_drol | 1.65 | sadInit = NULL; |
| 188 : | edgomez | 1.15 | |
| 189 : | edgomez | 1.16 | /* Restore FPU context : emms_c is a nop functions */ |
| 190 : | Isibaar | 1.1 | emms = emms_c; |
| 191 : | |||
| 192 : | edgomez | 1.48 | /* Qpel stuff */ |
| 193 : | xvid_QP_Funcs = &xvid_QP_Funcs_C; | ||
| 194 : | xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_C; | ||
| 195 : | xvid_Init_QP(); | ||
| 196 : | |||
| 197 : | edgomez | 1.16 | /* Quantization functions */ |
| 198 : | edgomez | 1.48 | quant_h263_intra = quant_h263_intra_c; |
| 199 : | quant_h263_inter = quant_h263_inter_c; | ||
| 200 : | dequant_h263_intra = dequant_h263_intra_c; | ||
| 201 : | dequant_h263_inter = dequant_h263_inter_c; | ||
| 202 : | |||
| 203 : | quant_mpeg_intra = quant_mpeg_intra_c; | ||
| 204 : | quant_mpeg_inter = quant_mpeg_inter_c; | ||
| 205 : | dequant_mpeg_intra = dequant_mpeg_intra_c; | ||
| 206 : | dequant_mpeg_inter = dequant_mpeg_inter_c; | ||
| 207 : | Isibaar | 1.1 | |
| 208 : | edgomez | 1.16 | /* Block transfer related functions */ |
| 209 : | Isibaar | 1.1 | transfer_8to16copy = transfer_8to16copy_c; |
| 210 : | transfer_16to8copy = transfer_16to8copy_c; | ||
| 211 : | edgomez | 1.16 | transfer_8to16sub = transfer_8to16sub_c; |
| 212 : | edgomez | 1.41 | transfer_8to16subro = transfer_8to16subro_c; |
| 213 : | suxen_drol | 1.11 | transfer_8to16sub2 = transfer_8to16sub2_c; |
| 214 : | syskin | 1.59 | transfer_8to16sub2ro = transfer_8to16sub2ro_c; |
| 215 : | edgomez | 1.16 | transfer_16to8add = transfer_16to8add_c; |
| 216 : | transfer8x8_copy = transfer8x8_copy_c; | ||
| 217 : | suxen_drol | 1.64 | transfer8x4_copy = transfer8x4_copy_c; |
| 218 : | Isibaar | 1.1 | |
| 219 : | edgomez | 1.41 | /* Interlacing functions */ |
| 220 : | MBFieldTest = MBFieldTest_c; | ||
| 221 : | |||
| 222 : | edgomez | 1.16 | /* Image interpolation related functions */ |
| 223 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_c; | ||
| 224 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_c; | ||
| 225 : | Isibaar | 1.1 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_c; |
| 226 : | |||
| 227 : | suxen_drol | 1.64 | interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_c; |
| 228 : | interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_c; | ||
| 229 : | interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_c; | ||
| 230 : | |||
| 231 : | edgomez | 1.55 | interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_c; |
| 232 : | interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_c; | ||
| 233 : | interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_c; | ||
| 234 : | interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_c; | ||
| 235 : | |||
| 236 : | edgomez | 1.41 | interpolate16x16_lowpass_h = interpolate16x16_lowpass_h_c; |
| 237 : | interpolate16x16_lowpass_v = interpolate16x16_lowpass_v_c; | ||
| 238 : | interpolate16x16_lowpass_hv = interpolate16x16_lowpass_hv_c; | ||
| 239 : | |||
| 240 : | interpolate8x8_lowpass_h = interpolate8x8_lowpass_h_c; | ||
| 241 : | interpolate8x8_lowpass_v = interpolate8x8_lowpass_v_c; | ||
| 242 : | interpolate8x8_lowpass_hv = interpolate8x8_lowpass_hv_c; | ||
| 243 : | |||
| 244 : | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_c; | ||
| 245 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_c; | ||
| 246 : | |||
| 247 : | interpolate8x8_avg2 = interpolate8x8_avg2_c; | ||
| 248 : | interpolate8x8_avg4 = interpolate8x8_avg4_c; | ||
| 249 : | |||
| 250 : | suxen_drol | 1.49 | /* postprocessing */ |
| 251 : | image_brightness = image_brightness_c; | ||
| 252 : | |||
| 253 : | edgomez | 1.16 | /* Initialize internal colorspace transformation tables */ |
| 254 : | Isibaar | 1.1 | colorspace_init(); |
| 255 : | |||
| 256 : | edgomez | 1.16 | /* All colorspace transformation functions User Format->YV12 */ |
| 257 : | edgomez | 1.41 | yv12_to_yv12 = yv12_to_yv12_c; |
| 258 : | rgb555_to_yv12 = rgb555_to_yv12_c; | ||
| 259 : | rgb565_to_yv12 = rgb565_to_yv12_c; | ||
| 260 : | bgr_to_yv12 = bgr_to_yv12_c; | ||
| 261 : | bgra_to_yv12 = bgra_to_yv12_c; | ||
| 262 : | abgr_to_yv12 = abgr_to_yv12_c; | ||
