Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cf/hc2cfdft2_16.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 08:07:16 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 -twiddle-log3 -precompute-twiddles -n 16 -dit -name hc2cfdft2_16 -include rdft/scalar/hc2cf.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 228 FP additions, 166 FP multiplications,
Chris@82 32 * (or, 136 additions, 74 multiplications, 92 fused multiply/add),
Chris@82 33 * 91 stack variables, 4 constants, and 64 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/hc2cf.h"
Chris@82 36
Chris@82 37 static void hc2cfdft2_16(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(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82 40 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@82 41 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 42 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 43 {
Chris@82 44 INT m;
Chris@82 45 for (m = mb, W = W + ((mb - 1) * 8); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 8, MAKE_VOLATILE_STRIDE(64, rs)) {
Chris@82 46 E T1, T2, Tw, Ty, Th, Tj, T4, T5, TY, T6, Tk, T1o, T1d, Tz, T1j;
Chris@82 47 E Tq, TF, T18, TR, TL, T13, T1A, T1K, T1E, T1H, Tc, T25, T2k, T29, T2h;
Chris@82 48 {
Chris@82 49 E Tx, TE, Ti, TK, Tp, TQ, Tb, T3;
Chris@82 50 T1 = W[0];
Chris@82 51 T2 = W[2];
Chris@82 52 T3 = T1 * T2;
Chris@82 53 Tw = W[6];
Chris@82 54 Tx = T1 * Tw;
Chris@82 55 Ty = W[7];
Chris@82 56 TE = T1 * Ty;
Chris@82 57 Th = W[4];
Chris@82 58 Ti = T1 * Th;
Chris@82 59 TK = T2 * Th;
Chris@82 60 Tj = W[5];
Chris@82 61 Tp = T1 * Tj;
Chris@82 62 TQ = T2 * Tj;
Chris@82 63 T4 = W[1];
Chris@82 64 T5 = W[3];
Chris@82 65 Tb = T1 * T5;
Chris@82 66 TY = FNMS(T4, T5, T3);
Chris@82 67 T6 = FMA(T4, T5, T3);
Chris@82 68 Tk = FNMS(T4, Tj, Ti);
Chris@82 69 T1o = FNMS(T4, Th, Tp);
Chris@82 70 T1d = FMA(T5, Th, TQ);
Chris@82 71 Tz = FMA(T4, Ty, Tx);
Chris@82 72 T1j = FMA(T4, Tj, Ti);
Chris@82 73 Tq = FMA(T4, Th, Tp);
Chris@82 74 TF = FNMS(T4, Tw, TE);
Chris@82 75 T18 = FNMS(T5, Tj, TK);
Chris@82 76 TR = FNMS(T5, Th, TQ);
Chris@82 77 TL = FMA(T5, Tj, TK);
Chris@82 78 {
Chris@82 79 E T1z, T1D, T24, T28;
Chris@82 80 T1z = TY * Th;
Chris@82 81 T1D = TY * Tj;
Chris@82 82 T13 = FMA(T4, T2, Tb);
Chris@82 83 T1A = FMA(T13, Tj, T1z);
Chris@82 84 T1K = FMA(T13, Th, T1D);
Chris@82 85 T1E = FNMS(T13, Th, T1D);
Chris@82 86 T1H = FNMS(T13, Tj, T1z);
Chris@82 87 T24 = T6 * Th;
Chris@82 88 T28 = T6 * Tj;
Chris@82 89 Tc = FNMS(T4, T2, Tb);
Chris@82 90 T25 = FNMS(Tc, Tj, T24);
Chris@82 91 T2k = FNMS(Tc, Th, T28);
Chris@82 92 T29 = FMA(Tc, Th, T28);
Chris@82 93 T2h = FMA(Tc, Tj, T24);
Chris@82 94 }
Chris@82 95 }
Chris@82 96 {
Chris@82 97 E T1v, T2q, T1s, T2s, T38, T3T, T1Y, T3P, T17, T1h, T2x, T2v, T33, T3Q, T1N;
Chris@82 98 E T3S, Tg, Tu, T3A, T2B, T2D, T3B, T2c, T3L, T2S, T3I, TJ, TV, T3E, T2G;
Chris@82 99 E T2I, T3D, T2n, T3J, T2X, T3M;
Chris@82 100 {
Chris@82 101 E T1t, T1u, T1W, T1m, T1Q, T1S, T1T, T1V, T36, T1r, T34, T1P, T1k, T1l, T1n;
Chris@82 102 E T2r;
Chris@82 103 T1t = Ip[0];
Chris@82 104 T1u = Im[0];
Chris@82 105 T1W = T1t + T1u;
Chris@82 106 T1k = Ip[WS(rs, 4)];
Chris@82 107 T1l = Im[WS(rs, 4)];
Chris@82 108 T1m = T1k - T1l;
Chris@82 109 T1Q = T1k + T1l;
Chris@82 110 {
Chris@82 111 E T1U, T1p, T1q, T1O;
Chris@82 112 T1S = Rm[0];
Chris@82 113 T1T = Rp[0];
Chris@82 114 T1U = T1S - T1T;
Chris@82 115 T1V = T1 * T1U;
Chris@82 116 T36 = T4 * T1U;
Chris@82 117 T1p = Rp[WS(rs, 4)];
Chris@82 118 T1q = Rm[WS(rs, 4)];
Chris@82 119 T1O = T1q - T1p;
Chris@82 120 T1r = T1p + T1q;
Chris@82 121 T34 = Tj * T1O;
Chris@82 122 T1P = Th * T1O;
Chris@82 123 }
Chris@82 124 T1v = T1t - T1u;
Chris@82 125 T2q = T1T + T1S;
Chris@82 126 T1n = T1j * T1m;
Chris@82 127 T1s = FNMS(T1o, T1r, T1n);
Chris@82 128 T2r = T1j * T1r;
Chris@82 129 T2s = FMA(T1o, T1m, T2r);
