Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cb/r2cbIII_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 08:07:44 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 16 -name r2cbIII_16 -dft-III -include rdft/scalar/r2cbIII.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 66 FP additions, 36 FP multiplications,
Chris@82 32 * (or, 46 additions, 16 multiplications, 20 fused multiply/add),
Chris@82 33 * 40 stack variables, 9 constants, and 32 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/r2cbIII.h"
Chris@82 36
Chris@82 37 static void r2cbIII_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(KP198912367, +0.198912367379658006911597622644676228597850501);
Chris@82 40 DK(KP1_961570560, +1.961570560806460898252364472268478073947867462);
Chris@82 41 DK(KP668178637, +0.668178637919298919997757686523080761552472251);
Chris@82 42 DK(KP1_662939224, +1.662939224605090474157576755235811513477121624);
Chris@82 43 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 44 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
Chris@82 45 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@82 46 DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
Chris@82 47 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 48 {
Chris@82 49 INT i;
Chris@82 50 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) {
Chris@82 51 E T7, TW, T13, Tj, TA, TK, TP, TH, Te, TX, T12, To, Tt, TC, TS;
Chris@82 52 E TB, TT, TY;
Chris@82 53 {
Chris@82 54 E T3, Tf, Tz, TU, T6, Tw, Ti, TV;
Chris@82 55 {
Chris@82 56 E T1, T2, Tx, Ty;
Chris@82 57 T1 = Cr[0];
Chris@82 58 T2 = Cr[WS(csr, 7)];
Chris@82 59 T3 = T1 + T2;
Chris@82 60 Tf = T1 - T2;
Chris@82 61 Tx = Ci[0];
Chris@82 62 Ty = Ci[WS(csi, 7)];
Chris@82 63 Tz = Tx + Ty;
Chris@82 64 TU = Ty - Tx;
Chris@82 65 }
Chris@82 66 {
Chris@82 67 E T4, T5, Tg, Th;
Chris@82 68 T4 = Cr[WS(csr, 4)];
Chris@82 69 T5 = Cr[WS(csr, 3)];
Chris@82 70 T6 = T4 + T5;
Chris@82 71 Tw = T4 - T5;
Chris@82 72 Tg = Ci[WS(csi, 4)];
Chris@82 73 Th = Ci[WS(csi, 3)];
Chris@82 74 Ti = Tg + Th;
Chris@82 75 TV = Th - Tg;
Chris@82 76 }
Chris@82 77 T7 = T3 + T6;
Chris@82 78 TW = TU - TV;
Chris@82 79 T13 = TV + TU;
Chris@82 80 Tj = Tf - Ti;
Chris@82 81 TA = Tw + Tz;
Chris@82 82 TK = Tw - Tz;
Chris@82 83 TP = T3 - T6;
Chris@82 84 TH = Tf + Ti;
Chris@82 85 }
Chris@82 86 {
Chris@82 87 E Ta, Tk, Tn, TR, Td, Tp, Ts, TQ;
Chris@82 88 {
Chris@82 89 E T8, T9, Tl, Tm;
Chris@82 90 T8 = Cr[WS(csr, 2)];
Chris@82 91 T9 = Cr[WS(csr, 5)];
Chris@82 92 Ta = T8 + T9;
Chris@82 93 Tk = T8 - T9;
Chris@82 94 Tl = Ci[WS(csi, 2)];
Chris@82 95 Tm = Ci[WS(csi, 5)];
Chris@82 96 Tn = Tl + Tm;
Chris@82 97 TR = Tl - Tm;
Chris@82 98 }
Chris@82 99 {
Chris@82 100 E Tb, Tc, Tq, Tr;
