Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/scalar/codelets/n1_15.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 16:35:53 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 15 -name n1_15 -include n.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 156 FP additions, 84 FP multiplications,
Chris@42 32 * (or, 72 additions, 0 multiplications, 84 fused multiply/add),
Chris@42 33 * 75 stack variables, 6 constants, and 60 memory accesses
Chris@42 34 */
Chris@42 35 #include "n.h"
Chris@42 36
Chris@42 37 static void n1_15(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(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42 40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42 41 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@42 42 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42 43 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 44 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 45 {
Chris@42 46 INT i;
Chris@42 47 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(60, is), MAKE_VOLATILE_STRIDE(60, os)) {
Chris@42 48 E T1r, T1g, T14, T13;
Chris@42 49 {
Chris@42 50 E T5, T2l, Tx, TV, T1z, T1X, T2s, Tr, T24, TT, T2e, T2n, T1Z, T1Q, T1B;
Chris@42 51 E T11, T1H, TW, T2t, Tg, TX, T25, TI, T2h, T2m, T1Y, T1T, T1A;
Chris@42 52 {
Chris@42 53 E T1, T1v, T2, T3, Tu, Tv, TZ, T10;
Chris@42 54 T1 = ri[0];
Chris@42 55 T1v = ii[0];
Chris@42 56 T2 = ri[WS(is, 5)];
Chris@42 57 T3 = ri[WS(is, 10)];
Chris@42 58 Tu = ii[WS(is, 5)];
Chris@42 59 Tv = ii[WS(is, 10)];
Chris@42 60 {
Chris@42 61 E T1k, Tm, TM, TJ, Tl, T2c, T1j, T1m, TP, T1p, Tp, TQ;
Chris@42 62 {
Chris@42 63 E Th, T1h, TK, TL, Tk, Tn, To, T1i;
Chris@42 64 {
Chris@42 65 E Ti, Tj, T1y, T4;
Chris@42 66 Th = ri[WS(is, 6)];
Chris@42 67 T1y = T3 - T2;
Chris@42 68 T4 = T2 + T3;
Chris@42 69 {
Chris@42 70 E T1w, Tw, Tt, T1x;
Chris@42 71 T1w = Tu + Tv;
Chris@42 72 Tw = Tu - Tv;
Chris@42 73 Ti = ri[WS(is, 11)];
Chris@42 74 T5 = T1 + T4;
Chris@42 75 Tt = FNMS(KP500000000, T4, T1);
Chris@42 76 T2l = T1v + T1w;
Chris@42 77 T1x = FNMS(KP500000000, T1w, T1v);
Chris@42 78 Tx = FNMS(KP866025403, Tw, Tt);
Chris@42 79 TV = FMA(KP866025403, Tw, Tt);
Chris@42 80 T1z = FMA(KP866025403, T1y, T1x);
Chris@42 81 T1X = FNMS(KP866025403, T1y, T1x);
Chris@42 82 Tj = ri[WS(is, 1)];
Chris@42 83 }
Chris@42 84 T1h = ii[WS(is, 6)];
Chris@42 85 TK = ii[WS(is, 11)];
Chris@42 86 TL = ii[WS(is, 1)];
Chris@42 87 Tk = Ti + Tj;
Chris@42 88 T1k = Tj - Ti;
Chris@42 89 }
Chris@42 90 Tm = ri[WS(is, 9)];
