Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/scalar/codelets/n1_8.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 8 -name n1_8 -include n.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 52 FP additions, 8 FP multiplications,
Chris@42 32 * (or, 44 additions, 0 multiplications, 8 fused multiply/add),
Chris@42 33 * 36 stack variables, 1 constants, and 32 memory accesses
Chris@42 34 */
Chris@42 35 #include "n.h"
Chris@42 36
Chris@42 37 static void n1_8(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(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 40 {
Chris@42 41 INT i;
Chris@42 42 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) {
Chris@42 43 E TF, TE, TD, TI;
Chris@42 44 {
Chris@42 45 E Tn, T3, TC, Ti, TB, T6, To, Tl, Td, TN, Tz, TH, Ta, Tq, Tt;
Chris@42 46 E TM;
Chris@42 47 {
Chris@42 48 E T4, T5, Tj, Tk;
Chris@42 49 {
Chris@42 50 E T1, T2, Tg, Th;
Chris@42 51 T1 = ri[0];
Chris@42 52 T2 = ri[WS(is, 4)];
Chris@42 53 Tg = ii[0];
Chris@42 54 Th = ii[WS(is, 4)];
Chris@42 55 T4 = ri[WS(is, 2)];
Chris@42 56 Tn = T1 - T2;
Chris@42 57 T3 = T1 + T2;
Chris@42 58 TC = Tg - Th;
Chris@42 59 Ti = Tg + Th;
Chris@42 60 T5 = ri[WS(is, 6)];
Chris@42 61 }
Chris@42 62 Tj = ii[WS(is, 2)];
Chris@42 63 Tk = ii[WS(is, 6)];
Chris@42 64 {
Chris@42 65 E Tb, Tc, Tw, Tx;
Chris@42 66 Tb = ri[WS(is, 7)];
Chris@42 67 TB = T4 - T5;
Chris@42 68 T6 = T4 + T5;
Chris@42 69 To = Tj - Tk;
Chris@42 70 Tl = Tj + Tk;
Chris@42 71 Tc = ri[WS(is, 3)];
Chris@42 72 Tw = ii[WS(is, 7)];
Chris@42 73 Tx = ii[WS(is, 3)];
Chris@42 74 {
Chris@42 75 E T8, Tv, Ty, T9, Tr, Ts;
Chris@42 76 T8 = ri[WS(is, 1)];
Chris@42 77 Td = Tb + Tc;
Chris@42 78 Tv = Tb - Tc;
Chris@42 79 TN = Tw + Tx;
Chris@42 80 Ty = Tw - Tx;
Chris@42 81 T9 = ri[WS(is, 5)];
Chris@42 82 Tr = ii[WS(is, 1)];
Chris@42 83 Ts = ii[WS(is, 5)];
Chris@42 84 Tz = Tv - Ty;
Chris@42 85 TH = Tv + Ty;
Chris@42 86 Ta = T8 + T9;
Chris@42 87 Tq = T8 - T9;
Chris@42 88 Tt = Tr - Ts;
Chris@42 89 TM = Tr + Ts;
Chris@42 90 }
Chris@42 91 }
Chris@42 92 }
Chris@42 93 {
Chris@42 94 E TL, TG, Tu, Tf, Tm, TO;
Chris@42 95 {
Chris@42 96 E T7, Te, TP, TQ;
Chris@42 97 TL = T3 - T6;
Chris@42 98 T7 = T3 + T6;
Chris@42 99 TG = Tt - Tq;
Chris@42 100 Tu = Tq + Tt;
Chris@42 101 Te = Ta + Td;
Chris@42 102 Tf = Td - Ta;
Chris@42 103 Tm = Ti - Tl;
Chris@42 104 TP = Ti + Tl;
Chris@42 105 TQ = TM + TN;
Chris@42 106 TO = TM - TN;
Chris@42 107 ro[0] = T7 + Te;
Chris@42 108 ro[WS(os, 4)] = T7 - Te;
Chris@42 109 io[0] = TP + TQ;
Chris@42 110 io[WS(os, 4)] = TP - TQ;
