Parent Directory
| 
Revision Log
Revision 1.6 - (view) (download)
| 1 : | edgomez | 1.4 | /***************************************************************************** |
| 2 : | * | ||
| 3 : | * XVID MPEG-4 VIDEO CODEC | ||
| 4 : | * - Inverse DCT (More precise version) - | ||
| 5 : | * | ||
| 6 : | * Copyright (c) 2001 Michael Niedermayer <michaelni@gmx.at> | ||
| 7 : | * | ||
| 8 : | * Originally distributed under the GNU LGPL License (ffmpeg). | ||
| 9 : | Isibaar | 1.6 | * It is licensed under the GNU GPL for the Xvid tree. |
| 10 : | edgomez | 1.4 | * |
| 11 : | * This program is free software ; you can redistribute it and/or modify | ||
| 12 : | * it under the terms of the GNU General Public License as published by | ||
| 13 : | * the Free Software Foundation ; either version 2 of the License, or | ||
| 14 : | * (at your option) any later version. | ||
| 15 : | * | ||
| 16 : | * This program is distributed in the hope that it will be useful, | ||
| 17 : | * but WITHOUT ANY WARRANTY ; without even the implied warranty of | ||
| 18 : | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
| 19 : | * GNU General Public License for more details. | ||
| 20 : | * | ||
| 21 : | * You should have received a copy of the GNU General Public License | ||
| 22 : | * along with this program ; if not, write to the Free Software | ||
| 23 : | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
| 24 : | edgomez | 1.2 | * |
| 25 : | Isibaar | 1.6 | * $Id: simple_idct.c,v 1.5 2004/04/05 20:36:36 edgomez Exp $ |
| 26 : | edgomez | 1.2 | * |
| 27 : | edgomez | 1.4 | ****************************************************************************/ |
| 28 : | |||
| 29 : | edgomez | 1.2 | /* |
| 30 : | based upon some outcommented c code from mpeg2dec (idct_mmx.c | ||
| 31 : | edgomez | 1.4 | written by Aaron Holtzman <aholtzma@ess.engr.uvic.ca>) |
| 32 : | edgomez | 1.2 | */ |
| 33 : | edgomez | 1.4 | |
| 34 : | edgomez | 1.2 | #include "../portab.h" |
| 35 : | #include "idct.h" | ||
| 36 : | |||
| 37 : | #if 0 | ||
| 38 : | #define W1 2841 /* 2048*sqrt (2)*cos (1*pi/16) */ | ||
| 39 : | #define W2 2676 /* 2048*sqrt (2)*cos (2*pi/16) */ | ||
| 40 : | #define W3 2408 /* 2048*sqrt (2)*cos (3*pi/16) */ | ||
| 41 : | #define W4 2048 /* 2048*sqrt (2)*cos (4*pi/16) */ | ||
| 42 : | #define W5 1609 /* 2048*sqrt (2)*cos (5*pi/16) */ | ||
| 43 : | #define W6 1108 /* 2048*sqrt (2)*cos (6*pi/16) */ | ||
| 44 : | #define W7 565 /* 2048*sqrt (2)*cos (7*pi/16) */ | ||
| 45 : | #define ROW_SHIFT 8 | ||
| 46 : | #define COL_SHIFT 17 | ||
| 47 : | #else | ||
| 48 : | edgomez | 1.4 | #define W1 22725 /* cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 */ |
| 49 : | #define W2 21407 /* cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 */ | ||
| 50 : | #define W3 19266 /* cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 */ | ||
| 51 : | #define W4 16383 /* cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 */ | ||
| 52 : | #define W5 12873 /* cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 */ | ||
| 53 : | #define W6 8867 /* cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 */ | ||
| 54 : | #define W7 4520 /* cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 */ | ||
| 55 : | edgomez | 1.2 | #define ROW_SHIFT 11 |
| 56 : | edgomez | 1.4 | #define COL_SHIFT 20 /* 6 */ |
| 57 : | edgomez | 1.2 | #endif |
| 58 : | |||
| 59 : | edgomez | 1.5 | /* |