Chris@42 91 TM = TK - TL;
Chris@42 92 T1i = TK + TL;
Chris@42 93 TJ = FNMS(KP500000000, Tk, Th);
Chris@42 94 Tl = Th + Tk;
Chris@42 95 Tn = ri[WS(is, 14)];
Chris@42 96 To = ri[WS(is, 4)];
Chris@42 97 T2c = T1h + T1i;
Chris@42 98 T1j = FNMS(KP500000000, T1i, T1h);
Chris@42 99 T1m = ii[WS(is, 9)];
Chris@42 100 TP = ii[WS(is, 14)];
Chris@42 101 T1p = To - Tn;
Chris@42 102 Tp = Tn + To;
Chris@42 103 TQ = ii[WS(is, 4)];
Chris@42 104 }
Chris@42 105 {
Chris@42 106 E TN, TS, T1o, T2d;
Chris@42 107 {
Chris@42 108 E TO, T1n, TR, Tq;
Chris@42 109 TN = FNMS(KP866025403, TM, TJ);
Chris@42 110 TZ = FMA(KP866025403, TM, TJ);
Chris@42 111 TO = FNMS(KP500000000, Tp, Tm);
Chris@42 112 Tq = Tm + Tp;
Chris@42 113 T1n = TP + TQ;
Chris@42 114 TR = TP - TQ;
Chris@42 115 T2s = Tl - Tq;
Chris@42 116 Tr = Tl + Tq;
Chris@42 117 T10 = FMA(KP866025403, TR, TO);
Chris@42 118 TS = FNMS(KP866025403, TR, TO);
Chris@42 119 T1o = FNMS(KP500000000, T1n, T1m);
Chris@42 120 T2d = T1m + T1n;
Chris@42 121 }
Chris@42 122 {
Chris@42 123 E T1O, T1l, T1P, T1q;
Chris@42 124 T1O = FNMS(KP866025403, T1k, T1j);
Chris@42 125 T1l = FMA(KP866025403, T1k, T1j);
Chris@42 126 T24 = TN - TS;
Chris@42 127 TT = TN + TS;
Chris@42 128 T1P = FNMS(KP866025403, T1p, T1o);
Chris@42 129 T1q = FMA(KP866025403, T1p, T1o);
Chris@42 130 T2e = T2c - T2d;
Chris@42 131 T2n = T2c + T2d;
Chris@42 132 T1Z = T1O + T1P;
Chris@42 133 T1Q = T1O - T1P;
Chris@42 134 T1r = T1l - T1q;
Chris@42 135 T1B = T1l + T1q;
Chris@42 136 }
Chris@42 137 }
Chris@42 138 }
Chris@42 139 {
Chris@42 140 E T19, Tb, TB, Ty, Ta, T2f, T18, T1b, TE, T1e, Te, TF;
Chris@42 141 {
Chris@42 142 E T6, T16, Tz, TA, T9, T7, T8, Tc, Td, T17;
Chris@42 143 T6 = ri[WS(is, 3)];
Chris@42 144 T7 = ri[WS(is, 8)];
Chris@42 145 T11 = TZ + T10;
Chris@42 146 T1H = TZ - T10;
Chris@42 147 T8 = ri[WS(is, 13)];
Chris@42 148 T16 = ii[WS(is, 3)];
Chris@42 149 Tz = ii[WS(is, 8)];
Chris@42 150 TA = ii[WS(is, 13)];
Chris@42 151 T9 = T7 + T8;
Chris@42 152 T19 = T8 - T7;
Chris@42 153 Tb = ri[WS(is, 12)];
Chris@42 154 TB = Tz - TA;
Chris@42 155 T17 = Tz + TA;
Chris@42 156 Ty = FNMS(KP500000000, T9, T6);
Chris@42 157 Ta = T6 + T9;
Chris@42 158 Tc = ri[WS(is, 2)];
Chris@42 159 Td = ri[WS(is, 7)];
Chris@42 160 T2f = T16 + T17;
Chris@42 161 T18 = FNMS(KP500000000, T17, T16);
Chris@42 162 T1b = ii[WS(is, 12)];
Chris@42 163 TE = ii[WS(is, 2)];
Chris@42 164 T1e = Td - Tc;
Chris@42 165 Te = Tc + Td;
Chris@42 166 TF = ii[WS(is, 7)];
Chris@42 167 }
Chris@42 168 {
Chris@42 169 E TC, TH, T1d, T2g;