Chris@82 101 Tb = Cr[WS(csr, 1)];
Chris@82 102 Tc = Cr[WS(csr, 6)];
Chris@82 103 Td = Tb + Tc;
Chris@82 104 Tp = Tb - Tc;
Chris@82 105 Tq = Ci[WS(csi, 1)];
Chris@82 106 Tr = Ci[WS(csi, 6)];
Chris@82 107 Ts = Tq + Tr;
Chris@82 108 TQ = Tr - Tq;
Chris@82 109 }
Chris@82 110 Te = Ta + Td;
Chris@82 111 TX = Ta - Td;
Chris@82 112 T12 = TR + TQ;
Chris@82 113 To = Tk - Tn;
Chris@82 114 Tt = Tp - Ts;
Chris@82 115 TC = Tk + Tn;
Chris@82 116 TS = TQ - TR;
Chris@82 117 TB = Tp + Ts;
Chris@82 118 }
Chris@82 119 R0[0] = KP2_000000000 * (T7 + Te);
Chris@82 120 R0[WS(rs, 4)] = KP2_000000000 * (T13 - T12);
Chris@82 121 TT = TP + TS;
Chris@82 122 TY = TW - TX;
Chris@82 123 R0[WS(rs, 1)] = KP1_847759065 * (FMA(KP414213562, TY, TT));
Chris@82 124 R0[WS(rs, 5)] = KP1_847759065 * (FNMS(KP414213562, TT, TY));
Chris@82 125 {
Chris@82 126 E T11, T14, TZ, T10;
Chris@82 127 T11 = T7 - Te;
Chris@82 128 T14 = T12 + T13;
Chris@82 129 R0[WS(rs, 2)] = KP1_414213562 * (T11 + T14);
Chris@82 130 R0[WS(rs, 6)] = KP1_414213562 * (T14 - T11);
Chris@82 131 TZ = TX + TW;
Chris@82 132 T10 = TP - TS;
Chris@82 133 R0[WS(rs, 3)] = KP1_847759065 * (FMA(KP414213562, T10, TZ));
Chris@82 134 R0[WS(rs, 7)] = -(KP1_847759065 * (FNMS(KP414213562, TZ, T10)));
Chris@82 135 }
Chris@82 136 {
Chris@82 137 E TJ, TO, TM, TN, TI, TL;
Chris@82 138 TI = TC + TB;
Chris@82 139 TJ = FNMS(KP707106781, TI, TH);
Chris@82 140 TO = FMA(KP707106781, TI, TH);
Chris@82 141 TL = To - Tt;
Chris@82 142 TM = FNMS(KP707106781, TL, TK);
Chris@82 143 TN = FMA(KP707106781, TL, TK);
Chris@82 144 R1[WS(rs, 1)] = KP1_662939224 * (FMA(KP668178637, TM, TJ));
Chris@82 145 R1[WS(rs, 7)] = -(KP1_961570560 * (FNMS(KP198912367, TN, TO)));
Chris@82 146 R1[WS(rs, 5)] = KP1_662939224 * (FNMS(KP668178637, TJ, TM));
Chris@82 147 R1[WS(rs, 3)] = KP1_961570560 * (FMA(KP198912367, TO, TN));
Chris@82 148 }
Chris@82 149 {
Chris@82 150 E Tv, TG, TE, TF, Tu, TD;
Chris@82 151 Tu = To + Tt;
Chris@82 152 Tv = FMA(KP707106781, Tu, Tj);
Chris@82 153 TG = FNMS(KP707106781, Tu, Tj);
Chris@82 154 TD = TB - TC;
Chris@82 155 TE = FNMS(KP707106781, TD, TA);
Chris@82 156 TF = FMA(KP707106781, TD, TA);
Chris@82 157 R1[0] = KP1_961570560 * (FNMS(KP198912367, TE, Tv));
Chris@82 158 R1[WS(rs, 6)] = -(KP1_662939224 * (FMA(KP668178637, TF, TG)));
Chris@82 159 R1[WS(rs, 4)] = -(KP1_961570560 * (FMA(KP198912367, Tv, TE)));
Chris@82 160 R1[WS(rs, 2)] = -(KP1_662939224 * (FNMS(KP668178637, TG, TF)));
Chris@82 161 }
Chris@82 162 }
Chris@82 163 }
Chris@82 164 }
Chris@82 165
Chris@82 166 static const kr2c_desc desc = { 16, "r2cbIII_16", {46, 16, 20, 0}, &GENUS };
Chris@82 167
Chris@82 168 void X(codelet_r2cbIII_16) (planner *p) {
Chris@82 169 X(kr2c_register) (p, r2cbIII_16, &desc);
Chris@82 170 }
