dct32_template.c (7526B)
1 /* 2 * Template for the Discrete Cosine Transform for 32 samples 3 * Copyright (c) 2001, 2002 Fabrice Bellard 4 * 5 * This file is part of FFmpeg. 6 * 7 * FFmpeg is free software; you can redistribute it and/or 8 * modify it under the terms of the GNU Lesser General Public 9 * License as published by the Free Software Foundation; either 10 * version 2.1 of the License, or (at your option) any later version. 11 * 12 * FFmpeg is distributed in the hope that it will be useful, 13 * but WITHOUT ANY WARRANTY; without even the implied warranty of 14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 15 * Lesser General Public License for more details. 16 * 17 * You should have received a copy of the GNU Lesser General Public 18 * License along with FFmpeg; if not, write to the Free Software 19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 20 */ 21 22 #include "dct32.h" 23 #include "mathops.h" 24 #include "libavutil/internal.h" 25 26 #ifdef CHECKED 27 #define SUINT int 28 #define SUINT32 int32_t 29 #else 30 #define SUINT unsigned 31 #define SUINT32 uint32_t 32 #endif 33 34 #if DCT32_FLOAT 35 # define dct32 ff_dct32_float 36 # define FIXHR(x) ((float)(x)) 37 # define MULH3(x, y, s) ((s)*(y)*(x)) 38 # define INTFLOAT float 39 # define SUINTFLOAT float 40 #else 41 # define dct32 ff_dct32_fixed 42 # define FIXHR(a) ((int)((a) * (1LL<<32) + 0.5)) 43 # define MULH3(x, y, s) MULH((s)*(x), y) 44 # define INTFLOAT int 45 # define SUINTFLOAT SUINT 46 #endif 47 48 49 /* tab[i][j] = 1.0 / (2.0 * cos(pi*(2*k+1) / 2^(6 - j))) */ 50 51 /* cos(i*pi/64) */ 52 53 #define COS0_0 FIXHR(0.50060299823519630134/2) 54 #define COS0_1 FIXHR(0.50547095989754365998/2) 55 #define COS0_2 FIXHR(0.51544730992262454697/2) 56 #define COS0_3 FIXHR(0.53104259108978417447/2) 57 #define COS0_4 FIXHR(0.55310389603444452782/2) 58 #define COS0_5 FIXHR(0.58293496820613387367/2) 59 #define COS0_6 FIXHR(0.62250412303566481615/2) 60 #define COS0_7 FIXHR(0.67480834145500574602/2) 61 #define COS0_8 FIXHR(0.74453627100229844977/2) 62 #define COS0_9 FIXHR(0.83934964541552703873/2) 63 #define COS0_10 FIXHR(0.97256823786196069369/2) 64 #define COS0_11 FIXHR(1.16943993343288495515/4) 65 #define COS0_12 FIXHR(1.48416461631416627724/4) 66 #define COS0_13 FIXHR(2.05778100995341155085/8) 67 #define COS0_14 FIXHR(3.40760841846871878570/8) 68 #define COS0_15 FIXHR(10.19000812354805681150/32) 69 70 #define COS1_0 FIXHR(0.50241928618815570551/2) 71 #define COS1_1 FIXHR(0.52249861493968888062/2) 72 #define COS1_2 FIXHR(0.56694403481635770368/2) 73 #define COS1_3 