Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cf/r2cf_16.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 08:06:26 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_r2cf.native -fma -compact -variables 4 -pipeline-latency 4 -n 16 -name r2cf_16 -include rdft/scalar/r2cf.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 58 FP additions, 20 FP multiplications,
Chris@82 32 * (or, 38 additions, 0 multiplications, 20 fused multiply/add),
Chris@82 33 * 34 stack variables, 3 constants, and 32 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/r2cf.h"
Chris@82 36
Chris@82 37 static void r2cf_16(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(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82 40 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 41 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@82 42 {
Chris@82 43 INT i;
Chris@82 44 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) {
Chris@82 45 E T3, T6, T7, TN, TB, Ta, Td, Te, TO, TE, Tm, TT, Ty, TI, Tt;
Chris@82 46 E TS, Tz, TL, TC, TD, TR, TU;
Chris@82 47 {
Chris@82 48 E T1, T2, T4, T5;
Chris@82 49 T1 = R0[0];
Chris@82 50 T2 = R0[WS(rs, 4)];
Chris@82 51 T3 = T1 + T2;
Chris@82 52 T4 = R0[WS(rs, 2)];
Chris@82 53 T5 = R0[WS(rs, 6)];
Chris@82 54 T6 = T4 + T5;
Chris@82 55 T7 = T3 + T6;
Chris@82 56 TN = T4 - T5;
Chris@82 57 TB = T1 - T2;
Chris@82 58 }
Chris@82 59 {
Chris@82 60 E T8, T9, Tb, Tc;
Chris@82 61 T8 = R0[WS(rs, 1)];
Chris@82 62 T9 = R0[WS(rs, 5)];
Chris@82 63 Ta = T8 + T9;
Chris@82 64 TC = T8 - T9;
Chris@82 65 Tb = R0[WS(rs, 7)];
Chris@82 66 Tc = R0[WS(rs, 3)];
Chris@82 67 Td = Tb + Tc;
Chris@82 68 TD = Tb - Tc;
Chris@82 69 }
Chris@82 70 Te = Ta + Td;
Chris@82 71 TO = TD - TC;
Chris@82 72 TE = TC + TD;
Chris@82 73 {
Chris@82 74 E Ti, TG, Tl, TH;
Chris@82 75 {
Chris@82 76 E Tg, Th, Tj, Tk;
Chris@82 77 Tg = R1[0];
Chris@82 78 Th = R1[WS(rs, 4)];
Chris@82 79 Ti = Tg + Th;
Chris@82 80 TG = Tg - Th;
Chris@82 81 Tj = R1[WS(rs, 2)];
Chris@82 82 Tk = R1[WS(rs, 6)];
Chris@82 83 Tl = Tj + Tk;
Chris@82 84 TH = Tj - Tk;
Chris@82 85 }
Chris@82 86 Tm = Ti - Tl;
Chris@82 87 TT = FMA(KP414213562, TG, TH);
Chris@82 88 Ty = Ti + Tl;
Chris@82 89 TI = FNMS(KP414213562, TH, TG);
Chris@82 90 }
Chris@82 91 {
Chris@82 92 E Tp, TJ, Ts, TK;
Chris@82 93 {
Chris@82 94 E Tn, To, Tq, Tr;
Chris@82 95 Tn = R1[WS(rs, 7)];
Chris@82 96 To = R1[WS(rs, 3)];
Chris@82 97 Tp = Tn + To;
Chris@82 98 TJ = Tn - To;
Chris@82 99 Tq = R1[WS(rs, 1)];
Chris@82 100 Tr = R1[WS(rs, 5)];
Chris@82 101 Ts = Tq + Tr;
Chris@82 102 TK = Tr - Tq;
Chris@82 103 }
Chris@82 104 Tt = Tp - Ts;
Chris@82 105 TS = FMA(KP414213562, TJ, TK);
Chris@82 106 Tz = Tp + Ts;
Chris@82 107 TL = FNMS(KP414213562, TK, TJ);
Chris@82 108 }
Chris@82 109 Cr[WS(csr, 4)] = T7 - Te;
Chris@82 110 Ci[WS(csi, 4)] = Tz - Ty;
Chris@82 111 {
Chris@82 112 E Tf, Tu, Tv, Tw;
Chris@82 113 Tf = T3 - T6;
Chris@82 114 Tu = Tm + Tt;
Chris@82 115 Cr[WS(csr, 6)] = FNMS(KP707106781, Tu, Tf);
