Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/scalar/codelets/n1_16.c @ 84:08ae793730bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +0000
parents 2cd0e3b3e1fd
children
rev   line source
Chris@42 1 /*
Chris@42 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42 4 *
Chris@42 5 * This program is free software; you can redistribute it and/or modify
Chris@42 6 * it under the terms of the GNU General Public License as published by
Chris@42 7 * the Free Software Foundation; either version 2 of the License, or
Chris@42 8 * (at your option) any later version.
Chris@42 9 *
Chris@42 10 * This program is distributed in the hope that it will be useful,
Chris@42 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@42 13 * GNU General Public License for more details.
Chris@42 14 *
Chris@42 15 * You should have received a copy of the GNU General Public License
Chris@42 16 * along with this program; if not, write to the Free Software
Chris@42 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@42 18 *
Chris@42 19 */
Chris@42 20
Chris@42 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@42 22 /* Generated on Sat Jul 30 16:35:56 EDT 2016 */
Chris@42 23
Chris@42 24 #include "codelet-dft.h"
Chris@42 25
Chris@42 26 #ifdef HAVE_FMA
Chris@42 27
Chris@42 28 /* Generated by: ../../../genfft/gen_notw.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 16 -name n1_16 -include n.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 144 FP additions, 40 FP multiplications,
Chris@42 32 * (or, 104 additions, 0 multiplications, 40 fused multiply/add),
Chris@42 33 * 82 stack variables, 3 constants, and 64 memory accesses
Chris@42 34 */
Chris@42 35 #include "n.h"
Chris@42 36
Chris@42 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@42 38 {
Chris@42 39 DK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@42 40 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@42 41 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 42 {
Chris@42 43 INT i;
Chris@42 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@42 45 E T1z, T1L, T1M, T1N, T1P, T1J, T1K, T1G, T1O, T1Q;
Chris@42 46 {
Chris@42 47 E T1l, T1H, T1R, T7, T1x, TN, TC, T25, T1E, T1b, T1Z, Tt, T2h, T22, T1D;
Chris@42 48 E T1g, T1n, TQ, Te, T26, TT, T1m, TJ, T1S, Tj, T11, Ti, T1V, TZ, Tk;
Chris@42 49 E T12, T13;
Chris@42 50 {
Chris@42 51 E Tq, T1c, Tp, T20, T1a, Tr, T1d, T1e;
Chris@42 52 {
Chris@42 53 E T4, TL, T3, T1k, Ty, T5, Tz, TA;
Chris@42 54 {
Chris@42 55 E T1, T2, Tw, Tx;
Chris@42 56 T1 = ri[0];
Chris@42 57 T2 = ri[WS(is, 8)];
Chris@42 58 Tw = ii[0];
Chris@42 59 Tx = ii[WS(is, 8)];
Chris@42 60 T4 = ri[WS(is, 4)];
Chris@42 61 TL = T1 - T2;
Chris@42 62 T3 = T1 + T2;
