Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cb/hc2cb_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 08:07:52 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 -sign 1 -n 10 -dif -name hc2cb_10 -include rdft/scalar/hc2cb.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/hc2cb.h"
Chris@82 36
Chris@82 37 static void hc2cb_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(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 42 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
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 TH, T1B, TB, T11, T1E, T1G, TK, TM, T1x, T1V, T3, T1g, Tl, T1I, T1J;
Chris@82 47 E TO, TP, T1p, Ti, Tk, T1n, T1o, TF, TG;
Chris@82 48 TF = Ip[0];
Chris@82 49 TG = Im[WS(rs, 4)];
Chris@82 50 TH = TF - TG;
Chris@82 51 T1B = TF + TG;
Chris@82 52 {
Chris@82 53 E Tp, T1u, Tz, T1s, Ts, T1v, Tw, T1r;
Chris@82 54 {
Chris@82 55 E Tn, To, Tx, Ty;
Chris@82 56 Tn = Ip[WS(rs, 4)];
Chris@82 57 To = Im[0];
Chris@82 58 Tp = Tn - To;
Chris@82 59 T1u = Tn + To;
Chris@82 60 Tx = Ip[WS(rs, 3)];
Chris@82 61 Ty = Im[WS(rs, 1)];
Chris@82 62 Tz = Tx - Ty;
Chris@82 63 T1s = Tx + Ty;
Chris@82 64 }
Chris@82 65 {
Chris@82 66 E Tq, Tr, Tu, Tv;
Chris@82 67 Tq = Ip[WS(rs, 1)];
Chris@82 68 Tr = Im[WS(rs, 3)];
Chris@82 69 Ts = Tq - Tr;
Chris@82 70 T1v = Tq + Tr;
Chris@82 71 Tu = Ip[WS(rs, 2)];
Chris@82 72 Tv = Im[WS(rs, 2)];
Chris@82 73 Tw = Tu - Tv;
Chris@82 74 T1r = Tu + Tv;
Chris@82 75 }
Chris@82 76 {
Chris@82 77 E Tt, TA, T1C, T1D;
Chris@82 78 Tt = Tp - Ts;
Chris@82 79 TA = Tw - Tz;
Chris@82 80 TB = FNMS(KP618033988, TA, Tt);
Chris@82 81 T11 = FMA(KP618033988, Tt, TA);
Chris@82 82 T1C = T1r - T1s;
Chris@82 83 T1D = T1u - T1v;
Chris@82 84 T1E = T1C + T1D;
Chris@82 85 T1G = T1C - T1D;
Chris@82 86 }
Chris@82 87 {
Chris@82 88 E TI, TJ, T1t, T1w;
Chris@82 89 TI = Tw + Tz;
Chris@82 90 TJ = Tp + Ts;
Chris@82 91 TK = TI + TJ;
Chris@82 92 TM = TI - TJ;
Chris@82 93 T1t = T1r + T1s;
Chris@82 94 T1w = T1u + T1v;
Chris@82 95 T1x = FMA(KP618033988, T1w, T1t);
Chris@82 96 T1V = FNMS(KP618033988, T1t, T1w);
Chris@82 97 }
Chris@82 98 }
Chris@82 99 {
Chris@82 100 E Td, T1k, Tg, T1l, Th, T1m, T6, T1h, T9, T1i, Ta, T1j, T1, T2;
Chris@82 101 T1 = Rp[0];
Chris@82 102 T2 = Rm[WS(rs, 4)];
Chris@82 103 T3 = T1 + T2;
Chris@82 104 T1g = T1 - T2;
Chris@82 105 {
Chris@82 106 E Tb, Tc, Te, Tf;
Chris@82 107 Tb = Rp[WS(rs, 4)];
Chris@82 108 Tc = Rm[0];
Chris@82 109 Td = Tb + Tc;
Chris@82 110 T1k = Tb - Tc;
Chris@82 111 Te = Rm[WS(rs, 3)];
Chris@82 112 Tf = Rp[WS(rs, 1)];
Chris@82 113 Tg = Te + Tf;
Chris@82 114 T1l = Te - Tf;
Chris@82 115 }
Chris@82 116 Th = Td + Tg;
