Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/t1fv_32.c @ 84:08ae793730bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +0000
parents 2cd0e3b3e1fd
children
rev   line source
Chris@42 1 /*
Chris@42 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42 4 *
Chris@42 5 * This program is free software; you can redistribute it and/or modify
Chris@42 6 * it under the terms of the GNU General Public License as published by
Chris@42 7 * the Free Software Foundation; either version 2 of the License, or
Chris@42 8 * (at your option) any later version.
Chris@42 9 *
Chris@42 10 * This program is distributed in the hope that it will be useful,
Chris@42 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@42 13 * GNU General Public License for more details.
Chris@42 14 *
Chris@42 15 * You should have received a copy of the GNU General Public License
Chris@42 16 * along with this program; if not, write to the Free Software
Chris@42 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@42 18 *
Chris@42 19 */
Chris@42 20
Chris@42 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@42 22 /* Generated on Sat Jul 30 16:42:07 EDT 2016 */
Chris@42 23
Chris@42 24 #include "codelet-dft.h"
Chris@42 25
Chris@42 26 #ifdef HAVE_FMA
Chris@42 27
Chris@42 28 /* Generated by: ../../../genfft/gen_twiddle_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 32 -name t1fv_32 -include t1f.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 217 FP additions, 160 FP multiplications,
Chris@42 32 * (or, 119 additions, 62 multiplications, 98 fused multiply/add),
Chris@42 33 * 112 stack variables, 7 constants, and 64 memory accesses
Chris@42 34 */
Chris@42 35 #include "t1f.h"
Chris@42 36
Chris@42 37 static void t1fv_32(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 38 {
Chris@42 39 DVK(KP831469612, +0.831469612302545237078788377617905756738560812);
Chris@42 40 DVK(KP980785280, +0.980785280403230449126182236134239036973933731);
Chris@42 41 DVK(KP668178637, +0.668178637919298919997757686523080761552472251);
Chris@42 42 DVK(KP198912367, +0.198912367379658006911597622644676228597850501);
Chris@42 43 DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@42 44 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 45 DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@42 46 {
Chris@42 47 INT m;
Chris@42 48 R *x;
Chris@42 49 x = ri;
Chris@42 50 for (m = mb, W = W + (mb * ((TWVL / VL) * 62)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 62), MAKE_VOLATILE_STRIDE(32, rs)) {
Chris@42 51 V T26, T25, T1Z, T22, T1W, T2a, T2k, T2g;
Chris@42 52 {
Chris@42 53 V T4, T1z, T2o, T32, T2r, T3f, Tf, T1A, T34, T2L, T1D, TC, T33, T2O, T1C;
Chris@42 54 V Tr, T2C, T3a, T2F, T3b, T1r, T21, T1k, T20, TQ, TM, TS, TL, T2t, TJ;
Chris@42 55 V T10, T2u;
Chris@42 56 {
Chris@42 57 V Tt, T9, T2p, Te, T2q, TA, Tu, Tx;
Chris@42 58 {
Chris@42 59 V T1, T1x, T2, T1v;
Chris@42 60 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 61 T1x = LD(&(x[WS(rs, 24)]), ms, &(x[0]));
Chris@42 62 T2 = LD(&(x[WS(rs, 16)]), ms, &(x[0]));
Chris@42 63 T1v = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@42 64 {
Chris@42 65 V T5, Tc, T7, Ta, T2m, T2n;
Chris@42 66 T5 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 67 Tc = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@42 68 T7 = LD(&(x[WS(rs, 20)]), ms, &(x[0]));
Chris@42 69 Ta = LD(&(x[WS(rs, 28)]), ms, &(x[0]));
Chris@42 70 {
Chris@42 71 V T1y, T3, T1w, T6, Td, T8, Tb, Ts, Tz;
Chris@42 72 Ts = LD(&(x[WS(rs, 30)]), ms, &(x[0]));
Chris@42 73 T1y = BYTWJ(&(W[TWVL * 46]), T1x);
Chris@42 74 T3 = BYTWJ(&(W[TWVL * 30]), T2);
Chris@42 75 T1w = BYTWJ(&(W[TWVL * 14]), T1v);
Chris@42 76 T6 = BYTWJ(&(W[TWVL * 6]), T5);
Chris@42 77 Td = BYTWJ(&(W[TWVL * 22]), Tc);
Chris@42 78 T8 = BYTWJ(&(W[TWVL * 38]), T7);
Chris@42 79 Tb = BYTWJ(&(W[TWVL * 54]), Ta);
Chris@42 80 Tt = BYTWJ(&(W[TWVL * 58]), Ts);
Chris@42 81 Tz = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 82 T4 = VSUB(T1, T3);
Chris@42 83 T2m = VADD(T1, T3);
Chris@42 84 T1z = VSUB(T1w, T1y);
Chris@42 85 T2n = VADD(T1w, T1y);
Chris@42 86 T9 = VSUB(T6, T8);
Chris@42 87 T2p = VADD(T6, T8);
Chris@42 88 Te = VSUB(Tb, Td);
Chris@42 89 T2q = VADD(Tb, Td);
Chris@42 90 TA = BYTWJ(&(W[TWVL * 10]), Tz);
Chris@42 91 }
Chris@42 92 Tu = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@42 93 T2o = VADD(T2m, T2n);
Chris@42 94 T32 = VSUB(T2m, T2n);
Chris@42 95 Tx = LD(&(x[WS(rs, 22)]), ms, &(x[0]));
