Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cf/hc2cf_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 08:06:56 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_hc2c.native -fma -compact -variables 4 -pipeline-latency 4 -n 10 -dit -name hc2cf_10 -include rdft/scalar/hc2cf.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 102 FP additions, 72 FP multiplications,
Chris@82 32 * (or, 48 additions, 18 multiplications, 54 fused multiply/add),
Chris@82 33 * 47 stack variables, 4 constants, and 40 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/hc2cf.h"
Chris@82 36
Chris@82 37 static void hc2cf_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(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82 42 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 43 {
Chris@82 44 INT m;
Chris@82 45 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 46 E T8, T26, T12, T1U, TM, TZ, T10, T1I, T1J, T24, T16, T17, T18, T1h, T1m;
Chris@82 47 E T1P, Tl, Ty, Tz, T1F, T1G, T23, T13, T14, T15, T1s, T1x, T1O;
Chris@82 48 {
Chris@82 49 E T1, T1T, T3, T6, T4, T1R, T2, T7, T1S, T5;
Chris@82 50 T1 = Rp[0];
Chris@82 51 T1T = Rm[0];
Chris@82 52 T3 = Ip[WS(rs, 2)];
Chris@82 53 T6 = Im[WS(rs, 2)];
Chris@82 54 T2 = W[8];
Chris@82 55 T4 = T2 * T3;
Chris@82 56 T1R = T2 * T6;
Chris@82 57 T5 = W[9];
Chris@82 58 T7 = FMA(T5, T6, T4);
Chris@82 59 T1S = FNMS(T5, T3, T1R);
Chris@82 60 T8 = T1 - T7;
Chris@82 61 T26 = T1T - T1S;
Chris@82 62 T12 = T1 + T7;
Chris@82 63 T1U = T1S + T1T;
Chris@82 64 }
Chris@82 65 {
Chris@82 66 E TF, T1e, TY, T1l, TL, T1g, TS, T1j;
Chris@82 67 {
Chris@82 68 E TB, TE, TC, T1d, TA, TD;
Chris@82 69 TB = Rp[WS(rs, 2)];
Chris@82 70 TE = Rm[WS(rs, 2)];
Chris@82 71 TA = W[6];
Chris@82 72 TC = TA * TB;
Chris@82 73 T1d = TA * TE;
Chris@82 74 TD = W[7];
Chris@82 75 TF = FMA(TD, TE, TC);
Chris@82 76 T1e = FNMS(TD, TB, T1d);
Chris@82 77 }
Chris@82 78 {
Chris@82 79 E TU, TX, TV, T1k, TT, TW;
Chris@82 80 TU = Ip[0];
Chris@82 81 TX = Im[0];
Chris@82 82 TT = W[0];
Chris@82 83 TV = TT * TU;
Chris@82 84 T1k = TT * TX;
Chris@82 85 TW = W[1];
Chris@82 86 TY = FMA(TW, TX, TV);
Chris@82 87 T1l = FNMS(TW, TU, T1k);
Chris@82 88 }
Chris@82 89 {
Chris@82 90 E TH, TK, TI, T1f, TG, TJ;
Chris@82 91 TH = Ip[WS(rs, 4)];
Chris@82 92 TK = Im[WS(rs, 4)];
Chris@82 93 TG = W[16];
Chris@82 94 TI = TG * TH;
Chris@82 95 T1f = TG * TK;
Chris@82 96 TJ = W[17];
Chris@82 97 TL = FMA(TJ, TK, TI);
Chris@82 98 T1g = FNMS(TJ, TH, T1f);
Chris@82 99 }
Chris@82 100 {
Chris@82 101 E TO, TR, TP, T1i, TN, TQ;
Chris@82 102 TO = Rp[WS(rs, 3)];
Chris@82 103 TR = Rm[WS(rs, 3)];
Chris@82 104 TN = W[10];
Chris@82 105 TP = TN * TO;
Chris@82 106 T1i = TN * TR;
Chris@82 107 TQ = W[11];
Chris@82 108 TS = FMA(TQ, TR, TP);
Chris@82 109 T1j = FNMS(TQ, TO, T1i);
Chris@82 110 }
Chris@82 111 TM = TF - TL;
Chris@82 112 TZ = TS - TY;
Chris@82 113 T10 = TM + TZ;
Chris@82 114 T1I = T1l - T1j;
Chris@82 115 T1J = T1g - T1e;
Chris@82 116 T24 = T1J + T1I;
Chris@82 117 T16 = TF + TL;
Chris@82 118 T17 = TS + TY;
Chris@82 119 T18 = T16 + T17;
Chris@82 120 T1h = T1e + T1g;
Chris@82 121 T1m = T1j + T1l;
Chris@82 122 T1P = T1h + T1m;
Chris@82 123 }
Chris@82 124 {
Chris@82 125 E Te, T1p, Tx, T1w, Tk, T1r, Tr, T1u;
Chris@82 126 {
Chris@82 127 E Ta, Td, Tb, T1o, T9, Tc;
Chris@82 128 Ta = Rp[WS(rs, 1)];
Chris@82 129 Td = Rm[WS(rs, 1)];
Chris@82 130 T9 = W[2];
Chris@82 131 Tb = T9 * Ta;
Chris@82 132 T1o = T9 * Td;
Chris@82 133 Tc = W[3];
Chris@82 134 Te = FMA(Tc, Td, Tb);
Chris@82 135 T1p = FNMS(Tc, Ta, T1o);
Chris@82 136 }
Chris@82 137 {
Chris@82 138 E Tt, Tw, Tu, T1v, Ts, Tv;
Chris@82 139 Tt = Ip[WS(rs, 1)];
Chris@82 140 Tw = Im[WS(rs, 1)];
Chris@82 141 Ts = W[4];
Chris@82 142 Tu = Ts * Tt;
Chris@82 143 T1v = Ts * Tw;
Chris@82 144 Tv = W[5];
Chris@82 145 Tx = FMA(Tv, Tw, Tu);
Chris@82 146 T1w = FNMS(Tv, Tt, T1v);
Chris@82 147 }
Chris@82 148 {
Chris@82 149 E Tg, Tj, Th, T1q, Tf, Ti;
Chris@82 150 Tg = Ip[WS(rs, 3)];
Chris@82 151 Tj = Im[WS(rs, 3)];
Chris@82 152 Tf = W[12];
Chris@82 153 Th = Tf * Tg;
Chris@82 154 T1q = Tf * Tj;
Chris@82 155 Ti = W[13];
Chris@82 156 Tk = FMA(Ti, Tj, Th);
Chris@82 157 T1r = FNMS(Ti, Tg, T1q);
Chris@82 158 }
Chris@82 159 {
Chris@82 160 E Tn, Tq, To, T1t, Tm, Tp;
Chris@82 161 Tn = Rp[WS(rs, 4)];
Chris@82 162 Tq = Rm[WS(rs, 4)];
Chris@82 163 Tm = W[14];
Chris@82 164 To = Tm * Tn;
Chris@82 165 T1t = Tm * Tq;
Chris@82 166 Tp = W[15];
Chris@82 167 Tr = FMA(Tp, Tq, To);
Chris@82 168 T1u = FNMS(Tp, Tn, T1t);
Chris@82 169 }
Chris@82 170 Tl = Te - Tk;
Chris@82 171 Ty = Tr - Tx;
Chris@82 172 Tz = Tl + Ty;
Chris@82 173 T1F = T1w - T1u;
Chris@82 174 T1G = T1r - T1p;
Chris@82 175 T23 = T1G + T1F;
Chris@82 176 T13 = Te + Tk;
Chris@82 177 T14 = Tr + Tx;
Chris@82 178 T15 = T13 + T14;
Chris@82 179 T1s = T1p + T1r;
Chris@82 180 T1x = T1u + T1w;
Chris@82 181 T1O = T1s + T1x;
Chris@82 182 }
Chris@82 183 {
Chris@82 184 E T1D, T11, T1C, T1L, T1N, T1H, T1K, T1M, T1E;
Chris@82 185 T1D = Tz - T10;
Chris@82 186 T11 = Tz + T10;
Chris@82 187 T1C = FNMS(KP250000000, T11, T8);
Chris@82 188 T1H = T1F - T1G;
Chris@82 189 T1K = T1I - T1J;
Chris@82 190 T1L = FMA(KP618033988, T1K, T1H);
Chris@82 191 T1N = FNMS(KP618033988, T1H, T1K);
Chris@82 192 Rm[WS(rs, 4)] = T8 + T11;
Chris@82 193 T1M = FNMS(KP559016994, T1D, T1C);
Chris@82 194 Rm[WS(rs, 2)] = FNMS(KP951056516, T1N, T1M);
Chris@82 195 Rp[WS(rs, 3)] = FMA(KP951056516, T1N, T1M);
Chris@82 196 T1E = FMA(KP559016994, T1D, T1C);
Chris@82 197 Rm[0] = FNMS(KP951056516, T1L, T1E);
Chris@82 198 Rp[WS(rs, 1)] = FMA(KP951056516, T1L, T1E);
Chris@82 199 }
Chris@82 200 {
Chris@82 201 E T28, T25, T27, T2c, T2e, T2a, T2b, T2d, T29;
Chris@82 202 T28 = T24 - T23;
Chris@82 203 T25 = T23 + T24;
Chris@82 204 T27 = FMA(KP250000000, T25, T26);
Chris@82 205 T2a = Ty - Tl;
Chris@82 206 T2b = TZ - TM;
Chris@82 207 T2c = FMA(KP618033988, T2b, T2a);
Chris@82 208 T2e = FNMS(KP618033988, T2a, T2b);
Chris@82 209 Im[WS(rs, 4)] = T25 - T26;
Chris@82 210 T2d = FNMS(KP559016994, T28, T27);
Chris@82 211 Im[WS(rs, 2)] = FMS(KP951056516, T2e, T2d);
Chris@82 212 Ip[WS(rs, 3)] = FMA(KP951056516, T2e, T2d);
Chris@82 213 T29 = FMA(KP559016994, T28, T27);
Chris@82 214 Im[0] = FMS(KP951056516, T2c, T29);
Chris@82 215 Ip[WS(rs, 1)] = FMA(KP951056516, T2c, T29);
Chris@82 216 }
Chris@82 217 {
Chris@82 218 E T1b, T19, T1a, T1z, T1B, T1n, T1y, T1A, T1c;
Chris@82 219 T1b = T15 - T18;
Chris@82 220 T19 = T15 + T18;
Chris@82 221 T1a = FNMS(KP250000000, T19, T12);
Chris@82 222 T1n = T1h - T1m;
Chris@82 223 T1y = T1s - T1x;
Chris@82 224 T1z = FNMS(KP618033988, T1y, T1n);
Chris@82 225 T1B = FMA(KP618033988, T1n, T1y);
Chris@82 226 Rp[0] = T12 + T19;
Chris@82 227 T1A = FMA(KP559016994, T1b, T1a);
Chris@82 228 Rp[WS(rs, 4)] = FNMS(KP951056516, T1B, T1A);
Chris@82 229 Rm[WS(rs, 3)] = FMA(KP951056516, T1B, T1A);
Chris@82 230 T1c = FNMS(KP559016994, T1b, T1a);
Chris@82 231 Rp[WS(rs, 2)] = FNMS(KP951056516, T1z, T1c);
Chris@82 232 Rm[WS(rs, 1)] = FMA(KP951056516, T1z, T1c);
Chris@82 233 }
Chris@82 234 {
Chris@82 235 E T1W, T1Q, T1V, T20, T22, T1Y, T1Z, T21, T1X;
Chris@82 236 T1W = T1O - T1P;
Chris@82 237 T1Q = T1O + T1P;
Chris@82 238 T1V = FNMS(KP250000000, T1Q, T1U);
Chris@82 239 T1Y = T16 - T17;
Chris@82 240 T1Z = T13 - T14;
Chris@82 241 T20 = FNMS(KP618033988, T1Z, T1Y);
Chris@82 242 T22 = FMA(KP618033988, T1Y, T1Z);
Chris@82 243 Ip[0] = T1Q + T1U;
Chris@82 244 T21 = FMA(KP559016994, T1W, T1V);
Chris@82 245 Im[WS(rs, 3)] = FMS(KP951056516, T22, T21);
Chris@82 246 Ip[WS(rs, 4)] = FMA(KP951056516, T22, T21);
Chris@82 247 T1X = FNMS(KP559016994, T1W, T1V);
Chris@82 248 Im[WS(rs, 1)] = FMS(KP951056516, T20, T1X);
Chris@82 249 Ip[WS(rs, 2)] = FMA(KP951056516, T20, T1X);
Chris@82 250 }
Chris@82 251 }
Chris@82 252 }
Chris@82 253 }
Chris@82 254
Chris@82 255 static const tw_instr twinstr[] = {
Chris@82 256 {TW_FULL, 1, 10},
Chris@82 257 {TW_NEXT, 1, 0}
Chris@82 258 };
Chris@82 259
Chris@82 260 static const hc2c_desc desc = { 10, "hc2cf_10", twinstr, &GENUS, {48, 18, 54, 0} };
Chris@82 261
Chris@82 262 void X(codelet_hc2cf_10) (planner *p) {
Chris@82 263 X(khc2c_register) (p, hc2cf_10, &desc, HC2C_VIA_RDFT);
Chris@82 264 }
Chris@82 265 #else
Chris@82 266
Chris@82 267 /* Generated by: ../../../genfft/gen_hc2c.native -compact -variables 4 -pipeline-latency 4 -n 10 -dit -name hc2cf_10 -include rdft/scalar/hc2cf.h */
Chris@82 268
Chris@82 269 /*
Chris@82 270 * This function contains 102 FP additions, 60 FP multiplications,
Chris@82 271 * (or, 72 additions, 30 multiplications, 30 fused multiply/add),
Chris@82 272 * 45 stack variables, 4 constants, and 40 memory accesses
Chris@82 273 */
Chris@82 274 #include "rdft/scalar/hc2cf.h"
Chris@82 275
Chris@82 276 static void hc2cf_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 277 {
Chris@82 278 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@82 279 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 280 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 281 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 282 {
Chris@82 283 INT m;
Chris@82 284 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@82 285 E T7, T1O, TT, T1C, TF, TQ, TR, T1r, T1s, T1L, TX, TY, TZ, T16, T19;
Chris@82 286 E T1y, Ti, Tt, Tu, T1o, T1p, T1M, TU, TV, TW, T1d, T1g, T1x;
Chris@82 287 {
Chris@82 288 E T1, T1B, T6, T1A;
Chris@82 289 T1 = Rp[0];
Chris@82 290 T1B = Rm[0];
Chris@82 291 {
Chris@82 292 E T3, T5, T2, T4;
Chris@82 293 T3 = Ip[WS(rs, 2)];
Chris@82 294 T5 = Im[WS(rs, 2)];
Chris@82 295 T2 = W[8];
Chris@82 296 T4 = W[9];
Chris@82 297 T6 = FMA(T2, T3, T4 * T5);
Chris@82 298 T1A = FNMS(T4, T3, T2 * T5);
Chris@82 299 }
Chris@82 300 T7 = T1 - T6;
Chris@82 301 T1O = T1B - T1A;
Chris@82 302 TT = T1 + T6;
Chris@82 303 T1C = T1A + T1B;
Chris@82 304 }
Chris@82 305 {
Chris@82 306 E Tz, T14, TP, T18, TE, T15, TK, T17;
Chris@82 307 {
Chris@82 308 E Tw, Ty, Tv, Tx;
Chris@82 309 Tw = Rp[WS(rs, 2)];
Chris@82 310 Ty = Rm[WS(rs, 2)];
Chris@82 311 Tv = W[6];
Chris@82 312 Tx = W[7];
Chris@82 313 Tz = FMA(Tv, Tw, Tx * Ty);
Chris@82 314 T14 = FNMS(Tx, Tw, Tv * Ty);
Chris@82 315 }
Chris@82 316 {
Chris@82 317 E TM, TO, TL, TN;
Chris@82 318 TM = Ip[0];
Chris@82 319 TO = Im[0];
Chris@82 320 TL = W[0];
Chris@82 321 TN = W[1];
Chris@82 322 TP = FMA(TL, TM, TN * TO);
Chris@82 323 T18 = FNMS(TN, TM, TL * TO);
Chris@82 324 }
Chris@82 325 {
Chris@82 326 E TB, TD, TA, TC;
Chris@82 327 TB = Ip[WS(rs, 4)];
Chris@82 328 TD = Im[WS(rs, 4)];
Chris@82 329 TA = W[16];
Chris@82 330 TC = W[17];
Chris@82 331 TE = FMA(TA, TB, TC * TD);
Chris@82 332 T15 = FNMS(TC, TB, TA * TD);
Chris@82 333 }
Chris@82 334 {
Chris@82 335 E TH, TJ, TG, TI;
Chris@82 336 TH = Rp[WS(rs, 3)];
Chris@82 337 TJ = Rm[WS(rs, 3)];
Chris@82 338 TG = W[10];
Chris@82 339 TI = W[11];
Chris@82 340 TK = FMA(TG, TH, TI * TJ);
Chris@82 341 T17 = FNMS(TI, TH, TG * TJ);
Chris@82 342 }
Chris@82 343 TF = Tz - TE;
Chris@82 344 TQ = TK - TP;
Chris@82 345 TR = TF + TQ;
Chris@82 346 T1r = T14 - T15;
Chris@82 347 T1s = T18 - T17;
Chris@82 348 T1L = T1s - T1r;
Chris@82 349 TX = Tz + TE;
Chris@82 350 TY = TK + TP;
Chris@82 351 TZ = TX + TY;
Chris@82 352 T16 = T14 + T15;
Chris@82 353 T19 = T17 + T18;
Chris@82 354 T1y = T16 + T19;
Chris@82 355 }
Chris@82 356 {
Chris@82 357 E Tc, T1b, Ts, T1f, Th, T1c, Tn, T1e;
Chris@82 358 {
Chris@82 359 E T9, Tb, T8, Ta;
Chris@82 360 T9 = Rp[WS(rs, 1)];
Chris@82 361 Tb = Rm[WS(rs, 1)];
Chris@82 362 T8 = W[2];
Chris@82 363 Ta = W[3];
Chris@82 364 Tc = FMA(T8, T9, Ta * Tb);
Chris@82 365 T1b = FNMS(Ta, T9, T8 * Tb);
Chris@82 366 }
Chris@82 367 {
Chris@82 368 E Tp, Tr, To, Tq;
Chris@82 369 Tp = Ip[WS(rs, 1)];
Chris@82 370 Tr = Im[WS(rs, 1)];
Chris@82 371 To = W[4];
Chris@82 372 Tq = W[5];
Chris@82 373 Ts = FMA(To, Tp, Tq * Tr);
Chris@82 374 T1f = FNMS(Tq, Tp, To * Tr);
Chris@82 375 }
Chris@82 376 {
Chris@82 377 E Te, Tg, Td, Tf;
Chris@82 378 Te = Ip[WS(rs, 3)];
Chris@82 379 Tg = Im[WS(rs, 3)];
Chris@82 380 Td = W[12];
Chris@82 381 Tf = W[13];
Chris@82 382 Th = FMA(Td, Te, Tf * Tg);
Chris@82 383 T1c = FNMS(Tf, Te, Td * Tg);
Chris@82 384 }
Chris@82 385 {
Chris@82 386 E Tk, Tm, Tj, Tl;
Chris@82 387 Tk = Rp[WS(rs, 4)];
Chris@82 388 Tm = Rm[WS(rs, 4)];
Chris@82 389 Tj = W[14];
Chris@82 390 Tl = W[15];
Chris@82 391 Tn = FMA(Tj, Tk, Tl * Tm);
Chris@82 392 T1e = FNMS(Tl, Tk, Tj * Tm);
Chris@82 393 }
Chris@82 394 Ti = Tc - Th;
Chris@82 395 Tt = Tn - Ts;
Chris@82 396 Tu = Ti + Tt;
Chris@82 397 T1o = T1b - T1c;
Chris@82 398 T1p = T1e - T1f;
Chris@82 399 T1M = T1o + T1p;
Chris@82 400 TU = Tc + Th;
Chris@82 401 TV = Tn + Ts;
Chris@82 402 TW = TU + TV;
Chris@82 403 T1d = T1b + T1c;
Chris@82 404 T1g = T1e + T1f;
Chris@82 405 T1x = T1d + T1g;
Chris@82 406 }
Chris@82 407 {
Chris@82 408 E T1l, TS, T1m, T1u, T1w, T1q, T1t, T1v, T1n;
Chris@82 409 T1l = KP559016994 * (Tu - TR);
Chris@82 410 TS = Tu + TR;
Chris@82 411 T1m = FNMS(KP250000000, TS, T7);
Chris@82 412 T1q = T1o - T1p;
Chris@82 413 T1t = T1r + T1s;
Chris@82 414 T1u = FMA(KP951056516, T1q, KP587785252 * T1t);
Chris@82 415 T1w = FNMS(KP587785252, T1q, KP951056516 * T1t);
Chris@82 416 Rm[WS(rs, 4)] = T7 + TS;
Chris@82 417 T1v = T1m - T1l;
Chris@82 418 Rm[WS(rs, 2)] = T1v - T1w;
Chris@82 419 Rp[WS(rs, 3)] = T1v + T1w;
Chris@82 420 T1n = T1l + T1m;
Chris@82 421 Rm[0] = T1n - T1u;
Chris@82 422 Rp[WS(rs, 1)] = T1n + T1u;
Chris@82 423 }
Chris@82 424 {
Chris@82 425 E T1S, T1N, T1T, T1R, T1V, T1P, T1Q, T1W, T1U;
Chris@82 426 T1S = KP559016994 * (T1M + T1L);
Chris@82 427 T1N = T1L - T1M;
Chris@82 428 T1T = FMA(KP250000000, T1N, T1O);
Chris@82 429 T1P = TQ - TF;
Chris@82 430 T1Q = Ti - Tt;
Chris@82 431 T1R = FNMS(KP951056516, T1Q, KP587785252 * T1P);
Chris@82 432 T1V = FMA(KP587785252, T1Q, KP951056516 * T1P);
Chris@82 433 Im[WS(rs, 4)] = T1N - T1O;
Chris@82 434 T1W = T1T - T1S;
Chris@82 435 Im[WS(rs, 2)] = T1V - T1W;
Chris@82 436 Ip[WS(rs, 3)] = T1V + T1W;
Chris@82 437 T1U = T1S + T1T;
Chris@82 438 Im[0] = T1R - T1U;
Chris@82 439 Ip[WS(rs, 1)] = T1R + T1U;
Chris@82 440 }
Chris@82 441 {
Chris@82 442 E T12, T10, T11, T1i, T1k, T1a, T1h, T1j, T13;
Chris@82 443 T12 = KP559016994 * (TW - TZ);
Chris@82 444 T10 = TW + TZ;
Chris@82 445 T11 = FNMS(KP250000000, T10, TT);
Chris@82 446 T1a = T16 - T19;
Chris@82 447 T1h = T1d - T1g;
Chris@82 448 T1i = FNMS(KP587785252, T1h, KP951056516 * T1a);
Chris@82 449 T1k = FMA(KP951056516, T1h, KP587785252 * T1a);
Chris@82 450 Rp[0] = TT + T10;
Chris@82 451 T1j = T12 + T11;
Chris@82 452 Rp[WS(rs, 4)] = T1j - T1k;
Chris@82 453 Rm[WS(rs, 3)] = T1j + T1k;
Chris@82 454 T13 = T11 - T12;
Chris@82 455 Rp[WS(rs, 2)] = T13 - T1i;
Chris@82 456 Rm[WS(rs, 1)] = T13 + T1i;
Chris@82 457 }
Chris@82 458 {
Chris@82 459 E T1H, T1z, T1G, T1F, T1J, T1D, T1E, T1K, T1I;
Chris@82 460 T1H = KP559016994 * (T1x - T1y);
Chris@82 461 T1z = T1x + T1y;
Chris@82 462 T1G = FNMS(KP250000000, T1z, T1C);
Chris@82 463 T1D = TX - TY;
Chris@82 464 T1E = TU - TV;
Chris@82 465 T1F = FNMS(KP587785252, T1E, KP951056516 * T1D);
Chris@82 466 T1J = FMA(KP951056516, T1E, KP587785252 * T1D);
Chris@82 467 Ip[0] = T1z + T1C;
Chris@82 468 T1K = T1H + T1G;
Chris@82 469 Im[WS(rs, 3)] = T1J - T1K;
Chris@82 470 Ip[WS(rs, 4)] = T1J + T1K;
Chris@82 471 T1I = T1G - T1H;
Chris@82 472 Im[WS(rs, 1)] = T1F - T1I;
Chris@82 473 Ip[WS(rs, 2)] = T1F + T1I;
Chris@82 474 }
Chris@82 475 }
Chris@82 476 }
Chris@82 477 }
Chris@82 478
Chris@82 479 static const tw_instr twinstr[] = {
Chris@82 480 {TW_FULL, 1, 10},
Chris@82 481 {TW_NEXT, 1, 0}
Chris@82 482 };
Chris@82 483
Chris@82 484 static const hc2c_desc desc = { 10, "hc2cf_10", twinstr, &GENUS, {72, 30, 30, 0} };
Chris@82 485
Chris@82 486 void X(codelet_hc2cf_10) (planner *p) {
Chris@82 487 X(khc2c_register) (p, hc2cf_10, &desc, HC2C_VIA_RDFT);
Chris@82 488 }
Chris@82 489 #endif