Chris@82 116 Cr[WS(csr, 2)] = FMA(KP707106781, Tu, Tf);
Chris@82 117 Tv = Td - Ta;
Chris@82 118 Tw = Tt - Tm;
Chris@82 119 Ci[WS(csi, 2)] = FMA(KP707106781, Tw, Tv);
Chris@82 120 Ci[WS(csi, 6)] = FMS(KP707106781, Tw, Tv);
Chris@82 121 }
Chris@82 122 {
Chris@82 123 E Tx, TA, TF, TM;
Chris@82 124 Tx = T7 + Te;
Chris@82 125 TA = Ty + Tz;
Chris@82 126 Cr[WS(csr, 8)] = Tx - TA;
Chris@82 127 Cr[0] = Tx + TA;
Chris@82 128 TF = FMA(KP707106781, TE, TB);
Chris@82 129 TM = TI + TL;
Chris@82 130 Cr[WS(csr, 7)] = FNMS(KP923879532, TM, TF);
Chris@82 131 Cr[WS(csr, 1)] = FMA(KP923879532, TM, TF);
Chris@82 132 }
Chris@82 133 TR = FNMS(KP707106781, TO, TN);
Chris@82 134 TU = TS - TT;
Chris@82 135 Ci[WS(csi, 1)] = FMS(KP923879532, TU, TR);
Chris@82 136 Ci[WS(csi, 7)] = FMA(KP923879532, TU, TR);
Chris@82 137 {
Chris@82 138 E TV, TW, TP, TQ;
Chris@82 139 TV = FNMS(KP707106781, TE, TB);
Chris@82 140 TW = TT + TS;
Chris@82 141 Cr[WS(csr, 5)] = FNMS(KP923879532, TW, TV);
Chris@82 142 Cr[WS(csr, 3)] = FMA(KP923879532, TW, TV);
Chris@82 143 TP = FMA(KP707106781, TO, TN);
Chris@82 144 TQ = TL - TI;
Chris@82 145 Ci[WS(csi, 3)] = FMA(KP923879532, TQ, TP);
Chris@82 146 Ci[WS(csi, 5)] = FMS(KP923879532, TQ, TP);
Chris@82 147 }
Chris@82 148 }
Chris@82 149 }
Chris@82 150 }
Chris@82 151
Chris@82 152 static const kr2c_desc desc = { 16, "r2cf_16", {38, 0, 20, 0}, &GENUS };
Chris@82 153
Chris@82 154 void X(codelet_r2cf_16) (planner *p) {
Chris@82 155 X(kr2c_register) (p, r2cf_16, &desc);
Chris@82 156 }
Chris@82 157
Chris@82 158 #else
Chris@82 159
Chris@82 160 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 16 -name r2cf_16 -include rdft/scalar/r2cf.h */
Chris@82 161
Chris@82 162 /*
Chris@82 163 * This function contains 58 FP additions, 12 FP multiplications,
Chris@82 164 * (or, 54 additions, 8 multiplications, 4 fused multiply/add),
Chris@82 165 * 34 stack variables, 3 constants, and 32 memory accesses
Chris@82 166 */
Chris@82 167 #include "rdft/scalar/r2cf.h"
Chris@82 168
Chris@82 169 static void r2cf_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 170 {
Chris@82 171 DK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82 172 DK(KP382683432, +0.382683432365089771728459984030398866761344562);
Chris@82 173 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 174 {
Chris@82 175 INT i;
Chris@82 176 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) {
Chris@82 177 E T3, T6, T7, Tz, Ti, Ta, Td, Te, TA, Th, Tq, TV, TF, TP, Tx;
Chris@82 178 E TU, TE, TM, Tg, Tf, TJ, TQ;
Chris@82 179 {
Chris@82 180 E T1, T2, T4, T5;
Chris@82 181 T1 = R0[0];
Chris@82 182 T2 = R0[WS(rs, 4)];
Chris@82 183 T3 = T1 + T2;
Chris@82 184 T4 = R0[WS(rs, 2)];
Chris@82 185 T5 = R0[WS(rs, 6)];
Chris@82 186 T6 = T4 + T5;
Chris@82 187 T7 = T3 + T6;
Chris@82 188 Tz = T1 - T2;
Chris@82 189 Ti = T4 - T5;
Chris@82 190 }
Chris@82 191 {
Chris@82 192 E T8, T9, Tb, Tc;
Chris@82 193 T8 = R0[WS(rs, 1)];
