Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/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 d0c2a83c1364
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 08:04:10 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 15 -name n1_15 -include dft/scalar/n.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 156 FP additions, 84 FP multiplications,
Chris@82 32 * (or, 72 additions, 0 multiplications, 84 fused multiply/add),
Chris@82 33 * 69 stack variables, 6 constants, and 60 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/scalar/n.h"
Chris@82 36
Chris@82 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@82 38 {
Chris@82 39 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 41 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82 42 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 43 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 44 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 45 {
Chris@82 46 INT i;
Chris@82 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@82 48 E T5, T2l, Tx, TV, T1z, T1X, Tl, Tq, Tr, TN, TS, TT, T2c, T2d, T2n;
Chris@82 49 E T1O, T1P, T1Z, T1l, T1q, T1B, TZ, T10, T11, Ta, Tf, Tg, TC, TH, TI;
Chris@82 50 E T2f, T2g, T2m, T1R, T1S, T1Y, T1a, T1f, T1A, TW, TX, TY;
Chris@82 51 {
Chris@82 52 E T1, T1v, T4, T1y, Tw, T1w, Tt, T1x;
Chris@82 53 T1 = ri[0];
Chris@82 54 T1v = ii[0];
Chris@82 55 {
Chris@82 56 E T2, T3, Tu, Tv;
Chris@82 57 T2 = ri[WS(is, 5)];
Chris@82 58 T3 = ri[WS(is, 10)];
Chris@82 59 T4 = T2 + T3;
Chris@82 60 T1y = T3 - T2;
Chris@82 61 Tu = ii[WS(is, 5)];
Chris@82 62 Tv = ii[WS(is, 10)];
Chris@82 63 Tw = Tu - Tv;
Chris@82 64 T1w = Tu + Tv;
Chris@82 65 }
Chris@82 66 T5 = T1 + T4;
Chris@82 67 T2l = T1v + T1w;
Chris@82 68 Tt = FNMS(KP500000000, T4, T1);
Chris@82 69 Tx = FNMS(KP866025403, Tw, Tt);
Chris@82 70 TV = FMA(KP866025403, Tw, Tt);
Chris@82 71 T1x = FNMS(KP500000000, T1w, T1v);
Chris@82 72 T1z = FMA(KP866025403, T1y, T1x);
Chris@82 73 T1X = FNMS(KP866025403, T1y, T1x);
Chris@82 74 }
Chris@82 75 {
Chris@82 76 E Th, Tk, TJ, T1k, T1h, T1i, TM, T1j, Tm, Tp, TO, T1p, T1m, T1n, TR;
Chris@82 77 E T1o;
Chris@82 78 {
Chris@82 79 E Ti, Tj, TK, TL;
Chris@82 80 Th = ri[WS(is, 6)];
Chris@82 81 Ti = ri[WS(is, 11)];
Chris@82 82 Tj = ri[WS(is, 1)];
Chris@82 83 Tk = Ti + Tj;
Chris@82 84 TJ = FNMS(KP500000000, Tk, Th);
Chris@82 85 T1k = Tj - Ti;
Chris@82 86 T1h = ii[WS(is, 6)];
Chris@82 87 TK = ii[WS(is, 11)];