Chris@42 170 {
Chris@42 171 E TD, T1c, TG, Tf;
Chris@42 172 TC = FNMS(KP866025403, TB, Ty);
Chris@42 173 TW = FMA(KP866025403, TB, Ty);
Chris@42 174 TD = FNMS(KP500000000, Te, Tb);
Chris@42 175 Tf = Tb + Te;
Chris@42 176 T1c = TE + TF;
Chris@42 177 TG = TE - TF;
Chris@42 178 T2t = Ta - Tf;
Chris@42 179 Tg = Ta + Tf;
Chris@42 180 TX = FMA(KP866025403, TG, TD);
Chris@42 181 TH = FNMS(KP866025403, TG, TD);
Chris@42 182 T1d = FNMS(KP500000000, T1c, T1b);
Chris@42 183 T2g = T1b + T1c;
Chris@42 184 }
Chris@42 185 {
Chris@42 186 E T1R, T1a, T1S, T1f;
Chris@42 187 T1R = FNMS(KP866025403, T19, T18);
Chris@42 188 T1a = FMA(KP866025403, T19, T18);
Chris@42 189 T25 = TC - TH;
Chris@42 190 TI = TC + TH;
Chris@42 191 T1S = FNMS(KP866025403, T1e, T1d);
Chris@42 192 T1f = FMA(KP866025403, T1e, T1d);
Chris@42 193 T2h = T2f - T2g;
Chris@42 194 T2m = T2f + T2g;
Chris@42 195 T1Y = T1R + T1S;
Chris@42 196 T1T = T1R - T1S;
Chris@42 197 T1g = T1a - T1f;
Chris@42 198 T1A = T1a + T1f;
Chris@42 199 }
Chris@42 200 }
Chris@42 201 }
Chris@42 202 }
Chris@42 203 {
Chris@42 204 E TY, T1G, T1M, T1L, T2a, T29, Ts, T22, T21, T20;
Chris@42 205 T2a = Tg - Tr;
Chris@42 206 Ts = Tg + Tr;
Chris@42 207 TY = TW + TX;
Chris@42 208 T1G = TW - TX;
Chris@42 209 T29 = FNMS(KP250000000, Ts, T5);
Chris@42 210 ro[0] = T5 + Ts;
Chris@42 211 {
Chris@42 212 E T2q, T2p, T2o, TU;
Chris@42 213 T2o = T2m + T2n;
Chris@42 214 T2q = T2m - T2n;
Chris@42 215 {
Chris@42 216 E T2k, T2i, T2b, T2j;
Chris@42 217 T2k = FMA(KP618033988, T2e, T2h);
Chris@42 218 T2i = FNMS(KP618033988, T2h, T2e);
Chris@42 219 T2b = FNMS(KP559016994, T2a, T29);
Chris@42 220 T2j = FMA(KP559016994, T2a, T29);
Chris@42 221 ro[WS(os, 3)] = FMA(KP951056516, T2i, T2b);
Chris@42 222 ro[WS(os, 12)] = FNMS(KP951056516, T2i, T2b);
Chris@42 223 ro[WS(os, 6)] = FMA(KP951056516, T2k, T2j);
Chris@42 224 ro[WS(os, 9)] = FNMS(KP951056516, T2k, T2j);
Chris@42 225 T2p = FNMS(KP250000000, T2o, T2l);
Chris@42 226 }
Chris@42 227 io[0] = T2l + T2o;
Chris@42 228 TU = TI + TT;
Chris@42 229 T1M = TI - TT;
Chris@42 230 {
Chris@42 231 E T2r, T2v, T2w, T2u;
Chris@42 232 T2r = FNMS(KP559016994, T2q, T2p);
Chris@42 233 T2v = FMA(KP559016994, T2q, T2p);
Chris@42 234 T2w = FMA(KP618033988, T2s, T2t);
Chris@42 235 T2u = FNMS(KP618033988, T2t, T2s);
Chris@42 236 io[WS(os, 9)] = FMA(KP951056516, T2w, T2v);
Chris@42 237 io[WS(os, 6)] = FNMS(KP951056516, T2w, T2v);
Chris@42 238 io[WS(os, 12)] = FMA(KP951056516, T2u, T2r);
Chris@42 239 io[WS(os, 3)] = FNMS(KP951056516, T2u, T2r);