Chris@82 130 {
Chris@82 131 E T35, T37, T1R, T1X;
Chris@82 132 T35 = FMA(Th, T1Q, T34);
Chris@82 133 T37 = FMA(T1, T1W, T36);
Chris@82 134 T38 = T35 + T37;
Chris@82 135 T3T = T37 - T35;
Chris@82 136 T1R = FNMS(Tj, T1Q, T1P);
Chris@82 137 T1X = FNMS(T4, T1W, T1V);
Chris@82 138 T1Y = T1R + T1X;
Chris@82 139 T3P = T1X - T1R;
Chris@82 140 }
Chris@82 141 }
Chris@82 142 {
Chris@82 143 E T11, T1F, T16, T2Z, T1C, T1b, T1L, T1g, T31, T1J;
Chris@82 144 {
Chris@82 145 E TZ, T10, T14, T15, T1B;
Chris@82 146 TZ = Ip[WS(rs, 2)];
Chris@82 147 T10 = Im[WS(rs, 2)];
Chris@82 148 T11 = TZ - T10;
Chris@82 149 T1F = TZ + T10;
Chris@82 150 T14 = Rp[WS(rs, 2)];
Chris@82 151 T15 = Rm[WS(rs, 2)];
Chris@82 152 T1B = T15 - T14;
Chris@82 153 T16 = T14 + T15;
Chris@82 154 T2Z = T1E * T1B;
Chris@82 155 T1C = T1A * T1B;
Chris@82 156 }
Chris@82 157 {
Chris@82 158 E T19, T1a, T1e, T1f, T1I;
Chris@82 159 T19 = Ip[WS(rs, 6)];
Chris@82 160 T1a = Im[WS(rs, 6)];
Chris@82 161 T1b = T19 - T1a;
Chris@82 162 T1L = T19 + T1a;
Chris@82 163 T1e = Rp[WS(rs, 6)];
Chris@82 164 T1f = Rm[WS(rs, 6)];
Chris@82 165 T1I = T1f - T1e;
Chris@82 166 T1g = T1e + T1f;
Chris@82 167 T31 = T1K * T1I;
Chris@82 168 T1J = T1H * T1I;
Chris@82 169 }
Chris@82 170 {
Chris@82 171 E T12, T1c, T2w, T2u;
Chris@82 172 T12 = TY * T11;
Chris@82 173 T17 = FNMS(T13, T16, T12);
Chris@82 174 T1c = T18 * T1b;
Chris@82 175 T1h = FNMS(T1d, T1g, T1c);
Chris@82 176 T2w = T18 * T1g;
Chris@82 177 T2x = FMA(T1d, T1b, T2w);
Chris@82 178 T2u = TY * T16;
Chris@82 179 T2v = FMA(T13, T11, T2u);
Chris@82 180 {
Chris@82 181 E T30, T32, T1G, T1M;
Chris@82 182 T30 = FMA(T1A, T1F, T2Z);
Chris@82 183 T32 = FMA(T1H, T1L, T31);
Chris@82 184 T33 = T30 + T32;
Chris@82 185 T3Q = T30 - T32;
Chris@82 186 T1G = FNMS(T1E, T1F, T1C);
Chris@82 187 T1M = FNMS(T1K, T1L, T1J);
Chris@82 188 T1N = T1G + T1M;
Chris@82 189 T3S = T1G - T1M;
Chris@82 190 }
Chris@82 191 }
Chris@82 192 }
Chris@82 193 {
Chris@82 194 E T9, T22, Ta, T2O, Tf, T20, T21, T2A, Tn, T2a, To, T2Q, Tt, T26, T27;
Chris@82 195 E T2C;
Chris@82 196 {
Chris@82 197 E T7, T8, Td, Te;
Chris@82 198 T7 = Ip[WS(rs, 1)];
Chris@82 199 T8 = Im[WS(rs, 1)];
Chris@82 200 T9 = T7 - T8;
Chris@82 201 T22 = T7 + T8;
Chris@82 202 Ta = T6 * T9;
Chris@82 203 T2O = T2 * T22;
Chris@82 204 Td = Rp[WS(rs, 1)];
Chris@82 205 Te = Rm[WS(rs, 1)];
Chris@82 206 Tf = Td + Te;
Chris@82 207 T20 = Td - Te;
Chris@82 208 T21 = T2 * T20;
Chris@82 209 T2A = T6 * Tf;
Chris@82 210 }
Chris@82 211 {
Chris@82 212 E Tl, Tm, Tr, Ts;
Chris@82 213 Tl = Ip[WS(rs, 5)];
Chris@82 214 Tm = Im[WS(rs, 5)];
Chris@82 215 Tn = Tl - Tm;
Chris@82 216 T2a = Tl + Tm;
Chris@82 217 To = Tk * Tn;
Chris@82 218 T2Q = T25 * T2a;
Chris@82 219 Tr = Rp[WS(rs, 5)];
Chris@82 220 Ts = Rm[WS(rs, 5)];
Chris@82 221 Tt = Tr + Ts;
Chris@82 222 T26 = Tr - Ts;
Chris@82 223 T27 = T25 * T26;
Chris@82 224 T2C = Tk * Tt;
Chris@82 225 }
Chris@82 226 Tg = FNMS(Tc, Tf, Ta);
Chris@82 227 Tu = FNMS(Tq, Tt, To);
Chris@82 228 T3A = Tg - Tu;
Chris@82 229 T2B = FMA(Tc, T9, T2A);
Chris@82 230 T2D = FMA(Tq, Tn, T2C);
Chris@82 231 T3B = T2B - T2D;
Chris@82 232 {
Chris@82 233 E T23, T2b, T2P, T2R;
Chris@82 234 T23 = FMA(T5, T22, T21);
Chris@82 235 T2b = FMA(T29, T2a, T27);
Chris@82 236 T2c = T23 + T2b;
Chris@82 237 T3L = T2b - T23;
Chris@82 238 T2P = FNMS(T5, T20, T2O);
Chris@82 239 T2R = FNMS(T29, T26, T2Q);
Chris@82 240 T2S = T2P + T2R;
Chris@82 241 T3I = T2R - T2P;
Chris@82 242 }
Chris@82 243 }
Chris@82 244 {
Chris@82 245 E TC, T2f, TD, T2T, TI, T2d, T2e, T2F, TO, T2l, TP, T2V, TU, T2i, T2j;
Chris@82 246 E T2H;
Chris@82 247 {
Chris@82 248 E TA, TB, TG, TH;
Chris@82 249 TA = Ip[WS(rs, 7)];
Chris@82 250 TB = Im[WS(rs, 7)];
Chris@82 251 TC = TA - TB;
Chris@82 252 T2f = TA + TB;
Chris@82 253 TD = Tz * TC;
Chris@82 254 T2T = Tw * T2f;
Chris@82 255 TG = Rp[WS(rs, 7)];