Chris@82 88 TL = ii[WS(is, 1)];
Chris@82 89 T1i = TK + TL;
Chris@82 90 TM = TK - TL;
Chris@82 91 T1j = FNMS(KP500000000, T1i, T1h);
Chris@82 92 }
Chris@82 93 {
Chris@82 94 E Tn, To, TP, TQ;
Chris@82 95 Tm = ri[WS(is, 9)];
Chris@82 96 Tn = ri[WS(is, 14)];
Chris@82 97 To = ri[WS(is, 4)];
Chris@82 98 Tp = Tn + To;
Chris@82 99 TO = FNMS(KP500000000, Tp, Tm);
Chris@82 100 T1p = To - Tn;
Chris@82 101 T1m = ii[WS(is, 9)];
Chris@82 102 TP = ii[WS(is, 14)];
Chris@82 103 TQ = ii[WS(is, 4)];
Chris@82 104 T1n = TP + TQ;
Chris@82 105 TR = TP - TQ;
Chris@82 106 T1o = FNMS(KP500000000, T1n, T1m);
Chris@82 107 }
Chris@82 108 Tl = Th + Tk;
Chris@82 109 Tq = Tm + Tp;
Chris@82 110 Tr = Tl + Tq;
Chris@82 111 TN = FNMS(KP866025403, TM, TJ);
Chris@82 112 TS = FNMS(KP866025403, TR, TO);
Chris@82 113 TT = TN + TS;
Chris@82 114 T2c = T1h + T1i;
Chris@82 115 T2d = T1m + T1n;
Chris@82 116 T2n = T2c + T2d;
Chris@82 117 T1O = FNMS(KP866025403, T1k, T1j);
Chris@82 118 T1P = FNMS(KP866025403, T1p, T1o);
Chris@82 119 T1Z = T1O + T1P;
Chris@82 120 T1l = FMA(KP866025403, T1k, T1j);
Chris@82 121 T1q = FMA(KP866025403, T1p, T1o);
Chris@82 122 T1B = T1l + T1q;
Chris@82 123 TZ = FMA(KP866025403, TM, TJ);
Chris@82 124 T10 = FMA(KP866025403, TR, TO);
Chris@82 125 T11 = TZ + T10;
Chris@82 126 }
Chris@82 127 {
Chris@82 128 E T6, T9, Ty, T19, T16, T17, TB, T18, Tb, Te, TD, T1e, T1b, T1c, TG;
Chris@82 129 E T1d;
Chris@82 130 {
Chris@82 131 E T7, T8, Tz, TA;
Chris@82 132 T6 = ri[WS(is, 3)];
Chris@82 133 T7 = ri[WS(is, 8)];
Chris@82 134 T8 = ri[WS(is, 13)];
Chris@82 135 T9 = T7 + T8;
Chris@82 136 Ty = FNMS(KP500000000, T9, T6);
Chris@82 137 T19 = T8 - T7;
Chris@82 138 T16 = ii[WS(is, 3)];
Chris@82 139 Tz = ii[WS(is, 8)];
Chris@82 140 TA = ii[WS(is, 13)];
Chris@82 141 T17 = Tz + TA;
Chris@82 142 TB = Tz - TA;
Chris@82 143 T18 = FNMS(KP500000000, T17, T16);
Chris@82 144 }
Chris@82 145 {
Chris@82 146 E Tc, Td, TE, TF;
Chris@82 147 Tb = ri[WS(is, 12)];
Chris@82 148 Tc = ri[WS(is, 2)];
Chris@82 149 Td = ri[WS(is, 7)];
Chris@82 150 Te = Tc + Td;
Chris@82 151 TD = FNMS(KP500000000, Te, Tb);
Chris@82 152 T1e = Td - Tc;
Chris@82 153 T1b = ii[WS(is, 12)];
Chris@82 154 TE = ii[WS(is, 2)];
Chris@82 155 TF = ii[WS(is, 7)];
Chris@82 156 T1c = TE + TF;
Chris@82 157 TG = TE - TF;
Chris@82 158 T1d = FNMS(KP500000000, T1c, T1b);
Chris@82 159 }
Chris@82 160 Ta = T6 + T9;
Chris@82 161 Tf = Tb + Te;
Chris@82 162 Tg = Ta + Tf;
Chris@82 163 TC = FNMS(KP866025403, TB, Ty);