| 263 : | rgba_to_yv12 = rgba_to_yv12_c; | ||
| 264 : | edgomez | 1.48 | argb_to_yv12 = argb_to_yv12_c; |
| 265 : | edgomez | 1.41 | yuyv_to_yv12 = yuyv_to_yv12_c; |
| 266 : | uyvy_to_yv12 = uyvy_to_yv12_c; | ||
| 267 : | |||
| 268 : | rgb555i_to_yv12 = rgb555i_to_yv12_c; | ||
| 269 : | rgb565i_to_yv12 = rgb565i_to_yv12_c; | ||
| 270 : | bgri_to_yv12 = bgri_to_yv12_c; | ||
| 271 : | bgrai_to_yv12 = bgrai_to_yv12_c; | ||
| 272 : | abgri_to_yv12 = abgri_to_yv12_c; | ||
| 273 : | rgbai_to_yv12 = rgbai_to_yv12_c; | ||
| 274 : | edgomez | 1.48 | argbi_to_yv12 = argbi_to_yv12_c; |
| 275 : | edgomez | 1.41 | yuyvi_to_yv12 = yuyvi_to_yv12_c; |
| 276 : | uyvyi_to_yv12 = uyvyi_to_yv12_c; | ||
| 277 : | |||
| 278 : | edgomez | 1.16 | /* All colorspace transformation functions YV12->User format */ |
| 279 : | edgomez | 1.41 | yv12_to_rgb555 = yv12_to_rgb555_c; |
| 280 : | yv12_to_rgb565 = yv12_to_rgb565_c; | ||
| 281 : | yv12_to_bgr = yv12_to_bgr_c; | ||
| 282 : | yv12_to_bgra = yv12_to_bgra_c; | ||
| 283 : | yv12_to_abgr = yv12_to_abgr_c; | ||
| 284 : | yv12_to_rgba = yv12_to_rgba_c; | ||
| 285 : | edgomez | 1.48 | yv12_to_argb = yv12_to_argb_c; |
| 286 : | edgomez | 1.41 | yv12_to_yuyv = yv12_to_yuyv_c; |
| 287 : | yv12_to_uyvy = yv12_to_uyvy_c; | ||
| 288 : | edgomez | 1.48 | |
| 289 : | edgomez | 1.41 | 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 : | edgomez | 1.48 | yv12_to_argbi = yv12_to_argbi_c; |
| 296 : | edgomez | 1.41 | yv12_to_yuyvi = yv12_to_yuyvi_c; |
| 297 : | yv12_to_uyvyi = yv12_to_uyvyi_c; | ||
| 298 : | Isibaar | 1.1 | |
| 299 : | edgomez | 1.16 | /* Functions used in motion estimation algorithms */ |
| 300 : | edgomez | 1.48 | calc_cbp = calc_cbp_c; |
| 301 : | sad16 = sad16_c; | ||
| 302 : | sad8 = sad8_c; | ||
| 303 : | sad16bi = sad16bi_c; | ||
| 304 : | sad8bi = sad8bi_c; | ||
| 305 : | dev16 = dev16_c; | ||
| 306 : | sad16v = sad16v_c; | ||
| 307 : | sse8_16bit = sse8_16bit_c; | ||
| 308 : | edgomez | 1.52 | sse8_8bit = sse8_8bit_c; |
| 309 : | Isibaar | 1.1 | |
| 310 : | Skal | 1.69 | init_GMC(cpu_flags); |
| 311 : | |||
| 312 : | edgomez | 1.41 | #if defined(ARCH_IS_IA32) |
| 313 : | |||
| 314 : | if ((cpu_flags & XVID_CPU_MMX) || (cpu_flags & XVID_CPU_MMXEXT) || | ||
| 315 : | (cpu_flags & XVID_CPU_3DNOW) || (cpu_flags & XVID_CPU_3DNOWEXT) || | ||
| 316 : | (cpu_flags & XVID_CPU_SSE) || (cpu_flags & XVID_CPU_SSE2)) | ||
| 317 : | { | ||
| 318 : | /* Restore FPU context : emms_c is a nop functions */ | ||
| 319 : | emms = emms_mmx; | ||
| 320 : | } | ||
| 321 : | |||
| 322 : | if ((cpu_flags & XVID_CPU_MMX)) { | ||
| 323 : | edgomez | 1.16 | |
| 324 : | /* Forward and Inverse Discrete Cosine Transformation functions */ | ||
| 325 : | edgomez | 1.48 | fdct = fdct_mmx_skal; |
| 326 : | Isibaar | 1.47 | idct = idct_mmx; |
| 327 : | Isibaar | 1.1 | |
| 328 : | edgomez | 1.48 | /* Qpel stuff */ |
| 329 : | xvid_QP_Funcs = &xvid_QP_Funcs_mmx; | ||
| 330 : | xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_mmx; | ||
| 331 : | |||
| 332 : | edgomez | 1.16 | /* Quantization related functions */ |
| 333 : | edgomez | 1.48 | quant_h263_intra = quant_h263_intra_mmx; |
| 334 : | quant_h263_inter = quant_h263_inter_mmx; | ||
| 335 : | dequant_h263_intra = dequant_h263_intra_mmx; | ||
| 336 : | dequant_h263_inter = dequant_h263_inter_mmx; | ||
| 337 : | |||
| 338 : | quant_mpeg_intra = quant_mpeg_intra_mmx; | ||
| 339 : | quant_mpeg_inter = quant_mpeg_inter_mmx; | ||
| 340 : | dequant_mpeg_intra = dequant_mpeg_intra_mmx; | ||
| 341 : | dequant_mpeg_inter = dequant_mpeg_inter_mmx; | ||
| 342 : | Isibaar | 1.1 | |
| 343 : | edgomez | 1.16 | /* Block related functions */ |
| 344 : | Isibaar | 1.1 | transfer_8to16copy = transfer_8to16copy_mmx; |
| 345 : | transfer_16to8copy = transfer_16to8copy_mmx; | ||
| 346 : | edgomez | 1.16 | transfer_8to16sub = transfer_8to16sub_mmx; |
| 347 : | edgomez | 1.41 | transfer_8to16subro = transfer_8to16subro_mmx; |
| 348 : | edgomez | 1.22 | transfer_8to16sub2 = transfer_8to16sub2_mmx; |
| 349 : | edgomez | 1.16 | transfer_16to8add = transfer_16to8add_mmx; |
| 350 : | transfer8x8_copy = transfer8x8_copy_mmx; | ||
| 351 : | suxen_drol | 1.64 | transfer8x4_copy = transfer8x4_copy_mmx; |