Chris@82 117 T1m = T1k + T1l;
Chris@82 118 {
Chris@82 119 E T4, T5, T7, T8;
Chris@82 120 T4 = Rp[WS(rs, 2)];
Chris@82 121 T5 = Rm[WS(rs, 2)];
Chris@82 122 T6 = T4 + T5;
Chris@82 123 T1h = T4 - T5;
Chris@82 124 T7 = Rm[WS(rs, 1)];
Chris@82 125 T8 = Rp[WS(rs, 3)];
Chris@82 126 T9 = T7 + T8;
Chris@82 127 T1i = T7 - T8;
Chris@82 128 }
Chris@82 129 Ta = T6 + T9;
Chris@82 130 T1j = T1h + T1i;
Chris@82 131 Tl = Ta - Th;
Chris@82 132 T1I = T1h - T1i;
Chris@82 133 T1J = T1k - T1l;
Chris@82 134 TO = Td - Tg;
Chris@82 135 TP = T6 - T9;
Chris@82 136 T1p = T1j - T1m;
Chris@82 137 Ti = Ta + Th;
Chris@82 138 Tk = FNMS(KP250000000, Ti, T3);
Chris@82 139 T1n = T1j + T1m;
Chris@82 140 T1o = FNMS(KP250000000, T1n, T1g);
Chris@82 141 }
Chris@82 142 Rp[0] = T3 + Ti;
Chris@82 143 Rm[0] = TH + TK;
Chris@82 144 {
Chris@82 145 E T2d, T29, T2b, T2c, T2e, T2a;
Chris@82 146 T2d = T1B + T1E;
Chris@82 147 T2a = T1g + T1n;
Chris@82 148 T29 = W[8];
Chris@82 149 T2b = T29 * T2a;
Chris@82 150 T2c = W[9];
Chris@82 151 T2e = T2c * T2a;
Chris@82 152 Ip[WS(rs, 2)] = FNMS(T2c, T2d, T2b);
Chris@82 153 Im[WS(rs, 2)] = FMA(T29, T2d, T2e);
Chris@82 154 }
Chris@82 155 {
Chris@82 156 E TQ, T16, TC, TU, TN, T15, T12, T1a, Tm, TL, T10;
Chris@82 157 TQ = FNMS(KP618033988, TP, TO);
Chris@82 158 T16 = FMA(KP618033988, TO, TP);
Chris@82 159 Tm = FNMS(KP559016994, Tl, Tk);
Chris@82 160 TC = FMA(KP951056516, TB, Tm);
Chris@82 161 TU = FNMS(KP951056516, TB, Tm);
Chris@82 162 TL = FNMS(KP250000000, TK, TH);
Chris@82 163 TN = FNMS(KP559016994, TM, TL);
Chris@82 164 T15 = FMA(KP559016994, TM, TL);
Chris@82 165 T10 = FMA(KP559016994, Tl, Tk);
Chris@82 166 T12 = FMA(KP951056516, T11, T10);
Chris@82 167 T1a = FNMS(KP951056516, T11, T10);
Chris@82 168 {
Chris@82 169 E TR, TE, TS, Tj, TD;
Chris@82 170 TR = FNMS(KP951056516, TQ, TN);
Chris@82 171 TE = W[3];
Chris@82 172 TS = TE * TC;
Chris@82 173 Tj = W[2];
Chris@82 174 TD = Tj * TC;
Chris@82 175 Rp[WS(rs, 1)] = FNMS(TE, TR, TD);
Chris@82 176 Rm[WS(rs, 1)] = FMA(Tj, TR, TS);
Chris@82 177 }
Chris@82 178 {
Chris@82 179 E T1d, T1c, T1e, T19, T1b;
Chris@82 180 T1d = FMA(KP951056516, T16, T15);
Chris@82 181 T1c = W[11];
Chris@82 182 T1e = T1c * T1a;
Chris@82 183 T19 = W[10];
Chris@82 184 T1b = T19 * T1a;
Chris@82 185 Rp[WS(rs, 3)] = FNMS(T1c, T1d, T1b);
Chris@82 186 Rm[WS(rs, 3)] = FMA(T19, T1d, T1e);
Chris@82 187 }
Chris@82 188 {
Chris@82 189 E TX, TW, TY, TT, TV;
Chris@82 190 TX = FMA(KP951056516, TQ, TN);
Chris@82 191 TW = W[15];
Chris@82 192 TY = TW * TU;
Chris@82 193 TT = W[14];
Chris@82 194 TV = TT * TU;
Chris@82 195 Rp[WS(rs, 4)] = FNMS(TW, TX, TV);