Chris@42 96 }
Chris@42 97 }
Chris@42 98 {
Chris@42 99 V Tv, To, Ty, Ti, Tj, Tm, Th;
Chris@42 100 Th = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 101 T2r = VADD(T2p, T2q);
Chris@42 102 T3f = VSUB(T2q, T2p);
Chris@42 103 Tf = VADD(T9, Te);
Chris@42 104 T1A = VSUB(Te, T9);
Chris@42 105 Tv = BYTWJ(&(W[TWVL * 26]), Tu);
Chris@42 106 To = LD(&(x[WS(rs, 26)]), ms, &(x[0]));
Chris@42 107 Ty = BYTWJ(&(W[TWVL * 42]), Tx);
Chris@42 108 Ti = BYTWJ(&(W[TWVL * 2]), Th);
Chris@42 109 Tj = LD(&(x[WS(rs, 18)]), ms, &(x[0]));
Chris@42 110 Tm = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@42 111 {
Chris@42 112 V T1f, T1h, T1a, T1c, T18, T2A, T2B, T1p;
Chris@42 113 {
Chris@42 114 V T15, T17, T1o, T1m;
Chris@42 115 {
Chris@42 116 V Tw, T2J, Tp, T2K, TB, Tk, Tn, T1n, T14, T16;
Chris@42 117 T14 = LD(&(x[WS(rs, 31)]), ms, &(x[WS(rs, 1)]));
Chris@42 118 T16 = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
Chris@42 119 Tw = VSUB(Tt, Tv);
Chris@42 120 T2J = VADD(Tt, Tv);
Chris@42 121 Tp = BYTWJ(&(W[TWVL * 50]), To);
Chris@42 122 T2K = VADD(TA, Ty);
Chris@42 123 TB = VSUB(Ty, TA);
Chris@42 124 Tk = BYTWJ(&(W[TWVL * 34]), Tj);
Chris@42 125 Tn = BYTWJ(&(W[TWVL * 18]), Tm);
Chris@42 126 T15 = BYTWJ(&(W[TWVL * 60]), T14);
Chris@42 127 T17 = BYTWJ(&(W[TWVL * 28]), T16);
Chris@42 128 T1n = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 129 {
Chris@42 130 V T2M, Tl, T2N, Tq, T1l;
Chris@42 131 T1l = LD(&(x[WS(rs, 23)]), ms, &(x[WS(rs, 1)]));
Chris@42 132 T34 = VSUB(T2J, T2K);
Chris@42 133 T2L = VADD(T2J, T2K);
Chris@42 134 T1D = VFMA(LDK(KP414213562), Tw, TB);
Chris@42 135 TC = VFNMS(LDK(KP414213562), TB, Tw);
Chris@42 136 T2M = VADD(Ti, Tk);
Chris@42 137 Tl = VSUB(Ti, Tk);
Chris@42 138 T2N = VADD(Tn, Tp);
Chris@42 139 Tq = VSUB(Tn, Tp);
Chris@42 140 T1o = BYTWJ(&(W[TWVL * 12]), T1n);
Chris@42 141 T1m = BYTWJ(&(W[TWVL * 44]), T1l);
Chris@42 142 {
Chris@42 143 V T1e, T1g, T19, T1b;
Chris@42 144 T1e = LD(&(x[WS(rs, 27)]), ms, &(x[WS(rs, 1)]));
Chris@42 145 T1g = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@42 146 T19 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 147 T1b = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)]));
Chris@42 148 T33 = VSUB(T2M, T2N);
Chris@42 149 T2O = VADD(T2M, T2N);
Chris@42 150 T1C = VFMA(LDK(KP414213562), Tl, Tq);
Chris@42 151 Tr = VFNMS(LDK(KP414213562), Tq, Tl);
Chris@42 152 T1f = BYTWJ(&(W[TWVL * 52]), T1e);
Chris@42 153 T1h = BYTWJ(&(W[TWVL * 20]), T1g);
Chris@42 154 T1a = BYTWJ(&(W[TWVL * 4]), T19);
Chris@42 155 T1c = BYTWJ(&(W[TWVL * 36]), T1b);
Chris@42 156 }
Chris@42 157 }
Chris@42 158 }
Chris@42 159 T18 = VSUB(T15, T17);
Chris@42 160 T2A = VADD(T15, T17);
Chris@42 161 T2B = VADD(T1o, T1m);
Chris@42 162 T1p = VSUB(T1m, T1o);
Chris@42 163 }
Chris@42 164 {
Chris@42 165 V TG, TI, TZ, TX;
Chris@42 166 {
Chris@42 167 V T1i, T2E, T1d, T2D, TH, TY, TF;
Chris@42 168 TF = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 169 T1i = VSUB(T1f, T1h);
Chris@42 170 T2E = VADD(T1f, T1h);
Chris@42 171 T1d = VSUB(T1a, T1c);
Chris@42 172 T2D = VADD(T1a, T1c);
Chris@42 173 TH = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)]));
Chris@42 174 TY = LD(&(x[WS(rs, 25)]), ms, &(x[WS(rs, 1)]));
Chris@42 175 T2C = VADD(T2A, T2B);
Chris@42 176 T3a = VSUB(T2A, T2B);
Chris@42 177 TG = BYTWJ(&(W[0]), TF);
Chris@42 178 {
Chris@42 179 V TW, T1j, T1q, TP, TR, TK;
Chris@42 180 TW = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@42 181 T2F = VADD(T2D, T2E);
Chris@42 182 T3b = VSUB(T2E, T2D);
Chris@42 183 T1j = VADD(T1d, T1i);
Chris@42 184 T1q = VSUB(T1i, T1d);
Chris@42 185 TI = BYTWJ(&(W[TWVL * 32]), TH);
Chris@42 186 TZ = BYTWJ(&(W[TWVL * 48]), TY);
Chris@42 187 TP = LD(&(x[WS(rs, 29)]), ms, &(x[WS(rs, 1)]));
Chris@42 188 TX = BYTWJ(&(W[TWVL * 16]), TW);
Chris@42 189 TR = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@42 190 TK = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 191 T1r = VFMA(LDK(KP707106781), T1q, T1p);
Chris@42 192 T21 = VFNMS(LDK(KP707106781), T1q, T1p);
Chris@42 193 T1k = VFMA(LDK(KP707106781), T1j, T18);
Chris@42 194 T20 = VFNMS(LDK(KP707106781), T1j, T18);
Chris@42 195 TQ = BYTWJ(&(W[TWVL * 56]), TP);
Chris@42 196 TM = LD(&(x[WS(rs, 21)]), ms, &(x[WS(rs, 1)]));
Chris@42 197 TS = BYTWJ(&(W[TWVL * 24]), TR);
Chris@42 198 TL = BYTWJ(&(W[TWVL * 8]), TK);
Chris@42 199 }
Chris@42 200 }
Chris@42 201 T2t = VADD(TG, TI);
Chris@42 202 TJ = VSUB(TG, TI);