| 352 : | edgomez | 1.22 | |
| 353 : | edgomez | 1.41 | /* Interlacing Functions */ |
| 354 : | MBFieldTest = MBFieldTest_mmx; | ||
| 355 : | edgomez | 1.16 | |
| 356 : | /* Image Interpolation related functions */ | ||
| 357 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_mmx; | ||
| 358 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_mmx; | ||
| 359 : | Isibaar | 1.1 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_mmx; |
| 360 : | |||
| 361 : | suxen_drol | 1.64 | interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_mmx; |
| 362 : | interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_mmx; | ||
| 363 : | interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_mmx; | ||
| 364 : | |||
| 365 : | edgomez | 1.55 | interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_mmx; |
| 366 : | interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_mmx; | ||
| 367 : | interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_mmx; | ||
| 368 : | interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_mmx; | ||
| 369 : | |||
| 370 : | edgomez | 1.41 | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_mmx; |
| 371 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_mmx; | ||
| 372 : | |||
| 373 : | interpolate8x8_avg2 = interpolate8x8_avg2_mmx; | ||
| 374 : | interpolate8x8_avg4 = interpolate8x8_avg4_mmx; | ||
| 375 : | suxen_drol | 1.49 | |
| 376 : | /* postprocessing */ | ||
| 377 : | image_brightness = image_brightness_mmx; | ||
| 378 : | edgomez | 1.41 | |
| 379 : | /* image input xxx_to_yv12 related functions */ | ||
| 380 : | yv12_to_yv12 = yv12_to_yv12_mmx; | ||
| 381 : | bgr_to_yv12 = bgr_to_yv12_mmx; | ||
| 382 : | bgra_to_yv12 = bgra_to_yv12_mmx; | ||
| 383 : | edgomez | 1.16 | yuyv_to_yv12 = yuyv_to_yv12_mmx; |
| 384 : | uyvy_to_yv12 = uyvy_to_yv12_mmx; | ||
| 385 : | Isibaar | 1.1 | |
| 386 : | edgomez | 1.41 | /* image output yv12_to_xxx related functions */ |
| 387 : | yv12_to_bgr = yv12_to_bgr_mmx; | ||
| 388 : | yv12_to_bgra = yv12_to_bgra_mmx; | ||
| 389 : | edgomez | 1.16 | yv12_to_yuyv = yv12_to_yuyv_mmx; |
| 390 : | yv12_to_uyvy = yv12_to_uyvy_mmx; | ||
| 391 : | edgomez | 1.48 | |
| 392 : | edgomez | 1.41 | yv12_to_yuyvi = yv12_to_yuyvi_mmx; |
| 393 : | yv12_to_uyvyi = yv12_to_uyvyi_mmx; | ||
| 394 : | Isibaar | 1.1 | |
| 395 : | edgomez | 1.16 | /* Motion estimation related functions */ |
| 396 : | edgomez | 1.52 | calc_cbp = calc_cbp_mmx; |
| 397 : | sad16 = sad16_mmx; | ||
| 398 : | sad8 = sad8_mmx; | ||
| 399 : | sad16bi = sad16bi_mmx; | ||
| 400 : | sad8bi = sad8bi_mmx; | ||
| 401 : | dev16 = dev16_mmx; | ||
| 402 : | sad16v = sad16v_mmx; | ||
| 403 : | edgomez | 1.48 | sse8_16bit = sse8_16bit_mmx; |
| 404 : | edgomez | 1.52 | sse8_8bit = sse8_8bit_mmx; |
| 405 : | Isibaar | 1.1 | } |
| 406 : | |||
| 407 : | suxen_drol | 1.33 | /* these 3dnow functions are faster than mmx, but slower than xmm. */ |
| 408 : | edgomez | 1.41 | if ((cpu_flags & XVID_CPU_3DNOW)) { |
| 409 : | |||
| 410 : | emms = emms_3dn; | ||
| 411 : | suxen_drol | 1.33 | |
| 412 : | /* ME functions */ | ||
| 413 : | sad16bi = sad16bi_3dn; | ||
| 414 : | sad8bi = sad8bi_3dn; | ||
| 415 : | edgomez | 1.41 | |
| 416 : | yuyv_to_yv12 = yuyv_to_yv12_3dn; | ||
| 417 : | uyvy_to_yv12 = uyvy_to_yv12_3dn; | ||
| 418 : | suxen_drol | 1.33 | } |
| 419 : | |||
| 420 : | |||
| 421 : | edgomez | 1.41 | if ((cpu_flags & XVID_CPU_MMXEXT)) { |
| 422 : | edgomez | 1.16 | |
| 423 : | edgomez | 1.48 | /* DCT */ |
| 424 : | fdct = fdct_xmm_skal; | ||
| 425 : | Isibaar | 1.47 | idct = idct_xmm; |
| 426 : | edgomez | 1.16 | |
| 427 : | /* Interpolation */ | ||
| 428 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_xmm; | ||
| 429 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_xmm; | ||
| 430 : | h | 1.3 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_xmm; |
| 431 : | edgomez | 1.55 | |
| 432 : | suxen_drol | 1.64 | interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_xmm; |
| 433 : | interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_xmm; | ||
| 434 : | interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_xmm; | ||
