Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/scalar/codelets/t1_7.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 16:36:07 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.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 7 -name t1_7 -include t.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 72 FP additions, 66 FP multiplications,
Chris@42 32 * (or, 18 additions, 12 multiplications, 54 fused multiply/add),
Chris@42 33 * 66 stack variables, 6 constants, and 28 memory accesses
Chris@42 34 */
Chris@42 35 #include "t.h"
Chris@42 36
Chris@42 37 static void t1_7(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 38 {
Chris@42 39 DK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@42 40 DK(KP801937735, +0.801937735804838252472204639014890102331838324);
Chris@42 41 DK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@42 42 DK(KP692021471, +0.692021471630095869627814897002069140197260599);
Chris@42 43 DK(KP554958132, +0.554958132087371191422194871006410481067288862);
Chris@42 44 DK(KP356895867, +0.356895867892209443894399510021300583399127187);
Chris@42 45 {
Chris@42 46 INT m;
Chris@42 47 for (m = mb, W = W + (mb * 12); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 12, MAKE_VOLATILE_STRIDE(14, rs)) {
Chris@42 48 E T1c, T19, T1i, T18, T16, T1q, T1t, T1r, T1u, T1s;
Chris@42 49 {
Chris@42 50 E T1, TR, T1h, Te, Tt, Tw, T1a, TM, T1g, Tr, Tu, TS, Tz, TC, Ty;
Chris@42 51 E Tv, TB;
Chris@42 52 T1 = ri[0];
Chris@42 53 T1c = ii[0];
Chris@42 54 {
Chris@42 55 E T9, Tc, TP, Ta, Tb, TO, T7;
Chris@42 56 {
Chris@42 57 E T3, T6, T8, TN, T4, T2, T5;
Chris@42 58 T3 = ri[WS(rs, 1)];
Chris@42 59 T6 = ii[WS(rs, 1)];
Chris@42 60 T2 = W[0];
Chris@42 61 T9 = ri[WS(rs, 6)];
Chris@42 62 Tc = ii[WS(rs, 6)];
Chris@42 63 T8 = W[10];
Chris@42 64 TN = T2 * T6;
Chris@42 65 T4 = T2 * T3;
Chris@42 66 T5 = W[1];
Chris@42 67 TP = T8 * Tc;
Chris@42 68 Ta = T8 * T9;
Chris@42 69 Tb = W[11];
Chris@42 70 TO = FNMS(T5, T3, TN);
Chris@42 71 T7 = FMA(T5, T6, T4);
Chris@42 72 }
Chris@42 73 {
Chris@42 74 E Tg, Tj, Th, TI, Tm, Tp, Tl, Ti, To, TQ, Td, Tf;
Chris@42 75 Tg = ri[WS(rs, 2)];
Chris@42 76 TQ = FNMS(Tb, T9, TP);
Chris@42 77 Td = FMA(Tb, Tc, Ta);
Chris@42 78 Tj = ii[WS(rs, 2)];
Chris@42 79 Tf = W[2];
Chris@42 80 T19 = TO + TQ;
Chris@42 81 TR = TO - TQ;
Chris@42 82 T1h = Td - T7;
Chris@42 83 Te = T7 + Td;
Chris@42 84 Th = Tf * Tg;
Chris@42 85 TI = Tf * Tj;
Chris@42 86 Tm = ri[WS(rs, 5)];