Chris@42 203 T10 = VSUB(TX, TZ);
Chris@42 204 T2u = VADD(TX, TZ);
Chris@42 205 }
Chris@42 206 }
Chris@42 207 }
Chris@42 208 }
Chris@42 209 {
Chris@42 210 V T2s, TT, T2x, T2P, T2Y, T2G, T37, T2v, T2w, TO, T2W, T30, T2U, TN, T2V;
Chris@42 211 T2s = VSUB(T2o, T2r);
Chris@42 212 T2U = VADD(T2o, T2r);
Chris@42 213 TN = BYTWJ(&(W[TWVL * 40]), TM);
Chris@42 214 TT = VSUB(TQ, TS);
Chris@42 215 T2x = VADD(TQ, TS);
Chris@42 216 T2P = VSUB(T2L, T2O);
Chris@42 217 T2V = VADD(T2O, T2L);
Chris@42 218 T2Y = VADD(T2C, T2F);
Chris@42 219 T2G = VSUB(T2C, T2F);
Chris@42 220 T37 = VSUB(T2t, T2u);
Chris@42 221 T2v = VADD(T2t, T2u);
Chris@42 222 T2w = VADD(TL, TN);
Chris@42 223 TO = VSUB(TL, TN);
Chris@42 224 T2W = VADD(T2U, T2V);
Chris@42 225 T30 = VSUB(T2U, T2V);
Chris@42 226 {
Chris@42 227 V T3i, T3o, T36, T3r, T3h, T3j, T12, T1Y, TV, T1X, T3s, T3d, T2Q, T2H, T31;
Chris@42 228 V T2Z;
Chris@42 229 {
Chris@42 230 V T35, T3g, T38, T2y, T11, TU;
Chris@42 231 T35 = VADD(T33, T34);
Chris@42 232 T3g = VSUB(T34, T33);
Chris@42 233 T38 = VSUB(T2w, T2x);
Chris@42 234 T2y = VADD(T2w, T2x);
Chris@42 235 T11 = VSUB(TO, TT);
Chris@42 236 TU = VADD(TO, TT);
Chris@42 237 {
Chris@42 238 V T3c, T39, T2X, T2z;
Chris@42 239 T3c = VFNMS(LDK(KP414213562), T3b, T3a);
Chris@42 240 T3i = VFMA(LDK(KP414213562), T3a, T3b);
Chris@42 241 T3o = VFNMS(LDK(KP707106781), T35, T32);
Chris@42 242 T36 = VFMA(LDK(KP707106781), T35, T32);
Chris@42 243 T3r = VFNMS(LDK(KP707106781), T3g, T3f);
Chris@42 244 T3h = VFMA(LDK(KP707106781), T3g, T3f);
Chris@42 245 T39 = VFNMS(LDK(KP414213562), T38, T37);
Chris@42 246 T3j = VFMA(LDK(KP414213562), T37, T38);
Chris@42 247 T2X = VADD(T2v, T2y);
Chris@42 248 T2z = VSUB(T2v, T2y);
Chris@42 249 T12 = VFMA(LDK(KP707106781), T11, T10);
Chris@42 250 T1Y = VFNMS(LDK(KP707106781), T11, T10);
Chris@42 251 TV = VFMA(LDK(KP707106781), TU, TJ);
Chris@42 252 T1X = VFNMS(LDK(KP707106781), TU, TJ);
Chris@42 253 T3s = VSUB(T3c, T39);
Chris@42 254 T3d = VADD(T39, T3c);
Chris@42 255 T2Q = VSUB(T2G, T2z);
Chris@42 256 T2H = VADD(T2z, T2G);
Chris@42 257 T31 = VSUB(T2Y, T2X);
Chris@42 258 T2Z = VADD(T2X, T2Y);
Chris@42 259 }
Chris@42 260 }
Chris@42 261 {
Chris@42 262 V Tg, T1U, TD, T1G, T13, T1s, T1H, T1B, T1V, T1E, T3k, T3p, T2e, T2f;
Chris@42 263 Tg = VFMA(LDK(KP707106781), Tf, T4);
Chris@42 264 T1U = VFNMS(LDK(KP707106781), Tf, T4);
Chris@42 265 T3k = VSUB(T3i, T3j);
Chris@42 266 T3p = VADD(T3j, T3i);
Chris@42 267 {
Chris@42 268 V T3v, T3t, T3e, T3m;
Chris@42 269 T3v = VFNMS(LDK(KP923879532), T3s, T3r);
Chris@42 270 T3t = VFMA(LDK(KP923879532), T3s, T3r);
Chris@42 271 T3e = VFNMS(LDK(KP923879532), T3d, T36);
Chris@42 272 T3m = VFMA(LDK(KP923879532), T3d, T36);
Chris@42 273 {
Chris@42 274 V T2R, T2T, T2I, T2S;
Chris@42 275 T2R = VFNMS(LDK(KP707106781), T2Q, T2P);
Chris@42 276 T2T = VFMA(LDK(KP707106781), T2Q, T2P);
Chris@42 277 T2I = VFNMS(LDK(KP707106781), T2H, T2s);
Chris@42 278 T2S = VFMA(LDK(KP707106781), T2H, T2s);
Chris@42 279 ST(&(x[WS(rs, 24)]), VFNMSI(T31, T30), ms, &(x[0]));
Chris@42 280 ST(&(x[WS(rs, 8)]), VFMAI(T31, T30), ms, &(x[0]));
Chris@42 281 ST(&(x[0]), VADD(T2W, T2Z), ms, &(x[0]));
Chris@42 282 ST(&(x[WS(rs, 16)]), VSUB(T2W, T2Z), ms, &(x[0]));
Chris@42 283 {
Chris@42 284 V T3u, T3q, T3l, T3n;
Chris@42 285 T3u = VFMA(LDK(KP923879532), T3p, T3o);
Chris@42 286 T3q = VFNMS(LDK(KP923879532), T3p, T3o);
Chris@42 287 T3l = VFNMS(LDK(KP923879532), T3k, T3h);
Chris@42 288 T3n = VFMA(LDK(KP923879532), T3k, T3h);
Chris@42 289 ST(&(x[WS(rs, 4)]), VFMAI(T2T, T2S), ms, &(x[0]));
Chris@42 290 ST(&(x[WS(rs, 28)]), VFNMSI(T2T, T2S), ms, &(x[0]));
Chris@42 291 ST(&(x[WS(rs, 20)]), VFMAI(T2R, T2I), ms, &(x[0]));
Chris@42 292 ST(&(x[WS(rs, 12)]), VFNMSI(T2R, T2I), ms, &(x[0]));
Chris@42 293 ST(&(x[WS(rs, 22)]), VFNMSI(T3t, T3q), ms, &(x[0]));
Chris@42 294 ST(&(x[WS(rs, 10)]), VFMAI(T3t, T3q), ms, &(x[0]));
Chris@42 295 ST(&(x[WS(rs, 26)]), VFMAI(T3v, T3u), ms, &(x[0]));
Chris@42 296 ST(&(x[WS(rs, 6)]), VFNMSI(T3v, T3u), ms, &(x[0]));
Chris@42 297 ST(&(x[WS(rs, 2)]), VFMAI(T3n, T3m), ms, &(x[0]));
Chris@42 298 ST(&(x[WS(rs, 30)]), VFNMSI(T3n, T3m), ms, &(x[0]));
Chris@42 299 ST(&(x[WS(rs, 18)]), VFMAI(T3l, T3e), ms, &(x[0]));
Chris@42 300 ST(&(x[WS(rs, 14)]), VFNMSI(T3l, T3e), ms, &(x[0]));
Chris@42 301 T26 = VSUB(TC, Tr);
Chris@42 302 TD = VADD(Tr, TC);
Chris@42 303 }
Chris@42 304 }
Chris@42 305 }
Chris@42 306 T1G = VFMA(LDK(KP198912367), TV, T12);