| 435 : | |||
| 436 : | edgomez | 1.55 | interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_xmm; |
| 437 : | interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_xmm; | ||
| 438 : | interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_xmm; | ||
| 439 : | interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_xmm; | ||
| 440 : | Isibaar | 1.25 | |
| 441 : | chenm001 | 1.29 | /* Quantization */ |
| 442 : | edgomez | 1.48 | quant_mpeg_intra = quant_mpeg_intra_xmm; |
| 443 : | quant_mpeg_inter = quant_mpeg_inter_xmm; | ||
| 444 : | edgomez | 1.41 | |
| 445 : | edgomez | 1.48 | dequant_h263_intra = dequant_h263_intra_xmm; |
| 446 : | dequant_h263_inter = dequant_h263_inter_xmm; | ||
| 447 : | chenm001 | 1.29 | |
| 448 : | edgomez | 1.19 | /* Buffer transfer */ |
| 449 : | transfer_8to16sub2 = transfer_8to16sub2_xmm; | ||
| 450 : | syskin | 1.59 | transfer_8to16sub2ro = transfer_8to16sub2ro_xmm; |
| 451 : | edgomez | 1.16 | |
| 452 : | /* Colorspace transformation */ | ||
| 453 : | edgomez | 1.41 | yv12_to_yv12 = yv12_to_yv12_xmm; |
| 454 : | yuyv_to_yv12 = yuyv_to_yv12_xmm; | ||
| 455 : | uyvy_to_yv12 = uyvy_to_yv12_xmm; | ||
| 456 : | Isibaar | 1.1 | |
| 457 : | edgomez | 1.16 | /* ME functions */ |
| 458 : | Isibaar | 1.1 | sad16 = sad16_xmm; |
| 459 : | suxen_drol | 1.33 | sad8 = sad8_xmm; |
| 460 : | chenm001 | 1.29 | sad16bi = sad16bi_xmm; |
| 461 : | suxen_drol | 1.33 | sad8bi = sad8bi_xmm; |
| 462 : | Isibaar | 1.1 | dev16 = dev16_xmm; |
| 463 : | edgomez | 1.41 | sad16v = sad16v_xmm; |
| 464 : | Isibaar | 1.1 | } |
| 465 : | |||
| 466 : | edgomez | 1.41 | if ((cpu_flags & XVID_CPU_3DNOW)) { |
| 467 : | edgomez | 1.16 | |
| 468 : | /* Interpolation */ | ||
| 469 : | Isibaar | 1.1 | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dn; |
| 470 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dn; | ||
| 471 : | h | 1.4 | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dn; |
| 472 : | suxen_drol | 1.64 | |
| 473 : | interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dn; | ||
| 474 : | interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dn; | ||
| 475 : | interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dn; | ||
| 476 : | Isibaar | 1.1 | } |
| 477 : | |||
| 478 : | edgomez | 1.41 | if ((cpu_flags & XVID_CPU_3DNOWEXT)) { |
| 479 : | |||
| 480 : | /* Buffer transfer */ | ||
| 481 : | transfer_8to16copy = transfer_8to16copy_3dne; | ||
| 482 : | transfer_16to8copy = transfer_16to8copy_3dne; | ||
| 483 : | transfer_8to16sub = transfer_8to16sub_3dne; | ||
| 484 : | transfer_8to16subro = transfer_8to16subro_3dne; | ||
| 485 : | transfer_16to8add = transfer_16to8add_3dne; | ||
| 486 : | transfer8x8_copy = transfer8x8_copy_3dne; | ||
| 487 : | suxen_drol | 1.64 | transfer8x4_copy = transfer8x4_copy_3dne; |
| 488 : | edgomez | 1.41 | |
| 489 : | edgomez | 1.51 | if ((cpu_flags & XVID_CPU_MMXEXT)) { |
| 490 : | /* Inverse DCT */ | ||
| 491 : | idct = idct_3dne; | ||
| 492 : | |||
| 493 : | /* Buffer transfer */ | ||
| 494 : | transfer_8to16sub2 = transfer_8to16sub2_3dne; | ||
| 495 : | |||
| 496 : | /* Interpolation */ | ||
| 497 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne; | ||
| 498 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne; | ||
| 499 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne; | ||
| 500 : | |||
| 501 : | suxen_drol | 1.64 | interpolate8x4_halfpel_h = interpolate8x4_halfpel_h_3dne; |
| 502 : | interpolate8x4_halfpel_v = interpolate8x4_halfpel_v_3dne; | ||
| 503 : | interpolate8x4_halfpel_hv = interpolate8x4_halfpel_hv_3dne; | ||
| 504 : | |||
| 505 : | edgomez | 1.51 | /* Quantization */ |
| 506 : | quant_h263_intra = quant_h263_intra_3dne; /* cmov only */ | ||
| 507 : | quant_h263_inter = quant_h263_inter_3dne; | ||
| 508 : | dequant_mpeg_intra = dequant_mpeg_intra_3dne; /* cmov only */ | ||
| 509 : | dequant_mpeg_inter = dequant_mpeg_inter_3dne; | ||
| 510 : | dequant_h263_intra = dequant_h263_intra_3dne; | ||
| 511 : | dequant_h263_inter = dequant_h263_inter_3dne; | ||
| 512 : | |||
| 513 : | /* ME functions */ | ||
| 514 : | calc_cbp = calc_cbp_3dne; | ||
| 515 : | |||