Chris@82 164 TH = FNMS(KP866025403, TG, TD);
Chris@82 165 TI = TC + TH;
Chris@82 166 T2f = T16 + T17;
Chris@82 167 T2g = T1b + T1c;
Chris@82 168 T2m = T2f + T2g;
Chris@82 169 T1R = FNMS(KP866025403, T19, T18);
Chris@82 170 T1S = FNMS(KP866025403, T1e, T1d);
Chris@82 171 T1Y = T1R + T1S;
Chris@82 172 T1a = FMA(KP866025403, T19, T18);
Chris@82 173 T1f = FMA(KP866025403, T1e, T1d);
Chris@82 174 T1A = T1a + T1f;
Chris@82 175 TW = FMA(KP866025403, TB, Ty);
Chris@82 176 TX = FMA(KP866025403, TG, TD);
Chris@82 177 TY = TW + TX;
Chris@82 178 }
Chris@82 179 {
Chris@82 180 E T2a, Ts, T29, T2i, T2k, T2e, T2h, T2j, T2b;
Chris@82 181 T2a = Tg - Tr;
Chris@82 182 Ts = Tg + Tr;
Chris@82 183 T29 = FNMS(KP250000000, Ts, T5);
Chris@82 184 T2e = T2c - T2d;
Chris@82 185 T2h = T2f - T2g;
Chris@82 186 T2i = FNMS(KP618033988, T2h, T2e);
Chris@82 187 T2k = FMA(KP618033988, T2e, T2h);
Chris@82 188 ro[0] = T5 + Ts;
Chris@82 189 T2j = FMA(KP559016994, T2a, T29);
Chris@82 190 ro[WS(os, 9)] = FNMS(KP951056516, T2k, T2j);
Chris@82 191 ro[WS(os, 6)] = FMA(KP951056516, T2k, T2j);
Chris@82 192 T2b = FNMS(KP559016994, T2a, T29);
Chris@82 193 ro[WS(os, 12)] = FNMS(KP951056516, T2i, T2b);
Chris@82 194 ro[WS(os, 3)] = FMA(KP951056516, T2i, T2b);
Chris@82 195 }
Chris@82 196 {
Chris@82 197 E T2q, T2o, T2p, T2u, T2w, T2s, T2t, T2v, T2r;
Chris@82 198 T2q = T2m - T2n;
Chris@82 199 T2o = T2m + T2n;
Chris@82 200 T2p = FNMS(KP250000000, T2o, T2l);
Chris@82 201 T2s = Tl - Tq;
Chris@82 202 T2t = Ta - Tf;
Chris@82 203 T2u = FNMS(KP618033988, T2t, T2s);
Chris@82 204 T2w = FMA(KP618033988, T2s, T2t);
Chris@82 205 io[0] = T2l + T2o;
Chris@82 206 T2v = FMA(KP559016994, T2q, T2p);
Chris@82 207 io[WS(os, 6)] = FNMS(KP951056516, T2w, T2v);
Chris@82 208 io[WS(os, 9)] = FMA(KP951056516, T2w, T2v);
Chris@82 209 T2r = FNMS(KP559016994, T2q, T2p);
Chris@82 210 io[WS(os, 3)] = FNMS(KP951056516, T2u, T2r);
Chris@82 211 io[WS(os, 12)] = FMA(KP951056516, T2u, T2r);
Chris@82 212 }
Chris@82 213 {
Chris@82 214 E T1M, TU, T1L, T1U, T1W, T1Q, T1T, T1V, T1N;
Chris@82 215 T1M = TI - TT;
Chris@82 216 TU = TI + TT;
Chris@82 217 T1L = FNMS(KP250000000, TU, Tx);
Chris@82 218 T1Q = T1O - T1P;
Chris@82 219 T1T = T1R - T1S;
Chris@82 220 T1U = FNMS(KP618033988, T1T, T1Q);
Chris@82 221 T1W = FMA(KP618033988, T1Q, T1T);
Chris@82 222 ro[WS(os, 5)] = Tx + TU;
Chris@82 223 T1V = FMA(KP559016994, T1M, T1L);
Chris@82 224 ro[WS(os, 14)] = FNMS(KP951056516, T1W, T1V);