Chris@42 111 }
Chris@42 112 {
Chris@42 113 E Tp, TA, TJ, TK;
Chris@42 114 TF = Tn - To;
Chris@42 115 Tp = Tn + To;
Chris@42 116 io[WS(os, 6)] = Tm - Tf;
Chris@42 117 io[WS(os, 2)] = Tf + Tm;
Chris@42 118 ro[WS(os, 2)] = TL + TO;
Chris@42 119 ro[WS(os, 6)] = TL - TO;
Chris@42 120 TA = Tu + Tz;
Chris@42 121 TE = Tz - Tu;
Chris@42 122 TD = TB + TC;
Chris@42 123 TJ = TC - TB;
Chris@42 124 TK = TG + TH;
Chris@42 125 TI = TG - TH;
Chris@42 126 ro[WS(os, 1)] = FMA(KP707106781, TA, Tp);
Chris@42 127 ro[WS(os, 5)] = FNMS(KP707106781, TA, Tp);
Chris@42 128 io[WS(os, 1)] = FMA(KP707106781, TK, TJ);
Chris@42 129 io[WS(os, 5)] = FNMS(KP707106781, TK, TJ);
Chris@42 130 }
Chris@42 131 }
Chris@42 132 }
Chris@42 133 io[WS(os, 3)] = FMA(KP707106781, TE, TD);
Chris@42 134 io[WS(os, 7)] = FNMS(KP707106781, TE, TD);
Chris@42 135 ro[WS(os, 3)] = FMA(KP707106781, TI, TF);
Chris@42 136 ro[WS(os, 7)] = FNMS(KP707106781, TI, TF);
Chris@42 137 }
Chris@42 138 }
Chris@42 139 }
Chris@42 140
Chris@42 141 static const kdft_desc desc = { 8, "n1_8", {44, 0, 8, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 142
Chris@42 143 void X(codelet_n1_8) (planner *p) {
Chris@42 144 X(kdft_register) (p, n1_8, &desc);
Chris@42 145 }
Chris@42 146
Chris@42 147 #else /* HAVE_FMA */
Chris@42 148
Chris@42 149 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 8 -name n1_8 -include n.h */
Chris@42 150
Chris@42 151 /*
Chris@42 152 * This function contains 52 FP additions, 4 FP multiplications,
Chris@42 153 * (or, 52 additions, 4 multiplications, 0 fused multiply/add),
Chris@42 154 * 28 stack variables, 1 constants, and 32 memory accesses
Chris@42 155 */
Chris@42 156 #include "n.h"
Chris@42 157
Chris@42 158 static void n1_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 159 {
Chris@42 160 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 161 {
Chris@42 162 INT i;
Chris@42 163 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) {
Chris@42 164 E T3, Tn, Ti, TC, T6, TB, Tl, To, Td, TN, Tz, TH, Ta, TM, Tu;
Chris@42 165 E TG;
Chris@42 166 {
Chris@42 167 E T1, T2, Tj, Tk;
Chris@42 168 T1 = ri[0];
Chris@42 169 T2 = ri[WS(is, 4)];
Chris@42 170 T3 = T1 + T2;
Chris@42 171 Tn = T1 - T2;
Chris@42 172 {
Chris@42 173 E Tg, Th, T4, T5;
Chris@42 174 Tg = ii[0];
Chris@42 175 Th = ii[WS(is, 4)];
Chris@42 176 Ti = Tg + Th;
Chris@42 177 TC = Tg - Th;
Chris@42 178 T4 = ri[WS(is, 2)];
Chris@42 179 T5 = ri[WS(is, 6)];
Chris@42 180 T6 = T4 + T5;
Chris@42 181 TB = T4 - T5;
Chris@42 182 }
Chris@42 183 Tj = ii[WS(is, 2)];