Chris@42 240 T1L = FNMS(KP250000000, TU, Tx);
Chris@42 241 }
Chris@42 242 ro[WS(os, 5)] = Tx + TU;
Chris@42 243 }
Chris@42 244 T20 = T1Y + T1Z;
Chris@42 245 T22 = T1Y - T1Z;
Chris@42 246 {
Chris@42 247 E T1N, T1V, T1W, T1U;
Chris@42 248 T1N = FNMS(KP559016994, T1M, T1L);
Chris@42 249 T1V = FMA(KP559016994, T1M, T1L);
Chris@42 250 T1W = FMA(KP618033988, T1Q, T1T);
Chris@42 251 T1U = FNMS(KP618033988, T1T, T1Q);
Chris@42 252 ro[WS(os, 11)] = FMA(KP951056516, T1W, T1V);
Chris@42 253 ro[WS(os, 14)] = FNMS(KP951056516, T1W, T1V);
Chris@42 254 ro[WS(os, 8)] = FMA(KP951056516, T1U, T1N);
Chris@42 255 ro[WS(os, 2)] = FNMS(KP951056516, T1U, T1N);
Chris@42 256 T21 = FNMS(KP250000000, T20, T1X);
Chris@42 257 }
Chris@42 258 io[WS(os, 5)] = T1X + T20;
Chris@42 259 {
Chris@42 260 E T1E, T1D, T1C, T12;
Chris@42 261 T1C = T1A + T1B;
Chris@42 262 T1E = T1A - T1B;
Chris@42 263 {
Chris@42 264 E T23, T27, T28, T26;
Chris@42 265 T23 = FNMS(KP559016994, T22, T21);
Chris@42 266 T27 = FMA(KP559016994, T22, T21);
Chris@42 267 T28 = FMA(KP618033988, T24, T25);
Chris@42 268 T26 = FNMS(KP618033988, T25, T24);
Chris@42 269 io[WS(os, 14)] = FMA(KP951056516, T28, T27);
Chris@42 270 io[WS(os, 11)] = FNMS(KP951056516, T28, T27);
Chris@42 271 io[WS(os, 8)] = FNMS(KP951056516, T26, T23);
Chris@42 272 io[WS(os, 2)] = FMA(KP951056516, T26, T23);
Chris@42 273 T1D = FNMS(KP250000000, T1C, T1z);
Chris@42 274 }
Chris@42 275 io[WS(os, 10)] = T1z + T1C;
Chris@42 276 T12 = TY + T11;
Chris@42 277 T14 = TY - T11;
Chris@42 278 {
Chris@42 279 E T1F, T1J, T1K, T1I;
Chris@42 280 T1F = FMA(KP559016994, T1E, T1D);
Chris@42 281 T1J = FNMS(KP559016994, T1E, T1D);
Chris@42 282 T1K = FNMS(KP618033988, T1G, T1H);
Chris@42 283 T1I = FMA(KP618033988, T1H, T1G);
Chris@42 284 io[WS(os, 13)] = FNMS(KP951056516, T1K, T1J);
Chris@42 285 io[WS(os, 7)] = FMA(KP951056516, T1K, T1J);
Chris@42 286 io[WS(os, 4)] = FMA(KP951056516, T1I, T1F);
Chris@42 287 io[WS(os, 1)] = FNMS(KP951056516, T1I, T1F);
Chris@42 288 T13 = FNMS(KP250000000, T12, TV);
Chris@42 289 }
Chris@42 290 ro[WS(os, 10)] = TV + T12;
Chris@42 291 }
Chris@42 292 }
Chris@42 293 }
Chris@42 294 {
Chris@42 295 E T1t, T15, T1s, T1u;
Chris@42 296 T1t = FNMS(KP559016994, T14, T13);
Chris@42 297 T15 = FMA(KP559016994, T14, T13);
Chris@42 298 T1s = FMA(KP618033988, T1r, T1g);
Chris@42 299 T1u = FNMS(KP618033988, T1g, T1r);
Chris@42 300 ro[WS(os, 13)] = FMA(KP951056516, T1u, T1t);
Chris@42 301 ro[WS(os, 7)] = FNMS(KP951056516, T1u, T1t);
