Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/scalar/codelets/n1_16.c @ 82:d0c2a83c1364

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW 3.3.8 source, and a Linux build
author Chris Cannam
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
rev   line source
Chris@82 1 /*
Chris@82 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@82 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@82 4 *
Chris@82 5 * This program is free software; you can redistribute it and/or modify
Chris@82 6 * it under the terms of the GNU General Public License as published by
Chris@82 7 * the Free Software Foundation; either version 2 of the License, or
Chris@82 8 * (at your option) any later version.
Chris@82 9 *
Chris@82 10 * This program is distributed in the hope that it will be useful,
Chris@82 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@82 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@82 13 * GNU General Public License for more details.
Chris@82 14 *
Chris@82 15 * You should have received a copy of the GNU General Public License
Chris@82 16 * along with this program; if not, write to the Free Software
Chris@82 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@82 18 *
Chris@82 19 */
Chris@82 20
Chris@82 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@82 22 /* Generated on Thu May 24 08:04:11 EDT 2018 */
Chris@82 23
Chris@82 24 #include "dft/codelet-dft.h"
Chris@82 25
Chris@82 26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
Chris@82 27
Chris@82 28 /* Generated by: ../../../genfft/gen_notw.native -fma -compact -variables 4 -pipeline-latency 4 -n 16 -name n1_16 -include dft/scalar/n.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 144 FP additions, 40 FP multiplications,
Chris@82 32 * (or, 104 additions, 0 multiplications, 40 fused multiply/add),
Chris@82 33 * 50 stack variables, 3 constants, and 64 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/scalar/n.h"
Chris@82 36
Chris@82 37 static void n1_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 38 {
Chris@82 39 DK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82 40 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@82 41 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 42 {
Chris@82 43 INT i;
Chris@82 44 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(64, is), MAKE_VOLATILE_STRIDE(64, os)) {
Chris@82 45 E T7, T1R, T25, TC, TN, T1x, T1H, T1l, Tt, T22, T2h, T1b, T1g, T1E, T1Z;
Chris@82 46 E T1D, Te, T1S, T26, TJ, TQ, T1m, T1n, TT, Tm, T1X, T2g, T10, T15, T1B;
Chris@82 47 E T1U, T1A;
Chris@82 48 {
Chris@82 49 E T3, TL, Ty, T1k, T6, T1j, TB, TM;
Chris@82 50 {
Chris@82 51 E T1, T2, Tw, Tx;
Chris@82 52 T1 = ri[0];
Chris@82 53 T2 = ri[WS(is, 8)];
Chris@82 54 T3 = T1 + T2;