| 516 : | sad16 = sad16_3dne; | ||
| 517 : | sad8 = sad8_3dne; | ||
| 518 : | sad16bi = sad16bi_3dne; | ||
| 519 : | sad8bi = sad8bi_3dne; | ||
| 520 : | dev16 = dev16_3dne; | ||
| 521 : | } | ||
| 522 : | edgomez | 1.41 | } |
| 523 : | |||
| 524 : | if ((cpu_flags & XVID_CPU_SSE2)) { | ||
| 525 : | edgomez | 1.16 | |
| 526 : | chenm001 | 1.29 | calc_cbp = calc_cbp_sse2; |
| 527 : | |||
| 528 : | edgomez | 1.16 | /* Quantization */ |
| 529 : | edgomez | 1.48 | quant_h263_intra = quant_h263_intra_sse2; |
| 530 : | quant_h263_inter = quant_h263_inter_sse2; | ||
| 531 : | dequant_h263_intra = dequant_h263_intra_sse2; | ||
| 532 : | dequant_h263_inter = dequant_h263_inter_sse2; | ||
| 533 : | h | 1.13 | |
| 534 : | edgomez | 1.48 | /* SAD operators */ |
| 535 : | edgomez | 1.16 | sad16 = sad16_sse2; |
| 536 : | dev16 = dev16_sse2; | ||
| 537 : | edgomez | 1.48 | |
| 538 : | Skal | 1.63 | /* DCT operators */ |
| 539 : | edgomez | 1.48 | fdct = fdct_sse2_skal; |
| 540 : | suxen_drol | 1.64 | /* idct = idct_sse2_skal; */ /* Is now IEEE1180 and Walken compliant. Disabled until fully tested. */ |
| 541 : | suxen_drol | 1.54 | |
| 542 : | /* postprocessing */ | ||
| 543 : | image_brightness = image_brightness_sse2; | ||
| 544 : | h | 1.12 | } |
| 545 : | edgomez | 1.48 | #endif /* ARCH_IS_IA32 */ |
| 546 : | Isibaar | 1.18 | |
| 547 : | edgomez | 1.41 | #if defined(ARCH_IS_IA64) |
| 548 : | edgomez | 1.43 | if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */ |
| 549 : | Isibaar | 1.18 | idct_ia64_init(); |
| 550 : | fdct = fdct_ia64; | ||
| 551 : | edgomez | 1.43 | idct = idct_ia64; /*not yet working, crashes */ |
| 552 : | Isibaar | 1.18 | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_ia64; |
| 553 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_ia64; | ||
| 554 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_ia64; | ||
| 555 : | sad16 = sad16_ia64; | ||
| 556 : | sad16bi = sad16bi_ia64; | ||
| 557 : | sad8 = sad8_ia64; | ||
| 558 : | dev16 = dev16_ia64; | ||
| 559 : | edgomez | 1.43 | /* Halfpel8_Refine = Halfpel8_Refine_ia64; */ |
| 560 : | edgomez | 1.48 | quant_h263_intra = quant_h263_intra_ia64; |
| 561 : | quant_h263_inter = quant_h263_inter_ia64; | ||
| 562 : | dequant_h263_intra = dequant_h263_intra_ia64; | ||
| 563 : | dequant_h263_inter = dequant_h263_inter_ia64; | ||
| 564 : | Isibaar | 1.18 | transfer_8to16copy = transfer_8to16copy_ia64; |
| 565 : | transfer_16to8copy = transfer_16to8copy_ia64; | ||
| 566 : | transfer_8to16sub = transfer_8to16sub_ia64; | ||
| 567 : | transfer_8to16sub2 = transfer_8to16sub2_ia64; | ||
| 568 : | transfer_16to8add = transfer_16to8add_ia64; | ||
| 569 : | transfer8x8_copy = transfer8x8_copy_ia64; | ||
| 570 : | } | ||
| 571 : | edgomez | 1.48 | #endif |
| 572 : | edgomez | 1.16 | |
| 573 : | edgomez | 1.41 | #if defined(ARCH_IS_PPC) |
| 574 : | edgomez | 1.50 | if ((cpu_flags & XVID_CPU_ALTIVEC)) { |
| 575 : | /* sad operators */ | ||
| 576 : | edgomez | 1.56 | sad16 = sad16_altivec_c; |
| 577 : | sad16bi = sad16bi_altivec_c; | ||
| 578 : | sad8 = sad8_altivec_c; | ||
| 579 : | dev16 = dev16_altivec_c; | ||
| 580 : | edgomez | 1.50 | |
| 581 : | sse8_16bit = sse8_16bit_altivec_c; | ||
| 582 : | |||
| 583 : | /* mem transfer */ | ||
| 584 : | transfer_8to16copy = transfer_8to16copy_altivec_c; | ||
| 585 : | transfer_16to8copy = transfer_16to8copy_altivec_c; | ||
| 586 : | transfer_8to16sub = transfer_8to16sub_altivec_c; | ||
| 587 : | transfer_8to16subro = transfer_8to16subro_altivec_c; | ||
| 588 : | transfer_8to16sub2 = transfer_8to16sub2_altivec_c; | ||
| 589 : | transfer_16to8add = transfer_16to8add_altivec_c; | ||
| 590 : | transfer8x8_copy = transfer8x8_copy_altivec_c; | ||
| 591 : | |||
| 592 : | /* Inverse DCT */ | ||
| 593 : | idct = idct_altivec_c; | ||
| 594 : | |||
| 595 : | /* Interpolation */ | ||
| 596 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_altivec_c; | ||
| 597 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_altivec_c; | ||
| 598 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_altivec_c; | ||
| 599 : | edgomez | 1.58 | |
| 600 : | edgomez | 1.50 | interpolate8x8_avg2 = interpolate8x8_avg2_altivec_c; |