Chris@82 225 ro[WS(os, 11)] = FMA(KP951056516, T1W, T1V);
Chris@82 226 T1N = FNMS(KP559016994, T1M, T1L);
Chris@82 227 ro[WS(os, 2)] = FNMS(KP951056516, T1U, T1N);
Chris@82 228 ro[WS(os, 8)] = FMA(KP951056516, T1U, T1N);
Chris@82 229 }
Chris@82 230 {
Chris@82 231 E T22, T20, T21, T26, T28, T24, T25, T27, T23;
Chris@82 232 T22 = T1Y - T1Z;
Chris@82 233 T20 = T1Y + T1Z;
Chris@82 234 T21 = FNMS(KP250000000, T20, T1X);
Chris@82 235 T24 = TN - TS;
Chris@82 236 T25 = TC - TH;
Chris@82 237 T26 = FNMS(KP618033988, T25, T24);
Chris@82 238 T28 = FMA(KP618033988, T24, T25);
Chris@82 239 io[WS(os, 5)] = T1X + T20;
Chris@82 240 T27 = FMA(KP559016994, T22, T21);
Chris@82 241 io[WS(os, 11)] = FNMS(KP951056516, T28, T27);
Chris@82 242 io[WS(os, 14)] = FMA(KP951056516, T28, T27);
Chris@82 243 T23 = FNMS(KP559016994, T22, T21);
Chris@82 244 io[WS(os, 2)] = FMA(KP951056516, T26, T23);
Chris@82 245 io[WS(os, 8)] = FNMS(KP951056516, T26, T23);
Chris@82 246 }
Chris@82 247 {
Chris@82 248 E T1E, T1C, T1D, T1I, T1K, T1G, T1H, T1J, T1F;
Chris@82 249 T1E = T1A - T1B;
Chris@82 250 T1C = T1A + T1B;
Chris@82 251 T1D = FNMS(KP250000000, T1C, T1z);
Chris@82 252 T1G = TW - TX;
Chris@82 253 T1H = TZ - T10;
Chris@82 254 T1I = FMA(KP618033988, T1H, T1G);
Chris@82 255 T1K = FNMS(KP618033988, T1G, T1H);
Chris@82 256 io[WS(os, 10)] = T1z + T1C;
Chris@82 257 T1J = FNMS(KP559016994, T1E, T1D);
Chris@82 258 io[WS(os, 7)] = FMA(KP951056516, T1K, T1J);
Chris@82 259 io[WS(os, 13)] = FNMS(KP951056516, T1K, T1J);
Chris@82 260 T1F = FMA(KP559016994, T1E, T1D);
Chris@82 261 io[WS(os, 1)] = FNMS(KP951056516, T1I, T1F);
Chris@82 262 io[WS(os, 4)] = FMA(KP951056516, T1I, T1F);
Chris@82 263 }
Chris@82 264 {
Chris@82 265 E T14, T12, T13, T1s, T1u, T1g, T1r, T1t, T15;
Chris@82 266 T14 = TY - T11;
Chris@82 267 T12 = TY + T11;
Chris@82 268 T13 = FNMS(KP250000000, T12, TV);
Chris@82 269 T1g = T1a - T1f;
Chris@82 270 T1r = T1l - T1q;
Chris@82 271 T1s = FMA(KP618033988, T1r, T1g);
Chris@82 272 T1u = FNMS(KP618033988, T1g, T1r);
Chris@82 273 ro[WS(os, 10)] = TV + T12;
Chris@82 274 T1t = FNMS(KP559016994, T14, T13);
Chris@82 275 ro[WS(os, 7)] = FNMS(KP951056516, T1u, T1t);
Chris@82 276 ro[WS(os, 13)] = FMA(KP951056516, T1u, T1t);
Chris@82 277 T15 = FMA(KP559016994, T14, T13);
Chris@82 278 ro[WS(os, 4)] = FNMS(KP951056516, T1s, T15);
Chris@82 279 ro[WS(os, 1)] = FMA(KP951056516, T1s, T15);
Chris@82 280 }
Chris@82 281 }
Chris@82 282 }
Chris@82 283 }
Chris@82 284