Chris@82 256 TH = Rm[WS(rs, 7)];
Chris@82 257 TI = TG + TH;
Chris@82 258 T2d = TG - TH;
Chris@82 259 T2e = Tw * T2d;
Chris@82 260 T2F = Tz * TI;
Chris@82 261 }
Chris@82 262 {
Chris@82 263 E TM, TN, TS, TT;
Chris@82 264 TM = Ip[WS(rs, 3)];
Chris@82 265 TN = Im[WS(rs, 3)];
Chris@82 266 TO = TM - TN;
Chris@82 267 T2l = TM + TN;
Chris@82 268 TP = TL * TO;
Chris@82 269 T2V = T2h * T2l;
Chris@82 270 TS = Rp[WS(rs, 3)];
Chris@82 271 TT = Rm[WS(rs, 3)];
Chris@82 272 TU = TS + TT;
Chris@82 273 T2i = TS - TT;
Chris@82 274 T2j = T2h * T2i;
Chris@82 275 T2H = TL * TU;
Chris@82 276 }
Chris@82 277 TJ = FNMS(TF, TI, TD);
Chris@82 278 TV = FNMS(TR, TU, TP);
Chris@82 279 T3E = TJ - TV;
Chris@82 280 T2G = FMA(TF, TC, T2F);
Chris@82 281 T2I = FMA(TR, TO, T2H);
Chris@82 282 T3D = T2G - T2I;
Chris@82 283 {
Chris@82 284 E T2g, T2m, T2U, T2W;
Chris@82 285 T2g = FMA(Ty, T2f, T2e);
Chris@82 286 T2m = FMA(T2k, T2l, T2j);
Chris@82 287 T2n = T2g + T2m;
Chris@82 288 T3J = T2m - T2g;
Chris@82 289 T2U = FNMS(Ty, T2d, T2T);
Chris@82 290 T2W = FNMS(T2k, T2i, T2V);
Chris@82 291 T2X = T2U + T2W;
Chris@82 292 T3M = T2U - T2W;
Chris@82 293 }
Chris@82 294 }
Chris@82 295 {
Chris@82 296 E TX, T3o, T3i, T3s, T3l, T3t, T1x, T3e, T2p, T2M, T2K, T3d, T3a, T3c, T2z;
Chris@82 297 E T3n;
Chris@82 298 {
Chris@82 299 E Tv, TW, T3g, T3h;
Chris@82 300 Tv = Tg + Tu;
Chris@82 301 TW = TJ + TV;
Chris@82 302 TX = Tv + TW;
Chris@82 303 T3o = Tv - TW;
Chris@82 304 T3g = T2X - T2S;
Chris@82 305 T3h = T2c - T2n;
Chris@82 306 T3i = T3g + T3h;
Chris@82 307 T3s = T3g - T3h;
Chris@82 308 }
Chris@82 309 {
Chris@82 310 E T3j, T3k, T1i, T1w;
Chris@82 311 T3j = T1Y - T1N;
Chris@82 312 T3k = T38 - T33;
Chris@82 313 T3l = T3j - T3k;
Chris@82 314 T3t = T3j + T3k;
Chris@82 315 T1i = T17 + T1h;
Chris@82 316 T1w = T1s + T1v;
Chris@82 317 T1x = T1i + T1w;
Chris@82 318 T3e = T1w - T1i;
Chris@82 319 }
Chris@82 320 {
Chris@82 321 E T1Z, T2o, T2E, T2J;
Chris@82 322 T1Z = T1N + T1Y;
Chris@82 323 T2o = T2c + T2n;
Chris@82 324 T2p = T1Z - T2o;
Chris@82 325 T2M = T2o + T1Z;
Chris@82 326 T2E = T2B + T2D;
Chris@82 327 T2J = T2G + T2I;
Chris@82 328 T2K = T2E + T2J;
Chris@82 329 T3d = T2J - T2E;
Chris@82 330 }
Chris@82 331 {
Chris@82 332 E T2Y, T39, T2t, T2y;
Chris@82 333 T2Y = T2S + T2X;
Chris@82 334 T39 = T33 + T38;
Chris@82 335 T3a = T2Y - T39;
Chris@82 336 T3c = T2Y + T39;
Chris@82 337 T2t = T2q + T2s;
Chris@82 338 T2y = T2v + T2x;
Chris@82 339 T2z = T2t + T2y;
Chris@82 340 T3n = T2t - T2y;
Chris@82 341 }
Chris@82 342 {
Chris@82 343 E T1y, T3b, T2L, T2N;
Chris@82 344 T1y = TX + T1x;
Chris@82 345 Ip[0] = KP500000000 * (T1y + T2p);
Chris@82 346 Im[WS(rs, 7)] = KP500000000 * (T2p - T1y);
Chris@82 347 T3b = T2z + T2K;
Chris@82 348 Rm[WS(rs, 7)] = KP500000000 * (T3b - T3c);
Chris@82 349 Rp[0] = KP500000000 * (T3b + T3c);
Chris@82 350 T2L = T2z - T2K;
Chris@82 351 Rm[WS(rs, 3)] = KP500000000 * (T2L - T2M);
Chris@82 352 Rp[WS(rs, 4)] = KP500000000 * (T2L + T2M);
Chris@82 353 T2N = T1x - TX;
Chris@82 354 Ip[WS(rs, 4)] = KP500000000 * (T2N + T3a);
Chris@82 355 Im[WS(rs, 3)] = KP500000000 * (T3a - T2N);
Chris@82 356 }
Chris@82 357 {
Chris@82 358 E T3f, T3m, T3v, T3w;
Chris@82 359 T3f = T3d + T3e;
Chris@82 360 T3m = T3i + T3l;
Chris@82 361 Ip[WS(rs, 2)] = KP500000000 * (FMA(KP707106781, T3m, T3f));
Chris@82 362 Im[WS(rs, 5)] = -(KP500000000 * (FNMS(KP707106781, T3m, T3f)));
Chris@82 363 T3v = T3n + T3o;
Chris@82 364 T3w = T3s + T3t;
Chris@82 365 Rm[WS(rs, 5)] = KP500000000 * (FNMS(KP707106781, T3w, T3v));
Chris@82 366 Rp[WS(rs, 2)] = KP500000000 * (FMA(KP707106781, T3w, T3v));
Chris@82 367 }
Chris@82 368 {
Chris@82 369 E T3p, T3q, T3r, T3u;
Chris@82 370 T3p = T3n - T3o;
Chris@82 371 T3q = T3l - T3i;
Chris@82 372 Rm[WS(rs, 1)] = KP500000000 * (FNMS(KP707106781, T3q, T3p));
