Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/scalar/codelets/n1_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 16:35:50 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_notw.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 7 -name n1_7 -include n.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 60 FP additions, 42 FP multiplications,
Chris@42 32 * (or, 18 additions, 0 multiplications, 42 fused multiply/add),
Chris@42 33 * 51 stack variables, 6 constants, and 28 memory accesses
Chris@42 34 */
Chris@42 35 #include "n.h"
Chris@42 36
Chris@42 37 static void n1_7(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 38 {
Chris@42 39 DK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@42 40 DK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@42 41 DK(KP801937735, +0.801937735804838252472204639014890102331838324);
Chris@42 42 DK(KP692021471, +0.692021471630095869627814897002069140197260599);
Chris@42 43 DK(KP356895867, +0.356895867892209443894399510021300583399127187);
Chris@42 44 DK(KP554958132, +0.554958132087371191422194871006410481067288862);
Chris@42 45 {
Chris@42 46 INT i;
Chris@42 47 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
Chris@42 48 E Tz, TP, Ty, TK, TN, TE, Tw, TF;
Chris@42 49 {
Chris@42 50 E T1, TI, T4, TG, Ta, TT, Tp, TH, T7, Tk, TJ, TO, Tu, Tb, TB;
Chris@42 51 E Tg, Tl, Th, Ti;
Chris@42 52 T1 = ri[0];
Chris@42 53 Tz = ii[0];
Chris@42 54 {
Chris@42 55 E T5, T6, Te, Tf;
Chris@42 56 {
Chris@42 57 E T2, T3, T8, T9;
Chris@42 58 T2 = ri[WS(is, 1)];
Chris@42 59 T3 = ri[WS(is, 6)];
Chris@42 60 T8 = ri[WS(is, 3)];
Chris@42 61 T9 = ri[WS(is, 4)];
Chris@42 62 T5 = ri[WS(is, 2)];
Chris@42 63 TI = T3 - T2;
Chris@42 64 T4 = T2 + T3;
Chris@42 65 TG = T9 - T8;
Chris@42 66 Ta = T8 + T9;
Chris@42 67 T6 = ri[WS(is, 5)];
Chris@42 68 }
Chris@42 69 Te = ii[WS(is, 2)];
Chris@42 70 TT = FMA(KP554958132, TG, TI);
Chris@42 71 Tp = FNMS(KP356895867, T4, Ta);
Chris@42 72 TH = T6 - T5;
Chris@42 73 T7 = T5 + T6;
Chris@42 74 Tf = ii[WS(is, 5)];
Chris@42 75 Tk = ii[WS(is, 3)];
Chris@42 76 TJ = FNMS(KP554958132, TI, TH);
Chris@42 77 TO = FMA(KP554958132, TH, TG);
Chris@42 78 Tu = FNMS(KP356895867, Ta, T7);
Chris@42 79 Tb = FNMS(KP356895867, T7, T4);
Chris@42 80 TB = Te + Tf;
Chris@42 81 Tg = Te - Tf;
Chris@42 82 Tl = ii[WS(is, 4)];
Chris@42 83 Th = ii[WS(is, 1)];
Chris@42 84 Ti = ii[WS(is, 6)];
Chris@42 85 }
Chris@42 86 {
Chris@42 87 E Tm, TA, Tj, TD, Ts, TL, Tx, TU, To, TR, Td, TM, Tv;
Chris@42 88 {
Chris@42 89 E TC, TQ, Tn, Tc;
Chris@42 90 ro[0] = T1 + T4 + T7 + Ta;
Chris@42 91 TC = Tk + Tl;
Chris@42 92 Tm = Tk - Tl;
Chris@42 93 TA = Th + Ti;
Chris@42 94 Tj = Th - Ti;
Chris@42 95 TD = FNMS(KP356895867, TC, TB);
Chris@42 96 Ts = FMA(KP554958132, Tg, Tm);
Chris@42 97 TL = FNMS(KP356895867, TA, TC);
Chris@42 98 TQ = FNMS(KP356895867, TB, TA);
Chris@42 99 Tx = FNMS(KP554958132, Tj, Tg);
Chris@42 100 Tn = FMA(KP554958132, Tm, Tj);
Chris@42 101 io[0] = Tz + TA + TB + TC;
Chris@42 102 Tc = FNMS(KP692021471, Tb, Ta);
Chris@42 103 TU = FMA(KP801937735, TT, TH);
Chris@42 104 To = FMA(KP801937735, Tn, Tg);
Chris@42 105 TR = FNMS(KP692021471, TQ, TC);
Chris@42 106 Td = FNMS(KP900968867, Tc, T1);
Chris@42 107 }
Chris@42 108 {
Chris@42 109 E Tt, Tr, TS, Tq;
Chris@42 110 Tt = FNMS(KP801937735, Ts, Tj);
Chris@42 111 Tq = FNMS(KP692021471, Tp, T7);
Chris@42 112 TS = FNMS(KP900968867, TR, Tz);
Chris@42 113 ro[WS(os, 1)] = FMA(KP974927912, To, Td);
Chris@42 114 ro[WS(os, 6)] = FNMS(KP974927912, To, Td);
Chris@42 115 Tr = FNMS(KP900968867, Tq, T1);
Chris@42 116 io[WS(os, 6)] = FNMS(KP974927912, TU, TS);
Chris@42 117 io[WS(os, 1)] = FMA(KP974927912, TU, TS);
Chris@42 118 TP = FNMS(KP801937735, TO, TI);
Chris@42 119 ro[WS(os, 2)] = FMA(KP974927912, Tt, Tr);
Chris@42 120 ro[WS(os, 5)] = FNMS(KP974927912, Tt, Tr);
Chris@42 121 TM = FNMS(KP692021471, TL, TB);
Chris@42 122 }
Chris@42 123 Ty = FNMS(KP801937735, Tx, Tm);
Chris@42 124 Tv = FNMS(KP692021471, Tu, T4);
Chris@42 125 TK = FNMS(KP801937735, TJ, TG);
Chris@42 126 TN = FNMS(KP900968867, TM, Tz);
Chris@42 127 TE = FNMS(KP692021471, TD, TA);
Chris@42 128 Tw = FNMS(KP900968867, Tv, T1);
Chris@42 129 }
Chris@42 130 }
Chris@42 131 io[WS(os, 5)] = FNMS(KP974927912, TP, TN);
Chris@42 132 io[WS(os, 2)] = FMA(KP974927912, TP, TN);
Chris@42 133 TF = FNMS(KP900968867, TE, Tz);
Chris@42 134 ro[WS(os, 3)] = FMA(KP974927912, Ty, Tw);
Chris@42 135 ro[WS(os, 4)] = FNMS(KP974927912, Ty, Tw);
Chris@42 136 io[WS(os, 4)] = FNMS(KP974927912, TK, TF);
Chris@42 137 io[WS(os, 3)] = FMA(KP974927912, TK, TF);
Chris@42 138 }
Chris@42 139 }
Chris@42 140 }
Chris@42 141
Chris@42 142 static const kdft_desc desc = { 7, "n1_7", {18, 0, 42, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 143
Chris@42 144 void X(codelet_n1_7) (planner *p) {
Chris@42 145 X(kdft_register) (p, n1_7, &desc);
Chris@42 146 }
Chris@42 147
Chris@42 148 #else /* HAVE_FMA */
Chris@42 149
Chris@42 150 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 7 -name n1_7 -include n.h */
Chris@42 151
Chris@42 152 /*
Chris@42 153 * This function contains 60 FP additions, 36 FP multiplications,
Chris@42 154 * (or, 36 additions, 12 multiplications, 24 fused multiply/add),
Chris@42 155 * 25 stack variables, 6 constants, and 28 memory accesses
Chris@42 156 */
Chris@42 157 #include "n.h"
Chris@42 158
Chris@42 159 static void n1_7(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 160 {
Chris@42 161 DK(KP222520933, +0.222520933956314404288902564496794759466355569);
Chris@42 162 DK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@42 163 DK(KP623489801, +0.623489801858733530525004884004239810632274731);
Chris@42 164 DK(KP433883739, +0.433883739117558120475768332848358754609990728);
Chris@42 165 DK(KP781831482, +0.781831482468029808708444526674057750232334519);
Chris@42 166 DK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@42 167 {
Chris@42 168 INT i;
Chris@42 169 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
Chris@42 170 E T1, Tu, T4, Tq, Te, Tx, T7, Ts, Tk, Tv, Ta, Tr, Th, Tw;
Chris@42 171 T1 = ri[0];
Chris@42 172 Tu = ii[0];
Chris@42 173 {
Chris@42 174 E T2, T3, Tc, Td;
Chris@42 175 T2 = ri[WS(is, 1)];
Chris@42 176 T3 = ri[WS(is, 6)];
Chris@42 177 T4 = T2 + T3;
Chris@42 178 Tq = T3 - T2;
Chris@42 179 Tc = ii[WS(is, 1)];
Chris@42 180 Td = ii[WS(is, 6)];
Chris@42 181 Te = Tc - Td;
Chris@42 182 Tx = Tc + Td;
Chris@42 183 }
Chris@42 184 {
Chris@42 185 E T5, T6, Ti, Tj;
Chris@42 186 T5 = ri[WS(is, 2)];
Chris@42 187 T6 = ri[WS(is, 5)];
Chris@42 188 T7 = T5 + T6;
Chris@42 189 Ts = T6 - T5;
Chris@42 190 Ti = ii[WS(is, 2)];
Chris@42 191 Tj = ii[WS(is, 5)];
Chris@42 192 Tk = Ti - Tj;
Chris@42 193 Tv = Ti + Tj;
Chris@42 194 }
Chris@42 195 {
Chris@42 196 E T8, T9, Tf, Tg;
Chris@42 197 T8 = ri[WS(is, 3)];
Chris@42 198 T9 = ri[WS(is, 4)];
Chris@42 199 Ta = T8 + T9;
Chris@42 200 Tr = T9 - T8;
Chris@42 201 Tf = ii[WS(is, 3)];
Chris@42 202 Tg = ii[WS(is, 4)];
Chris@42 203 Th = Tf - Tg;
Chris@42 204 Tw = Tf + Tg;
Chris@42 205 }
Chris@42 206 ro[0] = T1 + T4 + T7 + Ta;
Chris@42 207 io[0] = Tu + Tx + Tv + Tw;
Chris@42 208 {
Chris@42 209 E Tl, Tb, TB, TC;
Chris@42 210 Tl = FNMS(KP781831482, Th, KP974927912 * Te) - (KP433883739 * Tk);
Chris@42 211 Tb = FMA(KP623489801, Ta, T1) + FNMA(KP900968867, T7, KP222520933 * T4);
Chris@42 212 ro[WS(os, 5)] = Tb - Tl;
Chris@42 213 ro[WS(os, 2)] = Tb + Tl;
Chris@42 214 TB = FNMS(KP781831482, Tr, KP974927912 * Tq) - (KP433883739 * Ts);
Chris@42 215 TC = FMA(KP623489801, Tw, Tu) + FNMA(KP900968867, Tv, KP222520933 * Tx);
Chris@42 216 io[WS(os, 2)] = TB + TC;
Chris@42 217 io[WS(os, 5)] = TC - TB;
Chris@42 218 }
Chris@42 219 {
Chris@42 220 E Tn, Tm, Tz, TA;
Chris@42 221 Tn = FMA(KP781831482, Te, KP974927912 * Tk) + (KP433883739 * Th);
Chris@42 222 Tm = FMA(KP623489801, T4, T1) + FNMA(KP900968867, Ta, KP222520933 * T7);
Chris@42 223 ro[WS(os, 6)] = Tm - Tn;
Chris@42 224 ro[WS(os, 1)] = Tm + Tn;
Chris@42 225 Tz = FMA(KP781831482, Tq, KP974927912 * Ts) + (KP433883739 * Tr);
Chris@42 226 TA = FMA(KP623489801, Tx, Tu) + FNMA(KP900968867, Tw, KP222520933 * Tv);
Chris@42 227 io[WS(os, 1)] = Tz + TA;
Chris@42 228 io[WS(os, 6)] = TA - Tz;
Chris@42 229 }
Chris@42 230 {
Chris@42 231 E Tp, To, Tt, Ty;
Chris@42 232 Tp = FMA(KP433883739, Te, KP974927912 * Th) - (KP781831482 * Tk);
Chris@42 233 To = FMA(KP623489801, T7, T1) + FNMA(KP222520933, Ta, KP900968867 * T4);
Chris@42 234 ro[WS(os, 4)] = To - Tp;
Chris@42 235 ro[WS(os, 3)] = To + Tp;
Chris@42 236 Tt = FMA(KP433883739, Tq, KP974927912 * Tr) - (KP781831482 * Ts);
Chris@42 237 Ty = FMA(KP623489801, Tv, Tu) + FNMA(KP222520933, Tw, KP900968867 * Tx);
Chris@42 238 io[WS(os, 3)] = Tt + Ty;
Chris@42 239 io[WS(os, 4)] = Ty - Tt;
Chris@42 240 }
Chris@42 241 }
Chris@42 242 }
Chris@42 243 }
Chris@42 244
Chris@42 245 static const kdft_desc desc = { 7, "n1_7", {36, 12, 24, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 246
Chris@42 247 void X(codelet_n1_7) (planner *p) {
Chris@42 248 X(kdft_register) (p, n1_7, &desc);
Chris@42 249 }
Chris@42 250
Chris@42 251 #endif /* HAVE_FMA */