Chris@82 285 static const kdft_desc desc = { 15, "n1_15", {72, 0, 84, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 286
Chris@82 287 void X(codelet_n1_15) (planner *p) {
Chris@82 288 X(kdft_register) (p, n1_15, &desc);
Chris@82 289 }
Chris@82 290
Chris@82 291 #else
Chris@82 292
Chris@82 293 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 15 -name n1_15 -include dft/scalar/n.h */
Chris@82 294
Chris@82 295 /*
Chris@82 296 * This function contains 156 FP additions, 56 FP multiplications,
Chris@82 297 * (or, 128 additions, 28 multiplications, 28 fused multiply/add),
Chris@82 298 * 69 stack variables, 6 constants, and 60 memory accesses
Chris@82 299 */
Chris@82 300 #include "dft/scalar/n.h"
Chris@82 301
Chris@82 302 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@82 303 {
Chris@82 304 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@82 305 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 306 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 307 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 308 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 309 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
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(60, is), MAKE_VOLATILE_STRIDE(60, os)) {
Chris@82 313 E T5, T2l, Tx, TV, T1C, T20, Tl, Tq, Tr, TN, TS, TT, T2c, T2d, T2n;
Chris@82 314 E T1O, T1P, T22, T1l, T1q, T1w, TZ, T10, T11, Ta, Tf, Tg, TC, TH, TI;
Chris@82 315 E T2f, T2g, T2m, T1R, T1S, T21, T1a, T1f, T1v, TW, TX, TY;
Chris@82 316 {
Chris@82 317 E T1, T1z, T4, T1y, Tw, T1A, Tt, T1B;
Chris@82 318 T1 = ri[0];
Chris@82 319 T1z = ii[0];
Chris@82 320 {
Chris@82 321 E T2, T3, Tu, Tv;
Chris@82 322 T2 = ri[WS(is, 5)];
Chris@82 323 T3 = ri[WS(is, 10)];
Chris@82 324 T4 = T2 + T3;
Chris@82 325 T1y = KP866025403 * (T3 - T2);
Chris@82 326 Tu = ii[WS(is, 5)];
Chris@82 327 Tv = ii[WS(is, 10)];
Chris@82 328 Tw = KP866025403 * (Tu - Tv);
Chris@82 329 T1A = Tu + Tv;
Chris@82 330 }
Chris@82 331 T5 = T1 + T4;
Chris@82 332 T2l = T1z + T1A;
Chris@82 333 Tt = FNMS(KP500000000, T4, T1);
Chris@82 334 Tx = Tt - Tw;
Chris@82 335 TV = Tt + Tw;
Chris@82 336 T1B = FNMS(KP500000000, T1A, T1z);
Chris@82 337 T1C = T1y + T1B;
Chris@82 338 T20 = T1B - T1y;
Chris@82 339 }
Chris@82 340 {
Chris@82 341 E Th, Tk, TJ, T1h, T1i, T1j, TM, T1k, Tm, Tp, TO, T1m, T1n, T1o, TR;
Chris@82 342 E T1p;
Chris@82 343 {
Chris@82 344 E Ti, Tj, TK, TL;
Chris@82 345 Th = ri[WS(is, 6)];