Chris@42 184 Tk = ii[WS(is, 6)];
Chris@42 185 Tl = Tj + Tk;
Chris@42 186 To = Tj - Tk;
Chris@42 187 {
Chris@42 188 E Tb, Tc, Tv, Tw, Tx, Ty;
Chris@42 189 Tb = ri[WS(is, 7)];
Chris@42 190 Tc = ri[WS(is, 3)];
Chris@42 191 Tv = Tb - Tc;
Chris@42 192 Tw = ii[WS(is, 7)];
Chris@42 193 Tx = ii[WS(is, 3)];
Chris@42 194 Ty = Tw - Tx;
Chris@42 195 Td = Tb + Tc;
Chris@42 196 TN = Tw + Tx;
Chris@42 197 Tz = Tv - Ty;
Chris@42 198 TH = Tv + Ty;
Chris@42 199 }
Chris@42 200 {
Chris@42 201 E T8, T9, Tq, Tr, Ts, Tt;
Chris@42 202 T8 = ri[WS(is, 1)];
Chris@42 203 T9 = ri[WS(is, 5)];
Chris@42 204 Tq = T8 - T9;
Chris@42 205 Tr = ii[WS(is, 1)];
Chris@42 206 Ts = ii[WS(is, 5)];
Chris@42 207 Tt = Tr - Ts;
Chris@42 208 Ta = T8 + T9;
Chris@42 209 TM = Tr + Ts;
Chris@42 210 Tu = Tq + Tt;
Chris@42 211 TG = Tt - Tq;
Chris@42 212 }
Chris@42 213 }
Chris@42 214 {
Chris@42 215 E T7, Te, TP, TQ;
Chris@42 216 T7 = T3 + T6;
Chris@42 217 Te = Ta + Td;
Chris@42 218 ro[WS(os, 4)] = T7 - Te;
Chris@42 219 ro[0] = T7 + Te;
Chris@42 220 TP = Ti + Tl;
Chris@42 221 TQ = TM + TN;
Chris@42 222 io[WS(os, 4)] = TP - TQ;
Chris@42 223 io[0] = TP + TQ;
Chris@42 224 }
Chris@42 225 {
Chris@42 226 E Tf, Tm, TL, TO;
Chris@42 227 Tf = Td - Ta;
Chris@42 228 Tm = Ti - Tl;
Chris@42 229 io[WS(os, 2)] = Tf + Tm;
Chris@42 230 io[WS(os, 6)] = Tm - Tf;
Chris@42 231 TL = T3 - T6;
Chris@42 232 TO = TM - TN;
Chris@42 233 ro[WS(os, 6)] = TL - TO;
Chris@42 234 ro[WS(os, 2)] = TL + TO;
Chris@42 235 }
Chris@42 236 {
Chris@42 237 E Tp, TA, TJ, TK;
Chris@42 238 Tp = Tn + To;
Chris@42 239 TA = KP707106781 * (Tu + Tz);
Chris@42 240 ro[WS(os, 5)] = Tp - TA;
Chris@42 241 ro[WS(os, 1)] = Tp + TA;
Chris@42 242 TJ = TC - TB;
Chris@42 243 TK = KP707106781 * (TG + TH);
Chris@42 244 io[WS(os, 5)] = TJ - TK;
Chris@42 245 io[WS(os, 1)] = TJ + TK;
Chris@42 246 }
Chris@42 247 {
Chris@42 248 E TD, TE, TF, TI;
Chris@42 249 TD = TB + TC;
Chris@42 250 TE = KP707106781 * (Tz - Tu);
Chris@42 251 io[WS(os, 7)] = TD - TE;
Chris@42 252 io[WS(os, 3)] = TD + TE;
Chris@42 253 TF = Tn - To;
Chris@42 254 TI = KP707106781 * (TG - TH);
Chris@42 255 ro[WS(os, 7)] = TF - TI;
Chris@42 256 ro[WS(os, 3)] = TF + TI;
Chris@42 257 }
Chris@42 258 }
Chris@42 259 }
Chris@42 260 }
Chris@42 261
Chris@42 262 static const kdft_desc desc = { 8, "n1_8", {52, 4, 0, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 263
Chris@42 264 void X(codelet_n1_8) (planner *p) {
Chris@42 265 X(kdft_register) (p, n1_8, &desc);
Chris@42 266 }
Chris@42 267
Chris@42 268 #endif /* HAVE_FMA */