| 601 : | interpolate8x8_avg4 = interpolate8x8_avg4_altivec_c; | ||
| 602 : | edgomez | 1.56 | |
| 603 : | interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_altivec_c; | ||
| 604 : | interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_altivec_c; | ||
| 605 : | interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_altivec_c; | ||
| 606 : | interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_altivec_c; | ||
| 607 : | edgomez | 1.50 | |
| 608 : | /* Colorspace conversion */ | ||
| 609 : | bgra_to_yv12 = bgra_to_yv12_altivec_c; | ||
| 610 : | abgr_to_yv12 = abgr_to_yv12_altivec_c; | ||
| 611 : | rgba_to_yv12 = rgba_to_yv12_altivec_c; | ||
| 612 : | argb_to_yv12 = argb_to_yv12_altivec_c; | ||
| 613 : | |||
| 614 : | yuyv_to_yv12 = yuyv_to_yv12_altivec_c; | ||
| 615 : | uyvy_to_yv12 = uyvy_to_yv12_altivec_c; | ||
| 616 : | |||
| 617 : | yv12_to_yuyv = yv12_to_yuyv_altivec_c; | ||
| 618 : | yv12_to_uyvy = yv12_to_uyvy_altivec_c; | ||
| 619 : | |||
| 620 : | /* Quantization */ | ||
| 621 : | quant_h263_intra = quant_h263_intra_altivec_c; | ||
| 622 : | quant_h263_inter = quant_h263_inter_altivec_c; | ||
| 623 : | dequant_h263_intra = dequant_h263_intra_altivec_c; | ||
| 624 : | dequant_h263_inter = dequant_h263_inter_altivec_c; | ||
| 625 : | edgomez | 1.62 | |
| 626 : | dequant_mpeg_intra = dequant_mpeg_intra_altivec_c; | ||
| 627 : | dequant_mpeg_inter = dequant_mpeg_inter_altivec_c; | ||
| 628 : | edgomez | 1.56 | |
| 629 : | /* Qpel stuff */ | ||
| 630 : | xvid_QP_Funcs = &xvid_QP_Funcs_Altivec_C; | ||
| 631 : | xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_Altivec_C; | ||
| 632 : | edgomez | 1.50 | } |
| 633 : | canard | 1.6 | #endif |
| 634 : | edgomez | 1.41 | |
| 635 : | edgomez | 1.60 | #if defined(ARCH_IS_X86_64) |
| 636 : | /* For now, only XVID_CPU_ASM is looked for, so user can still | ||
| 637 : | * disable asm usage the usual way. When Intel EMT64 cpus will | ||
| 638 : | * be out, maybe we'll have to check more precisely what cpu | ||
| 639 : | * features there really are. */ | ||
| 640 : | if (cpu_flags & XVID_CPU_ASM) { | ||
| 641 : | /* SIMD state flusher */ | ||
| 642 : | Isibaar | 1.67 | emms = emms_mmx; |
| 643 : | edgomez | 1.60 | |
| 644 : | /* DCT operators */ | ||
| 645 : | fdct = fdct_skal_x86_64; | ||
| 646 : | idct = idct_x86_64; | ||
| 647 : | |||
| 648 : | /* SAD operators */ | ||
| 649 : | sad16 = sad16_x86_64; | ||
| 650 : | sad8 = sad8_x86_64; | ||
| 651 : | sad16bi = sad16bi_x86_64; | ||
| 652 : | sad8bi = sad8bi_x86_64; | ||
| 653 : | dev16 = dev16_x86_64; | ||
| 654 : | sad16v = sad16v_x86_64; | ||
| 655 : | sse8_16bit = sse8_16bit_x86_64; | ||
| 656 : | sse8_8bit = sse8_8bit_x86_64; | ||
| 657 : | |||
| 658 : | /* Interpolation operators */ | ||
| 659 : | interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_x86_64; | ||
| 660 : | interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_x86_64; | ||
| 661 : | interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_x86_64; | ||
| 662 : | |||
| 663 : | interpolate8x8_halfpel_add = interpolate8x8_halfpel_add_x86_64; | ||
| 664 : | interpolate8x8_halfpel_h_add = interpolate8x8_halfpel_h_add_x86_64; | ||
| 665 : | interpolate8x8_halfpel_v_add = interpolate8x8_halfpel_v_add_x86_64; | ||
| 666 : | interpolate8x8_halfpel_hv_add = interpolate8x8_halfpel_hv_add_x86_64; | ||
| 667 : | |||
| 668 : | interpolate8x8_6tap_lowpass_h = interpolate8x8_6tap_lowpass_h_x86_64; | ||
| 669 : | interpolate8x8_6tap_lowpass_v = interpolate8x8_6tap_lowpass_v_x86_64; | ||
| 670 : | |||
| 671 : | interpolate8x8_avg2 = interpolate8x8_avg2_x86_64; | ||
| 672 : | interpolate8x8_avg4 = interpolate8x8_avg4_x86_64; | ||
| 673 : | |||
| 674 : | /* Quantization related functions */ | ||
| 675 : | quant_h263_intra = quant_h263_intra_x86_64; | ||
| 676 : | quant_h263_inter = quant_h263_inter_x86_64; | ||
| 677 : | dequant_h263_intra = dequant_h263_intra_x86_64; | ||
| 678 : | dequant_h263_inter = dequant_h263_inter_x86_64; | ||
| 679 : | quant_mpeg_intra = quant_mpeg_intra_x86_64; | ||
| 680 : | quant_mpeg_inter = quant_mpeg_inter_x86_64; | ||
| 681 : | dequant_mpeg_intra = dequant_mpeg_intra_x86_64; | ||