Chris@82 346 Ti = ri[WS(is, 11)];
Chris@82 347 Tj = ri[WS(is, 1)];
Chris@82 348 Tk = Ti + Tj;
Chris@82 349 TJ = FNMS(KP500000000, Tk, Th);
Chris@82 350 T1h = KP866025403 * (Tj - Ti);
Chris@82 351 T1i = ii[WS(is, 6)];
Chris@82 352 TK = ii[WS(is, 11)];
Chris@82 353 TL = ii[WS(is, 1)];
Chris@82 354 T1j = TK + TL;
Chris@82 355 TM = KP866025403 * (TK - TL);
Chris@82 356 T1k = FNMS(KP500000000, T1j, T1i);
Chris@82 357 }
Chris@82 358 {
Chris@82 359 E Tn, To, TP, TQ;
Chris@82 360 Tm = ri[WS(is, 9)];
Chris@82 361 Tn = ri[WS(is, 14)];
Chris@82 362 To = ri[WS(is, 4)];
Chris@82 363 Tp = Tn + To;
Chris@82 364 TO = FNMS(KP500000000, Tp, Tm);
Chris@82 365 T1m = KP866025403 * (To - Tn);
Chris@82 366 T1n = ii[WS(is, 9)];
Chris@82 367 TP = ii[WS(is, 14)];
Chris@82 368 TQ = ii[WS(is, 4)];
Chris@82 369 T1o = TP + TQ;
Chris@82 370 TR = KP866025403 * (TP - TQ);
Chris@82 371 T1p = FNMS(KP500000000, T1o, T1n);
Chris@82 372 }
Chris@82 373 Tl = Th + Tk;
Chris@82 374 Tq = Tm + Tp;
Chris@82 375 Tr = Tl + Tq;
Chris@82 376 TN = TJ - TM;
Chris@82 377 TS = TO - TR;
Chris@82 378 TT = TN + TS;
Chris@82 379 T2c = T1i + T1j;
Chris@82 380 T2d = T1n + T1o;
Chris@82 381 T2n = T2c + T2d;
Chris@82 382 T1O = T1k - T1h;
Chris@82 383 T1P = T1p - T1m;
Chris@82 384 T22 = T1O + T1P;
Chris@82 385 T1l = T1h + T1k;
Chris@82 386 T1q = T1m + T1p;
Chris@82 387 T1w = T1l + T1q;
Chris@82 388 TZ = TJ + TM;
Chris@82 389 T10 = TO + TR;
Chris@82 390 T11 = TZ + T10;
Chris@82 391 }
Chris@82 392 {
Chris@82 393 E T6, T9, Ty, T16, T17, T18, TB, T19, Tb, Te, TD, T1b, T1c, T1d, TG;
Chris@82 394 E T1e;
Chris@82 395 {
Chris@82 396 E T7, T8, Tz, TA;
Chris@82 397 T6 = ri[WS(is, 3)];
Chris@82 398 T7 = ri[WS(is, 8)];
Chris@82 399 T8 = ri[WS(is, 13)];
Chris@82 400 T9 = T7 + T8;
Chris@82 401 Ty = FNMS(KP500000000, T9, T6);
Chris@82 402 T16 = KP866025403 * (T8 - T7);
Chris@82 403 T17 = ii[WS(is, 3)];
Chris@82 404 Tz = ii[WS(is, 8)];
Chris@82 405 TA = ii[WS(is, 13)];
Chris@82 406 T18 = Tz + TA;
Chris@82 407 TB = KP866025403 * (Tz - TA);
Chris@82 408 T19 = FNMS(KP500000000, T18, T17);
Chris@82 409 }
Chris@82 410 {
Chris@82 411 E Tc, Td, TE, TF;
Chris@82 412 Tb = ri[WS(is, 12)];
Chris@82 413 Tc = ri[WS(is, 2)];
Chris@82 414 Td = ri[WS(is, 7)];
Chris@82 415 Te = Tc + Td;
Chris@82 416 TD = FNMS(KP500000000, Te, Tb);
Chris@82 417 T1b = KP866025403 * (Td - Tc);
Chris@82 418 T1c = ii[WS(is, 12)];
Chris@82 419 TE = ii[WS(is, 2)];
Chris@82 420 TF = ii[WS(is, 7)];
Chris@82 421 T1d = TE + TF;