Chris@82 196 Rm[WS(rs, 4)] = FMA(TT, TX, TY);
Chris@82 197 }
Chris@82 198 {
Chris@82 199 E T17, T14, T18, TZ, T13;
Chris@82 200 T17 = FNMS(KP951056516, T16, T15);
Chris@82 201 T14 = W[7];
Chris@82 202 T18 = T14 * T12;
Chris@82 203 TZ = W[6];
Chris@82 204 T13 = TZ * T12;
Chris@82 205 Rp[WS(rs, 2)] = FNMS(T14, T17, T13);
Chris@82 206 Rm[WS(rs, 2)] = FMA(TZ, T17, T18);
Chris@82 207 }
Chris@82 208 }
Chris@82 209 {
Chris@82 210 E T1K, T20, T1y, T1O, T1H, T1Z, T1W, T24, T1q, T1F, T1U;
Chris@82 211 T1K = FMA(KP618033988, T1J, T1I);
Chris@82 212 T20 = FNMS(KP618033988, T1I, T1J);
Chris@82 213 T1q = FMA(KP559016994, T1p, T1o);
Chris@82 214 T1y = FNMS(KP951056516, T1x, T1q);
Chris@82 215 T1O = FMA(KP951056516, T1x, T1q);
Chris@82 216 T1F = FNMS(KP250000000, T1E, T1B);
Chris@82 217 T1H = FMA(KP559016994, T1G, T1F);
Chris@82 218 T1Z = FNMS(KP559016994, T1G, T1F);
Chris@82 219 T1U = FNMS(KP559016994, T1p, T1o);
Chris@82 220 T1W = FNMS(KP951056516, T1V, T1U);
Chris@82 221 T24 = FMA(KP951056516, T1V, T1U);
Chris@82 222 {
Chris@82 223 E T1L, T1A, T1M, T1f, T1z;
Chris@82 224 T1L = FMA(KP951056516, T1K, T1H);
Chris@82 225 T1A = W[1];
Chris@82 226 T1M = T1A * T1y;
Chris@82 227 T1f = W[0];
Chris@82 228 T1z = T1f * T1y;
Chris@82 229 Ip[0] = FNMS(T1A, T1L, T1z);
Chris@82 230 Im[0] = FMA(T1f, T1L, T1M);
Chris@82 231 }
Chris@82 232 {
Chris@82 233 E T27, T26, T28, T23, T25;
Chris@82 234 T27 = FNMS(KP951056516, T20, T1Z);
Chris@82 235 T26 = W[13];
Chris@82 236 T28 = T26 * T24;
Chris@82 237 T23 = W[12];
Chris@82 238 T25 = T23 * T24;
Chris@82 239 Ip[WS(rs, 3)] = FNMS(T26, T27, T25);
Chris@82 240 Im[WS(rs, 3)] = FMA(T23, T27, T28);
Chris@82 241 }
Chris@82 242 {
Chris@82 243 E T1R, T1Q, T1S, T1N, T1P;
Chris@82 244 T1R = FNMS(KP951056516, T1K, T1H);
Chris@82 245 T1Q = W[17];
Chris@82 246 T1S = T1Q * T1O;
Chris@82 247 T1N = W[16];
Chris@82 248 T1P = T1N * T1O;
Chris@82 249 Ip[WS(rs, 4)] = FNMS(T1Q, T1R, T1P);
Chris@82 250 Im[WS(rs, 4)] = FMA(T1N, T1R, T1S);
Chris@82 251 }
Chris@82 252 {
Chris@82 253 E T21, T1Y, T22, T1T, T1X;
Chris@82 254 T21 = FMA(KP951056516, T20, T1Z);
Chris@82 255 T1Y = W[5];
Chris@82 256 T22 = T1Y * T1W;
Chris@82 257 T1T = W[4];
Chris@82 258 T1X = T1T * T1W;
Chris@82 259 Ip[WS(rs, 1)] = FNMS(T1Y, T21, T1X);
Chris@82 260 Im[WS(rs, 1)] = FMA(T1T, T21, T22);
Chris@82 261 }
Chris@82 262 }
Chris@82 263 }
Chris@82 264 }
Chris@82 265 }
Chris@82 266
Chris@82 267 static const tw_instr twinstr[] = {
Chris@82 268 {TW_FULL, 1, 10},
Chris@82 269 {TW_NEXT, 1, 0}
Chris@82 270 };
Chris@82 271
Chris@82 272 static const hc2c_desc desc = { 10, "hc2cb_10", twinstr, &GENUS, {48, 18, 54, 0} };
Chris@82 273