Chris@82 55 TL = T1 - T2;
Chris@82 56 Tw = ii[0];
Chris@82 57 Tx = ii[WS(is, 8)];
Chris@82 58 Ty = Tw + Tx;
Chris@82 59 T1k = Tw - Tx;
Chris@82 60 }
Chris@82 61 {
Chris@82 62 E T4, T5, Tz, TA;
Chris@82 63 T4 = ri[WS(is, 4)];
Chris@82 64 T5 = ri[WS(is, 12)];
Chris@82 65 T6 = T4 + T5;
Chris@82 66 T1j = T4 - T5;
Chris@82 67 Tz = ii[WS(is, 4)];
Chris@82 68 TA = ii[WS(is, 12)];
Chris@82 69 TB = Tz + TA;
Chris@82 70 TM = Tz - TA;
Chris@82 71 }
Chris@82 72 T7 = T3 + T6;
Chris@82 73 T1R = T3 - T6;
Chris@82 74 T25 = Ty - TB;
Chris@82 75 TC = Ty + TB;
Chris@82 76 TN = TL - TM;
Chris@82 77 T1x = TL + TM;
Chris@82 78 T1H = T1k - T1j;
Chris@82 79 T1l = T1j + T1k;
Chris@82 80 }
Chris@82 81 {
Chris@82 82 E Tp, T1c, T1a, T20, Ts, T17, T1f, T21;
Chris@82 83 {
Chris@82 84 E Tn, To, T18, T19;
Chris@82 85 Tn = ri[WS(is, 15)];
Chris@82 86 To = ri[WS(is, 7)];
Chris@82 87 Tp = Tn + To;
Chris@82 88 T1c = Tn - To;
Chris@82 89 T18 = ii[WS(is, 15)];
Chris@82 90 T19 = ii[WS(is, 7)];
Chris@82 91 T1a = T18 - T19;
Chris@82 92 T20 = T18 + T19;
Chris@82 93 }
Chris@82 94 {
Chris@82 95 E Tq, Tr, T1d, T1e;
Chris@82 96 Tq = ri[WS(is, 3)];
Chris@82 97 Tr = ri[WS(is, 11)];
Chris@82 98 Ts = Tq + Tr;
Chris@82 99 T17 = Tq - Tr;
Chris@82 100 T1d = ii[WS(is, 3)];
Chris@82 101 T1e = ii[WS(is, 11)];
Chris@82 102 T1f = T1d - T1e;
Chris@82 103 T21 = T1d + T1e;
Chris@82 104 }
Chris@82 105 Tt = Tp + Ts;
Chris@82 106 T22 = T20 - T21;
Chris@82 107 T2h = T20 + T21;
Chris@82 108 T1b = T17 + T1a;
Chris@82 109 T1g = T1c - T1f;
Chris@82 110 T1E = T1a - T17;
Chris@82 111 T1Z = Tp - Ts;
Chris@82 112 T1D = T1c + T1f;
Chris@82 113 }
Chris@82 114 {
Chris@82 115 E Ta, TP, TF, TO, Td, TR, TI, TS;
Chris@82 116 {
Chris@82 117 E T8, T9, TD, TE;
Chris@82 118 T8 = ri[WS(is, 2)];
Chris@82 119 T9 = ri[WS(is, 10)];
Chris@82 120 Ta = T8 + T9;
Chris@82 121 TP = T8 - T9;
Chris@82 122 TD = ii[WS(is, 2)];
Chris@82 123 TE = ii[WS(is, 10)];
Chris@82 124 TF = TD + TE;
Chris@82 125 TO = TD - TE;
Chris@82 126 }
Chris@82 127 {
Chris@82 128 E Tb, Tc, TG, TH;
Chris@82 129 Tb = ri[WS(is, 14)];
Chris@82 130 Tc = ri[WS(is, 6)];
Chris@82 131 Td = Tb + Tc;
Chris@82 132 TR = Tb - Tc;
Chris@82 133 TG = ii[WS(is, 14)];
Chris@82 134 TH = ii[WS(is, 6)];
Chris@82 135 TI = TG + TH;
Chris@82 136 TS = TG - TH;
Chris@82 137 }
Chris@82 138 Te = Ta + Td;
Chris@82 139 T1S = TF - TI;
Chris@82 140 T26 = Td - Ta;
Chris@82 141 TJ = TF + TI;
Chris@82 142 TQ = TO - TP;
Chris@82 143 T1m = TR - TS;
Chris@82 144 T1n = TP + TO;
Chris@82 145 TT = TR + TS;
Chris@82 146 }
Chris@82 147 {
Chris@82 148 E Ti, T11, TZ, T1V, Tl, TW, T14, T1W;
Chris@82 149 {
Chris@82 150 E Tg, Th, TX, TY;
Chris@82 151 Tg = ri[WS(is, 1)];
Chris@82 152 Th = ri[WS(is, 9)];
Chris@82 153 Ti = Tg + Th;
Chris@82 154 T11 = Tg - Th;