Chris@82 422 TG = KP866025403 * (TE - TF);
Chris@82 423 T1e = FNMS(KP500000000, T1d, T1c);
Chris@82 424 }
Chris@82 425 Ta = T6 + T9;
Chris@82 426 Tf = Tb + Te;
Chris@82 427 Tg = Ta + Tf;
Chris@82 428 TC = Ty - TB;
Chris@82 429 TH = TD - TG;
Chris@82 430 TI = TC + TH;
Chris@82 431 T2f = T17 + T18;
Chris@82 432 T2g = T1c + T1d;
Chris@82 433 T2m = T2f + T2g;
Chris@82 434 T1R = T19 - T16;
Chris@82 435 T1S = T1e - T1b;
Chris@82 436 T21 = T1R + T1S;
Chris@82 437 T1a = T16 + T19;
Chris@82 438 T1f = T1b + T1e;
Chris@82 439 T1v = T1a + T1f;
Chris@82 440 TW = Ty + TB;
Chris@82 441 TX = TD + TG;
Chris@82 442 TY = TW + TX;
Chris@82 443 }
Chris@82 444 {
Chris@82 445 E T2a, Ts, T29, T2i, T2k, T2e, T2h, T2j, T2b;
Chris@82 446 T2a = KP559016994 * (Tg - Tr);
Chris@82 447 Ts = Tg + Tr;
Chris@82 448 T29 = FNMS(KP250000000, Ts, T5);
Chris@82 449 T2e = T2c - T2d;
Chris@82 450 T2h = T2f - T2g;
Chris@82 451 T2i = FNMS(KP587785252, T2h, KP951056516 * T2e);
Chris@82 452 T2k = FMA(KP951056516, T2h, KP587785252 * T2e);
Chris@82 453 ro[0] = T5 + Ts;
Chris@82 454 T2j = T2a + T29;
Chris@82 455 ro[WS(os, 9)] = T2j - T2k;
Chris@82 456 ro[WS(os, 6)] = T2j + T2k;
Chris@82 457 T2b = T29 - T2a;
Chris@82 458 ro[WS(os, 12)] = T2b - T2i;
Chris@82 459 ro[WS(os, 3)] = T2b + T2i;
Chris@82 460 }
Chris@82 461 {
Chris@82 462 E T2q, T2o, T2p, T2u, T2w, T2s, T2t, T2v, T2r;
Chris@82 463 T2q = KP559016994 * (T2m - T2n);
Chris@82 464 T2o = T2m + T2n;
Chris@82 465 T2p = FNMS(KP250000000, T2o, T2l);
Chris@82 466 T2s = Tl - Tq;
Chris@82 467 T2t = Ta - Tf;
Chris@82 468 T2u = FNMS(KP587785252, T2t, KP951056516 * T2s);
Chris@82 469 T2w = FMA(KP951056516, T2t, KP587785252 * T2s);
Chris@82 470 io[0] = T2l + T2o;
Chris@82 471 T2v = T2q + T2p;
Chris@82 472 io[WS(os, 6)] = T2v - T2w;
Chris@82 473 io[WS(os, 9)] = T2w + T2v;
Chris@82 474 T2r = T2p - T2q;
Chris@82 475 io[WS(os, 3)] = T2r - T2u;
Chris@82 476 io[WS(os, 12)] = T2u + T2r;
Chris@82 477 }
Chris@82 478 {
Chris@82 479 E T1M, TU, T1L, T1U, T1W, T1Q, T1T, T1V, T1N;
Chris@82 480 T1M = KP559016994 * (TI - TT);
Chris@82 481 TU = TI + TT;
Chris@82 482 T1L = FNMS(KP250000000, TU, Tx);
Chris@82 483 T1Q = T1O - T1P;
Chris@82 484 T1T = T1R - T1S;
Chris@82 485 T1U = FNMS(KP587785252, T1T, KP951056516 * T1Q);
Chris@82 486 T1W = FMA(KP951056516, T1T, KP587785252 * T1Q);
Chris@82 487 ro[WS(os, 5)] = Tx + TU;
Chris@82 488 T1V = T1M + T1L;
Chris@82 489 ro[WS(os, 14)] = T1V - T1W;
Chris@82 490 ro[WS(os, 11)] = T1V + T1W;