Chris@82 373 Rp[WS(rs, 6)] = KP500000000 * (FMA(KP707106781, T3q, T3p));
Chris@82 374 T3r = T3e - T3d;
Chris@82 375 T3u = T3s - T3t;
Chris@82 376 Ip[WS(rs, 6)] = KP500000000 * (FMA(KP707106781, T3u, T3r));
Chris@82 377 Im[WS(rs, 1)] = -(KP500000000 * (FNMS(KP707106781, T3u, T3r)));
Chris@82 378 }
Chris@82 379 }
Chris@82 380 {
Chris@82 381 E T3z, T4b, T4g, T4q, T4j, T4r, T3G, T4m, T3O, T46, T3Z, T4l, T42, T4c, T3V;
Chris@82 382 E T47;
Chris@82 383 {
Chris@82 384 E T3x, T3y, T4e, T4f;
Chris@82 385 T3x = T1v - T1s;
Chris@82 386 T3y = T2v - T2x;
Chris@82 387 T3z = T3x - T3y;
Chris@82 388 T4b = T3y + T3x;
Chris@82 389 T4e = T3I - T3J;
Chris@82 390 T4f = T3M - T3L;
Chris@82 391 T4g = FMA(KP414213562, T4f, T4e);
Chris@82 392 T4q = FNMS(KP414213562, T4e, T4f);
Chris@82 393 }
Chris@82 394 {
Chris@82 395 E T4h, T4i, T3C, T3F;
Chris@82 396 T4h = T3Q + T3P;
Chris@82 397 T4i = T3T - T3S;
Chris@82 398 T4j = FMA(KP414213562, T4i, T4h);
Chris@82 399 T4r = FNMS(KP414213562, T4h, T4i);
Chris@82 400 T3C = T3A - T3B;
Chris@82 401 T3F = T3D + T3E;
Chris@82 402 T3G = T3C + T3F;
Chris@82 403 T4m = T3C - T3F;
Chris@82 404 }
Chris@82 405 {
Chris@82 406 E T3K, T3N, T3X, T3Y;
Chris@82 407 T3K = T3I + T3J;
Chris@82 408 T3N = T3L + T3M;
Chris@82 409 T3O = FMA(KP414213562, T3N, T3K);
Chris@82 410 T46 = FNMS(KP414213562, T3K, T3N);
Chris@82 411 T3X = T2q - T2s;
Chris@82 412 T3Y = T17 - T1h;
Chris@82 413 T3Z = T3X + T3Y;
Chris@82 414 T4l = T3X - T3Y;
Chris@82 415 }
Chris@82 416 {
Chris@82 417 E T40, T41, T3R, T3U;
Chris@82 418 T40 = T3B + T3A;
Chris@82 419 T41 = T3D - T3E;
Chris@82 420 T42 = T40 + T41;
Chris@82 421 T4c = T41 - T40;
Chris@82 422 T3R = T3P - T3Q;
Chris@82 423 T3U = T3S + T3T;
Chris@82 424 T3V = FNMS(KP414213562, T3U, T3R);
Chris@82 425 T47 = FMA(KP414213562, T3R, T3U);
Chris@82 426 }
Chris@82 427 {
Chris@82 428 E T3H, T3W, T49, T4a;
Chris@82 429 T3H = FMA(KP707106781, T3G, T3z);
Chris@82 430 T3W = T3O + T3V;
Chris@82 431 Ip[WS(rs, 1)] = KP500000000 * (FMA(KP923879532, T3W, T3H));
Chris@82 432 Im[WS(rs, 6)] = -(KP500000000 * (FNMS(KP923879532, T3W, T3H)));
Chris@82 433 T49 = FMA(KP707106781, T42, T3Z);
Chris@82 434 T4a = T46 + T47;
Chris@82 435 Rm[WS(rs, 6)] = KP500000000 * (FNMS(KP923879532, T4a, T49));
Chris@82 436 Rp[WS(rs, 1)] = KP500000000 * (FMA(KP923879532, T4a, T49));
Chris@82 437 }
Chris@82 438 {
Chris@82 439 E T43, T44, T45, T48;
Chris@82 440 T43 = FNMS(KP707106781, T42, T3Z);
Chris@82 441 T44 = T3V - T3O;
Chris@82 442 Rm[WS(rs, 2)] = KP500000000 * (FNMS(KP923879532, T44, T43));
Chris@82 443 Rp[WS(rs, 5)] = KP500000000 * (FMA(KP923879532, T44, T43));
Chris@82 444 T45 = FNMS(KP707106781, T3G, T3z);
Chris@82 445 T48 = T46 - T47;
Chris@82 446 Ip[WS(rs, 5)] = KP500000000 * (FMA(KP923879532, T48, T45));
Chris@82 447 Im[WS(rs, 2)] = -(KP500000000 * (FNMS(KP923879532, T48, T45)));
Chris@82 448 }
Chris@82 449 {
Chris@82 450 E T4d, T4k, T4t, T4u;
Chris@82 451 T4d = FNMS(KP707106781, T4c, T4b);
Chris@82 452 T4k = T4g - T4j;
Chris@82 453 Ip[WS(rs, 7)] = KP500000000 * (FMA(KP923879532, T4k, T4d));
Chris@82 454 Im[0] = -(KP500000000 * (FNMS(KP923879532, T4k, T4d)));
Chris@82 455 T4t = FNMS(KP707106781, T4m, T4l);
Chris@82 456 T4u = T4q + T4r;
Chris@82 457 Rp[WS(rs, 7)] = KP500000000 * (FNMS(KP923879532, T4u, T4t));
Chris@82 458 Rm[0] = KP500000000 * (FMA(KP923879532, T4u, T4t));
Chris@82 459 }
Chris@82 460 {
Chris@82 461 E T4n, T4o, T4p, T4s;
Chris@82 462 T4n = FMA(KP707106781, T4m, T4l);
Chris@82 463 T4o = T4g + T4j;
Chris@82 464 Rm[WS(rs, 4)] = KP500000000 * (FNMS(KP923879532, T4o, T4n));
Chris@82 465 Rp[WS(rs, 3)] = KP500000000 * (FMA(KP923879532, T4o, T4n));
Chris@82 466 T4p = FMA(KP707106781, T4c, T4b);
Chris@82 467 T4s = T4q - T4r;
Chris@82 468 Ip[WS(rs, 3)] = KP500000000 * (FMA(KP923879532, T4s, T4p));
Chris@82 469 Im[WS(rs, 4)] = -(KP500000000 * (FNMS(KP923879532, T4s, T4p)));
Chris@82 470 }
Chris@82 471 }
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_CEXP, 1, 1},
Chris@82 479 {TW_CEXP, 1, 3},
Chris@82 480 {TW_CEXP, 1, 9},
Chris@82 481 {TW_CEXP, 1, 15},
Chris@82 482 {TW_NEXT, 1, 0}
Chris@82 483 };
Chris@82 484
Chris@82 485 static const hc2c_desc desc = { 16, "hc2cfdft2_16", twinstr, &GENUS, {136, 74, 92, 0} };
Chris@82 486
Chris@82 487 void X(codelet_hc2cfdft2_16) (planner *p) {
Chris@82 488 X(khc2c_register) (p, hc2cfdft2_16, &desc, HC2C_VIA_DFT);
Chris@82 489 }
Chris@82 490 #else
Chris@82 491
Chris@82 492 /* Generated by: ../../../genfft/gen_hc2cdft.native -compact -variables 4 -pipeline-latency 4 -twiddle-log3 -precompute-twiddles -n 16 -dit -name hc2cfdft2_16 -include rdft/scalar/hc2cf.h */
Chris@82 493
Chris@82 494 /*
Chris@82 495 * This function contains 228 FP additions, 124 FP multiplications,
Chris@82 496 * (or, 188 additions, 84 multiplications, 40 fused multiply/add),
Chris@82 497 * 91 stack variables, 4 constants, and 64 memory accesses
Chris@82 498 */
Chris@82 499 #include "rdft/scalar/hc2cf.h"
Chris@82 500
Chris@82 501 static void hc2cfdft2_16(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 502 {
Chris@82 503 DK(KP461939766, +0.461939766255643378064091594698394143411208313);
Chris@82 504 DK(KP191341716, +0.191341716182544885864229992015199433380672281);
Chris@82 505 DK(KP353553390, +0.353553390593273762200422181052424519642417969);
Chris@82 506 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 507 {
Chris@82 508 INT m;
Chris@82 509 for (m = mb, W = W + ((mb - 1) * 8); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 8, MAKE_VOLATILE_STRIDE(64, rs)) {
Chris@82 510 E T1, T4, T2, T5, T7, Td, T12, TY, Tk, Ti, Tm, T1l, T1b, TL, T1h;
Chris@82 511 E Ts, TR, T17, Ty, Tz, TA, TE, T1L, T1Q, T1H, T1O, T24, T2d, T20, T2b;
Chris@82 512 {
Chris@82 513 E Tl, TP, Tq, TK, Tj, TQ, Tr, TJ;
Chris@82 514 {
Chris@82 515 E T3, Tc, T6, Tb;
Chris@82 516 T1 = W[0];
Chris@82 517 T4 = W[1];
Chris@82 518 T2 = W[2];
Chris@82 519 T5 = W[3];
Chris@82 520 T3 = T1 * T2;
Chris@82 521 Tc = T4 * T2;
Chris@82 522 T6 = T4 * T5;
Chris@82 523 Tb = T1 * T5;
Chris@82 524 T7 = T3 + T6;
Chris@82 525 Td = Tb - Tc;
Chris@82 526 T12 = Tb + Tc;
Chris@82 527 TY = T3 - T6;
Chris@82 528 Tk = W[5];
Chris@82 529 Tl = T4 * Tk;
Chris@82 530 TP = T2 * Tk;
Chris@82 531 Tq = T1 * Tk;
Chris@82 532 TK = T5 * Tk;
Chris@82 533 Ti = W[4];
Chris@82 534 Tj = T1 * Ti;
Chris@82 535 TQ = T5 * Ti;
Chris@82 536 Tr = T4 * Ti;
Chris@82 537 TJ = T2 * Ti;
Chris@82 538 }
Chris@82 539 Tm = Tj - Tl;
Chris@82 540 T1l = Tq - Tr;
Chris@82 541 T1b = TP + TQ;
Chris@82 542 TL = TJ + TK;
Chris@82 543 T1h = Tj + Tl;
Chris@82 544 Ts = Tq + Tr;
Chris@82 545 TR = TP - TQ;
Chris@82 546 T17 = TJ - TK;
Chris@82 547 Ty = W[6];
Chris@82 548 Tz = W[7];
Chris@82 549 TA = FMA(T1, Ty, T4 * Tz);
Chris@82 550 TE = FNMS(T4, Ty, T1 * Tz);
Chris@82 551 {
Chris@82 552 E T1J, T1K, T1F, T1G;
Chris@82 553 T1J = TY * Tk;
Chris@82 554 T1K = T12 * Ti;
Chris@82 555 T1L = T1J - T1K;
Chris@82 556 T1Q = T1J + T1K;
Chris@82 557 T1F = TY * Ti;
Chris@82 558 T1G = T12 * Tk;
Chris@82 559 T1H = T1F + T1G;
Chris@82 560 T1O = T1F - T1G;
Chris@82 561 }
Chris@82 562 {
Chris@82 563 E T22, T23, T1Y, T1Z;
Chris@82 564 T22 = T7 * Tk;
Chris@82 565 T23 = Td * Ti;
Chris@82 566 T24 = T22 + T23;
Chris@82 567 T2d = T22 - T23;
Chris@82 568 T1Y = T7 * Ti;
Chris@82 569 T1Z = Td * Tk;
Chris@82 570 T20 = T1Y - T1Z;
Chris@82 571 T2b = T1Y + T1Z;
Chris@82 572 }
Chris@82 573 }
Chris@82 574 {
Chris@82 575 E T1t, T3i, T2l, T3B, T1E, T3t, T2M, T3x, T1g, T3C, T2J, T3u, T1T, T3w, T2o;
Chris@82 576 E T3j, Tx, T3b, T2C, T3q, T27, T3m, T2s, T3c, TW, T3f, T2F, T3n, T2g, T3p;
Chris@82 577 E T2v, T3e;
Chris@82 578 {
Chris@82 579 E T1k, T1C, T1o, T1B, T1s, T1z, T1y, T2j, T1p, T2k;
Chris@82 580 {
Chris@82 581 E T1i, T1j, T1m, T1n;
Chris@82 582 T1i = Ip[WS(rs, 4)];
Chris@82 583 T1j = Im[WS(rs, 4)];
Chris@82 584 T1k = T1i - T1j;
Chris@82 585 T1C = T1i + T1j;
Chris@82 586 T1m = Rp[WS(rs, 4)];
Chris@82 587 T1n = Rm[WS(rs, 4)];
Chris@82 588 T1o = T1m + T1n;
Chris@82 589 T1B = T1m - T1n;
Chris@82 590 }
Chris@82 591 {
Chris@82 592 E T1q, T1r, T1w, T1x;
Chris@82 593 T1q = Ip[0];
Chris@82 594 T1r = Im[0];
Chris@82 595 T1s = T1q - T1r;
Chris@82 596 T1z = T1q + T1r;
Chris@82 597 T1w = Rm[0];
Chris@82 598 T1x = Rp[0];
Chris@82 599 T1y = T1w - T1x;
Chris@82 600 T2j = T1x + T1w;
Chris@82 601 }
Chris@82 602 T1p = FNMS(T1l, T1o, T1h * T1k);
Chris@82 603 T1t = T1p + T1s;
Chris@82 604 T3i = T1s - T1p;
Chris@82 605 T2k = FMA(T1h, T1o, T1l * T1k);
Chris@82 606 T2l = T2j + T2k;
Chris@82 607 T3B = T2j - T2k;
Chris@82 608 {
Chris@82 609 E T1A, T1D, T2K, T2L;
Chris@82 610 T1A = FNMS(T4, T1z, T1 * T1y);
Chris@82 611 T1D = FMA(Ti, T1B, Tk * T1C);
Chris@82 612 T1E = T1A - T1D;
Chris@82 613 T3t = T1D + T1A;
Chris@82 614 T2K = FNMS(Tk, T1B, Ti * T1C);
Chris@82 615 T2L = FMA(T4, T1y, T1 * T1z);
Chris@82 616 T2M = T2K + T2L;
Chris@82 617 T3x = T2L - T2K;
Chris@82 618 }
Chris@82 619 }
Chris@82 620 {
Chris@82 621 E T11, T1M, T15, T1I, T1a, T1R, T1e, T1P;
Chris@82 622 {
Chris@82 623 E TZ, T10, T13, T14;
Chris@82 624 TZ = Ip[WS(rs, 2)];
Chris@82 625 T10 = Im[WS(rs, 2)];
Chris@82 626 T11 = TZ - T10;
Chris@82 627 T1M = TZ + T10;
Chris@82 628 T13 = Rp[WS(rs, 2)];
Chris@82 629 T14 = Rm[WS(rs, 2)];
Chris@82 630 T15 = T13 + T14;
Chris@82 631 T1I = T13 - T14;
Chris@82 632 }
Chris@82 633 {
Chris@82 634 E T18, T19, T1c, T1d;
Chris@82 635 T18 = Ip[WS(rs, 6)];
Chris@82 636 T19 = Im[WS(rs, 6)];
Chris@82 637 T1a = T18 - T19;
Chris@82 638 T1R = T18 + T19;
Chris@82 639 T1c = Rp[WS(rs, 6)];
Chris@82 640 T1d = Rm[WS(rs, 6)];
Chris@82 641 T1e = T1c + T1d;
Chris@82 642 T1P = T1c - T1d;
Chris@82 643 }
Chris@82 644 {
Chris@82 645 E T16, T1f, T2H, T2I;
Chris@82 646 T16 = FNMS(T12, T15, TY * T11);
Chris@82 647 T1f = FNMS(T1b, T1e, T17 * T1a);
Chris@82 648 T1g = T16 + T1f;
Chris@82 649 T3C = T16 - T1f;
Chris@82 650 T2H = FNMS(T1L, T1I, T1H * T1M);
Chris@82 651 T2I = FNMS(T1Q, T1P, T1O * T1R);
Chris@82 652 T2J = T2H + T2I;
Chris@82 653 T3u = T2H - T2I;
Chris@82 654 }
Chris@82 655 {
Chris@82 656 E T1N, T1S, T2m, T2n;
Chris@82 657 T1N = FMA(T1H, T1I, T1L * T1M);
Chris@82 658 T1S = FMA(T1O, T1P, T1Q * T1R);
Chris@82 659 T1T = T1N + T1S;
Chris@82 660 T3w = T1S - T1N;
Chris@82 661 T2m = FMA(TY, T15, T12 * T11);
Chris@82 662 T2n = FMA(T17, T1e, T1b * T1a);
Chris@82 663 T2o = T2m + T2n;
Chris@82 664 T3j = T2m - T2n;
Chris@82 665 }
Chris@82 666 }
Chris@82 667 {
Chris@82 668 E Ta, T1W, Tg, T1V, Tp, T25, Tv, T21;
Chris@82 669 {
Chris@82 670 E T8, T9, Te, Tf;
Chris@82 671 T8 = Ip[WS(rs, 1)];
Chris@82 672 T9 = Im[WS(rs, 1)];
Chris@82 673 Ta = T8 - T9;
Chris@82 674 T1W = T8 + T9;
Chris@82 675 Te = Rp[WS(rs, 1)];
Chris@82 676 Tf = Rm[WS(rs, 1)];
Chris@82 677 Tg = Te + Tf;
Chris@82 678 T1V = Te - Tf;
Chris@82 679 }
Chris@82 680 {
Chris@82 681 E Tn, To, Tt, Tu;
Chris@82 682 Tn = Ip[WS(rs, 5)];
Chris@82 683 To = Im[WS(rs, 5)];
Chris@82 684 Tp = Tn - To;
Chris@82 685 T25 = Tn + To;
Chris@82 686 Tt = Rp[WS(rs, 5)];
Chris@82 687 Tu = Rm[WS(rs, 5)];
Chris@82 688 Tv = Tt + Tu;
Chris@82 689 T21 = Tt - Tu;
Chris@82 690 }
Chris@82 691 {
Chris@82 692 E Th, Tw, T2A, T2B;
Chris@82 693 Th = FNMS(Td, Tg, T7 * Ta);
Chris@82 694 Tw = FNMS(Ts, Tv, Tm * Tp);
Chris@82 695 Tx = Th + Tw;
Chris@82 696 T3b = Th - Tw;
Chris@82 697 T2A = FNMS(T5, T1V, T2 * T1W);
Chris@82 698 T2B = FNMS(T24, T21, T20 * T25);
Chris@82 699 T2C = T2A + T2B;
Chris@82 700 T3q = T2A - T2B;
Chris@82 701 }
Chris@82 702 {
Chris@82 703 E T1X, T26, T2q, T2r;
Chris@82 704 T1X = FMA(T2, T1V, T5 * T1W);
Chris@82 705 T26 = FMA(T20, T21, T24 * T25);
Chris@82 706 T27 = T1X + T26;
Chris@82 707 T3m = T26 - T1X;
Chris@82 708 T2q = FMA(T7, Tg, Td * Ta);
Chris@82 709 T2r = FMA(Tm, Tv, Ts * Tp);
Chris@82 710 T2s = T2q + T2r;
Chris@82 711 T3c = T2q - T2r;
Chris@82 712 }
Chris@82 713 }
Chris@82 714 {
Chris@82 715 E TD, T29, TH, T28, TO, T2e, TU, T2c;
Chris@82 716 {
Chris@82 717 E TB, TC, TF, TG;
Chris@82 718 TB = Ip[WS(rs, 7)];
Chris@82 719 TC = Im[WS(rs, 7)];
Chris@82 720 TD = TB - TC;
Chris@82 721 T29 = TB + TC;
Chris@82 722 TF = Rp[WS(rs, 7)];
Chris@82 723 TG = Rm[WS(rs, 7)];
Chris@82 724 TH = TF + TG;
Chris@82 725 T28 = TF - TG;
Chris@82 726 }
Chris@82 727 {
Chris@82 728 E TM, TN, TS, TT;
Chris@82 729 TM = Ip[WS(rs, 3)];
Chris@82 730 TN = Im[WS(rs, 3)];
Chris@82 731 TO = TM - TN;
Chris@82 732 T2e = TM + TN;
Chris@82 733 TS = Rp[WS(rs, 3)];
Chris@82 734 TT = Rm[WS(rs, 3)];
Chris@82 735 TU = TS + TT;
Chris@82 736 T2c = TS - TT;
Chris@82 737 }
Chris@82 738 {
Chris@82 739 E TI, TV, T2D, T2E;
Chris@82 740 TI = FNMS(TE, TH, TA * TD);
Chris@82 741 TV = FNMS(TR, TU, TL * TO);
Chris@82 742 TW = TI + TV;
Chris@82 743 T3f = TI - TV;
Chris@82 744 T2D = FNMS(Tz, T28, Ty * T29);
Chris@82 745 T2E = FNMS(T2d, T2c, T2b * T2e);
Chris@82 746 T2F = T2D + T2E;
Chris@82 747 T3n = T2D - T2E;
Chris@82 748 }
Chris@82 749 {
Chris@82 750 E T2a, T2f, T2t, T2u;
Chris@82 751 T2a = FMA(Ty, T28, Tz * T29);
Chris@82 752 T2f = FMA(T2b, T2c, T2d * T2e);
Chris@82 753 T2g = T2a + T2f;
Chris@82 754 T3p = T2f - T2a;
Chris@82 755 T2t = FMA(TA, TH, TE * TD);
Chris@82 756 T2u = FMA(TL, TU, TR * TO);
Chris@82 757 T2v = T2t + T2u;
Chris@82 758 T3e = T2t - T2u;
Chris@82 759 }
Chris@82 760 }
Chris@82 761 {
Chris@82 762 E T1v, T2z, T2O, T2Q, T2i, T2y, T2x, T2P;
Chris@82 763 {
Chris@82 764 E TX, T1u, T2G, T2N;
Chris@82 765 TX = Tx + TW;
Chris@82 766 T1u = T1g + T1t;
Chris@82 767 T1v = TX + T1u;
Chris@82 768 T2z = T1u - TX;
Chris@82 769 T2G = T2C + T2F;
Chris@82 770 T2N = T2J + T2M;
Chris@82 771 T2O = T2G - T2N;
Chris@82 772 T2Q = T2G + T2N;
Chris@82 773 }
Chris@82 774 {
Chris@82 775 E T1U, T2h, T2p, T2w;
Chris@82 776 T1U = T1E - T1T;
Chris@82 777 T2h = T27 + T2g;
Chris@82 778 T2i = T1U - T2h;
Chris@82 779 T2y = T2h + T1U;
Chris@82 780 T2p = T2l + T2o;
Chris@82 781 T2w = T2s + T2v;
Chris@82 782 T2x = T2p - T2w;
Chris@82 783 T2P = T2p + T2w;
Chris@82 784 }
Chris@82 785 Ip[0] = KP500000000 * (T1v + T2i);
Chris@82 786 Rp[0] = KP500000000 * (T2P + T2Q);
Chris@82 787 Im[WS(rs, 7)] = KP500000000 * (T2i - T1v);
Chris@82 788 Rm[WS(rs, 7)] = KP500000000 * (T2P - T2Q);
Chris@82 789 Rm[WS(rs, 3)] = KP500000000 * (T2x - T2y);
Chris@82 790 Im[WS(rs, 3)] = KP500000000 * (T2O - T2z);
Chris@82 791 Rp[WS(rs, 4)] = KP500000000 * (T2x + T2y);
Chris@82 792 Ip[WS(rs, 4)] = KP500000000 * (T2z + T2O);
Chris@82 793 }
Chris@82 794 {
Chris@82 795 E T2T, T35, T33, T39, T2W, T36, T2Z, T37;
Chris@82 796 {
Chris@82 797 E T2R, T2S, T31, T32;
Chris@82 798 T2R = T2v - T2s;
Chris@82 799 T2S = T1t - T1g;
Chris@82 800 T2T = KP500000000 * (T2R + T2S);
Chris@82 801 T35 = KP500000000 * (T2S - T2R);
Chris@82 802 T31 = T2l - T2o;
Chris@82 803 T32 = Tx - TW;
Chris@82 804 T33 = KP500000000 * (T31 - T32);
Chris@82 805 T39 = KP500000000 * (T31 + T32);
Chris@82 806 }
Chris@82 807 {
Chris@82 808 E T2U, T2V, T2X, T2Y;
Chris@82 809 T2U = T2F - T2C;
Chris@82 810 T2V = T27 - T2g;
Chris@82 811 T2W = T2U + T2V;
Chris@82 812 T36 = T2U - T2V;
Chris@82 813 T2X = T1T + T1E;
Chris@82 814 T2Y = T2M - T2J;
Chris@82 815 T2Z = T2X - T2Y;
Chris@82 816 T37 = T2X + T2Y;
Chris@82 817 }
Chris@82 818 {
Chris@82 819 E T30, T3a, T34, T38;
Chris@82 820 T30 = KP353553390 * (T2W + T2Z);
Chris@82 821 Ip[WS(rs, 2)] = T2T + T30;
Chris@82 822 Im[WS(rs, 5)] = T30 - T2T;
Chris@82 823 T3a = KP353553390 * (T36 + T37);
Chris@82 824 Rm[WS(rs, 5)] = T39 - T3a;
Chris@82 825 Rp[WS(rs, 2)] = T39 + T3a;