Chris@42 302 ro[WS(os, 1)] = FMA(KP951056516, T1s, T15);
Chris@42 303 ro[WS(os, 4)] = FNMS(KP951056516, T1s, T15);
Chris@42 304 }
Chris@42 305 }
Chris@42 306 }
Chris@42 307 }
Chris@42 308
Chris@42 309 static const kdft_desc desc = { 15, "n1_15", {72, 0, 84, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 310
Chris@42 311 void X(codelet_n1_15) (planner *p) {
Chris@42 312 X(kdft_register) (p, n1_15, &desc);
Chris@42 313 }
Chris@42 314
Chris@42 315 #else /* HAVE_FMA */
Chris@42 316
Chris@42 317 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 15 -name n1_15 -include n.h */
Chris@42 318
Chris@42 319 /*
Chris@42 320 * This function contains 156 FP additions, 56 FP multiplications,
Chris@42 321 * (or, 128 additions, 28 multiplications, 28 fused multiply/add),
Chris@42 322 * 69 stack variables, 6 constants, and 60 memory accesses
Chris@42 323 */
Chris@42 324 #include "n.h"
Chris@42 325
Chris@42 326 static void n1_15(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 327 {
Chris@42 328 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@42 329 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42 330 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42 331 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42 332 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 333 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 334 {
Chris@42 335 INT i;
Chris@42 336 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(60, is), MAKE_VOLATILE_STRIDE(60, os)) {
Chris@42 337 E T5, T2l, Tx, TV, T1C, T20, Tl, Tq, Tr, TN, TS, TT, T2c, T2d, T2n;
Chris@42 338 E T1O, T1P, T22, T1l, T1q, T1w, TZ, T10, T11, Ta, Tf, Tg, TC, TH, TI;
Chris@42 339 E T2f, T2g, T2m, T1R, T1S, T21, T1a, T1f, T1v, TW, TX, TY;
Chris@42 340 {
Chris@42 341 E T1, T1z, T4, T1y, Tw, T1A, Tt, T1B;
Chris@42 342 T1 = ri[0];
Chris@42 343 T1z = ii[0];
Chris@42 344 {
Chris@42 345 E T2, T3, Tu, Tv;
Chris@42 346 T2 = ri[WS(is, 5)];
Chris@42 347 T3 = ri[WS(is, 10)];
Chris@42 348 T4 = T2 + T3;
Chris@42 349 T1y = KP866025403 * (T3 - T2);
Chris@42 350 Tu = ii[WS(is, 5)];
Chris@42 351 Tv = ii[WS(is, 10)];
Chris@42 352 Tw = KP866025403 * (Tu - Tv);
Chris@42 353 T1A = Tu + Tv;
Chris@42 354 }
Chris@42 355 T5 = T1 + T4;
Chris@42 356 T2l = T1z + T1A;
Chris@42 357 Tt = FNMS(KP500000000, T4, T1);
Chris@42 358 Tx = Tt - Tw;
Chris@42 359 TV = Tt + Tw;
Chris@42 360 T1B = FNMS(KP500000000, T1A, T1z);
Chris@42 361 T1C = T1y + T1B;
Chris@42 362 T20 = T1B - T1y;
Chris@42 363 }
Chris@42 364 {