Chris@82 155 TX = ii[WS(is, 1)];
Chris@82 156 TY = ii[WS(is, 9)];
Chris@82 157 TZ = TX - TY;
Chris@82 158 T1V = TX + TY;
Chris@82 159 }
Chris@82 160 {
Chris@82 161 E Tj, Tk, T12, T13;
Chris@82 162 Tj = ri[WS(is, 5)];
Chris@82 163 Tk = ri[WS(is, 13)];
Chris@82 164 Tl = Tj + Tk;
Chris@82 165 TW = Tj - Tk;
Chris@82 166 T12 = ii[WS(is, 5)];
Chris@82 167 T13 = ii[WS(is, 13)];
Chris@82 168 T14 = T12 - T13;
Chris@82 169 T1W = T12 + T13;
Chris@82 170 }
Chris@82 171 Tm = Ti + Tl;
Chris@82 172 T1X = T1V - T1W;
Chris@82 173 T2g = T1V + T1W;
Chris@82 174 T10 = TW + TZ;
Chris@82 175 T15 = T11 - T14;
Chris@82 176 T1B = TZ - TW;
Chris@82 177 T1U = Ti - Tl;
Chris@82 178 T1A = T11 + T14;
Chris@82 179 }
Chris@82 180 {
Chris@82 181 E Tf, Tu, T2j, T2k;
Chris@82 182 Tf = T7 + Te;
Chris@82 183 Tu = Tm + Tt;
Chris@82 184 ro[WS(os, 8)] = Tf - Tu;
Chris@82 185 ro[0] = Tf + Tu;
Chris@82 186 T2j = TC + TJ;
Chris@82 187 T2k = T2g + T2h;
Chris@82 188 io[WS(os, 8)] = T2j - T2k;
Chris@82 189 io[0] = T2j + T2k;
Chris@82 190 }
Chris@82 191 {
Chris@82 192 E Tv, TK, T2f, T2i;
Chris@82 193 Tv = Tt - Tm;
Chris@82 194 TK = TC - TJ;
Chris@82 195 io[WS(os, 4)] = Tv + TK;
Chris@82 196 io[WS(os, 12)] = TK - Tv;
Chris@82 197 T2f = T7 - Te;
Chris@82 198 T2i = T2g - T2h;
Chris@82 199 ro[WS(os, 12)] = T2f - T2i;
Chris@82 200 ro[WS(os, 4)] = T2f + T2i;
Chris@82 201 }
Chris@82 202 {
Chris@82 203 E T1T, T27, T24, T28, T1Y, T23;
Chris@82 204 T1T = T1R + T1S;
Chris@82 205 T27 = T25 - T26;
Chris@82 206 T1Y = T1U + T1X;
Chris@82 207 T23 = T1Z - T22;
Chris@82 208 T24 = T1Y + T23;
Chris@82 209 T28 = T23 - T1Y;
Chris@82 210 ro[WS(os, 10)] = FNMS(KP707106781, T24, T1T);
Chris@82 211 io[WS(os, 6)] = FMA(KP707106781, T28, T27);
Chris@82 212 ro[WS(os, 2)] = FMA(KP707106781, T24, T1T);
Chris@82 213 io[WS(os, 14)] = FNMS(KP707106781, T28, T27);
Chris@82 214 }
Chris@82 215 {
Chris@82 216 E T29, T2d, T2c, T2e, T2a, T2b;
Chris@82 217 T29 = T1R - T1S;
Chris@82 218 T2d = T26 + T25;
Chris@82 219 T2a = T1X - T1U;
Chris@82 220 T2b = T1Z + T22;
Chris@82 221 T2c = T2a - T2b;
Chris@82 222 T2e = T2a + T2b;
Chris@82 223 ro[WS(os, 14)] = FNMS(KP707106781, T2c, T29);
Chris@82 224 io[WS(os, 2)] = FMA(KP707106781, T2e, T2d);
Chris@82 225 ro[WS(os, 6)] = FMA(KP707106781, T2c, T29);
Chris@82 226 io[WS(os, 10)] = FNMS(KP707106781, T2e, T2d);
Chris@82 227 }
Chris@82 228 {
Chris@82 229 E TV, T1v, T1p, T1r, T1i, T1q, T1u, T1w, TU, T1o;
Chris@82 230 TU = TQ - TT;
Chris@82 231 TV = FMA(KP707106781, TU, TN);
Chris@82 232 T1v = FNMS(KP707106781, TU, TN);
Chris@82 233 T1o = T1m - T1n;
Chris@82 234 T1p = FNMS(KP707106781, T1o, T1l);
Chris@82 235 T1r = FMA(KP707106781, T1o, T1l);
Chris@82 236 {
Chris@82 237 E T16, T1h, T1s, T1t;
Chris@82 238 T16 = FMA(KP414213562, T15, T10);
Chris@82 239 T1h = FNMS(KP414213562, T1g, T1b);