Chris@82 274 void X(codelet_hc2cb_10) (planner *p) {
Chris@82 275 X(khc2c_register) (p, hc2cb_10, &desc, HC2C_VIA_RDFT);
Chris@82 276 }
Chris@82 277 #else
Chris@82 278
Chris@82 279 /* Generated by: ../../../genfft/gen_hc2c.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -dif -name hc2cb_10 -include rdft/scalar/hc2cb.h */
Chris@82 280
Chris@82 281 /*
Chris@82 282 * This function contains 102 FP additions, 60 FP multiplications,
Chris@82 283 * (or, 72 additions, 30 multiplications, 30 fused multiply/add),
Chris@82 284 * 39 stack variables, 4 constants, and 40 memory accesses
Chris@82 285 */
Chris@82 286 #include "rdft/scalar/hc2cb.h"
Chris@82 287
Chris@82 288 static void hc2cb_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 289 {
Chris@82 290 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 291 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 292 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@82 293 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 294 {
Chris@82 295 INT m;
Chris@82 296 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@82 297 E T3, T18, TJ, T1i, TE, TF, T1B, T1A, T1f, T1t, Ti, Tl, Tt, TA, T1w;
Chris@82 298 E T1v, T1p, T1E, TM, TO;
Chris@82 299 {
Chris@82 300 E T1, T2, TH, TI;
Chris@82 301 T1 = Rp[0];
Chris@82 302 T2 = Rm[WS(rs, 4)];
Chris@82 303 T3 = T1 + T2;
Chris@82 304 T18 = T1 - T2;
Chris@82 305 TH = Ip[0];
Chris@82 306 TI = Im[WS(rs, 4)];
Chris@82 307 TJ = TH - TI;
Chris@82 308 T1i = TH + TI;
Chris@82 309 }
Chris@82 310 {
Chris@82 311 E T6, T19, Tg, T1d, T9, T1a, Td, T1c;
Chris@82 312 {
Chris@82 313 E T4, T5, Te, Tf;
Chris@82 314 T4 = Rp[WS(rs, 2)];
Chris@82 315 T5 = Rm[WS(rs, 2)];
Chris@82 316 T6 = T4 + T5;
Chris@82 317 T19 = T4 - T5;
Chris@82 318 Te = Rm[WS(rs, 3)];
Chris@82 319 Tf = Rp[WS(rs, 1)];
Chris@82 320 Tg = Te + Tf;
Chris@82 321 T1d = Te - Tf;
Chris@82 322 }
Chris@82 323 {
Chris@82 324 E T7, T8, Tb, Tc;
Chris@82 325 T7 = Rm[WS(rs, 1)];
Chris@82 326 T8 = Rp[WS(rs, 3)];
Chris@82 327 T9 = T7 + T8;
Chris@82 328 T1a = T7 - T8;
Chris@82 329 Tb = Rp[WS(rs, 4)];
Chris@82 330 Tc = Rm[0];
Chris@82 331 Td = Tb + Tc;
Chris@82 332 T1c = Tb - Tc;
Chris@82 333 }
Chris@82 334 TE = T6 - T9;
Chris@82 335 TF = Td - Tg;
Chris@82 336 T1B = T1c - T1d;
Chris@82 337 T1A = T19 - T1a;
Chris@82 338 {
Chris@82 339 E T1b, T1e, Ta, Th;
Chris@82 340 T1b = T19 + T1a;
Chris@82 341 T1e = T1c + T1d;
Chris@82 342 T1f = T1b + T1e;
Chris@82 343 T1t = KP559016994 * (T1b - T1e);
Chris@82 344 Ta = T6 + T9;
Chris@82 345 Th = Td + Tg;
Chris@82 346 Ti = Ta + Th;
Chris@82 347 Tl = KP559016994 * (Ta - Th);
Chris@82 348 }
Chris@82 349 }
Chris@82 350 {
