Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/scalar/codelets/n1_5.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 5 -name n1_5 -include n.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 32 FP additions, 18 FP multiplications,
Chris@42 32 * (or, 14 additions, 0 multiplications, 18 fused multiply/add),
Chris@42 33 * 37 stack variables, 4 constants, and 20 memory accesses
Chris@42 34 */
Chris@42 35 #include "n.h"
Chris@42 36
Chris@42 37 static void n1_5(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(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42 40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42 41 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42 42 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@42 43 {
Chris@42 44 INT i;
Chris@42 45 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(20, is), MAKE_VOLATILE_STRIDE(20, os)) {
Chris@42 46 E Tq, Ti, Tk, Tu, Tw, Tp, Tb, Tj, Tr, Tv;
Chris@42 47 {
Chris@42 48 E T1, Tl, Ts, Tt, T8, Ta, Te, Tm, Tn, Th, To, T9;
Chris@42 49 T1 = ri[0];
Chris@42 50 Tl = ii[0];
Chris@42 51 {
Chris@42 52 E T2, T3, T5, T6;
Chris@42 53 T2 = ri[WS(is, 1)];
Chris@42 54 T3 = ri[WS(is, 4)];
Chris@42 55 T5 = ri[WS(is, 2)];
Chris@42 56 T6 = ri[WS(is, 3)];
Chris@42 57 {
Chris@42 58 E Tc, T4, T7, Td, Tf, Tg;
Chris@42 59 Tc = ii[WS(is, 1)];
Chris@42 60 Ts = T2 - T3;
Chris@42 61 T4 = T2 + T3;
Chris@42 62 Tt = T5 - T6;
Chris@42 63 T7 = T5 + T6;
Chris@42 64 Td = ii[WS(is, 4)];
Chris@42 65 Tf = ii[WS(is, 2)];
Chris@42 66 Tg = ii[WS(is, 3)];
Chris@42 67 T8 = T4 + T7;
Chris@42 68 Ta = T4 - T7;
Chris@42 69 Te = Tc - Td;
Chris@42 70 Tm = Tc + Td;
Chris@42 71 Tn = Tf + Tg;
Chris@42 72 Th = Tf - Tg;
Chris@42 73 }
Chris@42 74 }
Chris@42 75 ro[0] = T1 + T8;
Chris@42 76 To = Tm + Tn;
Chris@42 77 Tq = Tm - Tn;
Chris@42 78 Ti = FMA(KP618033988, Th, Te);
Chris@42 79 Tk = FNMS(KP618033988, Te, Th);
Chris@42 80 io[0] = Tl + To;
Chris@42 81 T9 = FNMS(KP250000000, T8, T1);
Chris@42 82 Tu = FMA(KP618033988, Tt, Ts);
Chris@42 83 Tw = FNMS(KP618033988, Ts, Tt);
Chris@42 84 Tp = FNMS(KP250000000, To, Tl);
Chris@42 85 Tb = FMA(KP559016994, Ta, T9);
Chris@42 86 Tj = FNMS(KP559016994, Ta, T9);
Chris@42 87 }
Chris@42 88 Tr = FMA(KP559016994, Tq, Tp);
Chris@42 89 Tv = FNMS(KP559016994, Tq, Tp);
Chris@42 90 ro[WS(os, 2)] = FNMS(KP951056516, Tk, Tj);
Chris@42 91 ro[WS(os, 3)] = FMA(KP951056516, Tk, Tj);
Chris@42 92 ro[WS(os, 1)] = FMA(KP951056516, Ti, Tb);
Chris@42 93 ro[WS(os, 4)] = FNMS(KP951056516, Ti, Tb);
Chris@42 94 io[WS(os, 2)] = FMA(KP951056516, Tw, Tv);
Chris@42 95 io[WS(os, 3)] = FNMS(KP951056516, Tw, Tv);
Chris@42 96 io[WS(os, 4)] = FMA(KP951056516, Tu, Tr);
Chris@42 97 io[WS(os, 1)] = FNMS(KP951056516, Tu, Tr);
Chris@42 98 }
Chris@42 99 }
Chris@42 100 }
Chris@42 101
Chris@42 102 static const kdft_desc desc = { 5, "n1_5", {14, 0, 18, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 103
Chris@42 104 void X(codelet_n1_5) (planner *p) {
Chris@42 105 X(kdft_register) (p, n1_5, &desc);
Chris@42 106 }
Chris@42 107
Chris@42 108 #else /* HAVE_FMA */
Chris@42 109
Chris@42 110 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 5 -name n1_5 -include n.h */
Chris@42 111
Chris@42 112 /*
Chris@42 113 * This function contains 32 FP additions, 12 FP multiplications,
Chris@42 114 * (or, 26 additions, 6 multiplications, 6 fused multiply/add),
Chris@42 115 * 21 stack variables, 4 constants, and 20 memory accesses
Chris@42 116 */
Chris@42 117 #include "n.h"
Chris@42 118
Chris@42 119 static void n1_5(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 120 {
Chris@42 121 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42 122 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@42 123 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42 124 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42 125 {
Chris@42 126 INT i;
Chris@42 127 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(20, is), MAKE_VOLATILE_STRIDE(20, os)) {
Chris@42 128 E T1, To, T8, Tt, T9, Ts, Te, Tp, Th, Tn;
Chris@42 129 T1 = ri[0];
Chris@42 130 To = ii[0];
Chris@42 131 {
Chris@42 132 E T2, T3, T4, T5, T6, T7;
Chris@42 133 T2 = ri[WS(is, 1)];
Chris@42 134 T3 = ri[WS(is, 4)];
Chris@42 135 T4 = T2 + T3;
Chris@42 136 T5 = ri[WS(is, 2)];
Chris@42 137 T6 = ri[WS(is, 3)];
Chris@42 138 T7 = T5 + T6;
Chris@42 139 T8 = T4 + T7;
Chris@42 140 Tt = T5 - T6;
Chris@42 141 T9 = KP559016994 * (T4 - T7);
Chris@42 142 Ts = T2 - T3;
Chris@42 143 }
Chris@42 144 {
Chris@42 145 E Tc, Td, Tl, Tf, Tg, Tm;
Chris@42 146 Tc = ii[WS(is, 1)];
Chris@42 147 Td = ii[WS(is, 4)];
Chris@42 148 Tl = Tc + Td;
Chris@42 149 Tf = ii[WS(is, 2)];
Chris@42 150 Tg = ii[WS(is, 3)];
Chris@42 151 Tm = Tf + Tg;
Chris@42 152 Te = Tc - Td;
Chris@42 153 Tp = Tl + Tm;
Chris@42 154 Th = Tf - Tg;
Chris@42 155 Tn = KP559016994 * (Tl - Tm);
Chris@42 156 }
Chris@42 157 ro[0] = T1 + T8;
Chris@42 158 io[0] = To + Tp;
Chris@42 159 {
Chris@42 160 E Ti, Tk, Tb, Tj, Ta;
Chris@42 161 Ti = FMA(KP951056516, Te, KP587785252 * Th);
Chris@42 162 Tk = FNMS(KP587785252, Te, KP951056516 * Th);
Chris@42 163 Ta = FNMS(KP250000000, T8, T1);
Chris@42 164 Tb = T9 + Ta;
Chris@42 165 Tj = Ta - T9;
Chris@42 166 ro[WS(os, 4)] = Tb - Ti;
Chris@42 167 ro[WS(os, 3)] = Tj + Tk;
Chris@42 168 ro[WS(os, 1)] = Tb + Ti;
Chris@42 169 ro[WS(os, 2)] = Tj - Tk;
Chris@42 170 }
Chris@42 171 {
Chris@42 172 E Tu, Tv, Tr, Tw, Tq;
Chris@42 173 Tu = FMA(KP951056516, Ts, KP587785252 * Tt);
Chris@42 174 Tv = FNMS(KP587785252, Ts, KP951056516 * Tt);
Chris@42 175 Tq = FNMS(KP250000000, Tp, To);
Chris@42 176 Tr = Tn + Tq;
Chris@42 177 Tw = Tq - Tn;
Chris@42 178 io[WS(os, 1)] = Tr - Tu;
Chris@42 179 io[WS(os, 3)] = Tw - Tv;
Chris@42 180 io[WS(os, 4)] = Tu + Tr;
Chris@42 181 io[WS(os, 2)] = Tv + Tw;
Chris@42 182 }
Chris@42 183 }
Chris@42 184 }
Chris@42 185 }
Chris@42 186
Chris@42 187 static const kdft_desc desc = { 5, "n1_5", {26, 6, 6, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 188
Chris@42 189 void X(codelet_n1_5) (planner *p) {
Chris@42 190 X(kdft_register) (p, n1_5, &desc);
Chris@42 191 }
Chris@42 192
Chris@42 193 #endif /* HAVE_FMA */