Chris@42 63 T1k = Tw - Tx;
Chris@42 64 Ty = Tw + Tx;
Chris@42 65 T5 = ri[WS(is, 12)];
Chris@42 66 Tz = ii[WS(is, 4)];
Chris@42 67 TA = ii[WS(is, 12)];
Chris@42 68 }
Chris@42 69 {
Chris@42 70 E Tn, To, T18, T19;
Chris@42 71 Tn = ri[WS(is, 15)];
Chris@42 72 {
Chris@42 73 E T1j, T6, TM, TB;
Chris@42 74 T1j = T4 - T5;
Chris@42 75 T6 = T4 + T5;
Chris@42 76 TM = Tz - TA;
Chris@42 77 TB = Tz + TA;
Chris@42 78 T1l = T1j + T1k;
Chris@42 79 T1H = T1k - T1j;
Chris@42 80 T1R = T3 - T6;
Chris@42 81 T7 = T3 + T6;
Chris@42 82 T1x = TL + TM;
Chris@42 83 TN = TL - TM;
Chris@42 84 TC = Ty + TB;
Chris@42 85 T25 = Ty - TB;
Chris@42 86 To = ri[WS(is, 7)];
Chris@42 87 }
Chris@42 88 T18 = ii[WS(is, 15)];
Chris@42 89 T19 = ii[WS(is, 7)];
Chris@42 90 Tq = ri[WS(is, 3)];
Chris@42 91 T1c = Tn - To;
Chris@42 92 Tp = Tn + To;
Chris@42 93 T20 = T18 + T19;
Chris@42 94 T1a = T18 - T19;
Chris@42 95 Tr = ri[WS(is, 11)];
Chris@42 96 T1d = ii[WS(is, 3)];
Chris@42 97 T1e = ii[WS(is, 11)];
Chris@42 98 }
Chris@42 99 }
Chris@42 100 {
Chris@42 101 E Tb, TP, Ta, TO, TF, Tc, TG, TH;
Chris@42 102 {
Chris@42 103 E T8, T9, TD, TE;
Chris@42 104 T8 = ri[WS(is, 2)];
Chris@42 105 {
Chris@42 106 E T17, Ts, T21, T1f;
Chris@42 107 T17 = Tq - Tr;
Chris@42 108 Ts = Tq + Tr;
Chris@42 109 T21 = T1d + T1e;
Chris@42 110 T1f = T1d - T1e;
Chris@42 111 T1E = T1a - T17;
Chris@42 112 T1b = T17 + T1a;
Chris@42 113 T1Z = Tp - Ts;
Chris@42 114 Tt = Tp + Ts;
Chris@42 115 T2h = T20 + T21;
Chris@42 116 T22 = T20 - T21;
Chris@42 117 T1D = T1c + T1f;
Chris@42 118 T1g = T1c - T1f;
Chris@42 119 T9 = ri[WS(is, 10)];
Chris@42 120 }
Chris@42 121 TD = ii[WS(is, 2)];
Chris@42 122 TE = ii[WS(is, 10)];
Chris@42 123 Tb = ri[WS(is, 14)];
Chris@42 124 TP = T8 - T9;
Chris@42 125 Ta = T8 + T9;
Chris@42 126 TO = TD - TE;
Chris@42 127 TF = TD + TE;
Chris@42 128 Tc = ri[WS(is, 6)];
Chris@42 129 TG = ii[WS(is, 14)];
Chris@42 130 TH = ii[WS(is, 6)];
Chris@42 131 }
Chris@42 132 {
Chris@42 133 E TR, Td, TS, TI;
Chris@42 134 T1n = TP + TO;
Chris@42 135 TQ = TO - TP;
Chris@42 136 TR = Tb - Tc;
Chris@42 137 Td = Tb + Tc;
Chris@42 138 TS = TG - TH;
Chris@42 139 TI = TG + TH;
Chris@42 140 Te = Ta + Td;
Chris@42 141 T26 = Td - Ta;
Chris@42 142 TT = TR + TS;
Chris@42 143 T1m = TR - TS;
Chris@42 144 TJ = TF + TI;
Chris@42 145 T1S = TF - TI;
Chris@42 146 }
Chris@42 147 }
Chris@42 148 {
Chris@42 149 E Tg, Th, TX, TY;
Chris@42 150 Tg = ri[WS(is, 1)];
Chris@42 151 Th = ri[WS(is, 9)];
Chris@42 152 TX = ii[WS(is, 1)];
Chris@42 153 TY = ii[WS(is, 9)];
Chris@42 154 Tj = ri[WS(is, 5)];
Chris@42 155 T11 = Tg - Th;
Chris@42 156 Ti = Tg + Th;
Chris@42 157 T1V = TX + TY;
Chris@42 158 TZ = TX - TY;
Chris@42 159 Tk = ri[WS(is, 13)];
Chris@42 160 T12 = ii[WS(is, 5)];