Chris@42 87 Tp = ii[WS(rs, 5)];
Chris@42 88 Tl = W[8];
Chris@42 89 Ti = W[3];
Chris@42 90 To = W[9];
Chris@42 91 {
Chris@42 92 E TJ, Tk, TL, Tq, TK, Tn, Ts;
Chris@42 93 Tt = ri[WS(rs, 3)];
Chris@42 94 TK = Tl * Tp;
Chris@42 95 Tn = Tl * Tm;
Chris@42 96 TJ = FNMS(Ti, Tg, TI);
Chris@42 97 Tk = FMA(Ti, Tj, Th);
Chris@42 98 TL = FNMS(To, Tm, TK);
Chris@42 99 Tq = FMA(To, Tp, Tn);
Chris@42 100 Tw = ii[WS(rs, 3)];
Chris@42 101 Ts = W[4];
Chris@42 102 T1a = TJ + TL;
Chris@42 103 TM = TJ - TL;
Chris@42 104 T1g = Tq - Tk;
Chris@42 105 Tr = Tk + Tq;
Chris@42 106 Tu = Ts * Tt;
Chris@42 107 TS = Ts * Tw;
Chris@42 108 }
Chris@42 109 Tz = ri[WS(rs, 4)];
Chris@42 110 TC = ii[WS(rs, 4)];
Chris@42 111 Ty = W[6];
Chris@42 112 Tv = W[5];
Chris@42 113 TB = W[7];
Chris@42 114 }
Chris@42 115 }
Chris@42 116 {
Chris@42 117 E TF, TT, Tx, TV, TD, T1d, TU, TA;
Chris@42 118 TF = FNMS(KP356895867, Tr, Te);
Chris@42 119 TU = Ty * TC;
Chris@42 120 TA = Ty * Tz;
Chris@42 121 TT = FNMS(Tv, Tt, TS);
Chris@42 122 Tx = FMA(Tv, Tw, Tu);
Chris@42 123 TV = FNMS(TB, Tz, TU);
Chris@42 124 TD = FMA(TB, TC, TA);
Chris@42 125 T1d = FNMS(KP356895867, T1a, T19);
Chris@42 126 {
Chris@42 127 E T1b, T15, T17, TW;
Chris@42 128 T17 = FNMS(KP554958132, TR, TM);
Chris@42 129 T1b = TT + TV;
Chris@42 130 TW = TT - TV;
Chris@42 131 {
Chris@42 132 E TE, T1l, T1e, T12;
Chris@42 133 T1i = TD - Tx;
Chris@42 134 TE = Tx + TD;
Chris@42 135 T1l = FNMS(KP356895867, T19, T1b);
Chris@42 136 T1e = FNMS(KP692021471, T1d, T1b);
Chris@42 137 ii[0] = T19 + T1a + T1b + T1c;
Chris@42 138 T12 = FMA(KP554958132, TM, TW);
Chris@42 139 {
Chris@42 140 E TX, T1o, T1j, T14;
Chris@42 141 TX = FMA(KP554958132, TW, TR);
Chris@42 142 T1o = FMA(KP554958132, T1g, T1i);
Chris@42 143 T1j = FMA(KP554958132, T1i, T1h);
Chris@42 144 T14 = FNMS(KP356895867, TE, Tr);
Chris@42 145 {
Chris@42 146 E TZ, TG, T1m, T1f;
Chris@42 147 TZ = FNMS(KP356895867, Te, TE);
Chris@42 148 TG = FNMS(KP692021471, TF, TE);
Chris@42 149 ri[0] = T1 + Te + Tr + TE;
Chris@42 150 T1m = FNMS(KP692021471, T1l, T1a);
Chris@42 151 T1f = FNMS(KP900968867, T1e, T1c);
Chris@42 152 {
Chris@42 153 E T13, TY, T1p, T1k;
Chris@42 154 T13 = FNMS(KP801937735, T12, TR);
Chris@42 155 TY = FMA(KP801937735, TX, TM);
Chris@42 156 T1p = FNMS(KP801937735, T1o, T1h);
Chris@42 157 T1k = FMA(KP801937735, T1j, T1g);
Chris@42 158 T15 = FNMS(KP692021471, T14, Te);
Chris@42 159 {
Chris@42 160 E T10, TH, T1n, T11;
Chris@42 161 T10 = FNMS(KP692021471, TZ, Tr);
Chris@42 162 TH = FNMS(KP900968867, TG, T1);
Chris@42 163 T1n = FNMS(KP900968867, T1m, T1c);
Chris@42 164 ii[WS(rs, 6)] = FNMS(KP974927912, T1k, T1f);
Chris@42 165 ii[WS(rs, 1)] = FMA(KP974927912, T1k, T1f);
Chris@42 166 T11 = FNMS(KP900968867, T10, T1);
Chris@42 167 ri[WS(rs, 1)] = FMA(KP974927912, TY, TH);
Chris@42 168 ri[WS(rs, 6)] = FNMS(KP974927912, TY, TH);
Chris@42 169 ii[WS(rs, 5)] = FNMS(KP974927912, T1p, T1n);
Chris@42 170 ii[WS(rs, 2)] = FMA(KP974927912, T1p, T1n);
Chris@42 171 ri[WS(rs, 2)] = FMA(KP974927912, T13, T11);
Chris@42 172 ri[WS(rs, 5)] = FNMS(KP974927912, T13, T11);
Chris@42 173 T18 = FNMS(KP801937735, T17, TW);
Chris@42 174 }
Chris@42 175 }
Chris@42 176 }
Chris@42 177 }
Chris@42 178 }
Chris@42 179 T16 = FNMS(KP900968867, T15, T1);
Chris@42 180 T1q = FNMS(KP356895867, T1b, T1a);
Chris@42 181 T1t = FNMS(KP554958132, T1h, T1g);
Chris@42 182 }
Chris@42 183 }
Chris@42 184 }
Chris@42 185 ri[WS(rs, 3)] = FMA(KP974927912, T18, T16);
Chris@42 186 ri[WS(rs, 4)] = FNMS(KP974927912, T18, T16);
Chris@42 187 T1r = FNMS(KP692021471, T1q, T19);
Chris@42 188 T1u = FNMS(KP801937735, T1t, T1i);
Chris@42 189 T1s = FNMS(KP900968867, T1r, T1c);
Chris@42 190 ii[WS(rs, 4)] = FNMS(KP974927912, T1u, T1s);
Chris@42 191 ii[WS(rs, 3)] = FMA(KP974927912, T1u, T1s);
Chris@42 192 }
Chris@42 193 }
Chris@42 194 }
Chris@42 195
Chris@42 196 static const tw_instr twinstr[] = {
Chris@42 197 {TW_FULL, 0, 7},
Chris@42 198 {TW_NEXT, 1, 0}
Chris@42 199 };
Chris@42 200
Chris@42 201 static const ct_desc desc = { 7, "t1_7", twinstr, &GENUS, {18, 12, 54, 0}, 0, 0, 0 };
Chris@42 202
Chris@42 203 void X(codelet_t1_7) (planner *p) {
Chris@42 204 X(kdft_dit_register) (p, t1_7, &desc);
Chris@42 205 }
Chris@42 206 #else /* HAVE_FMA */
Chris@42 207
Chris@42 208 /* Generated by: ../../../genfft/gen_twiddle.native -compact -variables 4 -pipeline-latency 4 -n 7 -name t1_7 -include t.h */
Chris@42 209
Chris@42 210 /*
Chris@42 211 * This function contains 72 FP additions, 60 FP multiplications,
Chris@42 212 * (or, 36 additions, 24 multiplications, 36 fused multiply/add),
Chris@42 213 * 29 stack variables, 6 constants, and 28 memory accesses
Chris@42 214 */
Chris@42 215 #include "t.h"
Chris@42 216
Chris@42 217 static void t1_7(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 218 {
Chris@42 219 DK(KP222520933, +0.222520933956314404288902564496794759466355569);
Chris@42 220 DK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@42 221 DK(KP623489801, +0.623489801858733530525004884004239810632274731);
Chris@42 222 DK(KP433883739, +0.433883739117558120475768332848358754609990728);
Chris@42 223 DK(KP781831482, +0.781831482468029808708444526674057750232334519);
Chris@42 224 DK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@42 225 {
Chris@42 226 INT m;
Chris@42 227 for (m = mb, W = W + (mb * 12); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 12, MAKE_VOLATILE_STRIDE(14, rs)) {
Chris@42 228 E T1, TR, Tc, TS, TC, TO, Tn, TT, TI, TP, Ty, TU, TF, TQ;
Chris@42 229 T1 = ri[0];
Chris@42 230 TR = ii[0];
Chris@42 231 {
Chris@42 232 E T6, TA, Tb, TB;
Chris@42 233 {
Chris@42 234 E T3, T5, T2, T4;
Chris@42 235 T3 = ri[WS(rs, 1)];
Chris@42 236 T5 = ii[WS(rs, 1)];
Chris@42 237 T2 = W[0];
Chris@42 238 T4 = W[1];
Chris@42 239 T6 = FMA(T2, T3, T4 * T5);
Chris@42 240 TA = FNMS(T4, T3, T2 * T5);
Chris@42 241 }
Chris@42 242 {
Chris@42 243 E T8, Ta, T7, T9;
Chris@42 244 T8 = ri[WS(rs, 6)];
Chris@42 245 Ta = ii[WS(rs, 6)];
Chris@42 246 T7 = W[10];
Chris@42 247 T9 = W[11];
Chris@42 248 Tb = FMA(T7, T8, T9 * Ta);
Chris@42 249 TB = FNMS(T9, T8, T7 * Ta);
Chris@42 250 }
Chris@42 251 Tc = T6 + Tb;
Chris@42 252 TS = Tb - T6;
Chris@42 253 TC = TA - TB;
Chris@42 254 TO = TA + TB;
Chris@42 255 }
Chris@42 256 {
Chris@42 257 E Th, TG, Tm, TH;
Chris@42 258 {
Chris@42 259 E Te, Tg, Td, Tf;
Chris@42 260 Te = ri[WS(rs, 2)];
Chris@42 261 Tg = ii[WS(rs, 2)];
Chris@42 262 Td = W[2];
Chris@42 263 Tf = W[3];
Chris@42 264 Th = FMA(Td, Te, Tf * Tg);
Chris@42 265 TG = FNMS(Tf, Te, Td * Tg);
Chris@42 266 }
Chris@42 267 {
Chris@42 268 E Tj, Tl, Ti, Tk;
Chris@42 269 Tj = ri[WS(rs, 5)];
Chris@42 270 Tl = ii[WS(rs, 5)];
Chris@42 271 Ti = W[8];
Chris@42 272 Tk = W[9];
Chris@42 273 Tm = FMA(Ti, Tj, Tk * Tl);
Chris@42 274 TH = FNMS(Tk, Tj, Ti * Tl);
Chris@42 275 }
Chris@42 276 Tn = Th + Tm;
Chris@42 277 TT = Tm - Th;
Chris@42 278 TI = TG - TH;
Chris@42 279 TP = TG + TH;
Chris@42 280 }
Chris@42 281 {
Chris@42 282 E Ts, TD, Tx, TE;
Chris@42 283 {
Chris@42 284 E Tp, Tr, To, Tq;
Chris@42 285 Tp = ri[WS(rs, 3)];
Chris@42 286 Tr = ii[WS(rs, 3)];
Chris@42 287 To = W[4];
Chris@42 288 Tq = W[5];
Chris@42 289 Ts = FMA(To, Tp, Tq * Tr);
Chris@42 290 TD = FNMS(Tq, Tp, To * Tr);
Chris@42 291 }
Chris@42 292 {
Chris@42 293 E Tu, Tw, Tt, Tv;
Chris@42 294 Tu = ri[WS(rs, 4)];
Chris@42 295 Tw = ii[WS(rs, 4)];
Chris@42 296 Tt = W[6];
Chris@42 297 Tv = W[7];
Chris@42 298 Tx = FMA(Tt, Tu, Tv * Tw);
Chris@42 299 TE = FNMS(Tv, Tu, Tt * Tw);
Chris@42 300 }
Chris@42 301 Ty = Ts + Tx;
Chris@42 302 TU = Tx - Ts;
Chris@42 303 TF = TD - TE;
Chris@42 304 TQ = TD + TE;
Chris@42 305 }
Chris@42 306 ri[0] = T1 + Tc + Tn + Ty;
Chris@42 307 ii[0] = TO + TP + TQ + TR;
Chris@42 308 {
Chris@42 309 E TJ, Tz, TX, TY;
Chris@42 310 TJ = FNMS(KP781831482, TF, KP974927912 * TC) - (KP433883739 * TI);
Chris@42 311 Tz = FMA(KP623489801, Ty, T1) + FNMA(KP900968867, Tn, KP222520933 * Tc);
Chris@42 312 ri[WS(rs, 5)] = Tz - TJ;
Chris@42 313 ri[WS(rs, 2)] = Tz + TJ;
Chris@42 314 TX = FNMS(KP781831482, TU, KP974927912 * TS) - (KP433883739 * TT);
Chris@42 315 TY = FMA(KP623489801, TQ, TR) + FNMA(KP900968867, TP, KP222520933 * TO);
Chris@42 316 ii[WS(rs, 2)] = TX + TY;
Chris@42 317 ii[WS(rs, 5)] = TY - TX;
Chris@42 318 }
Chris@42 319 {
Chris@42 320 E TL, TK, TV, TW;
Chris@42 321 TL = FMA(KP781831482, TC, KP974927912 * TI) + (KP433883739 * TF);
Chris@42 322 TK = FMA(KP623489801, Tc, T1) + FNMA(KP900968867, Ty, KP222520933 * Tn);
Chris@42 323 ri[WS(rs, 6)] = TK - TL;
Chris@42 324 ri[WS(rs, 1)] = TK + TL;
Chris@42 325 TV = FMA(KP781831482, TS, KP974927912 * TT) + (KP433883739 * TU);
Chris@42 326 TW = FMA(KP623489801, TO, TR) + FNMA(KP900968867, TQ, KP222520933 * TP);
Chris@42 327 ii[WS(rs, 1)] = TV + TW;
Chris@42 328 ii[WS(rs, 6)] = TW - TV;
Chris@42 329 }
Chris@42 330 {
Chris@42 331 E TN, TM, TZ, T10;
Chris@42 332 TN = FMA(KP433883739, TC, KP974927912 * TF) - (KP781831482 * TI);
Chris@42 333 TM = FMA(KP623489801, Tn, T1) + FNMA(KP222520933, Ty, KP900968867 * Tc);
Chris@42 334 ri[WS(rs, 4)] = TM - TN;
Chris@42 335 ri[WS(rs, 3)] = TM + TN;
Chris@42 336 TZ = FMA(KP433883739, TS, KP974927912 * TU) - (KP781831482 * TT);
Chris@42 337 T10 = FMA(KP623489801, TP, TR) + FNMA(KP222520933, TQ, KP900968867 * TO);
Chris@42 338 ii[WS(rs, 3)] = TZ + T10;
Chris@42 339 ii[WS(rs, 4)] = T10 - TZ;
Chris@42 340 }
Chris@42 341 }
Chris@42 342 }
Chris@42 343 }
Chris@42 344
Chris@42 345 static const tw_instr twinstr[] = {
Chris@42 346 {TW_FULL, 0, 7},
Chris@42 347 {TW_NEXT, 1, 0}
Chris@42 348 };
Chris@42 349
Chris@42 350 static const ct_desc desc = { 7, "t1_7", twinstr, &GENUS, {36, 24, 36, 0}, 0, 0, 0 };
Chris@42 351
Chris@42 352 void X(codelet_t1_7) (planner *p) {
Chris@42 353 X(kdft_dit_register) (p, t1_7, &desc);
Chris@42 354 }
Chris@42 355 #endif /* HAVE_FMA */