Chris@82 240 T1i = T16 - T1h;
Chris@82 241 T1q = T16 + T1h;
Chris@82 242 T1s = FMA(KP414213562, T1b, T1g);
Chris@82 243 T1t = FNMS(KP414213562, T10, T15);
Chris@82 244 T1u = T1s - T1t;
Chris@82 245 T1w = T1t + T1s;
Chris@82 246 }
Chris@82 247 ro[WS(os, 11)] = FNMS(KP923879532, T1i, TV);
Chris@82 248 io[WS(os, 11)] = FNMS(KP923879532, T1u, T1r);
Chris@82 249 ro[WS(os, 3)] = FMA(KP923879532, T1i, TV);
Chris@82 250 io[WS(os, 3)] = FMA(KP923879532, T1u, T1r);
Chris@82 251 io[WS(os, 7)] = FNMS(KP923879532, T1q, T1p);
Chris@82 252 ro[WS(os, 7)] = FNMS(KP923879532, T1w, T1v);
Chris@82 253 io[WS(os, 15)] = FMA(KP923879532, T1q, T1p);
Chris@82 254 ro[WS(os, 15)] = FMA(KP923879532, T1w, T1v);
Chris@82 255 }
Chris@82 256 {
Chris@82 257 E T1z, T1L, T1J, T1P, T1G, T1K, T1O, T1Q, T1y, T1I;
Chris@82 258 T1y = T1n + T1m;
Chris@82 259 T1z = FMA(KP707106781, T1y, T1x);
Chris@82 260 T1L = FNMS(KP707106781, T1y, T1x);
Chris@82 261 T1I = TQ + TT;
Chris@82 262 T1J = FNMS(KP707106781, T1I, T1H);
Chris@82 263 T1P = FMA(KP707106781, T1I, T1H);
Chris@82 264 {
Chris@82 265 E T1C, T1F, T1M, T1N;
Chris@82 266 T1C = FMA(KP414213562, T1B, T1A);
Chris@82 267 T1F = FNMS(KP414213562, T1E, T1D);
Chris@82 268 T1G = T1C + T1F;
Chris@82 269 T1K = T1F - T1C;
Chris@82 270 T1M = FNMS(KP414213562, T1A, T1B);
Chris@82 271 T1N = FMA(KP414213562, T1D, T1E);
Chris@82 272 T1O = T1M - T1N;
Chris@82 273 T1Q = T1M + T1N;
Chris@82 274 }
Chris@82 275 ro[WS(os, 9)] = FNMS(KP923879532, T1G, T1z);
Chris@82 276 io[WS(os, 9)] = FNMS(KP923879532, T1Q, T1P);
Chris@82 277 ro[WS(os, 1)] = FMA(KP923879532, T1G, T1z);
Chris@82 278 io[WS(os, 1)] = FMA(KP923879532, T1Q, T1P);
Chris@82 279 io[WS(os, 13)] = FNMS(KP923879532, T1K, T1J);
Chris@82 280 ro[WS(os, 13)] = FNMS(KP923879532, T1O, T1L);
Chris@82 281 io[WS(os, 5)] = FMA(KP923879532, T1K, T1J);
Chris@82 282 ro[WS(os, 5)] = FMA(KP923879532, T1O, T1L);
Chris@82 283 }
Chris@82 284 }
Chris@82 285 }
Chris@82 286 }
Chris@82 287
Chris@82 288 static const kdft_desc desc = { 16, "n1_16", {104, 0, 40, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 289
Chris@82 290 void X(codelet_n1_16) (planner *p) {
Chris@82 291 X(kdft_register) (p, n1_16, &desc);
Chris@82 292 }
Chris@82 293
Chris@82 294 #else
Chris@82 295
Chris@82 296 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 16 -name n1_16 -include dft/scalar/n.h */
Chris@82 297
Chris@82 298 /*
Chris@82 299 * This function contains 144 FP additions, 24 FP multiplications,
Chris@82 300 * (or, 136 additions, 16 multiplications, 8 fused multiply/add),
Chris@82 301 * 50 stack variables, 3 constants, and 64 memory accesses
Chris@82 302 */
Chris@82 303 #include "dft/scalar/n.h"
Chris@82 304
Chris@82 305 static void n1_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 306 {