FIXHR(0.64682178335999012954/2) 74 #define COS1_4 FIXHR(0.78815462345125022473/2) 75 #define COS1_5 FIXHR(1.06067768599034747134/4) 76 #define COS1_6 FIXHR(1.72244709823833392782/4) 77 #define COS1_7 FIXHR(5.10114861868916385802/16) 78 79 #define COS2_0 FIXHR(0.50979557910415916894/2) 80 #define COS2_1 FIXHR(0.60134488693504528054/2) 81 #define COS2_2 FIXHR(0.89997622313641570463/2) 82 #define COS2_3 FIXHR(2.56291544774150617881/8) 83 84 #define COS3_0 FIXHR(0.54119610014619698439/2) 85 #define COS3_1 FIXHR(1.30656296487637652785/4) 86 87 #define COS4_0 FIXHR(M_SQRT1_2/2) 88 89 /* butterfly operator */ 90 #define BF(a, b, c, s)\ 91 {\ 92 tmp0 = val##a + val##b;\ 93 tmp1 = val##a - val##b;\ 94 val##a = tmp0;\ 95 val##b = MULH3(tmp1, c, 1<<(s));\ 96 } 97 98 #define BF0(a, b, c, s)\ 99 {\ 100 tmp0 = tab[a] + tab[b];\ 101 tmp1 = tab[a] - tab[b];\ 102 val##a = tmp0;\ 103 val##b = MULH3(tmp1, c, 1<<(s));\ 104 } 105 106 #define BF1(a, b, c, d)\ 107 {\ 108 BF(a, b, COS4_0, 1);\ 109 BF(c, d,-COS4_0, 1);\ 110 val##c += val##d;\ 111 } 112 113 #define BF2(a, b, c, d)\ 114 {\ 115 BF(a, b, COS4_0, 1);\ 116 BF(c, d,-COS4_0, 1);\ 117 val##c += val##d;\ 118 val##a += val##c;\ 119 val##c += val##b;\ 120 val##b += val##d;\ 121 } 122 123 #define ADD(a, b) val##a += val##b 124 125 /* DCT32 without 1/sqrt(2) coef zero scaling. */ 126 void dct32(INTFLOAT *out, const INTFLOAT *tab_arg) 127 { 128 const SUINTFLOAT *tab = tab_arg; 129 SUINTFLOAT tmp0, tmp1; 130 131 SUINTFLOAT val0 , val1 , val2 , val3 , val4 , val5 , val6 , val7 , 132 val8 , val9 , val10, val11, val12, val13, val14, val15, 133 val16, val17, val18, val19, val20, val21, val22, val23, 134 val24, val25, val26, val27, val28, val29, val30, val31; 135 136 /* pass 1 */ 137 BF0( 0, 31, COS0_0 , 1); 138 BF0(15, 16, COS0_15, 5); 139 /* pass 2 */ 140 BF( 0, 15, COS1_0 , 1); 141 BF(16, 31,-COS1_0 , 1); 142 /* pass 1 */ 143 BF0( 7, 24, COS0_7 , 1); 144 BF0( 8, 23, COS0_8 , 1); 145 /* pass 2 */ 146 BF( 7, 8, COS1_7 , 4); 147 BF(23, 24,-COS1_7 , 4); 148 /* pass 3 */ 149 BF( 0, 7, COS2_0 , 1); 150 BF( 8, 15,-COS2_0 , 1); 151 BF(16, 23, COS2_0 , 1); 152 BF(24, 31,-COS2_0 , 1); 153 /* pass 1 */ 154 BF0( 3, 28, COS0_3 , 1); 155 BF0(12, 19, COS0_12, 2); 156 /* pass 2 */ 157 BF( 3, 12, COS1_3 , 1); 158 BF(19, 28,-COS1_3 , 1); 159 /* pass 1 */ 160 BF0( 4, 27, COS0_4 , 1); 161 BF0(11, 20, COS0_11, 2); 162 /* pass 2 */ 163 BF( 4, 11, COS1_4 , 1); 164 BF(20, 27,-COS1_4 , 1); 165 /* pass 3 */ 166 BF( 3, 4, COS2_3 , 3); 167 BF(11, 12,-COS2_3 , 3); 168 BF(19, 20, COS2_3 , 3); 169 BF(27, 28,-COS2_3 , 3); 170 /* pass 4 */ 171 BF( 0, 