Chris@82 194 T9 = R0[WS(rs, 5)];
Chris@82 195 Ta = T8 + T9;
Chris@82 196 Tg = T8 - T9;
Chris@82 197 Tb = R0[WS(rs, 7)];
Chris@82 198 Tc = R0[WS(rs, 3)];
Chris@82 199 Td = Tb + Tc;
Chris@82 200 Tf = Tb - Tc;
Chris@82 201 }
Chris@82 202 Te = Ta + Td;
Chris@82 203 TA = KP707106781 * (Tg + Tf);
Chris@82 204 Th = KP707106781 * (Tf - Tg);
Chris@82 205 {
Chris@82 206 E Tm, TN, Tp, TO;
Chris@82 207 {
Chris@82 208 E Tk, Tl, Tn, To;
Chris@82 209 Tk = R1[WS(rs, 7)];
Chris@82 210 Tl = R1[WS(rs, 3)];
Chris@82 211 Tm = Tk - Tl;
Chris@82 212 TN = Tk + Tl;
Chris@82 213 Tn = R1[WS(rs, 1)];
Chris@82 214 To = R1[WS(rs, 5)];
Chris@82 215 Tp = Tn - To;
Chris@82 216 TO = Tn + To;
Chris@82 217 }
Chris@82 218 Tq = FNMS(KP923879532, Tp, KP382683432 * Tm);
Chris@82 219 TV = TN + TO;
Chris@82 220 TF = FMA(KP923879532, Tm, KP382683432 * Tp);
Chris@82 221 TP = TN - TO;
Chris@82 222 }
Chris@82 223 {
Chris@82 224 E Tt, TK, Tw, TL;
Chris@82 225 {
Chris@82 226 E Tr, Ts, Tu, Tv;
Chris@82 227 Tr = R1[0];
Chris@82 228 Ts = R1[WS(rs, 4)];
Chris@82 229 Tt = Tr - Ts;
Chris@82 230 TK = Tr + Ts;
Chris@82 231 Tu = R1[WS(rs, 2)];
Chris@82 232 Tv = R1[WS(rs, 6)];
Chris@82 233 Tw = Tu - Tv;
Chris@82 234 TL = Tu + Tv;
Chris@82 235 }
Chris@82 236 Tx = FMA(KP382683432, Tt, KP923879532 * Tw);
Chris@82 237 TU = TK + TL;
Chris@82 238 TE = FNMS(KP382683432, Tw, KP923879532 * Tt);
Chris@82 239 TM = TK - TL;
Chris@82 240 }
Chris@82 241 Cr[WS(csr, 4)] = T7 - Te;
Chris@82 242 Ci[WS(csi, 4)] = TV - TU;
Chris@82 243 {
Chris@82 244 E Tj, Ty, TD, TG;
Chris@82 245 Tj = Th - Ti;
Chris@82 246 Ty = Tq - Tx;
Chris@82 247 Ci[WS(csi, 1)] = Tj + Ty;
Chris@82 248 Ci[WS(csi, 7)] = Ty - Tj;
Chris@82 249 TD = Tz + TA;
Chris@82 250 TG = TE + TF;
Chris@82 251 Cr[WS(csr, 7)] = TD - TG;
Chris@82 252 Cr[WS(csr, 1)] = TD + TG;
Chris@82 253 }
Chris@82 254 {
Chris@82 255 E TB, TC, TH, TI;
Chris@82 256 TB = Tz - TA;
Chris@82 257 TC = Tx + Tq;
Chris@82 258 Cr[WS(csr, 5)] = TB - TC;
Chris@82 259 Cr[WS(csr, 3)] = TB + TC;
Chris@82 260 TH = Ti + Th;
Chris@82 261 TI = TF - TE;
Chris@82 262 Ci[WS(csi, 3)] = TH + TI;
Chris@82 263 Ci[WS(csi, 5)] = TI - TH;
Chris@82 264 }
Chris@82 265 TJ = T3 - T6;
Chris@82 266 TQ = KP707106781 * (TM + TP);
Chris@82 267 Cr[WS(csr, 6)] = TJ - TQ;
Chris@82 268 Cr[WS(csr, 2)] = TJ + TQ;
Chris@82 269 {
Chris@82 270 E TR, TS, TT, TW;
Chris@82 271 TR = Td - Ta;
Chris@82 272 TS = KP707106781 * (TP - TM);
Chris@82 273 Ci[WS(csi, 2)] = TR + TS;
Chris@82 274 Ci[WS(csi, 6)] = TS - TR;
Chris@82 275 TT = T7 + Te;
Chris@82 276 TW = TU + TV;
Chris@82 277 Cr[WS(csr, 8)] = TT - TW;
Chris@82 278 Cr[0] = TT + TW;
Chris@82 279 }
Chris@82 280 }
Chris@82 281 }
Chris@82 282 }
Chris@82 283
Chris@82 284 static const kr2c_desc desc = { 16, "r2cf_16", {54, 8, 4, 0}, &GENUS };
Chris@82 285
Chris@82 286 void X(codelet_r2cf_16) (planner *p) {
Chris@82 287 X(kr2c_register) (p, r2cf_16, &desc);
Chris@82 288 }
Chris@82 289
Chris@82 290 #endif