Chris@82 307 DK(KP382683432, +0.382683432365089771728459984030398866761344562);
Chris@82 308 DK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82 309 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 310 {
Chris@82 311 INT i;
Chris@82 312 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(64, is), MAKE_VOLATILE_STRIDE(64, os)) {
Chris@82 313 E T7, T1R, T25, TC, TN, T1x, T1H, T1l, Tt, T22, T2h, T1b, T1g, T1E, T1Z;
Chris@82 314 E T1D, Te, T1S, T26, TJ, TQ, T1m, T1n, TT, Tm, T1X, T2g, T10, T15, T1B;
Chris@82 315 E T1U, T1A;
Chris@82 316 {
Chris@82 317 E T3, TL, Ty, T1k, T6, T1j, TB, TM;
Chris@82 318 {
Chris@82 319 E T1, T2, Tw, Tx;
Chris@82 320 T1 = ri[0];
Chris@82 321 T2 = ri[WS(is, 8)];
Chris@82 322 T3 = T1 + T2;
Chris@82 323 TL = T1 - T2;
Chris@82 324 Tw = ii[0];
Chris@82 325 Tx = ii[WS(is, 8)];
Chris@82 326 Ty = Tw + Tx;
Chris@82 327 T1k = Tw - Tx;
Chris@82 328 }
Chris@82 329 {
Chris@82 330 E T4, T5, Tz, TA;
Chris@82 331 T4 = ri[WS(is, 4)];
Chris@82 332 T5 = ri[WS(is, 12)];
Chris@82 333 T6 = T4 + T5;
Chris@82 334 T1j = T4 - T5;
Chris@82 335 Tz = ii[WS(is, 4)];
Chris@82 336 TA = ii[WS(is, 12)];
Chris@82 337 TB = Tz + TA;
Chris@82 338 TM = Tz - TA;
Chris@82 339 }
Chris@82 340 T7 = T3 + T6;
Chris@82 341 T1R = T3 - T6;
Chris@82 342 T25 = Ty - TB;
Chris@82 343 TC = Ty + TB;
Chris@82 344 TN = TL - TM;
Chris@82 345 T1x = TL + TM;
Chris@82 346 T1H = T1k - T1j;
Chris@82 347 T1l = T1j + T1k;
Chris@82 348 }
Chris@82 349 {
Chris@82 350 E Tp, T17, T1f, T20, Ts, T1c, T1a, T21;
Chris@82 351 {
Chris@82 352 E Tn, To, T1d, T1e;
Chris@82 353 Tn = ri[WS(is, 15)];
Chris@82 354 To = ri[WS(is, 7)];
Chris@82 355 Tp = Tn + To;
Chris@82 356 T17 = Tn - To;
Chris@82 357 T1d = ii[WS(is, 15)];
Chris@82 358 T1e = ii[WS(is, 7)];
Chris@82 359 T1f = T1d - T1e;
Chris@82 360 T20 = T1d + T1e;
Chris@82 361 }
Chris@82 362 {
Chris@82 363 E Tq, Tr, T18, T19;
Chris@82 364 Tq = ri[WS(is, 3)];
Chris@82 365 Tr = ri[WS(is, 11)];
Chris@82 366 Ts = Tq + Tr;
Chris@82 367 T1c = Tq - Tr;
Chris@82 368 T18 = ii[WS(is, 3)];
Chris@82 369 T19 = ii[WS(is, 11)];
Chris@82 370 T1a = T18 - T19;
Chris@82 371 T21 = T18 + T19;
Chris@82 372 }
Chris@82 373 Tt = Tp + Ts;
Chris@82 374 T22 = T20 - T21;
Chris@82 375 T2h = T20 + T21;
Chris@82 376 T1b = T17 - T1a;
Chris@82 377 T1g = T1c + T1f;
Chris@82 378 T1E = T1f - T1c;
Chris@82 379 T1Z = Tp - Ts;
Chris@82 380 T1D = T17 + T1a;
Chris@82 381 }
Chris@82 382 {
Chris@82 383 E Ta, TP, TF, TO, Td, TR, TI, TS;
Chris@82 384 {
Chris@82 385 E T8, T9, TD, TE;
Chris@82 386 T8 = ri[WS(is, 2)];
Chris@82 387 T9 = ri[WS(is, 10)];
Chris@82 388 Ta = T8 + T9;
Chris@82 389 TP = T8 - T9;
Chris@82 390 TD = ii[WS(is, 2)];
Chris@82 391 TE = ii[WS(is, 10)];
Chris@82 392 TF = TD + TE;