Chris@42 365 E Th, Tk, TJ, T1h, T1i, T1j, TM, T1k, Tm, Tp, TO, T1m, T1n, T1o, TR;
Chris@42 366 E T1p;
Chris@42 367 {
Chris@42 368 E Ti, Tj, TK, TL;
Chris@42 369 Th = ri[WS(is, 6)];
Chris@42 370 Ti = ri[WS(is, 11)];
Chris@42 371 Tj = ri[WS(is, 1)];
Chris@42 372 Tk = Ti + Tj;
Chris@42 373 TJ = FNMS(KP500000000, Tk, Th);
Chris@42 374 T1h = KP866025403 * (Tj - Ti);
Chris@42 375 T1i = ii[WS(is, 6)];
Chris@42 376 TK = ii[WS(is, 11)];
Chris@42 377 TL = ii[WS(is, 1)];
Chris@42 378 T1j = TK + TL;
Chris@42 379 TM = KP866025403 * (TK - TL);
Chris@42 380 T1k = FNMS(KP500000000, T1j, T1i);
Chris@42 381 }
Chris@42 382 {
Chris@42 383 E Tn, To, TP, TQ;
Chris@42 384 Tm = ri[WS(is, 9)];
Chris@42 385 Tn = ri[WS(is, 14)];
Chris@42 386 To = ri[WS(is, 4)];
Chris@42 387 Tp = Tn + To;
Chris@42 388 TO = FNMS(KP500000000, Tp, Tm);
Chris@42 389 T1m = KP866025403 * (To - Tn);
Chris@42 390 T1n = ii[WS(is, 9)];
Chris@42 391 TP = ii[WS(is, 14)];
Chris@42 392 TQ = ii[WS(is, 4)];
Chris@42 393 T1o = TP + TQ;
Chris@42 394 TR = KP866025403 * (TP - TQ);
Chris@42 395 T1p = FNMS(KP500000000, T1o, T1n);
Chris@42 396 }
Chris@42 397 Tl = Th + Tk;
Chris@42 398 Tq = Tm + Tp;
Chris@42 399 Tr = Tl + Tq;
Chris@42 400 TN = TJ - TM;
Chris@42 401 TS = TO - TR;
Chris@42 402 TT = TN + TS;
Chris@42 403 T2c = T1i + T1j;
Chris@42 404 T2d = T1n + T1o;
Chris@42 405 T2n = T2c + T2d;
Chris@42 406 T1O = T1k - T1h;
Chris@42 407 T1P = T1p - T1m;
Chris@42 408 T22 = T1O + T1P;
Chris@42 409 T1l = T1h + T1k;
Chris@42 410 T1q = T1m + T1p;
Chris@42 411 T1w = T1l + T1q;
Chris@42 412 TZ = TJ + TM;
Chris@42 413 T10 = TO + TR;
Chris@42 414 T11 = TZ + T10;
Chris@42 415 }
Chris@42 416 {
Chris@42 417 E T6, T9, Ty, T16, T17, T18, TB, T19, Tb, Te, TD, T1b, T1c, T1d, TG;
Chris@42 418 E T1e;
Chris@42 419 {
Chris@42 420 E T7, T8, Tz, TA;
Chris@42 421 T6 = ri[WS(is, 3)];
Chris@42 422 T7 = ri[WS(is, 8)];
Chris@42 423 T8 = ri[WS(is, 13)];
Chris@42 424 T9 = T7 + T8;
Chris@42 425 Ty = FNMS(KP500000000, T9, T6);
Chris@42 426 T16 = KP866025403 * (T8 - T7);
Chris@42 427 T17 = ii[WS(is, 3)];
Chris@42 428 Tz = ii[WS(is, 8)];
Chris@42 429 TA = ii[WS(is, 13)];
Chris@42 430 T18 = Tz + TA;
Chris@42 431 TB = KP866025403 * (Tz - TA);
Chris@42 432 T19 = FNMS(KP500000000, T18, T17);
Chris@42 433 }
Chris@42 434 {
Chris@42 435 E Tc, Td, TE, TF;
Chris@42 436 Tb = ri[WS(is, 12)];
Chris@42 437 Tc = ri[WS(is, 2)];
Chris@42 438 Td = ri[WS(is, 7)];
Chris@42 439 Te = Tc + Td;
Chris@42 440 TD = FNMS(KP500000000, Te, Tb);
Chris@42 441 T1b = KP866025403 * (Td - Tc);
Chris@42 442 T1c = ii[WS(is, 12)];