Chris@82 491 T1N = T1L - T1M;
Chris@82 492 ro[WS(os, 2)] = T1N - T1U;
Chris@82 493 ro[WS(os, 8)] = T1N + T1U;
Chris@82 494 }
Chris@82 495 {
Chris@82 496 E T25, T23, T24, T1Z, T28, T1X, T1Y, T27, T26;
Chris@82 497 T25 = KP559016994 * (T21 - T22);
Chris@82 498 T23 = T21 + T22;
Chris@82 499 T24 = FNMS(KP250000000, T23, T20);
Chris@82 500 T1X = TN - TS;
Chris@82 501 T1Y = TC - TH;
Chris@82 502 T1Z = FNMS(KP587785252, T1Y, KP951056516 * T1X);
Chris@82 503 T28 = FMA(KP951056516, T1Y, KP587785252 * T1X);
Chris@82 504 io[WS(os, 5)] = T20 + T23;
Chris@82 505 T27 = T25 + T24;
Chris@82 506 io[WS(os, 11)] = T27 - T28;
Chris@82 507 io[WS(os, 14)] = T28 + T27;
Chris@82 508 T26 = T24 - T25;
Chris@82 509 io[WS(os, 2)] = T1Z + T26;
Chris@82 510 io[WS(os, 8)] = T26 - T1Z;
Chris@82 511 }
Chris@82 512 {
Chris@82 513 E T1x, T1D, T1E, T1I, T1J, T1G, T1H, T1K, T1F;
Chris@82 514 T1x = KP559016994 * (T1v - T1w);
Chris@82 515 T1D = T1v + T1w;
Chris@82 516 T1E = FNMS(KP250000000, T1D, T1C);
Chris@82 517 T1G = TW - TX;
Chris@82 518 T1H = TZ - T10;
Chris@82 519 T1I = FMA(KP951056516, T1G, KP587785252 * T1H);
Chris@82 520 T1J = FNMS(KP587785252, T1G, KP951056516 * T1H);
Chris@82 521 io[WS(os, 10)] = T1C + T1D;
Chris@82 522 T1K = T1E - T1x;
Chris@82 523 io[WS(os, 7)] = T1J + T1K;
Chris@82 524 io[WS(os, 13)] = T1K - T1J;
Chris@82 525 T1F = T1x + T1E;
Chris@82 526 io[WS(os, 1)] = T1F - T1I;
Chris@82 527 io[WS(os, 4)] = T1I + T1F;
Chris@82 528 }
Chris@82 529 {
Chris@82 530 E T13, T12, T14, T1s, T1u, T1g, T1r, T1t, T15;
Chris@82 531 T13 = KP559016994 * (TY - T11);
Chris@82 532 T12 = TY + T11;
Chris@82 533 T14 = FNMS(KP250000000, T12, TV);
Chris@82 534 T1g = T1a - T1f;
Chris@82 535 T1r = T1l - T1q;
Chris@82 536 T1s = FMA(KP951056516, T1g, KP587785252 * T1r);
Chris@82 537 T1u = FNMS(KP587785252, T1g, KP951056516 * T1r);
Chris@82 538 ro[WS(os, 10)] = TV + T12;
Chris@82 539 T1t = T14 - T13;
Chris@82 540 ro[WS(os, 7)] = T1t - T1u;
Chris@82 541 ro[WS(os, 13)] = T1t + T1u;
Chris@82 542 T15 = T13 + T14;
Chris@82 543 ro[WS(os, 4)] = T15 - T1s;
Chris@82 544 ro[WS(os, 1)] = T15 + T1s;
Chris@82 545 }
Chris@82 546 }
Chris@82 547 }
Chris@82 548 }
Chris@82 549
Chris@82 550 static const kdft_desc desc = { 15, "n1_15", {128, 28, 28, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 551
Chris@82 552 void X(codelet_n1_15) (planner *p) {
Chris@82 553 X(kdft_register) (p, n1_15, &desc);
Chris@82 554 }
Chris@82 555
Chris@82 556 #endif