Chris@82 393 TO = TD - TE;
Chris@82 394 }
Chris@82 395 {
Chris@82 396 E Tb, Tc, TG, TH;
Chris@82 397 Tb = ri[WS(is, 14)];
Chris@82 398 Tc = ri[WS(is, 6)];
Chris@82 399 Td = Tb + Tc;
Chris@82 400 TR = Tb - Tc;
Chris@82 401 TG = ii[WS(is, 14)];
Chris@82 402 TH = ii[WS(is, 6)];
Chris@82 403 TI = TG + TH;
Chris@82 404 TS = TG - TH;
Chris@82 405 }
Chris@82 406 Te = Ta + Td;
Chris@82 407 T1S = TF - TI;
Chris@82 408 T26 = Td - Ta;
Chris@82 409 TJ = TF + TI;
Chris@82 410 TQ = TO - TP;
Chris@82 411 T1m = TR - TS;
Chris@82 412 T1n = TP + TO;
Chris@82 413 TT = TR + TS;
Chris@82 414 }
Chris@82 415 {
Chris@82 416 E Ti, T11, TZ, T1V, Tl, TW, T14, T1W;
Chris@82 417 {
Chris@82 418 E Tg, Th, TX, TY;
Chris@82 419 Tg = ri[WS(is, 1)];
Chris@82 420 Th = ri[WS(is, 9)];
Chris@82 421 Ti = Tg + Th;
Chris@82 422 T11 = Tg - Th;
Chris@82 423 TX = ii[WS(is, 1)];
Chris@82 424 TY = ii[WS(is, 9)];
Chris@82 425 TZ = TX - TY;
Chris@82 426 T1V = TX + TY;
Chris@82 427 }
Chris@82 428 {
Chris@82 429 E Tj, Tk, T12, T13;
Chris@82 430 Tj = ri[WS(is, 5)];
Chris@82 431 Tk = ri[WS(is, 13)];
Chris@82 432 Tl = Tj + Tk;
Chris@82 433 TW = Tj - Tk;
Chris@82 434 T12 = ii[WS(is, 5)];
Chris@82 435 T13 = ii[WS(is, 13)];
Chris@82 436 T14 = T12 - T13;
Chris@82 437 T1W = T12 + T13;
Chris@82 438 }
Chris@82 439 Tm = Ti + Tl;
Chris@82 440 T1X = T1V - T1W;
Chris@82 441 T2g = T1V + T1W;
Chris@82 442 T10 = TW + TZ;
Chris@82 443 T15 = T11 - T14;
Chris@82 444 T1B = T11 + T14;
Chris@82 445 T1U = Ti - Tl;
Chris@82 446 T1A = TZ - TW;
Chris@82 447 }
Chris@82 448 {
Chris@82 449 E Tf, Tu, T2j, T2k;
Chris@82 450 Tf = T7 + Te;
Chris@82 451 Tu = Tm + Tt;
Chris@82 452 ro[WS(os, 8)] = Tf - Tu;
Chris@82 453 ro[0] = Tf + Tu;
Chris@82 454 T2j = TC + TJ;
Chris@82 455 T2k = T2g + T2h;
Chris@82 456 io[WS(os, 8)] = T2j - T2k;
Chris@82 457 io[0] = T2j + T2k;
Chris@82 458 }
Chris@82 459 {
Chris@82 460 E Tv, TK, T2f, T2i;
Chris@82 461 Tv = Tt - Tm;
Chris@82 462 TK = TC - TJ;
Chris@82 463 io[WS(os, 4)] = Tv + TK;
Chris@82 464 io[WS(os, 12)] = TK - Tv;
Chris@82 465 T2f = T7 - Te;
Chris@82 466 T2i = T2g - T2h;
Chris@82 467 ro[WS(os, 12)] = T2f - T2i;
Chris@82 468 ro[WS(os, 4)] = T2f + T2i;
Chris@82 469 }
Chris@82 470 {
Chris@82 471 E T1T, T27, T24, T28, T1Y, T23;
Chris@82 472 T1T = T1R + T1S;
Chris@82 473 T27 = T25 - T26;
Chris@82 474 T1Y = T1U + T1X;
Chris@82 475 T23 = T1Z - T22;
Chris@82 476 T24 = KP707106781 * (T1Y + T23);
Chris@82 477 T28 = KP707106781 * (T23 - T1Y);
Chris@82 478 ro[WS(os, 10)] = T1T - T24;
Chris@82 479 io[WS(os, 6)] = T27 + T28;
Chris@82 480 ro[WS(os, 2)] = T1T + T24;
Chris@82 481 io[WS(os, 14)] = T27 - T28;
Chris@82 482 }
Chris@82 483 {
Chris@82 484 E T29, T2d, T2c, T2e, T2a, T2b;
Chris@82 485 T29 = T1R - T1S;
Chris@82 486 T2d = T26 + T25;
Chris@82 487 T2a = T1X - T1U;