Chris@42 443 TE = ii[WS(is, 2)];
Chris@42 444 TF = ii[WS(is, 7)];
Chris@42 445 T1d = TE + TF;
Chris@42 446 TG = KP866025403 * (TE - TF);
Chris@42 447 T1e = FNMS(KP500000000, T1d, T1c);
Chris@42 448 }
Chris@42 449 Ta = T6 + T9;
Chris@42 450 Tf = Tb + Te;
Chris@42 451 Tg = Ta + Tf;
Chris@42 452 TC = Ty - TB;
Chris@42 453 TH = TD - TG;
Chris@42 454 TI = TC + TH;
Chris@42 455 T2f = T17 + T18;
Chris@42 456 T2g = T1c + T1d;
Chris@42 457 T2m = T2f + T2g;
Chris@42 458 T1R = T19 - T16;
Chris@42 459 T1S = T1e - T1b;
Chris@42 460 T21 = T1R + T1S;
Chris@42 461 T1a = T16 + T19;
Chris@42 462 T1f = T1b + T1e;
Chris@42 463 T1v = T1a + T1f;
Chris@42 464 TW = Ty + TB;
Chris@42 465 TX = TD + TG;
Chris@42 466 TY = TW + TX;
Chris@42 467 }
Chris@42 468 {
Chris@42 469 E T2a, Ts, T29, T2i, T2k, T2e, T2h, T2j, T2b;
Chris@42 470 T2a = KP559016994 * (Tg - Tr);
Chris@42 471 Ts = Tg + Tr;
Chris@42 472 T29 = FNMS(KP250000000, Ts, T5);
Chris@42 473 T2e = T2c - T2d;
Chris@42 474 T2h = T2f - T2g;
Chris@42 475 T2i = FNMS(KP587785252, T2h, KP951056516 * T2e);
Chris@42 476 T2k = FMA(KP951056516, T2h, KP587785252 * T2e);
Chris@42 477 ro[0] = T5 + Ts;
Chris@42 478 T2j = T2a + T29;
Chris@42 479 ro[WS(os, 9)] = T2j - T2k;
Chris@42 480 ro[WS(os, 6)] = T2j + T2k;
Chris@42 481 T2b = T29 - T2a;
Chris@42 482 ro[WS(os, 12)] = T2b - T2i;
Chris@42 483 ro[WS(os, 3)] = T2b + T2i;
Chris@42 484 }
Chris@42 485 {
Chris@42 486 E T2q, T2o, T2p, T2u, T2w, T2s, T2t, T2v, T2r;
Chris@42 487 T2q = KP559016994 * (T2m - T2n);
Chris@42 488 T2o = T2m + T2n;
Chris@42 489 T2p = FNMS(KP250000000, T2o, T2l);
Chris@42 490 T2s = Tl - Tq;
Chris@42 491 T2t = Ta - Tf;
Chris@42 492 T2u = FNMS(KP587785252, T2t, KP951056516 * T2s);
Chris@42 493 T2w = FMA(KP951056516, T2t, KP587785252 * T2s);
Chris@42 494 io[0] = T2l + T2o;
Chris@42 495 T2v = T2q + T2p;
Chris@42 496 io[WS(os, 6)] = T2v - T2w;
Chris@42 497 io[WS(os, 9)] = T2w + T2v;
Chris@42 498 T2r = T2p - T2q;
Chris@42 499 io[WS(os, 3)] = T2r - T2u;
Chris@42 500 io[WS(os, 12)] = T2u + T2r;
Chris@42 501 }
Chris@42 502 {
Chris@42 503 E T1M, TU, T1L, T1U, T1W, T1Q, T1T, T1V, T1N;
Chris@42 504 T1M = KP559016994 * (TI - TT);
Chris@42 505 TU = TI + TT;
Chris@42 506 T1L = FNMS(KP250000000, TU, Tx);
Chris@42 507 T1Q = T1O - T1P;
Chris@42 508 T1T = T1R - T1S;
Chris@42 509 T1U = FNMS(KP587785252, T1T, KP951056516 * T1Q);
Chris@42 510 T1W = FMA(KP951056516, T1T, KP587785252 * T1Q);
Chris@42 511 ro[WS(os, 5)] = Tx + TU;
Chris@42 512 T1V = T1M + T1L;
Chris@42 513 ro[WS(os, 14)] = T1V - T1W;