Chris@82 171
Chris@82 172 #else
Chris@82 173
Chris@82 174 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 16 -name r2cbIII_16 -dft-III -include rdft/scalar/r2cbIII.h */
Chris@82 175
Chris@82 176 /*
Chris@82 177 * This function contains 66 FP additions, 32 FP multiplications,
Chris@82 178 * (or, 54 additions, 20 multiplications, 12 fused multiply/add),
Chris@82 179 * 40 stack variables, 9 constants, and 32 memory accesses
Chris@82 180 */
Chris@82 181 #include "rdft/scalar/r2cbIII.h"
Chris@82 182
Chris@82 183 static void r2cbIII_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 184 {
Chris@82 185 DK(KP1_961570560, +1.961570560806460898252364472268478073947867462);
Chris@82 186 DK(KP390180644, +0.390180644032256535696569736954044481855383236);
Chris@82 187 DK(KP1_111140466, +1.111140466039204449485661627897065748749874382);
Chris@82 188 DK(KP1_662939224, +1.662939224605090474157576755235811513477121624);
Chris@82 189 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 190 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
Chris@82 191 DK(KP765366864, +0.765366864730179543456919968060797733522689125);
Chris@82 192 DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
Chris@82 193 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 194 {
Chris@82 195 INT i;
Chris@82 196 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) {
Chris@82 197 E T7, TW, T13, Tj, TD, TK, TP, TH, Te, TX, T12, To, Tt, Tx, TS;
Chris@82 198 E Tw, TT, TY;
Chris@82 199 {
Chris@82 200 E T3, Tf, TC, TV, T6, Tz, Ti, TU;
Chris@82 201 {
Chris@82 202 E T1, T2, TA, TB;
Chris@82 203 T1 = Cr[0];
Chris@82 204 T2 = Cr[WS(csr, 7)];
Chris@82 205 T3 = T1 + T2;
Chris@82 206 Tf = T1 - T2;
Chris@82 207 TA = Ci[0];
Chris@82 208 TB = Ci[WS(csi, 7)];
Chris@82 209 TC = TA + TB;
Chris@82 210 TV = TB - TA;
Chris@82 211 }
Chris@82 212 {
Chris@82 213 E T4, T5, Tg, Th;
Chris@82 214 T4 = Cr[WS(csr, 4)];
Chris@82 215 T5 = Cr[WS(csr, 3)];
Chris@82 216 T6 = T4 + T5;
Chris@82 217 Tz = T4 - T5;
Chris@82 218 Tg = Ci[WS(csi, 4)];
Chris@82 219 Th = Ci[WS(csi, 3)];
Chris@82 220 Ti = Tg + Th;
Chris@82 221 TU = Tg - Th;
Chris@82 222 }
Chris@82 223 T7 = T3 + T6;
Chris@82 224 TW = TU + TV;
Chris@82 225 T13 = TV - TU;
Chris@82 226 Tj = Tf - Ti;
Chris@82 227 TD = Tz + TC;
Chris@82 228 TK = Tz - TC;
Chris@82 229 TP = T3 - T6;
Chris@82 230 TH = Tf + Ti;
Chris@82 231 }
Chris@82 232 {
Chris@82 233 E Ta, Tk, Tn, TR, Td, Tp, Ts, TQ;
Chris@82 234 {
Chris@82 235 E T8, T9, Tl, Tm;
Chris@82 236 T8 = Cr[WS(csr, 2)];
Chris@82 237 T9 = Cr[WS(csr, 5)];
Chris@82 238 Ta = T8 + T9;
Chris@82 239 Tk = T8 - T9;
Chris@82 240 Tl = Ci[WS(csi, 2)];
Chris@82 241 Tm = Ci[WS(csi, 5)];
Chris@82 242 Tn = Tl + Tm;
Chris@82 243 TR = Tl - Tm;