Chris@42 161 T13 = ii[WS(is, 13)];
Chris@42 162 }
Chris@42 163 }
Chris@42 164 {
Chris@42 165 E T2f, T1B, T10, T1U, T1X, T1A, T15, Tv, TK, T2i;
Chris@42 166 {
Chris@42 167 E Tf, Tu, T2j, T2k, T2g;
Chris@42 168 T2f = T7 - Te;
Chris@42 169 Tf = T7 + Te;
Chris@42 170 {
Chris@42 171 E TW, Tl, T1W, T14, Tm;
Chris@42 172 TW = Tj - Tk;
Chris@42 173 Tl = Tj + Tk;
Chris@42 174 T1W = T12 + T13;
Chris@42 175 T14 = T12 - T13;
Chris@42 176 T1B = TZ - TW;
Chris@42 177 T10 = TW + TZ;
Chris@42 178 T1U = Ti - Tl;
Chris@42 179 Tm = Ti + Tl;
Chris@42 180 T2g = T1V + T1W;
Chris@42 181 T1X = T1V - T1W;
Chris@42 182 T1A = T11 + T14;
Chris@42 183 T15 = T11 - T14;
Chris@42 184 Tu = Tm + Tt;
Chris@42 185 Tv = Tt - Tm;
Chris@42 186 }
Chris@42 187 TK = TC - TJ;
Chris@42 188 T2j = TC + TJ;
Chris@42 189 T2k = T2g + T2h;
Chris@42 190 T2i = T2g - T2h;
Chris@42 191 ro[0] = Tf + Tu;
Chris@42 192 ro[WS(os, 8)] = Tf - Tu;
Chris@42 193 io[0] = T2j + T2k;
Chris@42 194 io[WS(os, 8)] = T2j - T2k;
Chris@42 195 }
Chris@42 196 {
Chris@42 197 E T29, T1T, T27, T2d, T2a, T2b, T28, T24, T1Y, T23;
Chris@42 198 T29 = T1R - T1S;
Chris@42 199 T1T = T1R + T1S;
Chris@42 200 io[WS(os, 12)] = TK - Tv;
Chris@42 201 io[WS(os, 4)] = Tv + TK;
Chris@42 202 ro[WS(os, 4)] = T2f + T2i;
Chris@42 203 ro[WS(os, 12)] = T2f - T2i;
Chris@42 204 T27 = T25 - T26;
Chris@42 205 T2d = T26 + T25;
Chris@42 206 T2a = T1X - T1U;
Chris@42 207 T1Y = T1U + T1X;
Chris@42 208 T23 = T1Z - T22;
Chris@42 209 T2b = T1Z + T22;
Chris@42 210 T28 = T23 - T1Y;
Chris@42 211 T24 = T1Y + T23;
Chris@42 212 {
Chris@42 213 E T1I, TV, T1v, T1y, T1t, T1s, T1r, T1p, T1q, T1i;
Chris@42 214 {
Chris@42 215 E T1o, T2e, T2c, TU, T16, T1h;
Chris@42 216 T1I = TQ + TT;
Chris@42 217 TU = TQ - TT;
Chris@42 218 io[WS(os, 14)] = FNMS(KP707106781, T28, T27);
Chris@42 219 io[WS(os, 6)] = FMA(KP707106781, T28, T27);
Chris@42 220 ro[WS(os, 2)] = FMA(KP707106781, T24, T1T);
Chris@42 221 ro[WS(os, 10)] = FNMS(KP707106781, T24, T1T);
Chris@42 222 T2e = T2a + T2b;
Chris@42 223 T2c = T2a - T2b;
Chris@42 224 TV = FMA(KP707106781, TU, TN);
Chris@42 225 T1v = FNMS(KP707106781, TU, TN);
Chris@42 226 io[WS(os, 10)] = FNMS(KP707106781, T2e, T2d);
Chris@42 227 io[WS(os, 2)] = FMA(KP707106781, T2e, T2d);
Chris@42 228 ro[WS(os, 6)] = FMA(KP707106781, T2c, T29);
Chris@42 229 ro[WS(os, 14)] = FNMS(KP707106781, T2c, T29);
Chris@42 230 T1o = T1m - T1n;
Chris@42 231 T1y = T1n + T1m;
Chris@42 232 T1t = FNMS(KP414213562, T10, T15);
Chris@42 233 T16 = FMA(KP414213562, T15, T10);
Chris@42 234 T1h = FNMS(KP414213562, T1g, T1b);
Chris@42 235 T1s = FMA(KP414213562, T1b, T1g);
Chris@42 236 T1r = FMA(KP707106781, T1o, T1l);
Chris@42 237 T1p = FNMS(KP707106781, T1o, T1l);