| 682 : | dequant_mpeg_inter = dequant_mpeg_inter_x86_64; | ||
| 683 : | |||
| 684 : | /* Block related functions */ | ||
| 685 : | transfer_8to16copy = transfer_8to16copy_x86_64; | ||
| 686 : | transfer_16to8copy = transfer_16to8copy_x86_64; | ||
| 687 : | transfer_8to16sub = transfer_8to16sub_x86_64; | ||
| 688 : | transfer_8to16subro = transfer_8to16subro_x86_64; | ||
| 689 : | transfer_8to16sub2 = transfer_8to16sub2_x86_64; | ||
| 690 : | transfer_8to16sub2ro= transfer_8to16sub2ro_x86_64; | ||
| 691 : | transfer_16to8add = transfer_16to8add_x86_64; | ||
| 692 : | transfer8x8_copy = transfer8x8_copy_x86_64; | ||
| 693 : | |||
| 694 : | /* Qpel stuff */ | ||
| 695 : | xvid_QP_Funcs = &xvid_QP_Funcs_x86_64; | ||
| 696 : | xvid_QP_Add_Funcs = &xvid_QP_Add_Funcs_x86_64; | ||
| 697 : | |||
| 698 : | /* Interlacing Functions */ | ||
| 699 : | MBFieldTest = MBFieldTest_x86_64; | ||
| 700 : | } | ||
| 701 : | #endif | ||
| 702 : | |||
| 703 : | edgomez | 1.48 | #if defined(_DEBUG) |
| 704 : | xvid_debug = init->debug; | ||
| 705 : | #endif | ||
| 706 : | |||
| 707 : | edgomez | 1.50 | return(0); |
| 708 : | edgomez | 1.41 | } |
| 709 : | |||
| 710 : | |||
| 711 : | static int | ||
| 712 : | edgomez | 1.48 | xvid_gbl_info(xvid_gbl_info_t * info) |
| 713 : | edgomez | 1.41 | { |
| 714 : | edgomez | 1.48 | if (XVID_VERSION_MAJOR(info->version) != 1) /* v1.x.x */ |
| 715 : | return XVID_ERR_VERSION; | ||
| 716 : | edgomez | 1.41 | |
| 717 : | edgomez | 1.48 | info->actual_version = XVID_VERSION; |
| 718 : | suxen_drol | 1.66 | info->build = "xvid-1.2.0-dev"; |
| 719 : | edgomez | 1.48 | info->cpu_flags = detect_cpu_flags(); |
| 720 : | suxen_drol | 1.68 | info->num_threads = 0; |
| 721 : | edgomez | 1.41 | |
| 722 : | suxen_drol | 1.68 | #if defined(WIN32) |
| 723 : | { | ||
| 724 : | DWORD dwProcessAffinityMask, dwSystemAffinityMask; | ||
| 725 : | if (GetProcessAffinityMask(GetCurrentProcess(), &dwProcessAffinityMask, &dwSystemAffinityMask)) { | ||
| 726 : | int i; | ||
| 727 : | for(i=0; i<32; i++) { | ||
| 728 : | if ((dwProcessAffinityMask & (1<<i))) | ||
| 729 : | info->num_threads++; | ||
| 730 : | } | ||
| 731 : | } | ||
| 732 : | } | ||
| 733 : | edgomez | 1.48 | #endif |
| 734 : | edgomez | 1.41 | |
| 735 : | return 0; | ||
| 736 : | } | ||
| 737 : | |||
| 738 : | |||
| 739 : | edgomez | 1.48 | static int |
| 740 : | xvid_gbl_convert(xvid_gbl_convert_t* convert) | ||
| 741 : | edgomez | 1.41 | { |
| 742 : | edgomez | 1.48 | int width; |
| 743 : | int height; | ||
| 744 : | int width2; | ||
| 745 : | int height2; | ||
| 746 : | IMAGE img; | ||
| 747 : | edgomez | 1.41 | |
| 748 : | edgomez | 1.48 | if (XVID_VERSION_MAJOR(convert->version) != 1) /* v1.x.x */ |
| 749 : | return XVID_ERR_VERSION; | ||
| 750 : | edgomez | 1.41 | |
| 751 : | edgomez | 1.48 | #if 0 |
| 752 : | const int flip1 = (convert->input.colorspace & XVID_CSP_VFLIP) ^ (convert->output.colorspace & XVID_CSP_VFLIP); | ||
| 753 : | #endif | ||
| 754 : | width = convert->width; | ||
| 755 : | height = convert->height; | ||
| 756 : | width2 = convert->width/2; | ||
| 757 : | height2 = convert->height/2; | ||
| 758 : | edgomez | 1.41 | |
| 759 : | edgomez | 1.48 | switch (convert->input.csp & ~XVID_CSP_VFLIP) |
| 760 : | edgomez | 1.41 | { |
| 761 : | edgomez | 1.48 | case XVID_CSP_YV12 : |
| 762 : | img.y = convert->input.plane[0]; | ||
| 763 : | img.v = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height; | ||
| 764 : | img.u = (uint8_t*)convert->input.plane[0] + convert->input.stride[0]*height + (convert->input.stride[0]/2)*height2; | ||
| 765 : | image_output(&img, width, height, width, | ||
| 766 : | (uint8_t**)convert->output.plane, convert->output.stride, | ||
| 767 : | convert->output.csp, convert->interlacing); | ||
| 768 : | break; | ||
| 769 : | edgomez | 1.41 | |
| 770 : | edgomez | 1.48 | default : |
| 771 : | return XVID_ERR_FORMAT; | ||
| 772 : | edgomez | 1.41 | } |
| 773 : | |||
| 774 : | |||
| 775 : | edgomez | 1.48 | emms(); |
| 776 : | edgomez | 1.41 | return 0; |
| 777 : | } | ||
| 778 : | |||
| 779 : | edgomez | 1.48 | /***************************************************************************** |