Chris@82 244 }
Chris@82 245 {
Chris@82 246 E Tb, Tc, Tq, Tr;
Chris@82 247 Tb = Cr[WS(csr, 1)];
Chris@82 248 Tc = Cr[WS(csr, 6)];
Chris@82 249 Td = Tb + Tc;
Chris@82 250 Tp = Tb - Tc;
Chris@82 251 Tq = Ci[WS(csi, 1)];
Chris@82 252 Tr = Ci[WS(csi, 6)];
Chris@82 253 Ts = Tq + Tr;
Chris@82 254 TQ = Tr - Tq;
Chris@82 255 }
Chris@82 256 Te = Ta + Td;
Chris@82 257 TX = Ta - Td;
Chris@82 258 T12 = TR + TQ;
Chris@82 259 To = Tk - Tn;
Chris@82 260 Tt = Tp - Ts;
Chris@82 261 Tx = Tp + Ts;
Chris@82 262 TS = TQ - TR;
Chris@82 263 Tw = Tk + Tn;
Chris@82 264 }
Chris@82 265 R0[0] = KP2_000000000 * (T7 + Te);
Chris@82 266 R0[WS(rs, 4)] = KP2_000000000 * (T13 - T12);
Chris@82 267 TT = TP + TS;
Chris@82 268 TY = TW - TX;
Chris@82 269 R0[WS(rs, 1)] = FMA(KP1_847759065, TT, KP765366864 * TY);
Chris@82 270 R0[WS(rs, 5)] = FNMS(KP765366864, TT, KP1_847759065 * TY);
Chris@82 271 {
Chris@82 272 E T11, T14, TZ, T10;
Chris@82 273 T11 = T7 - Te;
Chris@82 274 T14 = T12 + T13;
Chris@82 275 R0[WS(rs, 2)] = KP1_414213562 * (T11 + T14);
Chris@82 276 R0[WS(rs, 6)] = KP1_414213562 * (T14 - T11);
Chris@82 277 TZ = TP - TS;
Chris@82 278 T10 = TX + TW;
Chris@82 279 R0[WS(rs, 3)] = FMA(KP765366864, TZ, KP1_847759065 * T10);
Chris@82 280 R0[WS(rs, 7)] = FNMS(KP1_847759065, TZ, KP765366864 * T10);
Chris@82 281 }
Chris@82 282 {
Chris@82 283 E TJ, TN, TM, TO, TI, TL;
Chris@82 284 TI = KP707106781 * (Tw + Tx);
Chris@82 285 TJ = TH - TI;
Chris@82 286 TN = TH + TI;
Chris@82 287 TL = KP707106781 * (To - Tt);
Chris@82 288 TM = TK - TL;
Chris@82 289 TO = TL + TK;
Chris@82 290 R1[WS(rs, 1)] = FMA(KP1_662939224, TJ, KP1_111140466 * TM);
Chris@82 291 R1[WS(rs, 7)] = FNMS(KP1_961570560, TN, KP390180644 * TO);
Chris@82 292 R1[WS(rs, 5)] = FNMS(KP1_111140466, TJ, KP1_662939224 * TM);
Chris@82 293 R1[WS(rs, 3)] = FMA(KP390180644, TN, KP1_961570560 * TO);
Chris@82 294 }
Chris@82 295 {
Chris@82 296 E Tv, TF, TE, TG, Tu, Ty;
Chris@82 297 Tu = KP707106781 * (To + Tt);
Chris@82 298 Tv = Tj + Tu;
Chris@82 299 TF = Tj - Tu;
Chris@82 300 Ty = KP707106781 * (Tw - Tx);
Chris@82 301 TE = Ty + TD;
Chris@82 302 TG = Ty - TD;
Chris@82 303 R1[0] = FNMS(KP390180644, TE, KP1_961570560 * Tv);
Chris@82 304 R1[WS(rs, 6)] = FNMS(KP1_662939224, TF, KP1_111140466 * TG);
Chris@82 305 R1[WS(rs, 4)] = -(FMA(KP390180644, Tv, KP1_961570560 * TE));
Chris@82 306 R1[WS(rs, 2)] = FMA(KP1_111140466, TF, KP1_662939224 * TG);
Chris@82 307 }
Chris@82 308 }
Chris@82 309 }
Chris@82 310 }
Chris@82 311
Chris@82 312 static const kr2c_desc desc = { 16, "r2cbIII_16", {54, 20, 12, 0}, &GENUS };
Chris@82 313
Chris@82 314 void X(codelet_r2cbIII_16) (planner *p) {
Chris@82 315 X(kr2c_register) (p, r2cbIII_16, &desc);
Chris@82 316 }
Chris@82 317
Chris@82 318 #endif