Chris@42 238 T1q = T16 + T1h;
Chris@42 239 T1i = T16 - T1h;
Chris@42 240 }
Chris@42 241 {
Chris@42 242 E T1w, T1u, T1C, T1F;
Chris@42 243 io[WS(os, 15)] = FMA(KP923879532, T1q, T1p);
Chris@42 244 io[WS(os, 7)] = FNMS(KP923879532, T1q, T1p);
Chris@42 245 ro[WS(os, 3)] = FMA(KP923879532, T1i, TV);
Chris@42 246 ro[WS(os, 11)] = FNMS(KP923879532, T1i, TV);
Chris@42 247 T1w = T1t + T1s;
Chris@42 248 T1u = T1s - T1t;
Chris@42 249 T1z = FMA(KP707106781, T1y, T1x);
Chris@42 250 T1L = FNMS(KP707106781, T1y, T1x);
Chris@42 251 ro[WS(os, 15)] = FMA(KP923879532, T1w, T1v);
Chris@42 252 ro[WS(os, 7)] = FNMS(KP923879532, T1w, T1v);
Chris@42 253 io[WS(os, 3)] = FMA(KP923879532, T1u, T1r);
Chris@42 254 io[WS(os, 11)] = FNMS(KP923879532, T1u, T1r);
Chris@42 255 T1M = FNMS(KP414213562, T1A, T1B);
Chris@42 256 T1C = FMA(KP414213562, T1B, T1A);
Chris@42 257 T1F = FNMS(KP414213562, T1E, T1D);
Chris@42 258 T1N = FMA(KP414213562, T1D, T1E);
Chris@42 259 T1P = FMA(KP707106781, T1I, T1H);
Chris@42 260 T1J = FNMS(KP707106781, T1I, T1H);
Chris@42 261 T1K = T1F - T1C;
Chris@42 262 T1G = T1C + T1F;
Chris@42 263 }
Chris@42 264 }
Chris@42 265 }
Chris@42 266 }
Chris@42 267 }
Chris@42 268 io[WS(os, 5)] = FMA(KP923879532, T1K, T1J);
Chris@42 269 io[WS(os, 13)] = FNMS(KP923879532, T1K, T1J);
Chris@42 270 ro[WS(os, 1)] = FMA(KP923879532, T1G, T1z);
Chris@42 271 ro[WS(os, 9)] = FNMS(KP923879532, T1G, T1z);
Chris@42 272 T1O = T1M - T1N;
Chris@42 273 T1Q = T1M + T1N;
Chris@42 274 io[WS(os, 1)] = FMA(KP923879532, T1Q, T1P);
Chris@42 275 io[WS(os, 9)] = FNMS(KP923879532, T1Q, T1P);
Chris@42 276 ro[WS(os, 5)] = FMA(KP923879532, T1O, T1L);
Chris@42 277 ro[WS(os, 13)] = FNMS(KP923879532, T1O, T1L);
Chris@42 278 }
Chris@42 279 }
Chris@42 280 }
Chris@42 281
Chris@42 282 static const kdft_desc desc = { 16, "n1_16", {104, 0, 40, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 283
Chris@42 284 void X(codelet_n1_16) (planner *p) {
Chris@42 285 X(kdft_register) (p, n1_16, &desc);
Chris@42 286 }
Chris@42 287
Chris@42 288 #else /* HAVE_FMA */
Chris@42 289
Chris@42 290 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 16 -name n1_16 -include n.h */
Chris@42 291
Chris@42 292 /*
Chris@42 293 * This function contains 144 FP additions, 24 FP multiplications,
Chris@42 294 * (or, 136 additions, 16 multiplications, 8 fused multiply/add),
Chris@42 295 * 50 stack variables, 3 constants, and 64 memory accesses
Chris@42 296 */
Chris@42 297 #include "n.h"
Chris@42 298
Chris@42 299 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@42 300 {
Chris@42 301 DK(KP382683432, +0.382683432365089771728459984030398866761344562);