Chris@42 514 ro[WS(os, 11)] = T1V + T1W;
Chris@42 515 T1N = T1L - T1M;
Chris@42 516 ro[WS(os, 2)] = T1N - T1U;
Chris@42 517 ro[WS(os, 8)] = T1N + T1U;
Chris@42 518 }
Chris@42 519 {
Chris@42 520 E T25, T23, T24, T1Z, T28, T1X, T1Y, T27, T26;
Chris@42 521 T25 = KP559016994 * (T21 - T22);
Chris@42 522 T23 = T21 + T22;
Chris@42 523 T24 = FNMS(KP250000000, T23, T20);
Chris@42 524 T1X = TN - TS;
Chris@42 525 T1Y = TC - TH;
Chris@42 526 T1Z = FNMS(KP587785252, T1Y, KP951056516 * T1X);
Chris@42 527 T28 = FMA(KP951056516, T1Y, KP587785252 * T1X);
Chris@42 528 io[WS(os, 5)] = T20 + T23;
Chris@42 529 T27 = T25 + T24;
Chris@42 530 io[WS(os, 11)] = T27 - T28;
Chris@42 531 io[WS(os, 14)] = T28 + T27;
Chris@42 532 T26 = T24 - T25;
Chris@42 533 io[WS(os, 2)] = T1Z + T26;
Chris@42 534 io[WS(os, 8)] = T26 - T1Z;
Chris@42 535 }
Chris@42 536 {
Chris@42 537 E T1x, T1D, T1E, T1I, T1J, T1G, T1H, T1K, T1F;
Chris@42 538 T1x = KP559016994 * (T1v - T1w);
Chris@42 539 T1D = T1v + T1w;
Chris@42 540 T1E = FNMS(KP250000000, T1D, T1C);
Chris@42 541 T1G = TW - TX;
Chris@42 542 T1H = TZ - T10;
Chris@42 543 T1I = FMA(KP951056516, T1G, KP587785252 * T1H);
Chris@42 544 T1J = FNMS(KP587785252, T1G, KP951056516 * T1H);
Chris@42 545 io[WS(os, 10)] = T1C + T1D;
Chris@42 546 T1K = T1E - T1x;
Chris@42 547 io[WS(os, 7)] = T1J + T1K;
Chris@42 548 io[WS(os, 13)] = T1K - T1J;
Chris@42 549 T1F = T1x + T1E;
Chris@42 550 io[WS(os, 1)] = T1F - T1I;
Chris@42 551 io[WS(os, 4)] = T1I + T1F;
Chris@42 552 }
Chris@42 553 {
Chris@42 554 E T13, T12, T14, T1s, T1u, T1g, T1r, T1t, T15;
Chris@42 555 T13 = KP559016994 * (TY - T11);
Chris@42 556 T12 = TY + T11;
Chris@42 557 T14 = FNMS(KP250000000, T12, TV);
Chris@42 558 T1g = T1a - T1f;
Chris@42 559 T1r = T1l - T1q;
Chris@42 560 T1s = FMA(KP951056516, T1g, KP587785252 * T1r);
Chris@42 561 T1u = FNMS(KP587785252, T1g, KP951056516 * T1r);
Chris@42 562 ro[WS(os, 10)] = TV + T12;
Chris@42 563 T1t = T14 - T13;
Chris@42 564 ro[WS(os, 7)] = T1t - T1u;
Chris@42 565 ro[WS(os, 13)] = T1t + T1u;
Chris@42 566 T15 = T13 + T14;
Chris@42 567 ro[WS(os, 4)] = T15 - T1s;
Chris@42 568 ro[WS(os, 1)] = T15 + T1s;
Chris@42 569 }
Chris@42 570 }
Chris@42 571 }
Chris@42 572 }
Chris@42 573
Chris@42 574 static const kdft_desc desc = { 15, "n1_15", {128, 28, 28, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 575
Chris@42 576 void X(codelet_n1_15) (planner *p) {
Chris@42 577 X(kdft_register) (p, n1_15, &desc);
Chris@42 578 }
Chris@42 579
Chris@42 580 #endif /* HAVE_FMA */