| 60 : | PPC mac operation. Causes compile problems on newer ppc targets | ||
| 61 : | |||
| 62 : | Was originally: #if defined(ARCH_IS_PPC) | ||
| 63 : | */ | ||
| 64 : | #if 0 | ||
| 65 : | edgomez | 1.2 | |
| 66 : | /* signed 16x16 -> 32 multiply add accumulate */ | ||
| 67 : | #define MAC16(rt, ra, rb) \ | ||
| 68 : | asm ("maclhw %0, %2, %3" : "=r" (rt) : "0" (rt), "r" (ra), "r" (rb)); | ||
| 69 : | |||
| 70 : | /* signed 16x16 -> 32 multiply */ | ||
| 71 : | #define MUL16(rt, ra, rb) \ | ||
| 72 : | asm ("mullhw %0, %1, %2" : "=r" (rt) : "r" (ra), "r" (rb)); | ||
| 73 : | |||
| 74 : | #else | ||
| 75 : | |||
| 76 : | /* signed 16x16 -> 32 multiply add accumulate */ | ||
| 77 : | #define MAC16(rt, ra, rb) rt += (ra) * (rb) | ||
| 78 : | |||
| 79 : | /* signed 16x16 -> 32 multiply */ | ||
| 80 : | #define MUL16(rt, ra, rb) rt = (ra) * (rb) | ||
| 81 : | |||
| 82 : | #endif | ||
| 83 : | |||
| 84 : | static __inline void idctRowCondDC (int16_t * const row) | ||
| 85 : | { | ||
| 86 : | int a0, a1, a2, a3, b0, b1, b2, b3; | ||
| 87 : | #ifdef FAST_64BIT | ||
| 88 : | uint64_t temp; | ||
| 89 : | #else | ||
| 90 : | uint32_t temp; | ||
| 91 : | #endif | ||
| 92 : | |||
| 93 : | #ifdef FAST_64BIT | ||
| 94 : | #ifdef ARCH_IS_BIG_ENDIAN | ||
| 95 : | #define ROW0_MASK 0xffff000000000000LL | ||
| 96 : | #else | ||
| 97 : | #define ROW0_MASK 0xffffLL | ||
| 98 : | #endif | ||
| 99 : | edgomez | 1.4 | if ( ((((uint64_t *)row)[0] & ~ROW0_MASK) | |
| 100 : | edgomez | 1.2 | ((uint64_t *)row)[1]) == 0) { |
| 101 : | temp = (row[0] << 3) & 0xffff; | ||
| 102 : | temp += temp << 16; | ||
| 103 : | temp += temp << 32; | ||
| 104 : | ((uint64_t *)row)[0] = temp; | ||
| 105 : | ((uint64_t *)row)[1] = temp; | ||
| 106 : | return; | ||
| 107 : | } | ||
| 108 : | #else | ||
| 109 : | if (!(((uint32_t*)row)[1] | | ||
| 110 : | ((uint32_t*)row)[2] | | ||
| 111 : | edgomez | 1.4 | ((uint32_t*)row)[3] | |
| 112 : | edgomez | 1.2 | row[1])) { |
| 113 : | temp = (row[0] << 3) & 0xffff; | ||
| 114 : | temp += temp << 16; | ||
| 115 : | ((uint32_t*)row)[0]=((uint32_t*)row)[1] = | ||
| 116 : | ((uint32_t*)row)[2]=((uint32_t*)row)[3] = temp; | ||
| 117 : | return; | ||
| 118 : | } | ||
| 119 : | #endif | ||
| 120 : | |||
| 121 : | a0 = (W4 * row[0]) + (1 << (ROW_SHIFT - 1)); | ||
| 122 : | a1 = a0; | ||
| 123 : | a2 = a0; | ||
| 124 : | a3 = a0; | ||
| 125 : | |||
| 126 : | /* no need to optimize : gcc does it */ | ||
| 127 : | a0 += W2 * row[2]; | ||
| 128 : | a1 += W6 * row[2]; | ||
| 129 : | a2 -= W6 * row[2]; | ||
| 130 : | a3 -= W2 * row[2]; | ||
| 131 : | |||
| 132 : | MUL16(b0, W1, row[1]); | ||
| 133 : | MAC16(b0, W3, row[3]); | ||
| 134 : | MUL16(b1, W3, row[1]); | ||
| 135 : | MAC16(b1, -W7, row[3]); | ||
| 136 : | MUL16(b2, W5, row[1]); | ||
| 137 : | MAC16(b2, -W1, row[3]); | ||
| 138 : | MUL16(b3, W7, row[1]); | ||
| 139 : | MAC16(b3, -W5, row[3]); | ||
| 140 : | |||
| 141 : | #ifdef FAST_64BIT | ||
| 142 : | temp = ((uint64_t*)row)[1]; | ||
| 143 : | #else | ||
| 144 : | temp = ((uint32_t*)row)[2] | ((uint32_t*)row)[3]; | ||
| 145 : | #endif | ||
| 146 : | if (temp != 0) { | ||
| 147 : | a0 += W4*row[4] + W6*row[6]; | ||
| 148 : | a1 += - W4*row[4] - W2*row[6]; | ||
| 149 : | a2 += - W4*row[4] + W2*row[6]; | ||
| 150 : | a3 += W4*row[4] - W6*row[6]; | ||
| 151 : | |||
| 152 : | MAC16(b0, W5, row[5]); | ||
| 153 : | MAC16(b0, W7, row[7]); | ||
