Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cf/hc2cfdft_10.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 08:07:11 EDT 2018 */
Chris@82 23
Chris@82 24 #include "rdft/codelet-rdft.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_hc2cdft.native -fma -compact -variables 4 -pipeline-latency 4 -n 10 -dit -name hc2cfdft_10 -include rdft/scalar/hc2cf.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 122 FP additions, 92 FP multiplications,
Chris@82 32 * (or, 68 additions, 38 multiplications, 54 fused multiply/add),
Chris@82 33 * 81 stack variables, 5 constants, and 40 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/hc2cf.h"
Chris@82 36
Chris@82 37 static void hc2cfdft_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 38 {
Chris@82 39 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 41 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 42 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 43 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82 44 {
Chris@82 45 INT m;
Chris@82 46 for (m = mb, W = W + ((mb - 1) * 18); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 18, MAKE_VOLATILE_STRIDE(40, rs)) {
Chris@82 47 E T3, T1u, Td, T1w, T1S, T2f, T14, T1p, T1j, T1q, T1N, T2e, TQ, T2i, T1n;
Chris@82 48 E T1H, Tz, T2h, T1m, T1C;
Chris@82 49 {
Chris@82 50 E T1, T2, T1h, Tc, TW, T1c, T1d, T1b, T1f, T1g, T1Q, T7, TV, T1J, TS;
Chris@82 51 E TU, Ts, Tx, T19, T18, T1O, T15, T17, Tt, T1A, Ti, Tn, TE, TD, T1F;
Chris@82 52 E TA, TC, Tj, T1y, TJ, TO, T12, T11, T1L, TY, T10, TK, T1D;
Chris@82 53 {
Chris@82 54 E Ta, Tb, T1e, T5, T6, TT;
Chris@82 55 T1 = Ip[0];
Chris@82 56 T2 = Im[0];
Chris@82 57 T1h = T1 + T2;
Chris@82 58 Ta = Rp[WS(rs, 2)];
Chris@82 59 Tb = Rm[WS(rs, 2)];
Chris@82 60 Tc = Ta - Tb;
Chris@82 61 TW = Ta + Tb;
Chris@82 62 T1c = Rm[0];
Chris@82 63 T1d = Rp[0];
Chris@82 64 T1e = T1c - T1d;
Chris@82 65 T1b = W[0];
Chris@82 66 T1f = T1b * T1e;
Chris@82 67 T1g = W[1];
Chris@82 68 T1Q = T1g * T1e;
Chris@82 69 T5 = Ip[WS(rs, 2)];
Chris@82 70 T6 = Im[WS(rs, 2)];
Chris@82 71 TT = T5 - T6;
Chris@82 72 T7 = T5 + T6;
Chris@82 73 TV = W[7];
Chris@82 74 T1J = TV * TT;
Chris@82 75 TS = W[6];
Chris@82 76 TU = TS * TT;
Chris@82 77 {
Chris@82 78 E Tq, Tr, T16, Tv, Tw, Tp;
Chris@82 79 Tq = Rm[WS(rs, 3)];
Chris@82 80 Tr = Rp[WS(rs, 3)];
Chris@82 81 Ts = Tq - Tr;
Chris@82 82 Tv = Ip[WS(rs, 3)];
Chris@82 83 Tw = Im[WS(rs, 3)];
Chris@82 84 Tx = Tv + Tw;
Chris@82 85 T16 = Tv - Tw;
Chris@82 86 T19 = Tr + Tq;
Chris@82 87 T18 = W[11];
Chris@82 88 T1O = T18 * T16;
Chris@82 89 T15 = W[10];
Chris@82 90 T17 = T15 * T16;
Chris@82 91 Tp = W[12];
Chris@82 92 Tt = Tp * Ts;
Chris@82 93 T1A = Tp * Tx;
Chris@82 94 }
Chris@82 95 {
Chris@82 96 E Tg, Th, TB, Tl, Tm, Tf;
Chris@82 97 Tg = Ip[WS(rs, 1)];
Chris@82 98 Th = Im[WS(rs, 1)];
Chris@82 99 Ti = Tg - Th;
Chris@82 100 Tl = Rp[WS(rs, 1)];
Chris@82 101 Tm = Rm[WS(rs, 1)];
Chris@82 102 Tn = Tl + Tm;
Chris@82 103 TB = Tm - Tl;
Chris@82 104 TE = Tg + Th;
Chris@82 105 TD = W[5];
Chris@82 106 T1F = TD * TB;
Chris@82 107 TA = W[4];
Chris@82 108 TC = TA * TB;
Chris@82 109 Tf = W[2];
Chris@82 110 Tj = Tf * Ti;
Chris@82 111 T1y = Tf * Tn;
Chris@82 112 }
Chris@82 113 {
Chris@82 114 E TH, TI, TZ, TM, TN, TG;
Chris@82 115 TH = Ip[WS(rs, 4)];
Chris@82 116 TI = Im[WS(rs, 4)];
Chris@82 117 TJ = TH - TI;
Chris@82 118 TM = Rp[WS(rs, 4)];
Chris@82 119 TN = Rm[WS(rs, 4)];
Chris@82 120 TO = TM + TN;
Chris@82 121 TZ = TN - TM;
Chris@82 122 T12 = TH + TI;
Chris@82 123 T11 = W[17];
Chris@82 124 T1L = T11 * TZ;
Chris@82 125 TY = W[16];
Chris@82 126 T10 = TY * TZ;
Chris@82 127 TG = W[14];
Chris@82 128 TK = TG * TJ;
Chris@82 129 T1D = TG * TO;
Chris@82 130 }
Chris@82 131 }
Chris@82 132 {
Chris@82 133 E T1P, T1R, T1K, T1M;
Chris@82 134 T3 = T1 - T2;
Chris@82 135 T1u = T1d + T1c;
Chris@82 136 {
Chris@82 137 E T4, T8, T9, T1v;
Chris@82 138 T4 = W[9];
Chris@82 139 T8 = T4 * T7;
Chris@82 140 T9 = W[8];
Chris@82 141 T1v = T9 * T7;
Chris@82 142 Td = FMA(T9, Tc, T8);
Chris@82 143 T1w = FNMS(T4, Tc, T1v);
Chris@82 144 }
Chris@82 145 T1P = FMA(T15, T19, T1O);
Chris@82 146 T1R = FMA(T1b, T1h, T1Q);
Chris@82 147 T1S = T1P - T1R;
Chris@82 148 T2f = T1P + T1R;
Chris@82 149 {
Chris@82 150 E TX, T13, T1a, T1i;
Chris@82 151 TX = FNMS(TV, TW, TU);
Chris@82 152 T13 = FNMS(T11, T12, T10);
Chris@82 153 T14 = TX + T13;
Chris@82 154 T1p = T13 - TX;
Chris@82 155 T1a = FNMS(T18, T19, T17);
Chris@82 156 T1i = FNMS(T1g, T1h, T1f);
Chris@82 157 T1j = T1a + T1i;
Chris@82 158 T1q = T1i - T1a;
Chris@82 159 }
Chris@82 160 T1K = FMA(TS, TW, T1J);
Chris@82 161 T1M = FMA(TY, T12, T1L);
Chris@82 162 T1N = T1K - T1M;
Chris@82 163 T2e = T1K + T1M;
Chris@82 164 {
Chris@82 165 E TF, T1G, TP, T1E, TL;
Chris@82 166 TF = FNMS(TD, TE, TC);
Chris@82 167 T1G = FMA(TA, TE, T1F);
Chris@82 168 TL = W[15];
Chris@82 169 TP = FNMS(TL, TO, TK);
Chris@82 170 T1E = FMA(TL, TJ, T1D);
Chris@82 171 TQ = TF + TP;
Chris@82 172 T2i = T1G + T1E;
Chris@82 173 T1n = TF - TP;
Chris@82 174 T1H = T1E - T1G;
Chris@82 175 }
Chris@82 176 {
Chris@82 177 E To, T1z, Ty, T1B, Tk, Tu;
Chris@82 178 Tk = W[3];
Chris@82 179 To = FNMS(Tk, Tn, Tj);
Chris@82 180 T1z = FMA(Tk, Ti, T1y);
Chris@82 181 Tu = W[13];
Chris@82 182 Ty = FNMS(Tu, Tx, Tt);
Chris@82 183 T1B = FMA(Tu, Ts, T1A);
Chris@82 184 Tz = To + Ty;
Chris@82 185 T2h = T1z + T1B;
Chris@82 186 T1m = Ty - To;
Chris@82 187 T1C = T1z - T1B;
Chris@82 188 }
Chris@82 189 }
Chris@82 190 }
Chris@82 191 {
Chris@82 192 E T2k, T2m, Te, T1l, T2b, T2c, T2l, T2d;
Chris@82 193 {
Chris@82 194 E T2g, T2j, TR, T1k;
Chris@82 195 T2g = T2e - T2f;
Chris@82 196 T2j = T2h - T2i;
Chris@82 197 T2k = FNMS(KP618033988, T2j, T2g);
Chris@82 198 T2m = FMA(KP618033988, T2g, T2j);
Chris@82 199 Te = T3 - Td;
Chris@82 200 TR = Tz + TQ;
Chris@82 201 T1k = T14 + T1j;
Chris@82 202 T1l = TR + T1k;
Chris@82 203 T2b = FNMS(KP250000000, T1l, Te);
Chris@82 204 T2c = TR - T1k;
Chris@82 205 }
Chris@82 206 Ip[0] = KP500000000 * (Te + T1l);
Chris@82 207 T2l = FMA(KP559016994, T2c, T2b);
Chris@82 208 Ip[WS(rs, 4)] = KP500000000 * (FMA(KP951056516, T2m, T2l));
Chris@82 209 Im[WS(rs, 3)] = -(KP500000000 * (FNMS(KP951056516, T2m, T2l)));
Chris@82 210 T2d = FNMS(KP559016994, T2c, T2b);
Chris@82 211 Ip[WS(rs, 2)] = KP500000000 * (FMA(KP951056516, T2k, T2d));
Chris@82 212 Im[WS(rs, 1)] = -(KP500000000 * (FNMS(KP951056516, T2k, T2d)));
Chris@82 213 }
Chris@82 214 {
Chris@82 215 E T2w, T2y, T2n, T2q, T2r, T2s, T2x, T2t;
Chris@82 216 {
Chris@82 217 E T2u, T2v, T2o, T2p;
Chris@82 218 T2u = T14 - T1j;
Chris@82 219 T2v = Tz - TQ;
Chris@82 220 T2w = FNMS(KP618033988, T2v, T2u);
Chris@82 221 T2y = FMA(KP618033988, T2u, T2v);
Chris@82 222 T2n = T1u + T1w;
Chris@82 223 T2o = T2h + T2i;
Chris@82 224 T2p = T2e + T2f;
Chris@82 225 T2q = T2o + T2p;
Chris@82 226 T2r = FNMS(KP250000000, T2q, T2n);
Chris@82 227 T2s = T2o - T2p;
Chris@82 228 }
Chris@82 229 Rp[0] = KP500000000 * (T2n + T2q);
Chris@82 230 T2x = FMA(KP559016994, T2s, T2r);
Chris@82 231 Rp[WS(rs, 4)] = KP500000000 * (FNMS(KP951056516, T2y, T2x));
Chris@82 232 Rm[WS(rs, 3)] = KP500000000 * (FMA(KP951056516, T2y, T2x));
Chris@82 233 T2t = FNMS(KP559016994, T2s, T2r);
Chris@82 234 Rp[WS(rs, 2)] = KP500000000 * (FNMS(KP951056516, T2w, T2t));
Chris@82 235 Rm[WS(rs, 1)] = KP500000000 * (FMA(KP951056516, T2w, T2t));
Chris@82 236 }
Chris@82 237 {
Chris@82 238 E T28, T2a, T1t, T1s, T23, T24, T29, T25;
Chris@82 239 {
Chris@82 240 E T26, T27, T1o, T1r;
Chris@82 241 T26 = T1H - T1C;
Chris@82 242 T27 = T1S - T1N;
Chris@82 243 T28 = FMA(KP618033988, T27, T26);
Chris@82 244 T2a = FNMS(KP618033988, T26, T27);
Chris@82 245 T1t = Td + T3;
Chris@82 246 T1o = T1m + T1n;
Chris@82 247 T1r = T1p + T1q;
Chris@82 248 T1s = T1o + T1r;
Chris@82 249 T23 = FMA(KP250000000, T1s, T1t);
Chris@82 250 T24 = T1r - T1o;
Chris@82 251 }
Chris@82 252 Im[WS(rs, 4)] = KP500000000 * (T1s - T1t);
Chris@82 253 T29 = FNMS(KP559016994, T24, T23);
Chris@82 254 Ip[WS(rs, 3)] = KP500000000 * (FMA(KP951056516, T2a, T29));
Chris@82 255 Im[WS(rs, 2)] = -(KP500000000 * (FNMS(KP951056516, T2a, T29)));
Chris@82 256 T25 = FMA(KP559016994, T24, T23);
Chris@82 257 Ip[WS(rs, 1)] = KP500000000 * (FMA(KP951056516, T28, T25));
Chris@82 258 Im[0] = -(KP500000000 * (FNMS(KP951056516, T28, T25)));
Chris@82 259 }
Chris@82 260 {
Chris@82 261 E T20, T22, T1x, T1U, T1V, T1W, T21, T1X;
Chris@82 262 {
Chris@82 263 E T1Y, T1Z, T1I, T1T;
Chris@82 264 T1Y = T1n - T1m;
Chris@82 265 T1Z = T1q - T1p;
Chris@82 266 T20 = FMA(KP618033988, T1Z, T1Y);
Chris@82 267 T22 = FNMS(KP618033988, T1Y, T1Z);
Chris@82 268 T1x = T1u - T1w;
Chris@82 269 T1I = T1C + T1H;
Chris@82 270 T1T = T1N + T1S;
Chris@82 271 T1U = T1I + T1T;
Chris@82 272 T1V = FNMS(KP250000000, T1U, T1x);
Chris@82 273 T1W = T1I - T1T;
Chris@82 274 }
Chris@82 275 Rm[WS(rs, 4)] = KP500000000 * (T1x + T1U);
Chris@82 276 T21 = FNMS(KP559016994, T1W, T1V);
Chris@82 277 Rp[WS(rs, 3)] = KP500000000 * (FMA(KP951056516, T22, T21));
Chris@82 278 Rm[WS(rs, 2)] = KP500000000 * (FNMS(KP951056516, T22, T21));
Chris@82 279 T1X = FMA(KP559016994, T1W, T1V);
Chris@82 280 Rp[WS(rs, 1)] = KP500000000 * (FMA(KP951056516, T20, T1X));
Chris@82 281 Rm[0] = KP500000000 * (FNMS(KP951056516, T20, T1X));
Chris@82 282 }
Chris@82 283 }
Chris@82 284 }
Chris@82 285 }
Chris@82 286
Chris@82 287 static const tw_instr twinstr[] = {
Chris@82 288 {TW_FULL, 1, 10},
Chris@82 289 {TW_NEXT, 1, 0}
Chris@82 290 };
Chris@82 291
Chris@82 292 static const hc2c_desc desc = { 10, "hc2cfdft_10", twinstr, &GENUS, {68, 38, 54, 0} };
Chris@82 293
Chris@82 294 void X(codelet_hc2cfdft_10) (planner *p) {
Chris@82 295 X(khc2c_register) (p, hc2cfdft_10, &desc, HC2C_VIA_DFT);
Chris@82 296 }
Chris@82 297 #else
Chris@82 298
Chris@82 299 /* Generated by: ../../../genfft/gen_hc2cdft.native -compact -variables 4 -pipeline-latency 4 -n 10 -dit -name hc2cfdft_10 -include rdft/scalar/hc2cf.h */
Chris@82 300
Chris@82 301 /*
Chris@82 302 * This function contains 122 FP additions, 68 FP multiplications,
Chris@82 303 * (or, 92 additions, 38 multiplications, 30 fused multiply/add),
Chris@82 304 * 62 stack variables, 5 constants, and 40 memory accesses
Chris@82 305 */
Chris@82 306 #include "rdft/scalar/hc2cf.h"
Chris@82 307
Chris@82 308 static void hc2cfdft_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 309 {
Chris@82 310 DK(KP293892626, +0.293892626146236564584352977319536384298826219);
Chris@82 311 DK(KP475528258, +0.475528258147576786058219666689691071702849317);
Chris@82 312 DK(KP125000000, +0.125000000000000000000000000000000000000000000);
Chris@82 313 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 314 DK(KP279508497, +0.279508497187473712051146708591409529430077295);
Chris@82 315 {
Chris@82 316 INT m;
Chris@82 317 for (m = mb, W = W + ((mb - 1) * 18); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 18, MAKE_VOLATILE_STRIDE(40, rs)) {
Chris@82 318 E Tw, TL, TM, T1W, T1X, T27, T1Z, T20, T26, TX, T1a, T1b, T1d, T1e, T1f;
Chris@82 319 E T1q, T1t, T1u, T1x, T1A, T1B, T1g, T1h, T1i, Td, T25, T1k, T1F;
Chris@82 320 {
Chris@82 321 E T3, T1D, T19, T1z, T7, Tb, TR, T1v, Tm, T1o, TK, T1s, Tv, T1p, T12;
Chris@82 322 E T1y, TF, T1r, TW, T1w;
Chris@82 323 {
Chris@82 324 E T1, T2, T18, T14, T15, T16, T13, T17;
Chris@82 325 T1 = Ip[0];
Chris@82 326 T2 = Im[0];
Chris@82 327 T18 = T1 + T2;
Chris@82 328 T14 = Rm[0];
Chris@82 329 T15 = Rp[0];
Chris@82 330 T16 = T14 - T15;
Chris@82 331 T3 = T1 - T2;
Chris@82 332 T1D = T15 + T14;
Chris@82 333 T13 = W[0];
Chris@82 334 T17 = W[1];
Chris@82 335 T19 = FNMS(T17, T18, T13 * T16);
Chris@82 336 T1z = FMA(T17, T16, T13 * T18);
Chris@82 337 }
Chris@82 338 {
Chris@82 339 E T5, T6, TO, T9, Ta, TQ, TN, TP;
Chris@82 340 T5 = Ip[WS(rs, 2)];
Chris@82 341 T6 = Im[WS(rs, 2)];
Chris@82 342 TO = T5 - T6;
Chris@82 343 T9 = Rp[WS(rs, 2)];
Chris@82 344 Ta = Rm[WS(rs, 2)];
Chris@82 345 TQ = T9 + Ta;
Chris@82 346 T7 = T5 + T6;
Chris@82 347 Tb = T9 - Ta;
Chris@82 348 TN = W[6];
Chris@82 349 TP = W[7];
Chris@82 350 TR = FNMS(TP, TQ, TN * TO);
Chris@82 351 T1v = FMA(TP, TO, TN * TQ);
Chris@82 352 }
Chris@82 353 {
Chris@82 354 E Th, TJ, Tl, TH;
Chris@82 355 {
Chris@82 356 E Tf, Tg, Tj, Tk;
Chris@82 357 Tf = Ip[WS(rs, 1)];
Chris@82 358 Tg = Im[WS(rs, 1)];
Chris@82 359 Th = Tf - Tg;
Chris@82 360 TJ = Tf + Tg;
Chris@82 361 Tj = Rp[WS(rs, 1)];
Chris@82 362 Tk = Rm[WS(rs, 1)];
Chris@82 363 Tl = Tj + Tk;
Chris@82 364 TH = Tj - Tk;
Chris@82 365 }
Chris@82 366 {
Chris@82 367 E Te, Ti, TG, TI;
Chris@82 368 Te = W[2];
Chris@82 369 Ti = W[3];
Chris@82 370 Tm = FNMS(Ti, Tl, Te * Th);
Chris@82 371 T1o = FMA(Te, Tl, Ti * Th);
Chris@82 372 TG = W[4];
Chris@82 373 TI = W[5];
Chris@82 374 TK = FMA(TG, TH, TI * TJ);
Chris@82 375 T1s = FNMS(TI, TH, TG * TJ);
Chris@82 376 }
Chris@82 377 }
Chris@82 378 {
Chris@82 379 E Tq, TZ, Tu, T11;
Chris@82 380 {
Chris@82 381 E To, Tp, Ts, Tt;
Chris@82 382 To = Ip[WS(rs, 3)];
Chris@82 383 Tp = Im[WS(rs, 3)];
Chris@82 384 Tq = To + Tp;
Chris@82 385 TZ = To - Tp;
Chris@82 386 Ts = Rp[WS(rs, 3)];
Chris@82 387 Tt = Rm[WS(rs, 3)];
Chris@82 388 Tu = Ts - Tt;
Chris@82 389 T11 = Ts + Tt;
Chris@82 390 }
Chris@82 391 {
Chris@82 392 E Tn, Tr, TY, T10;
Chris@82 393 Tn = W[13];
Chris@82 394 Tr = W[12];
Chris@82 395 Tv = FMA(Tn, Tq, Tr * Tu);
Chris@82 396 T1p = FNMS(Tn, Tu, Tr * Tq);
Chris@82 397 TY = W[10];
Chris@82 398 T10 = W[11];
Chris@82 399 T12 = FNMS(T10, T11, TY * TZ);
Chris@82 400 T1y = FMA(T10, TZ, TY * T11);
Chris@82 401 }
Chris@82 402 }
Chris@82 403 {
Chris@82 404 E TA, TV, TE, TT;
Chris@82 405 {
Chris@82 406 E Ty, Tz, TC, TD;
Chris@82 407 Ty = Ip[WS(rs, 4)];
Chris@82 408 Tz = Im[WS(rs, 4)];
Chris@82 409 TA = Ty - Tz;
Chris@82 410 TV = Ty + Tz;
Chris@82 411 TC = Rp[WS(rs, 4)];
Chris@82 412 TD = Rm[WS(rs, 4)];
Chris@82 413 TE = TC + TD;
Chris@82 414 TT = TC - TD;
Chris@82 415 }
Chris@82 416 {
Chris@82 417 E Tx, TB, TS, TU;
Chris@82 418 Tx = W[14];
Chris@82 419 TB = W[15];
Chris@82 420 TF = FNMS(TB, TE, Tx * TA);
Chris@82 421 T1r = FMA(Tx, TE, TB * TA);
Chris@82 422 TS = W[16];
Chris@82 423 TU = W[17];
Chris@82 424 TW = FMA(TS, TT, TU * TV);
Chris@82 425 T1w = FNMS(TU, TT, TS * TV);
Chris@82 426 }
Chris@82 427 }
Chris@82 428 Tw = Tm - Tv;
Chris@82 429 TL = TF - TK;
Chris@82 430 TM = Tw + TL;
Chris@82 431 T1W = T1v + T1w;
Chris@82 432 T1X = T1y + T1z;
Chris@82 433 T27 = T1W + T1X;
Chris@82 434 T1Z = T1o + T1p;
Chris@82 435 T20 = T1s + T1r;
Chris@82 436 T26 = T1Z + T20;
Chris@82 437 TX = TR - TW;
Chris@82 438 T1a = T12 + T19;
Chris@82 439 T1b = TX + T1a;
Chris@82 440 T1d = T19 - T12;
Chris@82 441 T1e = TR + TW;
Chris@82 442 T1f = T1d - T1e;
Chris@82 443 T1q = T1o - T1p;
Chris@82 444 T1t = T1r - T1s;
Chris@82 445 T1u = T1q + T1t;
Chris@82 446 T1x = T1v - T1w;
Chris@82 447 T1A = T1y - T1z;
Chris@82 448 T1B = T1x + T1A;
Chris@82 449 T1g = Tm + Tv;
Chris@82 450 T1h = TK + TF;
Chris@82 451 T1i = T1g + T1h;
Chris@82 452 {
Chris@82 453 E Tc, T1E, T4, T8;
Chris@82 454 T4 = W[9];
Chris@82 455 T8 = W[8];
Chris@82 456 Tc = FMA(T4, T7, T8 * Tb);
Chris@82 457 T1E = FNMS(T4, Tb, T8 * T7);
Chris@82 458 Td = T3 - Tc;
Chris@82 459 T25 = T1D + T1E;
Chris@82 460 T1k = Tc + T3;
Chris@82 461 T1F = T1D - T1E;
Chris@82 462 }
Chris@82 463 }
Chris@82 464 {
Chris@82 465 E T1U, T1c, T1T, T22, T24, T1Y, T21, T23, T1V;
Chris@82 466 T1U = KP279508497 * (TM - T1b);
Chris@82 467 T1c = TM + T1b;
Chris@82 468 T1T = FNMS(KP125000000, T1c, KP500000000 * Td);
Chris@82 469 T1Y = T1W - T1X;
Chris@82 470 T21 = T1Z - T20;
Chris@82 471 T22 = FNMS(KP293892626, T21, KP475528258 * T1Y);
Chris@82 472 T24 = FMA(KP475528258, T21, KP293892626 * T1Y);
Chris@82 473 Ip[0] = KP500000000 * (Td + T1c);
Chris@82 474 T23 = T1U + T1T;
Chris@82 475 Ip[WS(rs, 4)] = T23 + T24;
Chris@82 476 Im[WS(rs, 3)] = T24 - T23;
Chris@82 477 T1V = T1T - T1U;
Chris@82 478 Ip[WS(rs, 2)] = T1V + T22;
Chris@82 479 Im[WS(rs, 1)] = T22 - T1V;
Chris@82 480 }
Chris@82 481 {
Chris@82 482 E T2a, T28, T29, T2e, T2g, T2c, T2d, T2f, T2b;
Chris@82 483 T2a = KP279508497 * (T26 - T27);
Chris@82 484 T28 = T26 + T27;
Chris@82 485 T29 = FNMS(KP125000000, T28, KP500000000 * T25);
Chris@82 486 T2c = TX - T1a;
Chris@82 487 T2d = Tw - TL;
Chris@82 488 T2e = FNMS(KP293892626, T2d, KP475528258 * T2c);
Chris@82 489 T2g = FMA(KP475528258, T2d, KP293892626 * T2c);
Chris@82 490 Rp[0] = KP500000000 * (T25 + T28);
Chris@82 491 T2f = T2a + T29;
Chris@82 492 Rp[WS(rs, 4)] = T2f - T2g;
Chris@82 493 Rm[WS(rs, 3)] = T2g + T2f;
Chris@82 494 T2b = T29 - T2a;
Chris@82 495 Rp[WS(rs, 2)] = T2b - T2e;
Chris@82 496 Rm[WS(rs, 1)] = T2e + T2b;
Chris@82 497 }
Chris@82 498 {
Chris@82 499 E T1M, T1j, T1L, T1Q, T1S, T1O, T1P, T1R, T1N;
Chris@82 500 T1M = KP279508497 * (T1i + T1f);
Chris@82 501 T1j = T1f - T1i;
Chris@82 502 T1L = FMA(KP500000000, T1k, KP125000000 * T1j);
Chris@82 503 T1O = T1A - T1x;
Chris@82 504 T1P = T1q - T1t;
Chris@82 505 T1Q = FNMS(KP475528258, T1P, KP293892626 * T1O);
Chris@82 506 T1S = FMA(KP293892626, T1P, KP475528258 * T1O);
Chris@82 507 Im[WS(rs, 4)] = KP500000000 * (T1j - T1k);
Chris@82 508 T1R = T1L - T1M;
Chris@82 509 Ip[WS(rs, 3)] = T1R + T1S;
Chris@82 510 Im[WS(rs, 2)] = T1S - T1R;
Chris@82 511 T1N = T1L + T1M;
Chris@82 512 Ip[WS(rs, 1)] = T1N + T1Q;
Chris@82 513 Im[0] = T1Q - T1N;
Chris@82 514 }
Chris@82 515 {
Chris@82 516 E T1C, T1G, T1H, T1n, T1J, T1l, T1m, T1K, T1I;
Chris@82 517 T1C = KP279508497 * (T1u - T1B);
Chris@82 518 T1G = T1u + T1B;
Chris@82 519 T1H = FNMS(KP125000000, T1G, KP500000000 * T1F);
Chris@82 520 T1l = T1g - T1h;
Chris@82 521 T1m = T1e + T1d;
Chris@82 522 T1n = FMA(KP475528258, T1l, KP293892626 * T1m);
Chris@82 523 T1J = FNMS(KP293892626, T1l, KP475528258 * T1m);
Chris@82 524 Rm[WS(rs, 4)] = KP500000000 * (T1F + T1G);
Chris@82 525 T1K = T1H - T1C;
Chris@82 526 Rp[WS(rs, 3)] = T1J + T1K;
Chris@82 527 Rm[WS(rs, 2)] = T1K - T1J;
Chris@82 528 T1I = T1C + T1H;
Chris@82 529 Rp[WS(rs, 1)] = T1n + T1I;
Chris@82 530 Rm[0] = T1I - T1n;
Chris@82 531 }
Chris@82 532 }
Chris@82 533 }
Chris@82 534 }
Chris@82 535
Chris@82 536 static const tw_instr twinstr[] = {
Chris@82 537 {TW_FULL, 1, 10},
Chris@82 538 {TW_NEXT, 1, 0}
Chris@82 539 };
Chris@82 540
Chris@82 541 static const hc2c_desc desc = { 10, "hc2cfdft_10", twinstr, &GENUS, {92, 38, 30, 0} };
Chris@82 542
Chris@82 543 void X(codelet_hc2cfdft_10) (planner *p) {
Chris@82 544 X(khc2c_register) (p, hc2cfdft_10, &desc, HC2C_VIA_DFT);
Chris@82 545 }
Chris@82 546 #endif