Chris@42 307 T13 = VFNMS(LDK(KP198912367), T12, TV);
Chris@42 308 T1s = VFNMS(LDK(KP198912367), T1r, T1k);
Chris@42 309 T1H = VFMA(LDK(KP198912367), T1k, T1r);
Chris@42 310 T1B = VFNMS(LDK(KP707106781), T1A, T1z);
Chris@42 311 T25 = VFMA(LDK(KP707106781), T1A, T1z);
Chris@42 312 T1V = VADD(T1C, T1D);
Chris@42 313 T1E = VSUB(T1C, T1D);
Chris@42 314 {
Chris@42 315 V T1S, T1O, T1K, T1u, T1R, T1T, T1L, T1J;
Chris@42 316 {
Chris@42 317 V TE, T1M, T1I, T1N, T1t, T1Q, T1F, T1P, T28, T29;
Chris@42 318 TE = VFMA(LDK(KP923879532), TD, Tg);
Chris@42 319 T1M = VFNMS(LDK(KP923879532), TD, Tg);
Chris@42 320 T1I = VSUB(T1G, T1H);
Chris@42 321 T1N = VADD(T1G, T1H);
Chris@42 322 T1t = VADD(T13, T1s);
Chris@42 323 T1Q = VSUB(T1s, T13);
Chris@42 324 T1F = VFMA(LDK(KP923879532), T1E, T1B);
Chris@42 325 T1P = VFNMS(LDK(KP923879532), T1E, T1B);
Chris@42 326 T28 = VFNMS(LDK(KP668178637), T1X, T1Y);
Chris@42 327 T1Z = VFMA(LDK(KP668178637), T1Y, T1X);
Chris@42 328 T1S = VFMA(LDK(KP980785280), T1N, T1M);
Chris@42 329 T1O = VFNMS(LDK(KP980785280), T1N, T1M);
Chris@42 330 T22 = VFMA(LDK(KP668178637), T21, T20);
Chris@42 331 T29 = VFNMS(LDK(KP668178637), T20, T21);
Chris@42 332 T1K = VFMA(LDK(KP980785280), T1t, TE);
Chris@42 333 T1u = VFNMS(LDK(KP980785280), T1t, TE);
Chris@42 334 T1R = VFNMS(LDK(KP980785280), T1Q, T1P);
Chris@42 335 T1T = VFMA(LDK(KP980785280), T1Q, T1P);
Chris@42 336 T1L = VFMA(LDK(KP980785280), T1I, T1F);
Chris@42 337 T1J = VFNMS(LDK(KP980785280), T1I, T1F);
Chris@42 338 T2e = VFNMS(LDK(KP923879532), T1V, T1U);
Chris@42 339 T1W = VFMA(LDK(KP923879532), T1V, T1U);
Chris@42 340 T2a = VSUB(T28, T29);
Chris@42 341 T2f = VADD(T28, T29);
Chris@42 342 }
Chris@42 343 ST(&(x[WS(rs, 23)]), VFMAI(T1R, T1O), ms, &(x[WS(rs, 1)]));
Chris@42 344 ST(&(x[WS(rs, 9)]), VFNMSI(T1R, T1O), ms, &(x[WS(rs, 1)]));
Chris@42 345 ST(&(x[WS(rs, 25)]), VFNMSI(T1T, T1S), ms, &(x[WS(rs, 1)]));
Chris@42 346 ST(&(x[WS(rs, 7)]), VFMAI(T1T, T1S), ms, &(x[WS(rs, 1)]));
Chris@42 347 ST(&(x[WS(rs, 31)]), VFMAI(T1L, T1K), ms, &(x[WS(rs, 1)]));
Chris@42 348 ST(&(x[WS(rs, 1)]), VFNMSI(T1L, T1K), ms, &(x[WS(rs, 1)]));
Chris@42 349 ST(&(x[WS(rs, 15)]), VFMAI(T1J, T1u), ms, &(x[WS(rs, 1)]));
Chris@42 350 ST(&(x[WS(rs, 17)]), VFNMSI(T1J, T1u), ms, &(x[WS(rs, 1)]));
Chris@42 351 }
Chris@42 352 T2k = VFNMS(LDK(KP831469612), T2f, T2e);
Chris@42 353 T2g = VFMA(LDK(KP831469612), T2f, T2e);
Chris@42 354 }
Chris@42 355 }
Chris@42 356 }
Chris@42 357 }
Chris@42 358 {
Chris@42 359 V T2i, T23, T2h, T27;
Chris@42 360 T2i = VSUB(T22, T1Z);
Chris@42 361 T23 = VADD(T1Z, T22);
Chris@42 362 T2h = VFNMS(LDK(KP923879532), T26, T25);
Chris@42 363 T27 = VFMA(LDK(KP923879532), T26, T25);
Chris@42 364 {
Chris@42 365 V T2c, T24, T2j, T2l, T2d, T2b;
Chris@42 366 T2c = VFMA(LDK(KP831469612), T23, T1W);
Chris@42 367 T24 = VFNMS(LDK(KP831469612), T23, T1W);
Chris@42 368 T2j = VFMA(LDK(KP831469612), T2i, T2h);
Chris@42 369 T2l = VFNMS(LDK(KP831469612), T2i, T2h);
Chris@42 370 T2d = VFMA(LDK(KP831469612), T2a, T27);
Chris@42 371 T2b = VFNMS(LDK(KP831469612), T2a, T27);
Chris@42 372 ST(&(x[WS(rs, 21)]), VFNMSI(T2j, T2g), ms, &(x[WS(rs, 1)]));
Chris@42 373 ST(&(x[WS(rs, 11)]), VFMAI(T2j, T2g), ms, &(x[WS(rs, 1)]));
Chris@42 374 ST(&(x[WS(rs, 27)]), VFMAI(T2l, T2k), ms, &(x[WS(rs, 1)]));
Chris@42 375 ST(&(x[WS(rs, 5)]), VFNMSI(T2l, T2k), ms, &(x[WS(rs, 1)]));
Chris@42 376 ST(&(x[WS(rs, 3)]), VFMAI(T2d, T2c), ms, &(x[WS(rs, 1)]));
Chris@42 377 ST(&(x[WS(rs, 29)]), VFNMSI(T2d, T2c), ms, &(x[WS(rs, 1)]));
Chris@42 378 ST(&(x[WS(rs, 19)]), VFMAI(T2b, T24), ms, &(x[WS(rs, 1)]));
Chris@42 379 ST(&(x[WS(rs, 13)]), VFNMSI(T2b, T24), ms, &(x[WS(rs, 1)]));
Chris@42 380 }
Chris@42 381 }
Chris@42 382 }
Chris@42 383 }
Chris@42 384 VLEAVE();
Chris@42 385 }
Chris@42 386
Chris@42 387 static const tw_instr twinstr[] = {
Chris@42 388 VTW(0, 1),
Chris@42 389 VTW(0, 2),
Chris@42 390 VTW(0, 3),
Chris@42 391 VTW(0, 4),
Chris@42 392 VTW(0, 5),
Chris@42 393 VTW(0, 6),
Chris@42 394 VTW(0, 7),
Chris@42 395 VTW(0, 8),
Chris@42 396 VTW(0, 9),
Chris@42 397 VTW(0, 10),
Chris@42 398 VTW(0, 11),
Chris@42 399 VTW(0, 12),
Chris@42 400 VTW(0, 13),
Chris@42 401 VTW(0, 14),
Chris@42 402 VTW(0, 15),
Chris@42 403 VTW(0, 16),
Chris@42 404 VTW(0, 17),
Chris@42 405 VTW(0, 18),
Chris@42 406 VTW(0, 19),
Chris@42 407 VTW(0, 20),
Chris@42 408 VTW(0, 21),
Chris@42 409 VTW(0, 22),
Chris@42 410 VTW(0, 23),
Chris@42 411 VTW(0, 24),
Chris@42 412 VTW(0, 25),
Chris@42 413 VTW(0, 26),
Chris@42 414 VTW(0, 27),
Chris@42 415 VTW(0, 28),
Chris@42 416 VTW(0, 29),
Chris@42 417 VTW(0, 30),
Chris@42 418 VTW(0, 31),
Chris@42 419 {TW_NEXT, VL, 0}
Chris@42 420 };
Chris@42 421
Chris@42 422 static const ct_desc desc = { 32, XSIMD_STRING("t1fv_32"), twinstr, &GENUS, {119, 62, 98, 0}, 0, 0, 0 };
Chris@42 423
Chris@42 424 void XSIMD(codelet_t1fv_32) (planner *p) {
Chris@42 425 X(kdft_dit_register) (p, t1fv_32, &desc);
Chris@42 426 }
Chris@42 427 #else /* HAVE_FMA */
Chris@42 428
Chris@42 429 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 32 -name t1fv_32 -include t1f.h */
Chris@42 430
Chris@42 431 /*
Chris@42 432 * This function contains 217 FP additions, 104 FP multiplications,
Chris@42 433 * (or, 201 additions, 88 multiplications, 16 fused multiply/add),
Chris@42 434 * 59 stack variables, 7 constants, and 64 memory accesses
Chris@42 435 */
Chris@42 436 #include "t1f.h"
Chris@42 437
Chris@42 438 static void t1fv_32(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 439 {
Chris@42 440 DVK(KP555570233, +0.555570233019602224742830813948532874374937191);
Chris@42 441 DVK(KP831469612, +0.831469612302545237078788377617905756738560812);
Chris@42 442 DVK(KP195090322, +0.195090322016128267848284868477022240927691618);
Chris@42 443 DVK(KP980785280, +0.980785280403230449126182236134239036973933731);
Chris@42 444 DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
Chris@42 445 DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@42 446 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 447 {
Chris@42 448 INT m;
Chris@42 449 R *x;
Chris@42 450 x = ri;
Chris@42 451 for (m = mb, W = W + (mb * ((TWVL / VL) * 62)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 62), MAKE_VOLATILE_STRIDE(32, rs)) {
Chris@42 452 V T4, T1A, T2o, T32, Tf, T1v, T2r, T3f, TC, T1C, T2L, T34, Tr, T1D, T2O;
Chris@42 453 V T33, T1k, T20, T2F, T3b, T1r, T21, T2C, T3a, TV, T1X, T2y, T38, T12, T1Y;
Chris@42 454 V T2v, T37;
Chris@42 455 {
Chris@42 456 V T1, T1z, T3, T1x, T1y, T2, T1w, T2m, T2n;
Chris@42 457 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 458 T1y = LD(&(x[WS(rs, 24)]), ms, &(x[0]));
Chris@42 459 T1z = BYTWJ(&(W[TWVL * 46]), T1y);
Chris@42 460 T2 = LD(&(x[WS(rs, 16)]), ms, &(x[0]));
Chris@42 461 T3 = BYTWJ(&(W[TWVL * 30]), T2);
Chris@42 462 T1w = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@42 463 T1x = BYTWJ(&(W[TWVL * 14]), T1w);
Chris@42 464 T4 = VSUB(T1, T3);
Chris@42 465 T1A = VSUB(T1x, T1z);
Chris@42 466 T2m = VADD(T1, T3);
Chris@42 467 T2n = VADD(T1x, T1z);
Chris@42 468 T2o = VADD(T2m, T2n);
Chris@42 469 T32 = VSUB(T2m, T2n);
Chris@42 470 }
Chris@42 471 {
Chris@42 472 V T6, Td, T8, Tb;
Chris@42 473 {
Chris@42 474 V T5, Tc, T7, Ta;
Chris@42 475 T5 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 476 T6 = BYTWJ(&(W[TWVL * 6]), T5);
Chris@42 477 Tc = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@42 478 Td = BYTWJ(&(W[TWVL * 22]), Tc);
Chris@42 479 T7 = LD(&(x[WS(rs, 20)]), ms, &(x[0]));
Chris@42 480 T8 = BYTWJ(&(W[TWVL * 38]), T7);
Chris@42 481 Ta = LD(&(x[WS(rs, 28)]), ms, &(x[0]));
Chris@42 482 Tb = BYTWJ(&(W[TWVL * 54]), Ta);
Chris@42 483 }
Chris@42 484 {
Chris@42 485 V T9, Te, T2p, T2q;
Chris@42 486 T9 = VSUB(T6, T8);
Chris@42 487 Te = VSUB(Tb, Td);
Chris@42 488 Tf = VMUL(LDK(KP707106781), VADD(T9, Te));
Chris@42 489 T1v = VMUL(LDK(KP707106781), VSUB(Te, T9));
Chris@42 490 T2p = VADD(T6, T8);
Chris@42 491 T2q = VADD(Tb, Td);
Chris@42 492 T2r = VADD(T2p, T2q);
Chris@42 493 T3f = VSUB(T2q, T2p);
Chris@42 494 }
Chris@42 495 }
Chris@42 496 {
Chris@42 497 V Tt, TA, Tv, Ty;
Chris@42 498 {
Chris@42 499 V Ts, Tz, Tu, Tx;
Chris@42 500 Ts = LD(&(x[WS(rs, 30)]), ms, &(x[0]));
Chris@42 501 Tt = BYTWJ(&(W[TWVL * 58]), Ts);
Chris@42 502 Tz = LD(&(x[WS(rs, 22)]), ms, &(x[0]));
Chris@42 503 TA = BYTWJ(&(W[TWVL * 42]), Tz);
Chris@42 504 Tu = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@42 505 Tv = BYTWJ(&(W[TWVL * 26]), Tu);
Chris@42 506 Tx = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 507 Ty = BYTWJ(&(W[TWVL * 10]), Tx);
Chris@42 508 }
Chris@42 509 {
Chris@42 510 V Tw, TB, T2J, T2K;
Chris@42 511 Tw = VSUB(Tt, Tv);
Chris@42 512 TB = VSUB(Ty, TA);
Chris@42 513 TC = VFMA(LDK(KP923879532), Tw, VMUL(LDK(KP382683432), TB));
Chris@42 514 T1C = VFNMS(LDK(KP923879532), TB, VMUL(LDK(KP382683432), Tw));
Chris@42 515 T2J = VADD(Tt, Tv);
Chris@42 516 T2K = VADD(Ty, TA);
Chris@42 517 T2L = VADD(T2J, T2K);
Chris@42 518 T34 = VSUB(T2J, T2K);
Chris@42 519 }
Chris@42 520 }
Chris@42 521 {
Chris@42 522 V Ti, Tp, Tk, Tn;
Chris@42 523 {
Chris@42 524 V Th, To, Tj, Tm;
Chris@42 525 Th = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 526 Ti = BYTWJ(&(W[TWVL * 2]), Th);
Chris@42 527 To = LD(&(x[WS(rs, 26)]), ms, &(x[0]));
Chris@42 528 Tp = BYTWJ(&(W[TWVL * 50]), To);
Chris@42 529 Tj = LD(&(x[WS(rs, 18)]), ms, &(x[0]));
Chris@42 530 Tk = BYTWJ(&(W[TWVL * 34]), Tj);
Chris@42 531 Tm = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@42 532 Tn = BYTWJ(&(W[TWVL * 18]), Tm);
Chris@42 533 }
Chris@42 534 {
Chris@42 535 V Tl, Tq, T2M, T2N;
Chris@42 536 Tl = VSUB(Ti, Tk);
Chris@42 537 Tq = VSUB(Tn, Tp);
Chris@42 538 Tr = VFNMS(LDK(KP382683432), Tq, VMUL(LDK(KP923879532), Tl));
Chris@42 539 T1D = VFMA(LDK(KP382683432), Tl, VMUL(LDK(KP923879532), Tq));
Chris@42 540 T2M = VADD(Ti, Tk);
Chris@42 541 T2N = VADD(Tn, Tp);
Chris@42 542 T2O = VADD(T2M, T2N);
Chris@42 543 T33 = VSUB(T2M, T2N);
Chris@42 544 }
Chris@42 545 }
Chris@42 546 {
Chris@42 547 V T15, T17, T1p, T1n, T1f, T1h, T1i, T1a, T1c, T1d;
Chris@42 548 {
Chris@42 549 V T14, T16, T1o, T1m;
Chris@42 550 T14 = LD(&(x[WS(rs, 31)]), ms, &(x[WS(rs, 1)]));
Chris@42 551 T15 = BYTWJ(&(W[TWVL * 60]), T14);
Chris@42 552 T16 = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
Chris@42 553 T17 = BYTWJ(&(W[TWVL * 28]), T16);
Chris@42 554 T1o = LD(&(x[WS(rs, 23)]), ms, &(x[WS(rs, 1)]));
Chris@42 555 T1p = BYTWJ(&(W[TWVL * 44]), T1o);
Chris@42 556 T1m = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 557 T1n = BYTWJ(&(W[TWVL * 12]), T1m);
Chris@42 558 {
Chris@42 559 V T1e, T1g, T19, T1b;
Chris@42 560 T1e = LD(&(x[WS(rs, 27)]), ms, &(x[WS(rs, 1)]));
Chris@42 561 T1f = BYTWJ(&(W[TWVL * 52]), T1e);
Chris@42 562 T1g = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@42 563 T1h = BYTWJ(&(W[TWVL * 20]), T1g);
Chris@42 564 T1i = VSUB(T1f, T1h);
Chris@42 565 T19 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 566 T1a = BYTWJ(&(W[TWVL * 4]), T19);
Chris@42 567 T1b = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)]));
Chris@42 568 T1c = BYTWJ(&(W[TWVL * 36]), T1b);
Chris@42 569 T1d = VSUB(T1a, T1c);
Chris@42 570 }
Chris@42 571 }
Chris@42 572 {
Chris@42 573 V T18, T1j, T2D, T2E;
Chris@42 574 T18 = VSUB(T15, T17);
Chris@42 575 T1j = VMUL(LDK(KP707106781), VADD(T1d, T1i));
Chris@42 576 T1k = VADD(T18, T1j);
Chris@42 577 T20 = VSUB(T18, T1j);
Chris@42 578 T2D = VADD(T1a, T1c);
Chris@42 579 T2E = VADD(T1f, T1h);
Chris@42 580 T2F = VADD(T2D, T2E);
Chris@42 581 T3b = VSUB(T2E, T2D);
Chris@42 582 }
Chris@42 583 {
Chris@42 584 V T1l, T1q, T2A, T2B;
Chris@42 585 T1l = VMUL(LDK(KP707106781), VSUB(T1i, T1d));
Chris@42 586 T1q = VSUB(T1n, T1p);
Chris@42 587 T1r = VSUB(T1l, T1q);
Chris@42 588 T21 = VADD(T1q, T1l);
Chris@42 589 T2A = VADD(T15, T17);
Chris@42 590 T2B = VADD(T1n, T1p);
Chris@42 591 T2C = VADD(T2A, T2B);
Chris@42 592 T3a = VSUB(T2A, T2B);
Chris@42 593 }
Chris@42 594 }
Chris@42 595 {
Chris@42 596 V TG, TI, T10, TY, TQ, TS, TT, TL, TN, TO;
Chris@42 597 {
Chris@42 598 V TF, TH, TZ, TX;
Chris@42 599 TF = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 600 TG = BYTWJ(&(W[0]), TF);
Chris@42 601 TH = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)]));
Chris@42 602 TI = BYTWJ(&(W[TWVL * 32]), TH);
Chris@42 603 TZ = LD(&(x[WS(rs, 25)]), ms, &(x[WS(rs, 1)]));
Chris@42 604 T10 = BYTWJ(&(W[TWVL * 48]), TZ);
Chris@42 605 TX = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@42 606 TY = BYTWJ(&(W[TWVL * 16]), TX);
Chris@42 607 {
Chris@42 608 V TP, TR, TK, TM;
Chris@42 609 TP = LD(&(x[WS(rs, 29)]), ms, &(x[WS(rs, 1)]));
Chris@42 610 TQ = BYTWJ(&(W[TWVL * 56]), TP);
Chris@42 611 TR = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@42 612 TS = BYTWJ(&(W[TWVL * 24]), TR);
Chris@42 613 TT = VSUB(TQ, TS);
Chris@42 614 TK = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 615 TL = BYTWJ(&(W[TWVL * 8]), TK);
Chris@42 616 TM = LD(&(x[WS(rs, 21)]), ms, &(x[WS(rs, 1)]));
Chris@42 617 TN = BYTWJ(&(W[TWVL * 40]), TM);
Chris@42 618 TO = VSUB(TL, TN);
Chris@42 619 }
Chris@42 620 }
Chris@42 621 {
Chris@42 622 V TJ, TU, T2w, T2x;
Chris@42 623 TJ = VSUB(TG, TI);
Chris@42 624 TU = VMUL(LDK(KP707106781), VADD(TO, TT));
Chris@42 625 TV = VADD(TJ, TU);
Chris@42 626 T1X = VSUB(TJ, TU);
Chris@42 627 T2w = VADD(TL, TN);
Chris@42 628 T2x = VADD(TQ, TS);
Chris@42 629 T2y = VADD(T2w, T2x);
Chris@42 630 T38 = VSUB(T2x, T2w);
Chris@42 631 }
Chris@42 632 {
Chris@42 633 V TW, T11, T2t, T2u;
Chris@42 634 TW = VMUL(LDK(KP707106781), VSUB(TT, TO));
Chris@42 635 T11 = VSUB(TY, T10);
Chris@42 636 T12 = VSUB(TW, T11);
Chris@42 637 T1Y = VADD(T11, TW);
Chris@42 638 T2t = VADD(TG, TI);
Chris@42 639 T2u = VADD(TY, T10);
Chris@42 640 T2v = VADD(T2t, T2u);
Chris@42 641 T37 = VSUB(T2t, T2u);
Chris@42 642 }
Chris@42 643 }
Chris@42 644 {
Chris@42 645 V T2W, T30, T2Z, T31;
Chris@42 646 {
Chris@42 647 V T2U, T2V, T2X, T2Y;
Chris@42 648 T2U = VADD(T2o, T2r);
Chris@42 649 T2V = VADD(T2O, T2L);
Chris@42 650 T2W = VADD(T2U, T2V);
Chris@42 651 T30 = VSUB(T2U, T2V);
Chris@42 652 T2X = VADD(T2v, T2y);
Chris@42 653 T2Y = VADD(T2C, T2F);
Chris@42 654 T2Z = VADD(T2X, T2Y);
Chris@42 655 T31 = VBYI(VSUB(T2Y, T2X));
Chris@42 656 }
Chris@42 657 ST(&(x[WS(rs, 16)]), VSUB(T2W, T2Z), ms, &(x[0]));
Chris@42 658 ST(&(x[WS(rs, 8)]), VADD(T30, T31), ms, &(x[0]));
Chris@42 659 ST(&(x[0]), VADD(T2W, T2Z), ms, &(x[0]));
Chris@42 660 ST(&(x[WS(rs, 24)]), VSUB(T30, T31), ms, &(x[0]));
Chris@42 661 }
Chris@42 662 {
Chris@42 663 V T2s, T2P, T2H, T2Q, T2z, T2G;
Chris@42 664 T2s = VSUB(T2o, T2r);
Chris@42 665 T2P = VSUB(T2L, T2O);
Chris@42 666 T2z = VSUB(T2v, T2y);
Chris@42 667 T2G = VSUB(T2C, T2F);
Chris@42 668 T2H = VMUL(LDK(KP707106781), VADD(T2z, T2G));
Chris@42 669 T2Q = VMUL(LDK(KP707106781), VSUB(T2G, T2z));
Chris@42 670 {
Chris@42 671 V T2I, T2R, T2S, T2T;
Chris@42 672 T2I = VADD(T2s, T2H);
Chris@42 673 T2R = VBYI(VADD(T2P, T2Q));
Chris@42 674 ST(&(x[WS(rs, 28)]), VSUB(T2I, T2R), ms, &(x[0]));
Chris@42 675 ST(&(x[WS(rs, 4)]), VADD(T2I, T2R), ms, &(x[0]));
Chris@42 676 T2S = VSUB(T2s, T2H);
Chris@42 677 T2T = VBYI(VSUB(T2Q, T2P));
Chris@42 678 ST(&(x[WS(rs, 20)]), VSUB(T2S, T2T), ms, &(x[0]));
Chris@42 679 ST(&(x[WS(rs, 12)]), VADD(T2S, T2T), ms, &(x[0]));
Chris@42 680 }
Chris@42 681 }
Chris@42 682 {
Chris@42 683 V T36, T3r, T3h, T3p, T3d, T3o, T3k, T3s, T35, T3g;
Chris@42 684 T35 = VMUL(LDK(KP707106781), VADD(T33, T34));
Chris@42 685 T36 = VADD(T32, T35);
Chris@42 686 T3r = VSUB(T32, T35);
Chris@42 687 T3g = VMUL(LDK(KP707106781), VSUB(T34, T33));
Chris@42 688 T3h = VADD(T3f, T3g);
Chris@42 689 T3p = VSUB(T3g, T3f);
Chris@42 690 {
Chris@42 691 V T39, T3c, T3i, T3j;
Chris@42 692 T39 = VFMA(LDK(KP923879532), T37, VMUL(LDK(KP382683432), T38));
Chris@42 693 T3c = VFNMS(LDK(KP382683432), T3b, VMUL(LDK(KP923879532), T3a));
Chris@42 694 T3d = VADD(T39, T3c);
Chris@42 695 T3o = VSUB(T3c, T39);
Chris@42 696 T3i = VFNMS(LDK(KP382683432), T37, VMUL(LDK(KP923879532), T38));
Chris@42 697 T3j = VFMA(LDK(KP382683432), T3a, VMUL(LDK(KP923879532), T3b));
Chris@42 698 T3k = VADD(T3i, T3j);
Chris@42 699 T3s = VSUB(T3j, T3i);
Chris@42 700 }
Chris@42 701 {
Chris@42 702 V T3e, T3l, T3u, T3v;
Chris@42 703 T3e = VADD(T36, T3d);
Chris@42 704 T3l = VBYI(VADD(T3h, T3k));
Chris@42 705 ST(&(x[WS(rs, 30)]), VSUB(T3e, T3l), ms, &(x[0]));
Chris@42 706 ST(&(x[WS(rs, 2)]), VADD(T3e, T3l), ms, &(x[0]));
Chris@42 707 T3u = VBYI(VADD(T3p, T3o));
Chris@42 708 T3v = VADD(T3r, T3s);
Chris@42 709 ST(&(x[WS(rs, 6)]), VADD(T3u, T3v), ms, &(x[0]));
Chris@42 710 ST(&(x[WS(rs, 26)]), VSUB(T3v, T3u), ms, &(x[0]));
Chris@42 711 }
Chris@42 712 {
Chris@42 713 V T3m, T3n, T3q, T3t;
Chris@42 714 T3m = VSUB(T36, T3d);
Chris@42 715 T3n = VBYI(VSUB(T3k, T3h));
Chris@42 716 ST(&(x[WS(rs, 18)]), VSUB(T3m, T3n), ms, &(x[0]));
Chris@42 717 ST(&(x[WS(rs, 14)]), VADD(T3m, T3n), ms, &(x[0]));
Chris@42 718 T3q = VBYI(VSUB(T3o, T3p));
Chris@42 719 T3t = VSUB(T3r, T3s);
Chris@42 720 ST(&(x[WS(rs, 10)]), VADD(T3q, T3t), ms, &(x[0]));
Chris@42 721 ST(&(x[WS(rs, 22)]), VSUB(T3t, T3q), ms, &(x[0]));
Chris@42 722 }
Chris@42 723 }
Chris@42 724 {
Chris@42 725 V TE, T1P, T1I, T1Q, T1t, T1M, T1F, T1N;
Chris@42 726 {
Chris@42 727 V Tg, TD, T1G, T1H;
Chris@42 728 Tg = VADD(T4, Tf);
Chris@42 729 TD = VADD(Tr, TC);
Chris@42 730 TE = VADD(Tg, TD);
Chris@42 731 T1P = VSUB(Tg, TD);
Chris@42 732 T1G = VFNMS(LDK(KP195090322), TV, VMUL(LDK(KP980785280), T12));
Chris@42 733 T1H = VFMA(LDK(KP195090322), T1k, VMUL(LDK(KP980785280), T1r));
Chris@42 734 T1I = VADD(T1G, T1H);
Chris@42 735 T1Q = VSUB(T1H, T1G);
Chris@42 736 }
Chris@42 737 {
Chris@42 738 V T13, T1s, T1B, T1E;
Chris@42 739 T13 = VFMA(LDK(KP980785280), TV, VMUL(LDK(KP195090322), T12));
Chris@42 740 T1s = VFNMS(LDK(KP195090322), T1r, VMUL(LDK(KP980785280), T1k));
Chris@42 741 T1t = VADD(T13, T1s);