Chris@82 826 T34 = KP353553390 * (T2Z - T2W);
Chris@82 827 Rm[WS(rs, 1)] = T33 - T34;
Chris@82 828 Rp[WS(rs, 6)] = T33 + T34;
Chris@82 829 T38 = KP353553390 * (T36 - T37);
Chris@82 830 Ip[WS(rs, 6)] = T35 + T38;
Chris@82 831 Im[WS(rs, 1)] = T38 - T35;
Chris@82 832 }
Chris@82 833 }
Chris@82 834 {
Chris@82 835 E T3k, T3Q, T3Z, T3D, T3h, T40, T3X, T45, T3G, T3P, T3s, T3K, T3U, T44, T3z;
Chris@82 836 E T3L;
Chris@82 837 {
Chris@82 838 E T3d, T3g, T3o, T3r;
Chris@82 839 T3k = KP500000000 * (T3i - T3j);
Chris@82 840 T3Q = KP500000000 * (T3j + T3i);
Chris@82 841 T3Z = KP500000000 * (T3B - T3C);
Chris@82 842 T3D = KP500000000 * (T3B + T3C);
Chris@82 843 T3d = T3b - T3c;
Chris@82 844 T3g = T3e + T3f;
Chris@82 845 T3h = KP353553390 * (T3d + T3g);
Chris@82 846 T40 = KP353553390 * (T3d - T3g);
Chris@82 847 {
Chris@82 848 E T3V, T3W, T3E, T3F;
Chris@82 849 T3V = T3u + T3t;
Chris@82 850 T3W = T3x - T3w;
Chris@82 851 T3X = FNMS(KP461939766, T3W, KP191341716 * T3V);
Chris@82 852 T45 = FMA(KP461939766, T3V, KP191341716 * T3W);
Chris@82 853 T3E = T3c + T3b;
Chris@82 854 T3F = T3e - T3f;
Chris@82 855 T3G = KP353553390 * (T3E + T3F);
Chris@82 856 T3P = KP353553390 * (T3F - T3E);
Chris@82 857 }
Chris@82 858 T3o = T3m + T3n;
Chris@82 859 T3r = T3p - T3q;
Chris@82 860 T3s = FMA(KP191341716, T3o, KP461939766 * T3r);
Chris@82 861 T3K = FNMS(KP191341716, T3r, KP461939766 * T3o);
Chris@82 862 {
Chris@82 863 E T3S, T3T, T3v, T3y;
Chris@82 864 T3S = T3n - T3m;
Chris@82 865 T3T = T3q + T3p;
Chris@82 866 T3U = FMA(KP461939766, T3S, KP191341716 * T3T);
Chris@82 867 T44 = FNMS(KP461939766, T3T, KP191341716 * T3S);
Chris@82 868 T3v = T3t - T3u;
Chris@82 869 T3y = T3w + T3x;
Chris@82 870 T3z = FNMS(KP191341716, T3y, KP461939766 * T3v);
Chris@82 871 T3L = FMA(KP191341716, T3v, KP461939766 * T3y);
Chris@82 872 }
Chris@82 873 }
Chris@82 874 {
Chris@82 875 E T3l, T3A, T3N, T3O;
Chris@82 876 T3l = T3h + T3k;
Chris@82 877 T3A = T3s + T3z;
Chris@82 878 Ip[WS(rs, 1)] = T3l + T3A;
Chris@82 879 Im[WS(rs, 6)] = T3A - T3l;
Chris@82 880 T3N = T3D + T3G;
Chris@82 881 T3O = T3K + T3L;
Chris@82 882 Rm[WS(rs, 6)] = T3N - T3O;
Chris@82 883 Rp[WS(rs, 1)] = T3N + T3O;
Chris@82 884 }
Chris@82 885 {
Chris@82 886 E T3H, T3I, T3J, T3M;
Chris@82 887 T3H = T3D - T3G;
Chris@82 888 T3I = T3z - T3s;
Chris@82 889 Rm[WS(rs, 2)] = T3H - T3I;
Chris@82 890 Rp[WS(rs, 5)] = T3H + T3I;
Chris@82 891 T3J = T3k - T3h;
Chris@82 892 T3M = T3K - T3L;
Chris@82 893 Ip[WS(rs, 5)] = T3J + T3M;
Chris@82 894 Im[WS(rs, 2)] = T3M - T3J;
Chris@82 895 }
Chris@82 896 {
Chris@82 897 E T3R, T3Y, T47, T48;
Chris@82 898 T3R = T3P + T3Q;
Chris@82 899 T3Y = T3U + T3X;
Chris@82 900 Ip[WS(rs, 3)] = T3R + T3Y;
Chris@82 901 Im[WS(rs, 4)] = T3Y - T3R;
Chris@82 902 T47 = T3Z + T40;
Chris@82 903 T48 = T44 + T45;
Chris@82 904 Rm[WS(rs, 4)] = T47 - T48;
Chris@82 905 Rp[WS(rs, 3)] = T47 + T48;
Chris@82 906 }
Chris@82 907 {
Chris@82 908 E T41, T42, T43, T46;
Chris@82 909 T41 = T3Z - T40;
Chris@82 910 T42 = T3X - T3U;
Chris@82 911 Rm[0] = T41 - T42;
Chris@82 912 Rp[WS(rs, 7)] = T41 + T42;
Chris@82 913 T43 = T3Q - T3P;
Chris@82 914 T46 = T44 - T45;
Chris@82 915 Ip[WS(rs, 7)] = T43 + T46;
Chris@82 916 Im[0] = T46 - T43;
Chris@82 917 }
Chris@82 918 }
Chris@82 919 }
Chris@82 920 }
Chris@82 921 }
Chris@82 922 }
Chris@82 923
Chris@82 924 static const tw_instr twinstr[] = {
Chris@82 925 {TW_CEXP, 1, 1},
Chris@82 926 {TW_CEXP, 1, 3},
Chris@82 927 {TW_CEXP, 1, 9},
Chris@82 928 {TW_CEXP, 1, 15},
Chris@82 929 {TW_NEXT, 1, 0}
Chris@82 930 };
Chris@82 931
Chris@82 932 static const hc2c_desc desc = { 16, "hc2cfdft2_16", twinstr, &GENUS, {188, 84, 40, 0} };
Chris@82 933
Chris@82 934 void X(codelet_hc2cfdft2_16) (planner *p) {
Chris@82 935 X(khc2c_register) (p, hc2cfdft2_16, &desc, HC2C_VIA_DFT);
Chris@82 936 }
Chris@82 937 #endif