| 780 : | * XviD Global Entry point | ||
| 781 : | * | ||
| 782 : | * Well this function initialize all internal function pointers according | ||
| 783 : | * to the CPU features forced by the library client or autodetected (depending | ||
| 784 : | * on the XVID_CPU_FORCE flag). It also initializes vlc coding tables and all | ||
| 785 : | * image colorspace transformation tables. | ||
| 786 : | * | ||
| 787 : | ****************************************************************************/ | ||
| 788 : | Isibaar | 1.1 | |
| 789 : | edgomez | 1.41 | |
| 790 : | int | ||
| 791 : | edgomez | 1.48 | xvid_global(void *handle, |
| 792 : | edgomez | 1.41 | int opt, |
| 793 : | void *param1, | ||
| 794 : | void *param2) | ||
| 795 : | { | ||
| 796 : | switch(opt) | ||
| 797 : | { | ||
| 798 : | edgomez | 1.48 | case XVID_GBL_INIT : |
| 799 : | return xvid_gbl_init((xvid_gbl_init_t*)param1); | ||
| 800 : | edgomez | 1.41 | |
| 801 : | edgomez | 1.48 | case XVID_GBL_INFO : |
| 802 : | return xvid_gbl_info((xvid_gbl_info_t*)param1); | ||
| 803 : | |||
| 804 : | case XVID_GBL_CONVERT : | ||
| 805 : | return xvid_gbl_convert((xvid_gbl_convert_t*)param1); | ||
| 806 : | edgomez | 1.41 | |
| 807 : | default : | ||
| 808 : | return XVID_ERR_FAIL; | ||
| 809 : | } | ||
| 810 : | } | ||
| 811 : | |||
| 812 : | edgomez | 1.16 | /***************************************************************************** |
| 813 : | * XviD Native decoder entry point | ||
| 814 : | * | ||
| 815 : | * This function is just a wrapper to all the option cases. | ||
| 816 : | * | ||
| 817 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
| 818 : | * else returns the wrapped function result | ||
| 819 : | * | ||
| 820 : | ****************************************************************************/ | ||
| 821 : | |||
| 822 : | edgomez | 1.15 | int |
| 823 : | xvid_decore(void *handle, | ||
| 824 : | int opt, | ||
| 825 : | void *param1, | ||
| 826 : | void *param2) | ||
| 827 : | Isibaar | 1.1 | { |
| 828 : | edgomez | 1.15 | switch (opt) { |
| 829 : | case XVID_DEC_CREATE: | ||
| 830 : | edgomez | 1.48 | return decoder_create((xvid_dec_create_t *) param1); |
| 831 : | edgomez | 1.15 | |
| 832 : | case XVID_DEC_DESTROY: | ||
| 833 : | return decoder_destroy((DECODER *) handle); | ||
| 834 : | Isibaar | 1.1 | |
| 835 : | edgomez | 1.48 | case XVID_DEC_DECODE: |
| 836 : | return decoder_decode((DECODER *) handle, (xvid_dec_frame_t *) param1, (xvid_dec_stats_t*) param2); | ||
| 837 : | |||
| 838 : | Isibaar | 1.1 | default: |
| 839 : | edgomez | 1.15 | return XVID_ERR_FAIL; |
| 840 : | } | ||
| 841 : | Isibaar | 1.1 | } |
| 842 : | |||
| 843 : | edgomez | 1.16 | |
| 844 : | /***************************************************************************** | ||
| 845 : | * XviD Native encoder entry point | ||
| 846 : | * | ||
| 847 : | * This function is just a wrapper to all the option cases. | ||
| 848 : | * | ||
| 849 : | * Returned values : XVID_ERR_FAIL when opt is invalid | ||
| 850 : | * else returns the wrapped function result | ||
| 851 : | * | ||
| 852 : | ****************************************************************************/ | ||
| 853 : | Isibaar | 1.1 | |
| 854 : | edgomez | 1.15 | int |
| 855 : | xvid_encore(void *handle, | ||
| 856 : | int opt, | ||
| 857 : | void *param1, | ||
| 858 : | void *param2) | ||
| 859 : | Isibaar | 1.1 | { |
| 860 : | edgomez | 1.15 | switch (opt) { |
| 861 : | case XVID_ENC_ENCODE: | ||
| 862 : | edgomez | 1.41 | |
| 863 : | edgomez | 1.48 | return enc_encode((Encoder *) handle, |
| 864 : | (xvid_enc_frame_t *) param1, | ||
| 865 : | (xvid_enc_stats_t *) param2); | ||
| 866 : | edgomez | 1.15 | |
| 867 : | case XVID_ENC_CREATE: | ||
| 868 : | edgomez | 1.48 | return enc_create((xvid_enc_create_t *) param1); |
| 869 : | edgomez | 1.15 | |
| 870 : | case XVID_ENC_DESTROY: | ||
| 871 : | edgomez | 1.48 | return enc_destroy((Encoder *) handle); |
| 872 : | Isibaar | 1.1 | |
| 873 : | default: | ||
| 874 : | edgomez | 1.15 | return XVID_ERR_FAIL; |
| 875 : | } | ||
| 876 : | Isibaar | 1.1 | } |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |