Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cb/r2cbIII_10.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 08:07:44 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_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cbIII_10 -dft-III -include rdft/scalar/r2cbIII.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 32 FP additions, 28 FP multiplications,
Chris@82 32 * (or, 14 additions, 10 multiplications, 18 fused multiply/add),
Chris@82 33 * 22 stack variables, 5 constants, and 20 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/r2cbIII.h"
Chris@82 36
Chris@82 37 static void r2cbIII_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 38 {
Chris@82 39 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 41 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 42 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82 43 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 44 {
Chris@82 45 INT i;
Chris@82 46 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) {
Chris@82 47 E T1, To, T8, Tt, Ta, Ts, Te, Tq, Th, Tn;
Chris@82 48 T1 = Cr[WS(csr, 2)];
Chris@82 49 To = Ci[WS(csi, 2)];
Chris@82 50 {
Chris@82 51 E T2, T3, T4, T5, T6, T7;
Chris@82 52 T2 = Cr[WS(csr, 4)];
Chris@82 53 T3 = Cr[0];
Chris@82 54 T4 = T2 + T3;
Chris@82 55 T5 = Cr[WS(csr, 3)];
Chris@82 56 T6 = Cr[WS(csr, 1)];
Chris@82 57 T7 = T5 + T6;
Chris@82 58 T8 = T4 + T7;
Chris@82 59 Tt = T5 - T6;
Chris@82 60 Ta = T7 - T4;
Chris@82 61 Ts = T2 - T3;
Chris@82 62 }
Chris@82 63 {
Chris@82 64 E Tc, Td, Tl, Tf, Tg, Tm;
Chris@82 65 Tc = Ci[WS(csi, 3)];
Chris@82 66 Td = Ci[WS(csi, 1)];
Chris@82 67 Tl = Tc + Td;
Chris@82 68 Tf = Ci[WS(csi, 4)];
Chris@82 69 Tg = Ci[0];
Chris@82 70 Tm = Tf + Tg;
Chris@82 71 Te = Tc - Td;
Chris@82 72 Tq = Tl + Tm;
Chris@82 73 Th = Tf - Tg;
Chris@82 74 Tn = Tl - Tm;
Chris@82 75 }
Chris@82 76 R0[0] = KP2_000000000 * (T1 + T8);
Chris@82 77 R1[WS(rs, 2)] = KP2_000000000 * (Tn - To);
Chris@82 78 {
Chris@82 79 E Ti, Tk, Tb, Tj, T9;
Chris@82 80 Ti = FMA(KP618033988, Th, Te);
Chris@82 81 Tk = FNMS(KP618033988, Te, Th);
Chris@82 82 T9 = FMS(KP250000000, T8, T1);
Chris@82 83 Tb = FNMS(KP559016994, Ta, T9);
Chris@82 84 Tj = FMA(KP559016994, Ta, T9);
Chris@82 85 R0[WS(rs, 1)] = KP2_000000000 * (FMA(KP951056516, Ti, Tb));
Chris@82 86 R0[WS(rs, 3)] = KP2_000000000 * (FMA(KP951056516, Tk, Tj));
Chris@82 87 R0[WS(rs, 4)] = -(KP2_000000000 * (FNMS(KP951056516, Ti, Tb)));
Chris@82 88 R0[WS(rs, 2)] = -(KP2_000000000 * (FNMS(KP951056516, Tk, Tj)));
Chris@82 89 }
Chris@82 90 {
Chris@82 91 E Tu, Tw, Tr, Tv, Tp;
Chris@82 92 Tu = FMA(KP618033988, Tt, Ts);
Chris@82 93 Tw = FNMS(KP618033988, Ts, Tt);
Chris@82 94 Tp = FMA(KP250000000, Tn, To);
Chris@82 95 Tr = FMA(KP559016994, Tq, Tp);
Chris@82 96 Tv = FNMS(KP559016994, Tq, Tp);
Chris@82 97 R1[0] = -(KP2_000000000 * (FMA(KP951056516, Tu, Tr)));
Chris@82 98 R1[WS(rs, 3)] = KP2_000000000 * (FNMS(KP951056516, Tw, Tv));
Chris@82 99 R1[WS(rs, 4)] = -(KP2_000000000 * (FNMS(KP951056516, Tu, Tr)));
Chris@82 100 R1[WS(rs, 1)] = KP2_000000000 * (FMA(KP951056516, Tw, Tv));
Chris@82 101 }
Chris@82 102 }
Chris@82 103 }
Chris@82 104 }
Chris@82 105
Chris@82 106 static const kr2c_desc desc = { 10, "r2cbIII_10", {14, 10, 18, 0}, &GENUS };
Chris@82 107
Chris@82 108 void X(codelet_r2cbIII_10) (planner *p) {
Chris@82 109 X(kr2c_register) (p, r2cbIII_10, &desc);
Chris@82 110 }
Chris@82 111
Chris@82 112 #else
Chris@82 113
Chris@82 114 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cbIII_10 -dft-III -include rdft/scalar/r2cbIII.h */
Chris@82 115
Chris@82 116 /*
Chris@82 117 * This function contains 32 FP additions, 16 FP multiplications,
Chris@82 118 * (or, 26 additions, 10 multiplications, 6 fused multiply/add),
Chris@82 119 * 22 stack variables, 5 constants, and 20 memory accesses
Chris@82 120 */
Chris@82 121 #include "rdft/scalar/r2cbIII.h"
Chris@82 122
Chris@82 123 static void r2cbIII_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 124 {
Chris@82 125 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 126 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
Chris@82 127 DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
Chris@82 128 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 129 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
Chris@82 130 {
Chris@82 131 INT i;
Chris@82 132 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) {
Chris@82 133 E T1, To, T8, Tq, Ta, Tp, Te, Ts, Th, Tn;
Chris@82 134 T1 = Cr[WS(csr, 2)];
Chris@82 135 To = Ci[WS(csi, 2)];
Chris@82 136 {
Chris@82 137 E T2, T3, T4, T5, T6, T7;
Chris@82 138 T2 = Cr[WS(csr, 4)];
Chris@82 139 T3 = Cr[0];
Chris@82 140 T4 = T2 + T3;
Chris@82 141 T5 = Cr[WS(csr, 3)];
Chris@82 142 T6 = Cr[WS(csr, 1)];
Chris@82 143 T7 = T5 + T6;
Chris@82 144 T8 = T4 + T7;
Chris@82 145 Tq = T5 - T6;
Chris@82 146 Ta = KP1_118033988 * (T7 - T4);
Chris@82 147 Tp = T2 - T3;
Chris@82 148 }
Chris@82 149 {
Chris@82 150 E Tc, Td, Tm, Tf, Tg, Tl;
Chris@82 151 Tc = Ci[WS(csi, 4)];
Chris@82 152 Td = Ci[0];
Chris@82 153 Tm = Tc + Td;
Chris@82 154 Tf = Ci[WS(csi, 1)];
Chris@82 155 Tg = Ci[WS(csi, 3)];
Chris@82 156 Tl = Tg + Tf;
Chris@82 157 Te = Tc - Td;
Chris@82 158 Ts = KP1_118033988 * (Tl + Tm);
Chris@82 159 Th = Tf - Tg;
Chris@82 160 Tn = Tl - Tm;
Chris@82 161 }
Chris@82 162 R0[0] = KP2_000000000 * (T1 + T8);
Chris@82 163 R1[WS(rs, 2)] = KP2_000000000 * (Tn - To);
Chris@82 164 {
Chris@82 165 E Ti, Tj, Tb, Tk, T9;
Chris@82 166 Ti = FNMS(KP1_902113032, Th, KP1_175570504 * Te);
Chris@82 167 Tj = FMA(KP1_175570504, Th, KP1_902113032 * Te);
Chris@82 168 T9 = FNMS(KP2_000000000, T1, KP500000000 * T8);
Chris@82 169 Tb = T9 - Ta;
Chris@82 170 Tk = T9 + Ta;
Chris@82 171 R0[WS(rs, 1)] = Tb + Ti;
Chris@82 172 R0[WS(rs, 3)] = Tk + Tj;
Chris@82 173 R0[WS(rs, 4)] = Ti - Tb;
Chris@82 174 R0[WS(rs, 2)] = Tj - Tk;
Chris@82 175 }
Chris@82 176 {
Chris@82 177 E Tr, Tv, Tu, Tw, Tt;
Chris@82 178 Tr = FMA(KP1_902113032, Tp, KP1_175570504 * Tq);
Chris@82 179 Tv = FNMS(KP1_175570504, Tp, KP1_902113032 * Tq);
Chris@82 180 Tt = FMA(KP500000000, Tn, KP2_000000000 * To);
Chris@82 181 Tu = Ts + Tt;
Chris@82 182 Tw = Tt - Ts;
Chris@82 183 R1[0] = -(Tr + Tu);
Chris@82 184 R1[WS(rs, 3)] = Tw - Tv;
Chris@82 185 R1[WS(rs, 4)] = Tr - Tu;
Chris@82 186 R1[WS(rs, 1)] = Tv + Tw;
Chris@82 187 }
Chris@82 188 }
Chris@82 189 }
Chris@82 190 }
Chris@82 191
Chris@82 192 static const kr2c_desc desc = { 10, "r2cbIII_10", {26, 10, 6, 0}, &GENUS };
Chris@82 193
Chris@82 194 void X(codelet_r2cbIII_10) (planner *p) {
Chris@82 195 X(kr2c_register) (p, r2cbIII_10, &desc);
Chris@82 196 }
Chris@82 197
Chris@82 198 #endif