Chris@42 302 DK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@42 303 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 304 {
Chris@42 305 INT i;
Chris@42 306 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@42 307 E T7, T1R, T25, TC, TN, T1x, T1H, T1l, Tt, T22, T2h, T1b, T1g, T1E, T1Z;
Chris@42 308 E T1D, Te, T1S, T26, TJ, TQ, T1m, T1n, TT, Tm, T1X, T2g, T10, T15, T1B;
Chris@42 309 E T1U, T1A;
Chris@42 310 {
Chris@42 311 E T3, TL, Ty, T1k, T6, T1j, TB, TM;
Chris@42 312 {
Chris@42 313 E T1, T2, Tw, Tx;
Chris@42 314 T1 = ri[0];
Chris@42 315 T2 = ri[WS(is, 8)];
Chris@42 316 T3 = T1 + T2;
Chris@42 317 TL = T1 - T2;
Chris@42 318 Tw = ii[0];
Chris@42 319 Tx = ii[WS(is, 8)];
Chris@42 320 Ty = Tw + Tx;
Chris@42 321 T1k = Tw - Tx;
Chris@42 322 }
Chris@42 323 {
Chris@42 324 E T4, T5, Tz, TA;
Chris@42 325 T4 = ri[WS(is, 4)];
Chris@42 326 T5 = ri[WS(is, 12)];
Chris@42 327 T6 = T4 + T5;
Chris@42 328 T1j = T4 - T5;
Chris@42 329 Tz = ii[WS(is, 4)];
Chris@42 330 TA = ii[WS(is, 12)];
Chris@42 331 TB = Tz + TA;
Chris@42 332 TM = Tz - TA;
Chris@42 333 }
Chris@42 334 T7 = T3 + T6;
Chris@42 335 T1R = T3 - T6;
Chris@42 336 T25 = Ty - TB;
Chris@42 337 TC = Ty + TB;
Chris@42 338 TN = TL - TM;
Chris@42 339 T1x = TL + TM;
Chris@42 340 T1H = T1k - T1j;
Chris@42 341 T1l = T1j + T1k;
Chris@42 342 }
Chris@42 343 {
Chris@42 344 E Tp, T17, T1f, T20, Ts, T1c, T1a, T21;
Chris@42 345 {
Chris@42 346 E Tn, To, T1d, T1e;
Chris@42 347 Tn = ri[WS(is, 15)];
Chris@42 348 To = ri[WS(is, 7)];
Chris@42 349 Tp = Tn + To;
Chris@42 350 T17 = Tn - To;
Chris@42 351 T1d = ii[WS(is, 15)];
Chris@42 352 T1e = ii[WS(is, 7)];
Chris@42 353 T1f = T1d - T1e;
Chris@42 354 T20 = T1d + T1e;
Chris@42 355 }
Chris@42 356 {
Chris@42 357 E Tq, Tr, T18, T19;
Chris@42 358 Tq = ri[WS(is, 3)];
Chris@42 359 Tr = ri[WS(is, 11)];
Chris@42 360 Ts = Tq + Tr;
Chris@42 361 T1c = Tq - Tr;
Chris@42 362 T18 = ii[WS(is, 3)];
Chris@42 363 T19 = ii[WS(is, 11)];
Chris@42 364 T1a = T18 - T19;
Chris@42 365 T21 = T18 + T19;
Chris@42 366 }
Chris@42 367 Tt = Tp + Ts;
Chris@42 368 T22 = T20 - T21;
Chris@42 369 T2h = T20 + T21;
Chris@42 370 T1b = T17 - T1a;
Chris@42 371 T1g = T1c + T1f;
Chris@42 372 T1E = T1f - T1c;
Chris@42 373 T1Z = Tp - Ts;
Chris@42 374 T1D = T17 + T1a;
Chris@42 375 }
Chris@42 376 {
Chris@42 377 E Ta, TP, TF, TO, Td, TR, TI, TS;
Chris@42 378 {
Chris@42 379 E T8, T9, TD, TE;
Chris@42 380 T8 = ri[WS(is, 2)];
Chris@42 381 T9 = ri[WS(is, 10)];
Chris@42 382 Ta = T8 + T9;
Chris@42 383 TP = T8 - T9;
Chris@42 384 TD = ii[WS(is, 2)];
Chris@42 385 TE = ii[WS(is, 10)];
Chris@42 386 TF = TD + TE;
Chris@42 387 TO = TD - TE;
Chris@42 388 }
Chris@42 389 {
Chris@42 390 E Tb, Tc, TG, TH;
Chris@42 391 Tb = ri[WS(is, 14)];
Chris@42 392 Tc = ri[WS(is, 6)];
Chris@42 393 Td = Tb + Tc;
Chris@42 394 TR = Tb - Tc;
Chris@42 395 TG = ii[WS(is, 14)];
Chris@42 396 TH = ii[WS(is, 6)];
Chris@42 397 TI = TG + TH;
Chris@42 398 TS = TG - TH;
Chris@42 399 }
Chris@42 400 Te = Ta + Td;
Chris@42 401 T1S = TF - TI;
Chris@42 402 T26 = Td - Ta;
Chris@42 403 TJ = TF + TI;
Chris@42 404 TQ = TO - TP;
Chris@42 405 T1m = TR - TS;
Chris@42 406 T1n = TP + TO;
Chris@42 407 TT = TR + TS;
Chris@42 408 }
Chris@42 409 {
Chris@42 410 E Ti, T11, TZ, T1V, Tl, TW, T14, T1W;
Chris@42 411 {
Chris@42 412 E Tg, Th, TX, TY;
Chris@42 413 Tg = ri[WS(is, 1)];
Chris@42 414 Th = ri[WS(is, 9)];
Chris@42 415 Ti = Tg + Th;
Chris@42 416 T11 = Tg - Th;
Chris@42 417 TX = ii[WS(is, 1)];
Chris@42 418 TY = ii[WS(is, 9)];
Chris@42 419 TZ = TX - TY;
Chris@42 420 T1V = TX + TY;
Chris@42 421 }
Chris@42 422 {
Chris@42 423 E Tj, Tk, T12, T13;
Chris@42 424 Tj = ri[WS(is, 5)];
Chris@42 425 Tk = ri[WS(is, 13)];
Chris@42 426 Tl = Tj + Tk;
Chris@42 427 TW = Tj - Tk;
Chris@42 428 T12 = ii[WS(is, 5)];
Chris@42 429 T13 = ii[WS(is, 13)];
Chris@42 430 T14 = T12 - T13;
Chris@42 431 T1W = T12 + T13;
Chris@42 432 }
Chris@42 433 Tm = Ti + Tl;
Chris@42 434 T1X = T1V - T1W;
Chris@42 435 T2g = T1V + T1W;
Chris@42 436 T10 = TW + TZ;
Chris@42 437 T15 = T11 - T14;
Chris@42 438 T1B = T11 + T14;
Chris@42 439 T1U = Ti - Tl;
Chris@42 440 T1A = TZ - TW;
Chris@42 441 }
Chris@42 442 {
Chris@42 443 E Tf, Tu, T2j, T2k;
Chris@42 444 Tf = T7 + Te;
Chris@42 445 Tu = Tm + Tt;
Chris@42 446 ro[WS(os, 8)] = Tf - Tu;
Chris@42 447 ro[0] = Tf + Tu;
Chris@42 448 T2j = TC + TJ;
Chris@42 449 T2k = T2g + T2h;
Chris@42 450 io[WS(os, 8)] = T2j - T2k;
Chris@42 451 io[0] = T2j + T2k;
Chris@42 452 }
Chris@42 453 {
Chris@42 454 E Tv, TK, T2f, T2i;
Chris@42 455 Tv = Tt - Tm;
Chris@42 456 TK = TC - TJ;
Chris@42 457 io[WS(os, 4)] = Tv + TK;
Chris@42 458 io[WS(os, 12)] = TK - Tv;
Chris@42 459 T2f = T7 - Te;
Chris@42 460 T2i = T2g - T2h;
Chris@42 461 ro[WS(os, 12)] = T2f - T2i;
Chris@42 462 ro[WS(os, 4)] = T2f + T2i;
Chris@42 463 }
Chris@42 464 {
Chris@42 465 E T1T, T27, T24, T28, T1Y, T23;
Chris@42 466 T1T = T1R + T1S;
Chris@42 467 T27 = T25 - T26;
Chris@42 468 T1Y = T1U + T1X;
Chris@42 469 T23 = T1Z - T22;
Chris@42 470 T24 = KP707106781 * (T1Y + T23);
Chris@42 471 T28 = KP707106781 * (T23 - T1Y);
Chris@42 472 ro[WS(os, 10)] = T1T - T24;
Chris@42 473 io[WS(os, 6)] = T27 + T28;
Chris@42 474 ro[WS(os, 2)] = T1T + T24;
Chris@42 475 io[WS(os, 14)] = T27 - T28;
Chris@42 476 }
Chris@42 477 {
Chris@42 478 E T29, T2d, T2c, T2e, T2a, T2b;
Chris@42 479 T29 = T1R - T1S;
Chris@42 480 T2d = T26 + T25;
Chris@42 481 T2a = T1X - T1U;
Chris@42 482 T2b = T1Z + T22;
Chris@42 483 T2c = KP707106781 * (T2a - T2b);