Chris@82 488 T2b = T1Z + T22;
Chris@82 489 T2c = KP707106781 * (T2a - T2b);
Chris@82 490 T2e = KP707106781 * (T2a + T2b);
Chris@82 491 ro[WS(os, 14)] = T29 - T2c;
Chris@82 492 io[WS(os, 2)] = T2d + T2e;
Chris@82 493 ro[WS(os, 6)] = T29 + T2c;
Chris@82 494 io[WS(os, 10)] = T2d - T2e;
Chris@82 495 }
Chris@82 496 {
Chris@82 497 E TV, T1r, T1p, T1v, T1i, T1q, T1u, T1w, TU, T1o;
Chris@82 498 TU = KP707106781 * (TQ - TT);
Chris@82 499 TV = TN + TU;
Chris@82 500 T1r = TN - TU;
Chris@82 501 T1o = KP707106781 * (T1m - T1n);
Chris@82 502 T1p = T1l - T1o;
Chris@82 503 T1v = T1l + T1o;
Chris@82 504 {
Chris@82 505 E T16, T1h, T1s, T1t;
Chris@82 506 T16 = FMA(KP923879532, T10, KP382683432 * T15);
Chris@82 507 T1h = FNMS(KP923879532, T1g, KP382683432 * T1b);
Chris@82 508 T1i = T16 + T1h;
Chris@82 509 T1q = T1h - T16;
Chris@82 510 T1s = FNMS(KP923879532, T15, KP382683432 * T10);
Chris@82 511 T1t = FMA(KP382683432, T1g, KP923879532 * T1b);
Chris@82 512 T1u = T1s - T1t;
Chris@82 513 T1w = T1s + T1t;
Chris@82 514 }
Chris@82 515 ro[WS(os, 11)] = TV - T1i;
Chris@82 516 io[WS(os, 11)] = T1v - T1w;
Chris@82 517 ro[WS(os, 3)] = TV + T1i;
Chris@82 518 io[WS(os, 3)] = T1v + T1w;
Chris@82 519 io[WS(os, 15)] = T1p - T1q;
Chris@82 520 ro[WS(os, 15)] = T1r - T1u;
Chris@82 521 io[WS(os, 7)] = T1p + T1q;
Chris@82 522 ro[WS(os, 7)] = T1r + T1u;
Chris@82 523 }
Chris@82 524 {
Chris@82 525 E T1z, T1L, T1J, T1P, T1G, T1K, T1O, T1Q, T1y, T1I;
Chris@82 526 T1y = KP707106781 * (T1n + T1m);
Chris@82 527 T1z = T1x + T1y;
Chris@82 528 T1L = T1x - T1y;
Chris@82 529 T1I = KP707106781 * (TQ + TT);
Chris@82 530 T1J = T1H - T1I;
Chris@82 531 T1P = T1H + T1I;
Chris@82 532 {
Chris@82 533 E T1C, T1F, T1M, T1N;
Chris@82 534 T1C = FMA(KP382683432, T1A, KP923879532 * T1B);
Chris@82 535 T1F = FNMS(KP382683432, T1E, KP923879532 * T1D);
Chris@82 536 T1G = T1C + T1F;
Chris@82 537 T1K = T1F - T1C;
Chris@82 538 T1M = FNMS(KP382683432, T1B, KP923879532 * T1A);
Chris@82 539 T1N = FMA(KP923879532, T1E, KP382683432 * T1D);
Chris@82 540 T1O = T1M - T1N;
Chris@82 541 T1Q = T1M + T1N;
Chris@82 542 }
Chris@82 543 ro[WS(os, 9)] = T1z - T1G;
Chris@82 544 io[WS(os, 9)] = T1P - T1Q;
Chris@82 545 ro[WS(os, 1)] = T1z + T1G;
Chris@82 546 io[WS(os, 1)] = T1P + T1Q;
Chris@82 547 io[WS(os, 13)] = T1J - T1K;
Chris@82 548 ro[WS(os, 13)] = T1L - T1O;
Chris@82 549 io[WS(os, 5)] = T1J + T1K;
Chris@82 550 ro[WS(os, 5)] = T1L + T1O;
Chris@82 551 }
Chris@82 552 }
Chris@82 553 }
Chris@82 554 }
Chris@82 555
Chris@82 556 static const kdft_desc desc = { 16, "n1_16", {136, 16, 8, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 557
Chris@82 558 void X(codelet_n1_16) (planner *p) {
Chris@82 559 X(kdft_register) (p, n1_16, &desc);
Chris@82 560 }
Chris@82 561
Chris@82 562 #endif