| 154 : | edgomez | 1.4 | |
| 155 : | edgomez | 1.2 | MAC16(b1, -W1, row[5]); |
| 156 : | MAC16(b1, -W5, row[7]); | ||
| 157 : | edgomez | 1.4 | |
| 158 : | edgomez | 1.2 | MAC16(b2, W7, row[5]); |
| 159 : | MAC16(b2, W3, row[7]); | ||
| 160 : | edgomez | 1.4 | |
| 161 : | edgomez | 1.2 | MAC16(b3, W3, row[5]); |
| 162 : | MAC16(b3, -W1, row[7]); | ||
| 163 : | } | ||
| 164 : | |||
| 165 : | row[0] = (a0 + b0) >> ROW_SHIFT; | ||
| 166 : | row[7] = (a0 - b0) >> ROW_SHIFT; | ||
| 167 : | row[1] = (a1 + b1) >> ROW_SHIFT; | ||
| 168 : | row[6] = (a1 - b1) >> ROW_SHIFT; | ||
| 169 : | row[2] = (a2 + b2) >> ROW_SHIFT; | ||
| 170 : | row[5] = (a2 - b2) >> ROW_SHIFT; | ||
| 171 : | row[3] = (a3 + b3) >> ROW_SHIFT; | ||
| 172 : | row[4] = (a3 - b3) >> ROW_SHIFT; | ||
| 173 : | } | ||
| 174 : | |||
| 175 : | |||
| 176 : | static __inline void idctSparseCol (int16_t * const col) | ||
| 177 : | { | ||
| 178 : | int a0, a1, a2, a3, b0, b1, b2, b3; | ||
| 179 : | |||
| 180 : | /* XXX: I did that only to give same values as previous code */ | ||
| 181 : | a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); | ||
| 182 : | a1 = a0; | ||
| 183 : | a2 = a0; | ||
| 184 : | a3 = a0; | ||
| 185 : | |||
| 186 : | a0 += + W2*col[8*2]; | ||
| 187 : | a1 += + W6*col[8*2]; | ||
| 188 : | a2 += - W6*col[8*2]; | ||
| 189 : | a3 += - W2*col[8*2]; | ||
| 190 : | |||
| 191 : | MUL16(b0, W1, col[8*1]); | ||
| 192 : | MUL16(b1, W3, col[8*1]); | ||
| 193 : | MUL16(b2, W5, col[8*1]); | ||
| 194 : | MUL16(b3, W7, col[8*1]); | ||
| 195 : | |||
| 196 : | MAC16(b0, + W3, col[8*3]); | ||
| 197 : | MAC16(b1, - W7, col[8*3]); | ||
| 198 : | MAC16(b2, - W1, col[8*3]); | ||
| 199 : | MAC16(b3, - W5, col[8*3]); | ||
| 200 : | |||
| 201 : | if(col[8*4]){ | ||
| 202 : | a0 += + W4*col[8*4]; | ||
| 203 : | a1 += - W4*col[8*4]; | ||
| 204 : | a2 += - W4*col[8*4]; | ||
| 205 : | a3 += + W4*col[8*4]; | ||
| 206 : | } | ||
| 207 : | |||
| 208 : | if (col[8*5]) { | ||
| 209 : | MAC16(b0, + W5, col[8*5]); | ||
| 210 : | MAC16(b1, - W1, col[8*5]); | ||
| 211 : | MAC16(b2, + W7, col[8*5]); | ||
| 212 : | MAC16(b3, + W3, col[8*5]); | ||
| 213 : | } | ||
| 214 : | |||
| 215 : | if(col[8*6]){ | ||
| 216 : | a0 += + W6*col[8*6]; | ||
| 217 : | a1 += - W2*col[8*6]; | ||
| 218 : | a2 += + W2*col[8*6]; | ||
| 219 : | a3 += - W6*col[8*6]; | ||
| 220 : | } | ||
| 221 : | |||
| 222 : | if (col[8*7]) { | ||
| 223 : | MAC16(b0, + W7, col[8*7]); | ||
| 224 : | MAC16(b1, - W5, col[8*7]); | ||
| 225 : | MAC16(b2, + W3, col[8*7]); | ||
| 226 : | MAC16(b3, - W1, col[8*7]); | ||
| 227 : | } | ||
| 228 : | |||
| 229 : | col[0 ] = ((a0 + b0) >> COL_SHIFT); | ||
| 230 : | col[8 ] = ((a1 + b1) >> COL_SHIFT); | ||
| 231 : | col[16] = ((a2 + b2) >> COL_SHIFT); | ||
| 232 : | col[24] = ((a3 + b3) >> COL_SHIFT); | ||
| 233 : | col[32] = ((a3 - b3) >> COL_SHIFT); | ||
| 234 : | col[40] = ((a2 - b2) >> COL_SHIFT); | ||
| 235 : | col[48] = ((a1 - b1) >> COL_SHIFT); | ||
| 236 : | col[56] = ((a0 - b0) >> COL_SHIFT); | ||
| 237 : | } | ||
| 238 : | |||
| 239 : | void simple_idct_c(int16_t * const block) | ||
| 240 : | { | ||
| 241 : | int i; | ||
| 242 : | for(i=0; i<8; i++) | ||
| 243 : | idctRowCondDC(block + i*8); | ||
| 244 : | edgomez | 1.4 | |
| 245 : | edgomez | 1.2 | for(i=0; i<8; i++) |
| 246 : | idctSparseCol(block + i); | ||
| 247 : | edgomez | 1.4 | } |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |