Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cf/hf_9.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 08:06:29 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_hc2hc.native -fma -compact -variables 4 -pipeline-latency 4 -n 9 -dit -name hf_9 -include rdft/scalar/hf.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 96 FP additions, 88 FP multiplications,
Chris@82 32 * (or, 24 additions, 16 multiplications, 72 fused multiply/add),
Chris@82 33 * 55 stack variables, 10 constants, and 36 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/hf.h"
Chris@82 36
Chris@82 37 static void hf_9(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 38 {
Chris@82 39 DK(KP852868531, +0.852868531952443209628250963940074071936020296);
Chris@82 40 DK(KP492403876, +0.492403876506104029683371512294761506835321626);
Chris@82 41 DK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@82 42 DK(KP777861913, +0.777861913430206160028177977318626690410586096);
Chris@82 43 DK(KP839099631, +0.839099631177280011763127298123181364687434283);
Chris@82 44 DK(KP954188894, +0.954188894138671133499268364187245676532219158);
Chris@82 45 DK(KP363970234, +0.363970234266202361351047882776834043890471784);
Chris@82 46 DK(KP176326980, +0.176326980708464973471090386868618986121633062);
Chris@82 47 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 48 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 49 {
Chris@82 50 INT m;
Chris@82 51 for (m = mb, W = W + ((mb - 1) * 16); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 16, MAKE_VOLATILE_STRIDE(18, rs)) {
Chris@82 52 E T1, T1P, Te, T1S, T10, T1Q, T1a, T1d, Ty, T18, Tl, T13, T19, T1c, T1l;
Chris@82 53 E T1r, TS, T1p, TF, T1o, T1g, T1q;
Chris@82 54 T1 = cr[0];
Chris@82 55 T1P = ci[0];
Chris@82 56 {
Chris@82 57 E T3, T6, T4, TW, T9, Tc, Ta, TY, T2, T8;
Chris@82 58 T3 = cr[WS(rs, 3)];
Chris@82 59 T6 = ci[WS(rs, 3)];
Chris@82 60 T2 = W[4];
Chris@82 61 T4 = T2 * T3;
Chris@82 62 TW = T2 * T6;
Chris@82 63 T9 = cr[WS(rs, 6)];
Chris@82 64 Tc = ci[WS(rs, 6)];
Chris@82 65 T8 = W[10];
Chris@82 66 Ta = T8 * T9;
Chris@82 67 TY = T8 * Tc;
Chris@82 68 {
Chris@82 69 E T7, TX, Td, TZ, T5, Tb;
Chris@82 70 T5 = W[5];
Chris@82 71 T7 = FMA(T5, T6, T4);
Chris@82 72 TX = FNMS(T5, T3, TW);
Chris@82 73 Tb = W[11];
Chris@82 74 Td = FMA(Tb, Tc, Ta);
Chris@82 75 TZ = FNMS(Tb, T9, TY);
Chris@82 76 Te = T7 + Td;
Chris@82 77 T1S = Td - T7;
Chris@82 78 T10 = TX - TZ;
Chris@82 79 T1Q = TX + TZ;
Chris@82 80 }
Chris@82 81 }
Chris@82 82 {
Chris@82 83 E Th, Tk, Ti, T12, Tx, T17, Tr, T15, Tg, Tj;
Chris@82 84 Th = cr[WS(rs, 1)];
Chris@82 85 Tk = ci[WS(rs, 1)];
Chris@82 86 Tg = W[0];
Chris@82 87 Ti = Tg * Th;
Chris@82 88 T12 = Tg * Tk;
Chris@82 89 {
Chris@82 90 E Tt, Tw, Tu, T16, Ts, Tv;
Chris@82 91 Tt = cr[WS(rs, 7)];
Chris@82 92 Tw = ci[WS(rs, 7)];
Chris@82 93 Ts = W[12];
Chris@82 94 Tu = Ts * Tt;
Chris@82 95 T16 = Ts * Tw;
Chris@82 96 Tv = W[13];
Chris@82 97 Tx = FMA(Tv, Tw, Tu);
Chris@82 98 T17 = FNMS(Tv, Tt, T16);
Chris@82 99 }
Chris@82 100 {
Chris@82 101 E Tn, Tq, To, T14, Tm, Tp;
Chris@82 102 Tn = cr[WS(rs, 4)];
Chris@82 103 Tq = ci[WS(rs, 4)];
Chris@82 104 Tm = W[6];
Chris@82 105 To = Tm * Tn;
Chris@82 106 T14 = Tm * Tq;
Chris@82 107 Tp = W[7];
Chris@82 108 Tr = FMA(Tp, Tq, To);
Chris@82 109 T15 = FNMS(Tp, Tn, T14);
Chris@82 110 }
Chris@82 111 T1a = Tr - Tx;
Chris@82 112 T1d = T15 - T17;
Chris@82 113 Ty = Tr + Tx;
Chris@82 114 T18 = T15 + T17;
Chris@82 115 Tj = W[1];
Chris@82 116 Tl = FMA(Tj, Tk, Ti);
Chris@82 117 T13 = FNMS(Tj, Th, T12);
Chris@82 118 T19 = FNMS(KP500000000, T18, T13);
Chris@82 119 T1c = FNMS(KP500000000, Ty, Tl);
Chris@82 120 }
Chris@82 121 {
Chris@82 122 E TB, TE, TC, T1n, TR, T1k, TL, T1i, TA, TD;
Chris@82 123 TB = cr[WS(rs, 2)];
Chris@82 124 TE = ci[WS(rs, 2)];
Chris@82 125 TA = W[2];
Chris@82 126 TC = TA * TB;
Chris@82 127 T1n = TA * TE;
Chris@82 128 {
Chris@82 129 E TN, TQ, TO, T1j, TM, TP;
Chris@82 130 TN = cr[WS(rs, 8)];
Chris@82 131 TQ = ci[WS(rs, 8)];
Chris@82 132 TM = W[14];
Chris@82 133 TO = TM * TN;
Chris@82 134 T1j = TM * TQ;
Chris@82 135 TP = W[15];
Chris@82 136 TR = FMA(TP, TQ, TO);
Chris@82 137 T1k = FNMS(TP, TN, T1j);
Chris@82 138 }
Chris@82 139 {
Chris@82 140 E TH, TK, TI, T1h, TG, TJ;
Chris@82 141 TH = cr[WS(rs, 5)];
Chris@82 142 TK = ci[WS(rs, 5)];
Chris@82 143 TG = W[8];
Chris@82 144 TI = TG * TH;
Chris@82 145 T1h = TG * TK;
Chris@82 146 TJ = W[9];
Chris@82 147 TL = FMA(TJ, TK, TI);
Chris@82 148 T1i = FNMS(TJ, TH, T1h);
Chris@82 149 }
Chris@82 150 T1l = T1i - T1k;
Chris@82 151 T1r = TR - TL;
Chris@82 152 TS = TL + TR;
Chris@82 153 T1p = T1i + T1k;
Chris@82 154 TD = W[3];
Chris@82 155 TF = FMA(TD, TE, TC);
Chris@82 156 T1o = FNMS(TD, TB, T1n);
Chris@82 157 T1g = FNMS(KP500000000, TS, TF);
Chris@82 158 T1q = FNMS(KP500000000, T1p, T1o);
Chris@82 159 }
Chris@82 160 {
Chris@82 161 E Tf, T21, TU, T24, T1O, T22, T1L, T23;
Chris@82 162 Tf = T1 + Te;
Chris@82 163 T21 = T1Q + T1P;
Chris@82 164 {
Chris@82 165 E Tz, TT, T1M, T1N;
Chris@82 166 Tz = Tl + Ty;
Chris@82 167 TT = TF + TS;
Chris@82 168 TU = Tz + TT;
Chris@82 169 T24 = TT - Tz;
Chris@82 170 T1M = T13 + T18;
Chris@82 171 T1N = T1o + T1p;
Chris@82 172 T1O = T1M - T1N;
Chris@82 173 T22 = T1M + T1N;
Chris@82 174 }
Chris@82 175 cr[0] = Tf + TU;
Chris@82 176 ci[WS(rs, 8)] = T22 + T21;
Chris@82 177 T1L = FNMS(KP500000000, TU, Tf);
Chris@82 178 ci[WS(rs, 2)] = FNMS(KP866025403, T1O, T1L);
Chris@82 179 cr[WS(rs, 3)] = FMA(KP866025403, T1O, T1L);
Chris@82 180 T23 = FNMS(KP500000000, T22, T21);
Chris@82 181 cr[WS(rs, 6)] = FMS(KP866025403, T24, T23);
Chris@82 182 ci[WS(rs, 5)] = FMA(KP866025403, T24, T23);
Chris@82 183 }
Chris@82 184 {
Chris@82 185 E T11, T1z, T1T, T1X, T1f, T1w, T1t, T1x, T1u, T1Y, T1C, T1I, T1F, T1J, T1G;
Chris@82 186 E T1U, TV, T1R;
Chris@82 187 TV = FNMS(KP500000000, Te, T1);
Chris@82 188 T11 = FNMS(KP866025403, T10, TV);
Chris@82 189 T1z = FMA(KP866025403, T10, TV);
Chris@82 190 T1R = FNMS(KP500000000, T1Q, T1P);
Chris@82 191 T1T = FMA(KP866025403, T1S, T1R);
Chris@82 192 T1X = FNMS(KP866025403, T1S, T1R);
Chris@82 193 {
Chris@82 194 E T1b, T1e, T1m, T1s;
Chris@82 195 T1b = FMA(KP866025403, T1a, T19);
Chris@82 196 T1e = FNMS(KP866025403, T1d, T1c);
Chris@82 197 T1f = FMA(KP176326980, T1e, T1b);
Chris@82 198 T1w = FNMS(KP176326980, T1b, T1e);
Chris@82 199 T1m = FNMS(KP866025403, T1l, T1g);
Chris@82 200 T1s = FNMS(KP866025403, T1r, T1q);
Chris@82 201 T1t = FNMS(KP363970234, T1s, T1m);
Chris@82 202 T1x = FMA(KP363970234, T1m, T1s);
Chris@82 203 }
Chris@82 204 T1u = FNMS(KP954188894, T1t, T1f);
Chris@82 205 T1Y = FMA(KP954188894, T1x, T1w);
Chris@82 206 {
Chris@82 207 E T1A, T1B, T1D, T1E;
Chris@82 208 T1A = FMA(KP866025403, T1r, T1q);
Chris@82 209 T1B = FMA(KP866025403, T1l, T1g);
Chris@82 210 T1C = FMA(KP176326980, T1B, T1A);
Chris@82 211 T1I = FNMS(KP176326980, T1A, T1B);
Chris@82 212 T1D = FMA(KP866025403, T1d, T1c);
Chris@82 213 T1E = FNMS(KP866025403, T1a, T19);
Chris@82 214 T1F = FMA(KP839099631, T1E, T1D);
Chris@82 215 T1J = FNMS(KP839099631, T1D, T1E);
Chris@82 216 }
Chris@82 217 T1G = FMA(KP777861913, T1F, T1C);
Chris@82 218 T1U = FNMS(KP777861913, T1J, T1I);
Chris@82 219 cr[WS(rs, 2)] = FMA(KP984807753, T1u, T11);
Chris@82 220 ci[WS(rs, 7)] = FNMS(KP984807753, T1U, T1T);
Chris@82 221 ci[WS(rs, 6)] = FNMS(KP984807753, T1Y, T1X);
Chris@82 222 cr[WS(rs, 1)] = FMA(KP984807753, T1G, T1z);
Chris@82 223 {
Chris@82 224 E T1V, T1W, T1H, T1K;
Chris@82 225 T1V = FMA(KP492403876, T1U, T1T);
Chris@82 226 T1W = FNMS(KP777861913, T1F, T1C);
Chris@82 227 cr[WS(rs, 7)] = FMS(KP852868531, T1W, T1V);
Chris@82 228 ci[WS(rs, 4)] = FMA(KP852868531, T1W, T1V);
Chris@82 229 T1H = FNMS(KP492403876, T1G, T1z);
Chris@82 230 T1K = FMA(KP777861913, T1J, T1I);
Chris@82 231 ci[WS(rs, 1)] = FNMS(KP852868531, T1K, T1H);
Chris@82 232 cr[WS(rs, 4)] = FMA(KP852868531, T1K, T1H);
Chris@82 233 }
Chris@82 234 {
Chris@82 235 E T1v, T1y, T1Z, T20;
Chris@82 236 T1v = FNMS(KP492403876, T1u, T11);
Chris@82 237 T1y = FNMS(KP954188894, T1x, T1w);
Chris@82 238 ci[WS(rs, 3)] = FNMS(KP852868531, T1y, T1v);
Chris@82 239 ci[0] = FMA(KP852868531, T1y, T1v);
Chris@82 240 T1Z = FMA(KP492403876, T1Y, T1X);
Chris@82 241 T20 = FMA(KP954188894, T1t, T1f);
Chris@82 242 cr[WS(rs, 5)] = FMS(KP852868531, T20, T1Z);
Chris@82 243 cr[WS(rs, 8)] = -(FMA(KP852868531, T20, T1Z));
Chris@82 244 }
Chris@82 245 }
Chris@82 246 }
Chris@82 247 }
Chris@82 248 }
Chris@82 249
Chris@82 250 static const tw_instr twinstr[] = {
Chris@82 251 {TW_FULL, 1, 9},
Chris@82 252 {TW_NEXT, 1, 0}
Chris@82 253 };
Chris@82 254
Chris@82 255 static const hc2hc_desc desc = { 9, "hf_9", twinstr, &GENUS, {24, 16, 72, 0} };
Chris@82 256
Chris@82 257 void X(codelet_hf_9) (planner *p) {
Chris@82 258 X(khc2hc_register) (p, hf_9, &desc);
Chris@82 259 }
Chris@82 260 #else
Chris@82 261
Chris@82 262 /* Generated by: ../../../genfft/gen_hc2hc.native -compact -variables 4 -pipeline-latency 4 -n 9 -dit -name hf_9 -include rdft/scalar/hf.h */
Chris@82 263
Chris@82 264 /*
Chris@82 265 * This function contains 96 FP additions, 72 FP multiplications,
Chris@82 266 * (or, 60 additions, 36 multiplications, 36 fused multiply/add),
Chris@82 267 * 41 stack variables, 8 constants, and 36 memory accesses
Chris@82 268 */
Chris@82 269 #include "rdft/scalar/hf.h"
Chris@82 270
Chris@82 271 static void hf_9(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 272 {
Chris@82 273 DK(KP642787609, +0.642787609686539326322643409907263432907559884);
Chris@82 274 DK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@82 275 DK(KP939692620, +0.939692620785908384054109277324731469936208134);
Chris@82 276 DK(KP342020143, +0.342020143325668733044099614682259580763083368);
Chris@82 277 DK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@82 278 DK(KP173648177, +0.173648177666930348851716626769314796000375677);
Chris@82 279 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 280 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 281 {
Chris@82 282 INT m;
Chris@82 283 for (m = mb, W = W + ((mb - 1) * 16); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 16, MAKE_VOLATILE_STRIDE(18, rs)) {
Chris@82 284 E T1, T1B, TQ, T1A, Tc, TN, T1C, T1D, TL, T1x, T19, T1o, T1c, T1n, Tu;
Chris@82 285 E T1w, TW, T1k, T11, T1l;
Chris@82 286 {
Chris@82 287 E T6, TO, Tb, TP;
Chris@82 288 T1 = cr[0];
Chris@82 289 T1B = ci[0];
Chris@82 290 {
Chris@82 291 E T3, T5, T2, T4;
Chris@82 292 T3 = cr[WS(rs, 3)];
Chris@82 293 T5 = ci[WS(rs, 3)];
Chris@82 294 T2 = W[4];
Chris@82 295 T4 = W[5];
Chris@82 296 T6 = FMA(T2, T3, T4 * T5);
Chris@82 297 TO = FNMS(T4, T3, T2 * T5);
Chris@82 298 }
Chris@82 299 {
Chris@82 300 E T8, Ta, T7, T9;
Chris@82 301 T8 = cr[WS(rs, 6)];
Chris@82 302 Ta = ci[WS(rs, 6)];
Chris@82 303 T7 = W[10];
Chris@82 304 T9 = W[11];
Chris@82 305 Tb = FMA(T7, T8, T9 * Ta);
Chris@82 306 TP = FNMS(T9, T8, T7 * Ta);
Chris@82 307 }
Chris@82 308 TQ = KP866025403 * (TO - TP);
Chris@82 309 T1A = KP866025403 * (Tb - T6);
Chris@82 310 Tc = T6 + Tb;
Chris@82 311 TN = FNMS(KP500000000, Tc, T1);
Chris@82 312 T1C = TO + TP;
Chris@82 313 T1D = FNMS(KP500000000, T1C, T1B);
Chris@82 314 }
Chris@82 315 {
Chris@82 316 E Tz, T13, TE, T14, TJ, T15, TK, T16;
Chris@82 317 {
Chris@82 318 E Tw, Ty, Tv, Tx;
Chris@82 319 Tw = cr[WS(rs, 2)];
Chris@82 320 Ty = ci[WS(rs, 2)];
Chris@82 321 Tv = W[2];
Chris@82 322 Tx = W[3];
Chris@82 323 Tz = FMA(Tv, Tw, Tx * Ty);
Chris@82 324 T13 = FNMS(Tx, Tw, Tv * Ty);
Chris@82 325 }
Chris@82 326 {
Chris@82 327 E TB, TD, TA, TC;
Chris@82 328 TB = cr[WS(rs, 5)];
Chris@82 329 TD = ci[WS(rs, 5)];
Chris@82 330 TA = W[8];
Chris@82 331 TC = W[9];
Chris@82 332 TE = FMA(TA, TB, TC * TD);
Chris@82 333 T14 = FNMS(TC, TB, TA * TD);
Chris@82 334 }
Chris@82 335 {
Chris@82 336 E TG, TI, TF, TH;
Chris@82 337 TG = cr[WS(rs, 8)];
Chris@82 338 TI = ci[WS(rs, 8)];
Chris@82 339 TF = W[14];
Chris@82 340 TH = W[15];
Chris@82 341 TJ = FMA(TF, TG, TH * TI);
Chris@82 342 T15 = FNMS(TH, TG, TF * TI);
Chris@82 343 }
Chris@82 344 TK = TE + TJ;
Chris@82 345 T16 = T14 + T15;
Chris@82 346 TL = Tz + TK;
Chris@82 347 T1x = T13 + T16;
Chris@82 348 {
Chris@82 349 E T17, T18, T1a, T1b;
Chris@82 350 T17 = FNMS(KP500000000, T16, T13);
Chris@82 351 T18 = KP866025403 * (TJ - TE);
Chris@82 352 T19 = T17 - T18;
Chris@82 353 T1o = T18 + T17;
Chris@82 354 T1a = FNMS(KP500000000, TK, Tz);
Chris@82 355 T1b = KP866025403 * (T14 - T15);
Chris@82 356 T1c = T1a - T1b;
Chris@82 357 T1n = T1a + T1b;
Chris@82 358 }
Chris@82 359 }
Chris@82 360 {
Chris@82 361 E Ti, TX, Tn, TT, Ts, TU, Tt, TY;
Chris@82 362 {
Chris@82 363 E Tf, Th, Te, Tg;
Chris@82 364 Tf = cr[WS(rs, 1)];
Chris@82 365 Th = ci[WS(rs, 1)];
Chris@82 366 Te = W[0];
Chris@82 367 Tg = W[1];
Chris@82 368 Ti = FMA(Te, Tf, Tg * Th);
Chris@82 369 TX = FNMS(Tg, Tf, Te * Th);
Chris@82 370 }
Chris@82 371 {
Chris@82 372 E Tk, Tm, Tj, Tl;
Chris@82 373 Tk = cr[WS(rs, 4)];
Chris@82 374 Tm = ci[WS(rs, 4)];
Chris@82 375 Tj = W[6];
Chris@82 376 Tl = W[7];
Chris@82 377 Tn = FMA(Tj, Tk, Tl * Tm);
Chris@82 378 TT = FNMS(Tl, Tk, Tj * Tm);
Chris@82 379 }
Chris@82 380 {
Chris@82 381 E Tp, Tr, To, Tq;
Chris@82 382 Tp = cr[WS(rs, 7)];
Chris@82 383 Tr = ci[WS(rs, 7)];
Chris@82 384 To = W[12];
Chris@82 385 Tq = W[13];
Chris@82 386 Ts = FMA(To, Tp, Tq * Tr);
Chris@82 387 TU = FNMS(Tq, Tp, To * Tr);
Chris@82 388 }
Chris@82 389 Tt = Tn + Ts;
Chris@82 390 TY = TT + TU;
Chris@82 391 Tu = Ti + Tt;
Chris@82 392 T1w = TX + TY;
Chris@82 393 {
Chris@82 394 E TS, TV, TZ, T10;
Chris@82 395 TS = FNMS(KP500000000, Tt, Ti);
Chris@82 396 TV = KP866025403 * (TT - TU);
Chris@82 397 TW = TS - TV;
Chris@82 398 T1k = TS + TV;
Chris@82 399 TZ = FNMS(KP500000000, TY, TX);
Chris@82 400 T10 = KP866025403 * (Ts - Tn);
Chris@82 401 T11 = TZ - T10;
Chris@82 402 T1l = T10 + TZ;
Chris@82 403 }
Chris@82 404 }
Chris@82 405 {
Chris@82 406 E T1y, Td, TM, T1v;
Chris@82 407 T1y = KP866025403 * (T1w - T1x);
Chris@82 408 Td = T1 + Tc;
Chris@82 409 TM = Tu + TL;
Chris@82 410 T1v = FNMS(KP500000000, TM, Td);
Chris@82 411 cr[0] = Td + TM;
Chris@82 412 cr[WS(rs, 3)] = T1v + T1y;
Chris@82 413 ci[WS(rs, 2)] = T1v - T1y;
Chris@82 414 }
Chris@82 415 {
Chris@82 416 E TR, T1I, T1e, T1K, T1i, T1H, T1f, T1J;
Chris@82 417 TR = TN - TQ;
Chris@82 418 T1I = T1D - T1A;
Chris@82 419 {
Chris@82 420 E T12, T1d, T1g, T1h;
Chris@82 421 T12 = FMA(KP173648177, TW, KP984807753 * T11);
Chris@82 422 T1d = FNMS(KP939692620, T1c, KP342020143 * T19);
Chris@82 423 T1e = T12 + T1d;
Chris@82 424 T1K = KP866025403 * (T1d - T12);
Chris@82 425 T1g = FNMS(KP984807753, TW, KP173648177 * T11);
Chris@82 426 T1h = FMA(KP342020143, T1c, KP939692620 * T19);
Chris@82 427 T1i = KP866025403 * (T1g + T1h);
Chris@82 428 T1H = T1g - T1h;
Chris@82 429 }
Chris@82 430 cr[WS(rs, 2)] = TR + T1e;
Chris@82 431 ci[WS(rs, 6)] = T1H + T1I;
Chris@82 432 T1f = FNMS(KP500000000, T1e, TR);
Chris@82 433 ci[0] = T1f - T1i;
Chris@82 434 ci[WS(rs, 3)] = T1f + T1i;
Chris@82 435 T1J = FMS(KP500000000, T1H, T1I);
Chris@82 436 cr[WS(rs, 5)] = T1J - T1K;
Chris@82 437 cr[WS(rs, 8)] = T1K + T1J;
Chris@82 438 }
Chris@82 439 {
Chris@82 440 E T1L, T1M, T1N, T1O;
Chris@82 441 T1L = KP866025403 * (TL - Tu);
Chris@82 442 T1M = T1C + T1B;
Chris@82 443 T1N = T1w + T1x;
Chris@82 444 T1O = FNMS(KP500000000, T1N, T1M);
Chris@82 445 cr[WS(rs, 6)] = T1L - T1O;
Chris@82 446 ci[WS(rs, 8)] = T1N + T1M;
Chris@82 447 ci[WS(rs, 5)] = T1L + T1O;
Chris@82 448 }
Chris@82 449 {
Chris@82 450 E T1j, T1E, T1q, T1z, T1u, T1F, T1r, T1G;
Chris@82 451 T1j = TN + TQ;
Chris@82 452 T1E = T1A + T1D;
Chris@82 453 {
Chris@82 454 E T1m, T1p, T1s, T1t;
Chris@82 455 T1m = FMA(KP766044443, T1k, KP642787609 * T1l);
Chris@82 456 T1p = FMA(KP173648177, T1n, KP984807753 * T1o);
Chris@82 457 T1q = T1m + T1p;
Chris@82 458 T1z = KP866025403 * (T1p - T1m);
Chris@82 459 T1s = FNMS(KP642787609, T1k, KP766044443 * T1l);
Chris@82 460 T1t = FNMS(KP984807753, T1n, KP173648177 * T1o);
Chris@82 461 T1u = KP866025403 * (T1s - T1t);
Chris@82 462 T1F = T1s + T1t;
Chris@82 463 }
Chris@82 464 cr[WS(rs, 1)] = T1j + T1q;
Chris@82 465 T1r = FNMS(KP500000000, T1q, T1j);
Chris@82 466 ci[WS(rs, 1)] = T1r - T1u;
Chris@82 467 cr[WS(rs, 4)] = T1r + T1u;
Chris@82 468 ci[WS(rs, 7)] = T1F + T1E;
Chris@82 469 T1G = FNMS(KP500000000, T1F, T1E);
Chris@82 470 cr[WS(rs, 7)] = T1z - T1G;
Chris@82 471 ci[WS(rs, 4)] = T1z + T1G;
Chris@82 472 }
Chris@82 473 }
Chris@82 474 }
Chris@82 475 }
Chris@82 476
Chris@82 477 static const tw_instr twinstr[] = {
Chris@82 478 {TW_FULL, 1, 9},
Chris@82 479 {TW_NEXT, 1, 0}
Chris@82 480 };
Chris@82 481
Chris@82 482 static const hc2hc_desc desc = { 9, "hf_9", twinstr, &GENUS, {60, 36, 36, 0} };
Chris@82 483
Chris@82 484 void X(codelet_hf_9) (planner *p) {
Chris@82 485 X(khc2hc_register) (p, hf_9, &desc);
Chris@82 486 }
Chris@82 487 #endif