3, COS3_0 , 1); 172 BF( 4, 7,-COS3_0 , 1); 173 BF( 8, 11, COS3_0 , 1); 174 BF(12, 15,-COS3_0 , 1); 175 BF(16, 19, COS3_0 , 1); 176 BF(20, 23,-COS3_0 , 1); 177 BF(24, 27, COS3_0 , 1); 178 BF(28, 31,-COS3_0 , 1); 179 180 181 182 /* pass 1 */ 183 BF0( 1, 30, COS0_1 , 1); 184 BF0(14, 17, COS0_14, 3); 185 /* pass 2 */ 186 BF( 1, 14, COS1_1 , 1); 187 BF(17, 30,-COS1_1 , 1); 188 /* pass 1 */ 189 BF0( 6, 25, COS0_6 , 1); 190 BF0( 9, 22, COS0_9 , 1); 191 /* pass 2 */ 192 BF( 6, 9, COS1_6 , 2); 193 BF(22, 25,-COS1_6 , 2); 194 /* pass 3 */ 195 BF( 1, 6, COS2_1 , 1); 196 BF( 9, 14,-COS2_1 , 1); 197 BF(17, 22, COS2_1 , 1); 198 BF(25, 30,-COS2_1 , 1); 199 200 /* pass 1 */ 201 BF0( 2, 29, COS0_2 , 1); 202 BF0(13, 18, COS0_13, 3); 203 /* pass 2 */ 204 BF( 2, 13, COS1_2 , 1); 205 BF(18, 29,-COS1_2 , 1); 206 /* pass 1 */ 207 BF0( 5, 26, COS0_5 , 1); 208 BF0(10, 21, COS0_10, 1); 209 /* pass 2 */ 210 BF( 5, 10, COS1_5 , 2); 211 BF(21, 26,-COS1_5 , 2); 212 /* pass 3 */ 213 BF( 2, 5, COS2_2 , 1); 214 BF(10, 13,-COS2_2 , 1); 215 BF(18, 21, COS2_2 , 1); 216 BF(26, 29,-COS2_2 , 1); 217 /* pass 4 */ 218 BF( 1, 2, COS3_1 , 2); 219 BF( 5, 6,-COS3_1 , 2); 220 BF( 9, 10, COS3_1 , 2); 221 BF(13, 14,-COS3_1 , 2); 222 BF(17, 18, COS3_1 , 2); 223 BF(21, 22,-COS3_1 , 2); 224 BF(25, 26, COS3_1 , 2); 225 BF(29, 30,-COS3_1 , 2); 226 227 /* pass 5 */ 228 BF1( 0, 1, 2, 3); 229 BF2( 4, 5, 6, 7); 230 BF1( 8, 9, 10, 11); 231 BF2(12, 13, 14, 15); 232 BF1(16, 17, 18, 19); 233 BF2(20, 21, 22, 23); 234 BF1(24, 25, 26, 27); 235 BF2(28, 29, 30, 31); 236 237 /* pass 6 */ 238 239 ADD( 8, 12); 240 ADD(12, 10); 241 ADD(10, 14); 242 ADD(14, 9); 243 ADD( 9, 13); 244 ADD(13, 11); 245 ADD(11, 15); 246 247 out[ 0] = val0; 248 out[16] = val1; 249 out[ 8] = val2; 250 out[24] = val3; 251 out[ 4] = val4; 252 out[20] = val5; 253 out[12] = val6; 254 out[28] = val7; 255 out[ 2] = val8; 256 out[18] = val9; 257 out[10] = val10; 258 out[26] = val11; 259 out[ 6] = val12; 260 out[22] = val13; 261 out[14] = val14; 262 out[30] = val15; 263 264 ADD(24, 28); 265 ADD(28, 26); 266 ADD(26, 30); 267 ADD(30, 25); 268 ADD(25, 29); 269 ADD(29, 27); 270 ADD(27, 31); 271 272 out[ 1] = val16 + val24; 273 out[17] = val17 + val25; 274 out[ 9] = val18 + val26; 275 out[25] = val19 + val27; 276 out[ 5] = val20 + val28; 277 out[21] = val21 + val29; 278 out[13] = val22 + val30; 279 out[29] = val23 + val31; 280 out[ 3] = val24 + val20; 281 out[19] = val25 + val21; 282 out[11] = val26 + val22; 283 out[27] = val27 + val23; 284 out[ 7] = val28 + val18; 285 out[23] = val29 + val19; 286 out[15] = val30 + val17; 287 out[31] = val31; 288 }