Chris@42 742 T1M = VSUB(T1s, T13);
Chris@42 743 T1B = VSUB(T1v, T1A);
Chris@42 744 T1E = VSUB(T1C, T1D);
Chris@42 745 T1F = VADD(T1B, T1E);
Chris@42 746 T1N = VSUB(T1E, T1B);
Chris@42 747 }
Chris@42 748 {
Chris@42 749 V T1u, T1J, T1S, T1T;
Chris@42 750 T1u = VADD(TE, T1t);
Chris@42 751 T1J = VBYI(VADD(T1F, T1I));
Chris@42 752 ST(&(x[WS(rs, 31)]), VSUB(T1u, T1J), ms, &(x[WS(rs, 1)]));
Chris@42 753 ST(&(x[WS(rs, 1)]), VADD(T1u, T1J), ms, &(x[WS(rs, 1)]));
Chris@42 754 T1S = VBYI(VADD(T1N, T1M));
Chris@42 755 T1T = VADD(T1P, T1Q);
Chris@42 756 ST(&(x[WS(rs, 7)]), VADD(T1S, T1T), ms, &(x[WS(rs, 1)]));
Chris@42 757 ST(&(x[WS(rs, 25)]), VSUB(T1T, T1S), ms, &(x[WS(rs, 1)]));
Chris@42 758 }
Chris@42 759 {
Chris@42 760 V T1K, T1L, T1O, T1R;
Chris@42 761 T1K = VSUB(TE, T1t);
Chris@42 762 T1L = VBYI(VSUB(T1I, T1F));
Chris@42 763 ST(&(x[WS(rs, 17)]), VSUB(T1K, T1L), ms, &(x[WS(rs, 1)]));
Chris@42 764 ST(&(x[WS(rs, 15)]), VADD(T1K, T1L), ms, &(x[WS(rs, 1)]));
Chris@42 765 T1O = VBYI(VSUB(T1M, T1N));
Chris@42 766 T1R = VSUB(T1P, T1Q);
Chris@42 767 ST(&(x[WS(rs, 9)]), VADD(T1O, T1R), ms, &(x[WS(rs, 1)]));
Chris@42 768 ST(&(x[WS(rs, 23)]), VSUB(T1R, T1O), ms, &(x[WS(rs, 1)]));
Chris@42 769 }
Chris@42 770 }
Chris@42 771 {
Chris@42 772 V T1W, T2h, T2a, T2i, T23, T2e, T27, T2f;
Chris@42 773 {
Chris@42 774 V T1U, T1V, T28, T29;
Chris@42 775 T1U = VSUB(T4, Tf);
Chris@42 776 T1V = VADD(T1D, T1C);
Chris@42 777 T1W = VADD(T1U, T1V);
Chris@42 778 T2h = VSUB(T1U, T1V);
Chris@42 779 T28 = VFNMS(LDK(KP555570233), T1X, VMUL(LDK(KP831469612), T1Y));
Chris@42 780 T29 = VFMA(LDK(KP555570233), T20, VMUL(LDK(KP831469612), T21));
Chris@42 781 T2a = VADD(T28, T29);
Chris@42 782 T2i = VSUB(T29, T28);
Chris@42 783 }
Chris@42 784 {
Chris@42 785 V T1Z, T22, T25, T26;
Chris@42 786 T1Z = VFMA(LDK(KP831469612), T1X, VMUL(LDK(KP555570233), T1Y));
Chris@42 787 T22 = VFNMS(LDK(KP555570233), T21, VMUL(LDK(KP831469612), T20));
Chris@42 788 T23 = VADD(T1Z, T22);
Chris@42 789 T2e = VSUB(T22, T1Z);
Chris@42 790 T25 = VADD(T1A, T1v);
Chris@42 791 T26 = VSUB(TC, Tr);
Chris@42 792 T27 = VADD(T25, T26);
Chris@42 793 T2f = VSUB(T26, T25);
Chris@42 794 }
Chris@42 795 {
Chris@42 796 V T24, T2b, T2k, T2l;
Chris@42 797 T24 = VADD(T1W, T23);
Chris@42 798 T2b = VBYI(VADD(T27, T2a));
Chris@42 799 ST(&(x[WS(rs, 29)]), VSUB(T24, T2b), ms, &(x[WS(rs, 1)]));
Chris@42 800 ST(&(x[WS(rs, 3)]), VADD(T24, T2b), ms, &(x[WS(rs, 1)]));
Chris@42 801 T2k = VBYI(VADD(T2f, T2e));
Chris@42 802 T2l = VADD(T2h, T2i);
Chris@42 803 ST(&(x[WS(rs, 5)]), VADD(T2k, T2l), ms, &(x[WS(rs, 1)]));
Chris@42 804 ST(&(x[WS(rs, 27)]), VSUB(T2l, T2k), ms, &(x[WS(rs, 1)]));
Chris@42 805 }
Chris@42 806 {
Chris@42 807 V T2c, T2d, T2g, T2j;
Chris@42 808 T2c = VSUB(T1W, T23);
Chris@42 809 T2d = VBYI(VSUB(T2a, T27));
Chris@42 810 ST(&(x[WS(rs, 19)]), VSUB(T2c, T2d), ms, &(x[WS(rs, 1)]));
Chris@42 811 ST(&(x[WS(rs, 13)]), VADD(T2c, T2d), ms, &(x[WS(rs, 1)]));
Chris@42 812 T2g = VBYI(VSUB(T2e, T2f));
Chris@42 813 T2j = VSUB(T2h, T2i);
Chris@42 814 ST(&(x[WS(rs, 11)]), VADD(T2g, T2j), ms, &(x[WS(rs, 1)]));
Chris@42 815 ST(&(x[WS(rs, 21)]), VSUB(T2j, T2g), ms, &(x[WS(rs, 1)]));
Chris@42 816 }
Chris@42 817 }
Chris@42 818 }
Chris@42 819 }
Chris@42 820 VLEAVE();
Chris@42 821 }
Chris@42 822
Chris@42 823 static const tw_instr twinstr[] = {
Chris@42 824 VTW(0, 1),
Chris@42 825 VTW(0, 2),
Chris@42 826 VTW(0, 3),
Chris@42 827 VTW(0, 4),
Chris@42 828 VTW(0, 5),
Chris@42 829 VTW(0, 6),
Chris@42 830 VTW(0, 7),
Chris@42 831 VTW(0, 8),
Chris@42 832 VTW(0, 9),
Chris@42 833 VTW(0, 10),
Chris@42 834 VTW(0, 11),
Chris@42 835 VTW(0, 12),
Chris@42 836 VTW(0, 13),
Chris@42 837 VTW(0, 14),
Chris@42 838 VTW(0, 15),
Chris@42 839 VTW(0, 16),
Chris@42 840 VTW(0, 17),
Chris@42 841 VTW(0, 18),
Chris@42 842 VTW(0, 19),
Chris@42 843 VTW(0, 20),
Chris@42 844 VTW(0, 21),
Chris@42 845 VTW(0, 22),
Chris@42 846 VTW(0, 23),
Chris@42 847 VTW(0, 24),
Chris@42 848 VTW(0, 25),
Chris@42 849 VTW(0, 26),
Chris@42 850 VTW(0, 27),
Chris@42 851 VTW(0, 28),
Chris@42 852 VTW(0, 29),
Chris@42 853 VTW(0, 30),
Chris@42 854 VTW(0, 31),
Chris@42 855 {TW_NEXT, VL, 0}
Chris@42 856 };
Chris@42 857
Chris@42 858 static const ct_desc desc = { 32, XSIMD_STRING("t1fv_32"), twinstr, &GENUS, {201, 88, 16, 0}, 0, 0, 0 };
Chris@42 859
Chris@42 860 void XSIMD(codelet_t1fv_32) (planner *p) {
Chris@42 861 X(kdft_dit_register) (p, t1fv_32, &desc);
Chris@42 862 }
Chris@42 863 #endif /* HAVE_FMA */