Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cb/r2cb_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 08:07:28 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 r2cb_10 -include rdft/scalar/r2cb.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 34 FP additions, 20 FP multiplications,
Chris@82 32 * (or, 14 additions, 0 multiplications, 20 fused multiply/add),
Chris@82 33 * 26 stack variables, 5 constants, and 20 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/r2cb.h"
Chris@82 36
Chris@82 37 static void r2cb_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(KP1_902113032, +1.902113032590307144232878666758764286811397268);
Chris@82 40 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
Chris@82 41 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
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 T3, Tb, Tn, Tu, Tk, Tv, Ta, Ts, Te, Tg, Ti, Tj;
Chris@82 48 {
Chris@82 49 E T1, T2, Tl, Tm;
Chris@82 50 T1 = Cr[0];
Chris@82 51 T2 = Cr[WS(csr, 5)];
Chris@82 52 T3 = T1 - T2;
Chris@82 53 Tb = T1 + T2;
Chris@82 54 Tl = Ci[WS(csi, 2)];
Chris@82 55 Tm = Ci[WS(csi, 3)];
Chris@82 56 Tn = Tl - Tm;
Chris@82 57 Tu = Tl + Tm;
Chris@82 58 }
Chris@82 59 Ti = Ci[WS(csi, 4)];
Chris@82 60 Tj = Ci[WS(csi, 1)];
Chris@82 61 Tk = Ti - Tj;
Chris@82 62 Tv = Ti + Tj;
Chris@82 63 {
Chris@82 64 E T6, Tc, T9, Td;
Chris@82 65 {
Chris@82 66 E T4, T5, T7, T8;
Chris@82 67 T4 = Cr[WS(csr, 2)];
Chris@82 68 T5 = Cr[WS(csr, 3)];
Chris@82 69 T6 = T4 - T5;
Chris@82 70 Tc = T4 + T5;
Chris@82 71 T7 = Cr[WS(csr, 4)];
Chris@82 72 T8 = Cr[WS(csr, 1)];
Chris@82 73 T9 = T7 - T8;
Chris@82 74 Td = T7 + T8;
Chris@82 75 }
Chris@82 76 Ta = T6 + T9;
Chris@82 77 Ts = T6 - T9;
Chris@82 78 Te = Tc + Td;
Chris@82 79 Tg = Tc - Td;
Chris@82 80 }
Chris@82 81 R1[WS(rs, 2)] = FMA(KP2_000000000, Ta, T3);
Chris@82 82 R0[0] = FMA(KP2_000000000, Te, Tb);
Chris@82 83 {
Chris@82 84 E To, Tq, Th, Tp, Tf;
Chris@82 85 To = FNMS(KP618033988, Tn, Tk);
Chris@82 86 Tq = FMA(KP618033988, Tk, Tn);
Chris@82 87 Tf = FNMS(KP500000000, Te, Tb);
Chris@82 88 Th = FNMS(KP1_118033988, Tg, Tf);
Chris@82 89 Tp = FMA(KP1_118033988, Tg, Tf);
Chris@82 90 R0[WS(rs, 4)] = FNMS(KP1_902113032, To, Th);
Chris@82 91 R0[WS(rs, 2)] = FMA(KP1_902113032, Tq, Tp);
Chris@82 92 R0[WS(rs, 1)] = FMA(KP1_902113032, To, Th);
Chris@82 93 R0[WS(rs, 3)] = FNMS(KP1_902113032, Tq, Tp);
Chris@82 94 }
Chris@82 95 {
Chris@82 96 E Tw, Ty, Tt, Tx, Tr;
Chris@82 97 Tw = FMA(KP618033988, Tv, Tu);
Chris@82 98 Ty = FNMS(KP618033988, Tu, Tv);
Chris@82 99 Tr = FNMS(KP500000000, Ta, T3);
Chris@82 100 Tt = FMA(KP1_118033988, Ts, Tr);
Chris@82 101 Tx = FNMS(KP1_118033988, Ts, Tr);
Chris@82 102 R1[0] = FNMS(KP1_902113032, Tw, Tt);
Chris@82 103 R1[WS(rs, 3)] = FMA(KP1_902113032, Ty, Tx);
Chris@82 104 R1[WS(rs, 4)] = FMA(KP1_902113032, Tw, Tt);
Chris@82 105 R1[WS(rs, 1)] = FNMS(KP1_902113032, Ty, Tx);
Chris@82 106 }
Chris@82 107 }
Chris@82 108 }
Chris@82 109 }
Chris@82 110
Chris@82 111 static const kr2c_desc desc = { 10, "r2cb_10", {14, 0, 20, 0}, &GENUS };
Chris@82 112
Chris@82 113 void X(codelet_r2cb_10) (planner *p) {
Chris@82 114 X(kr2c_register) (p, r2cb_10, &desc);
Chris@82 115 }
Chris@82 116
Chris@82 117 #else
Chris@82 118
Chris@82 119 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cb_10 -include rdft/scalar/r2cb.h */
Chris@82 120
Chris@82 121 /*
Chris@82 122 * This function contains 34 FP additions, 14 FP multiplications,
Chris@82 123 * (or, 26 additions, 6 multiplications, 8 fused multiply/add),
Chris@82 124 * 26 stack variables, 5 constants, and 20 memory accesses
Chris@82 125 */
Chris@82 126 #include "rdft/scalar/r2cb.h"
Chris@82 127
Chris@82 128 static void r2cb_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 129 {
Chris@82 130 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 131 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
Chris@82 132 DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
Chris@82 133 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 134 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
Chris@82 135 {
Chris@82 136 INT i;
Chris@82 137 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@82 138 E T3, Tb, Tn, Tv, Tk, Tu, Ta, Ts, Te, Tg, Ti, Tj;
Chris@82 139 {
Chris@82 140 E T1, T2, Tl, Tm;
Chris@82 141 T1 = Cr[0];
Chris@82 142 T2 = Cr[WS(csr, 5)];
Chris@82 143 T3 = T1 - T2;
Chris@82 144 Tb = T1 + T2;
Chris@82 145 Tl = Ci[WS(csi, 4)];
Chris@82 146 Tm = Ci[WS(csi, 1)];
Chris@82 147 Tn = Tl - Tm;
Chris@82 148 Tv = Tl + Tm;
Chris@82 149 }
Chris@82 150 Ti = Ci[WS(csi, 2)];
Chris@82 151 Tj = Ci[WS(csi, 3)];
Chris@82 152 Tk = Ti - Tj;
Chris@82 153 Tu = Ti + Tj;
Chris@82 154 {
Chris@82 155 E T6, Tc, T9, Td;
Chris@82 156 {
Chris@82 157 E T4, T5, T7, T8;
Chris@82 158 T4 = Cr[WS(csr, 2)];
Chris@82 159 T5 = Cr[WS(csr, 3)];
Chris@82 160 T6 = T4 - T5;
Chris@82 161 Tc = T4 + T5;
Chris@82 162 T7 = Cr[WS(csr, 4)];
Chris@82 163 T8 = Cr[WS(csr, 1)];
Chris@82 164 T9 = T7 - T8;
Chris@82 165 Td = T7 + T8;
Chris@82 166 }
Chris@82 167 Ta = T6 + T9;
Chris@82 168 Ts = KP1_118033988 * (T6 - T9);
Chris@82 169 Te = Tc + Td;
Chris@82 170 Tg = KP1_118033988 * (Tc - Td);
Chris@82 171 }
Chris@82 172 R1[WS(rs, 2)] = FMA(KP2_000000000, Ta, T3);
Chris@82 173 R0[0] = FMA(KP2_000000000, Te, Tb);
Chris@82 174 {
Chris@82 175 E To, Tq, Th, Tp, Tf;
Chris@82 176 To = FNMS(KP1_902113032, Tn, KP1_175570504 * Tk);
Chris@82 177 Tq = FMA(KP1_902113032, Tk, KP1_175570504 * Tn);
Chris@82 178 Tf = FNMS(KP500000000, Te, Tb);
Chris@82 179 Th = Tf - Tg;
Chris@82 180 Tp = Tg + Tf;
Chris@82 181 R0[WS(rs, 1)] = Th - To;
Chris@82 182 R0[WS(rs, 2)] = Tp + Tq;
Chris@82 183 R0[WS(rs, 4)] = Th + To;
Chris@82 184 R0[WS(rs, 3)] = Tp - Tq;
Chris@82 185 }
Chris@82 186 {
Chris@82 187 E Tw, Ty, Tt, Tx, Tr;
Chris@82 188 Tw = FNMS(KP1_902113032, Tv, KP1_175570504 * Tu);
Chris@82 189 Ty = FMA(KP1_902113032, Tu, KP1_175570504 * Tv);
Chris@82 190 Tr = FNMS(KP500000000, Ta, T3);
Chris@82 191 Tt = Tr - Ts;
Chris@82 192 Tx = Ts + Tr;
Chris@82 193 R1[WS(rs, 3)] = Tt - Tw;
Chris@82 194 R1[WS(rs, 4)] = Tx + Ty;
Chris@82 195 R1[WS(rs, 1)] = Tt + Tw;
Chris@82 196 R1[0] = Tx - Ty;
Chris@82 197 }
Chris@82 198 }
Chris@82 199 }
Chris@82 200 }
Chris@82 201
Chris@82 202 static const kr2c_desc desc = { 10, "r2cb_10", {26, 6, 8, 0}, &GENUS };
Chris@82 203
Chris@82 204 void X(codelet_r2cb_10) (planner *p) {
Chris@82 205 X(kr2c_register) (p, r2cb_10, &desc);
Chris@82 206 }
Chris@82 207
Chris@82 208 #endif