Chris@42 484 T2e = KP707106781 * (T2a + T2b);
Chris@42 485 ro[WS(os, 14)] = T29 - T2c;
Chris@42 486 io[WS(os, 2)] = T2d + T2e;
Chris@42 487 ro[WS(os, 6)] = T29 + T2c;
Chris@42 488 io[WS(os, 10)] = T2d - T2e;
Chris@42 489 }
Chris@42 490 {
Chris@42 491 E TV, T1r, T1p, T1v, T1i, T1q, T1u, T1w, TU, T1o;
Chris@42 492 TU = KP707106781 * (TQ - TT);
Chris@42 493 TV = TN + TU;
Chris@42 494 T1r = TN - TU;
Chris@42 495 T1o = KP707106781 * (T1m - T1n);
Chris@42 496 T1p = T1l - T1o;
Chris@42 497 T1v = T1l + T1o;
Chris@42 498 {
Chris@42 499 E T16, T1h, T1s, T1t;
Chris@42 500 T16 = FMA(KP923879532, T10, KP382683432 * T15);
Chris@42 501 T1h = FNMS(KP923879532, T1g, KP382683432 * T1b);
Chris@42 502 T1i = T16 + T1h;
Chris@42 503 T1q = T1h - T16;
Chris@42 504 T1s = FNMS(KP923879532, T15, KP382683432 * T10);
Chris@42 505 T1t = FMA(KP382683432, T1g, KP923879532 * T1b);
Chris@42 506 T1u = T1s - T1t;
Chris@42 507 T1w = T1s + T1t;
Chris@42 508 }
Chris@42 509 ro[WS(os, 11)] = TV - T1i;
Chris@42 510 io[WS(os, 11)] = T1v - T1w;
Chris@42 511 ro[WS(os, 3)] = TV + T1i;
Chris@42 512 io[WS(os, 3)] = T1v + T1w;
Chris@42 513 io[WS(os, 15)] = T1p - T1q;
Chris@42 514 ro[WS(os, 15)] = T1r - T1u;
Chris@42 515 io[WS(os, 7)] = T1p + T1q;
Chris@42 516 ro[WS(os, 7)] = T1r + T1u;
Chris@42 517 }
Chris@42 518 {
Chris@42 519 E T1z, T1L, T1J, T1P, T1G, T1K, T1O, T1Q, T1y, T1I;
Chris@42 520 T1y = KP707106781 * (T1n + T1m);
Chris@42 521 T1z = T1x + T1y;
Chris@42 522 T1L = T1x - T1y;
Chris@42 523 T1I = KP707106781 * (TQ + TT);
Chris@42 524 T1J = T1H - T1I;
Chris@42 525 T1P = T1H + T1I;
Chris@42 526 {
Chris@42 527 E T1C, T1F, T1M, T1N;
Chris@42 528 T1C = FMA(KP382683432, T1A, KP923879532 * T1B);
Chris@42 529 T1F = FNMS(KP382683432, T1E, KP923879532 * T1D);
Chris@42 530 T1G = T1C + T1F;
Chris@42 531 T1K = T1F - T1C;
Chris@42 532 T1M = FNMS(KP382683432, T1B, KP923879532 * T1A);
Chris@42 533 T1N = FMA(KP923879532, T1E, KP382683432 * T1D);
Chris@42 534 T1O = T1M - T1N;
Chris@42 535 T1Q = T1M + T1N;
Chris@42 536 }
Chris@42 537 ro[WS(os, 9)] = T1z - T1G;
Chris@42 538 io[WS(os, 9)] = T1P - T1Q;
Chris@42 539 ro[WS(os, 1)] = T1z + T1G;
Chris@42 540 io[WS(os, 1)] = T1P + T1Q;
Chris@42 541 io[WS(os, 13)] = T1J - T1K;
Chris@42 542 ro[WS(os, 13)] = T1L - T1O;
Chris@42 543 io[WS(os, 5)] = T1J + T1K;
Chris@42 544 ro[WS(os, 5)] = T1L + T1O;
Chris@42 545 }
Chris@42 546 }
Chris@42 547 }
Chris@42 548 }
Chris@42 549
Chris@42 550 static const kdft_desc desc = { 16, "n1_16", {136, 16, 8, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 551
Chris@42 552 void X(codelet_n1_16) (planner *p) {
Chris@42 553 X(kdft_register) (p, n1_16, &desc);
Chris@42 554 }
Chris@42 555
Chris@42 556 #endif /* HAVE_FMA */