Chris@82 351 E Tp, T1j, Tz, T1n, Ts, T1k, Tw, T1m;
Chris@82 352 {
Chris@82 353 E Tn, To, Tx, Ty;
Chris@82 354 Tn = Ip[WS(rs, 2)];
Chris@82 355 To = Im[WS(rs, 2)];
Chris@82 356 Tp = Tn - To;
Chris@82 357 T1j = Tn + To;
Chris@82 358 Tx = Ip[WS(rs, 1)];
Chris@82 359 Ty = Im[WS(rs, 3)];
Chris@82 360 Tz = Tx - Ty;
Chris@82 361 T1n = Tx + Ty;
Chris@82 362 }
Chris@82 363 {
Chris@82 364 E Tq, Tr, Tu, Tv;
Chris@82 365 Tq = Ip[WS(rs, 3)];
Chris@82 366 Tr = Im[WS(rs, 1)];
Chris@82 367 Ts = Tq - Tr;
Chris@82 368 T1k = Tq + Tr;
Chris@82 369 Tu = Ip[WS(rs, 4)];
Chris@82 370 Tv = Im[0];
Chris@82 371 Tw = Tu - Tv;
Chris@82 372 T1m = Tu + Tv;
Chris@82 373 }
Chris@82 374 Tt = Tp - Ts;
Chris@82 375 TA = Tw - Tz;
Chris@82 376 T1w = T1m + T1n;
Chris@82 377 T1v = T1j + T1k;
Chris@82 378 {
Chris@82 379 E T1l, T1o, TK, TL;
Chris@82 380 T1l = T1j - T1k;
Chris@82 381 T1o = T1m - T1n;
Chris@82 382 T1p = T1l + T1o;
Chris@82 383 T1E = KP559016994 * (T1l - T1o);
Chris@82 384 TK = Tp + Ts;
Chris@82 385 TL = Tw + Tz;
Chris@82 386 TM = TK + TL;
Chris@82 387 TO = KP559016994 * (TK - TL);
Chris@82 388 }
Chris@82 389 }
Chris@82 390 Rp[0] = T3 + Ti;
Chris@82 391 Rm[0] = TJ + TM;
Chris@82 392 {
Chris@82 393 E T1g, T1q, T17, T1h;
Chris@82 394 T1g = T18 + T1f;
Chris@82 395 T1q = T1i + T1p;
Chris@82 396 T17 = W[8];
Chris@82 397 T1h = W[9];
Chris@82 398 Ip[WS(rs, 2)] = FNMS(T1h, T1q, T17 * T1g);
Chris@82 399 Im[WS(rs, 2)] = FMA(T1h, T1g, T17 * T1q);
Chris@82 400 }
Chris@82 401 {
Chris@82 402 E TB, TG, T11, TX, TP, T10, Tm, TW, TN, Tk;
Chris@82 403 TB = FNMS(KP951056516, TA, KP587785252 * Tt);
Chris@82 404 TG = FNMS(KP951056516, TF, KP587785252 * TE);
Chris@82 405 T11 = FMA(KP951056516, TE, KP587785252 * TF);
Chris@82 406 TX = FMA(KP951056516, Tt, KP587785252 * TA);
Chris@82 407 TN = FNMS(KP250000000, TM, TJ);
Chris@82 408 TP = TN - TO;
Chris@82 409 T10 = TO + TN;
Chris@82 410 Tk = FNMS(KP250000000, Ti, T3);
Chris@82 411 Tm = Tk - Tl;
Chris@82 412 TW = Tl + Tk;
Chris@82 413 {
Chris@82 414 E TC, TQ, Tj, TD;
Chris@82 415 TC = Tm - TB;
Chris@82 416 TQ = TG + TP;
Chris@82 417 Tj = W[2];
Chris@82 418 TD = W[3];
Chris@82 419 Rp[WS(rs, 1)] = FNMS(TD, TQ, Tj * TC);
Chris@82 420 Rm[WS(rs, 1)] = FMA(TD, TC, Tj * TQ);
Chris@82 421 }
Chris@82 422 {
Chris@82 423 E T14, T16, T13, T15;
Chris@82 424 T14 = TW - TX;
Chris@82 425 T16 = T11 + T10;
Chris@82 426 T13 = W[10];
Chris@82 427 T15 = W[11];
Chris@82 428 Rp[WS(rs, 3)] = FNMS(T15, T16, T13 * T14);
Chris@82 429 Rm[WS(rs, 3)] = FMA(T15, T14, T13 * T16);
Chris@82 430 }
Chris@82 431 {
Chris@82 432 E TS, TU, TR, TT;
Chris@82 433 TS = Tm + TB;
Chris@82 434 TU = TP - TG;
Chris@82 435 TR = W[14];
Chris@82 436 TT = W[15];
Chris@82 437 Rp[WS(rs, 4)] = FNMS(TT, TU, TR * TS);
Chris@82 438 Rm[WS(rs, 4)] = FMA(TT, TS, TR * TU);
Chris@82 439 }
Chris@82 440 {
Chris@82 441 E TY, T12, TV, TZ;
Chris@82 442 TY = TW + TX;
Chris@82 443 T12 = T10 - T11;
Chris@82 444 TV = W[6];
Chris@82 445 TZ = W[7];
Chris@82 446 Rp[WS(rs, 2)] = FNMS(TZ, T12, TV * TY);
Chris@82 447 Rm[WS(rs, 2)] = FMA(TZ, TY, TV * T12);
Chris@82 448 }
Chris@82 449 }
Chris@82 450 {
Chris@82 451 E T1x, T1C, T1Q, T1N, T1F, T1R, T1u, T1M, T1D, T1s;
Chris@82 452 T1x = FNMS(KP951056516, T1w, KP587785252 * T1v);
Chris@82 453 T1C = FNMS(KP951056516, T1B, KP587785252 * T1A);
Chris@82 454 T1Q = FMA(KP951056516, T1A, KP587785252 * T1B);
Chris@82 455 T1N = FMA(KP951056516, T1v, KP587785252 * T1w);
Chris@82 456 T1D = FNMS(KP250000000, T1p, T1i);
Chris@82 457 T1F = T1D - T1E;
Chris@82 458 T1R = T1E + T1D;
Chris@82 459 T1s = FNMS(KP250000000, T1f, T18);
Chris@82 460 T1u = T1s - T1t;
Chris@82 461 T1M = T1t + T1s;
Chris@82 462 {
Chris@82 463 E T1y, T1G, T1r, T1z;
Chris@82 464 T1y = T1u - T1x;
Chris@82 465 T1G = T1C + T1F;
Chris@82 466 T1r = W[12];
Chris@82 467 T1z = W[13];
Chris@82 468 Ip[WS(rs, 3)] = FNMS(T1z, T1G, T1r * T1y);
Chris@82 469 Im[WS(rs, 3)] = FMA(T1r, T1G, T1z * T1y);
Chris@82 470 }
Chris@82 471 {
Chris@82 472 E T1U, T1W, T1T, T1V;
Chris@82 473 T1U = T1M + T1N;
Chris@82 474 T1W = T1R - T1Q;
Chris@82 475 T1T = W[16];
Chris@82 476 T1V = W[17];
Chris@82 477 Ip[WS(rs, 4)] = FNMS(T1V, T1W, T1T * T1U);
Chris@82 478 Im[WS(rs, 4)] = FMA(T1T, T1W, T1V * T1U);
Chris@82 479 }
Chris@82 480 {
Chris@82 481 E T1I, T1K, T1H, T1J;
Chris@82 482 T1I = T1u + T1x;
Chris@82 483 T1K = T1F - T1C;
Chris@82 484 T1H = W[4];
Chris@82 485 T1J = W[5];
Chris@82 486 Ip[WS(rs, 1)] = FNMS(T1J, T1K, T1H * T1I);
Chris@82 487 Im[WS(rs, 1)] = FMA(T1H, T1K, T1J * T1I);
Chris@82 488 }
Chris@82 489 {
Chris@82 490 E T1O, T1S, T1L, T1P;
Chris@82 491 T1O = T1M - T1N;
Chris@82 492 T1S = T1Q + T1R;
Chris@82 493 T1L = W[0];
Chris@82 494 T1P = W[1];
Chris@82 495 Ip[0] = FNMS(T1P, T1S, T1L * T1O);
Chris@82 496 Im[0] = FMA(T1L, T1S, T1P * T1O);
Chris@82 497 }
Chris@82 498 }
Chris@82 499 }
Chris@82 500 }
Chris@82 501 }
Chris@82 502
Chris@82 503 static const tw_instr twinstr[] = {
Chris@82 504 {TW_FULL, 1, 10},
Chris@82 505 {TW_NEXT, 1, 0}
Chris@82 506 };
Chris@82 507
Chris@82 508 static const hc2c_desc desc = { 10, "hc2cb_10", twinstr, &GENUS, {72, 30, 30, 0} };
Chris@82 509
Chris@82 510 void X(codelet_hc2cb_10) (planner *p) {
Chris@82 511 X(khc2c_register) (p, hc2cb_10, &desc, HC2C_VIA_RDFT);
Chris@82 512 }
Chris@82 513 #endif