Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cb/r2cbIII_12.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 12 -name r2cbIII_12 -dft-III -include rdft/scalar/r2cbIII.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 42 FP additions, 20 FP multiplications,
Chris@82 32 * (or, 30 additions, 8 multiplications, 12 fused multiply/add),
Chris@82 33 * 25 stack variables, 4 constants, and 24 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/r2cbIII.h"
Chris@82 36
Chris@82 37 static void r2cbIII_12(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(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 40 DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
Chris@82 41 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
Chris@82 42 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 43 {
Chris@82 44 INT i;
Chris@82 45 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) {
Chris@82 46 E T5, Tx, Tb, Te, Tw, Ts, Ta, TA, Tg, Tj, Tz, Tp, Tt, Tu;
Chris@82 47 {
Chris@82 48 E T1, T2, T3, T4;
Chris@82 49 T1 = Cr[WS(csr, 1)];
Chris@82 50 T2 = Cr[WS(csr, 5)];
Chris@82 51 T3 = Cr[WS(csr, 2)];
Chris@82 52 T4 = T2 + T3;
Chris@82 53 T5 = T1 + T4;
Chris@82 54 Tx = T2 - T3;
Chris@82 55 Tb = FNMS(KP2_000000000, T1, T4);
Chris@82 56 }
Chris@82 57 {
Chris@82 58 E Tq, Tc, Td, Tr;
Chris@82 59 Tq = Ci[WS(csi, 1)];
Chris@82 60 Tc = Ci[WS(csi, 5)];
Chris@82 61 Td = Ci[WS(csi, 2)];
Chris@82 62 Tr = Td - Tc;
Chris@82 63 Te = Tc + Td;
Chris@82 64 Tw = FMA(KP2_000000000, Tq, Tr);
Chris@82 65 Ts = Tq - Tr;
Chris@82 66 }
Chris@82 67 {
Chris@82 68 E T6, T7, T8, T9;
Chris@82 69 T6 = Cr[WS(csr, 4)];
Chris@82 70 T7 = Cr[0];
Chris@82 71 T8 = Cr[WS(csr, 3)];
Chris@82 72 T9 = T7 + T8;
Chris@82 73 Ta = T6 + T9;
Chris@82 74 TA = T7 - T8;
Chris@82 75 Tg = FNMS(KP2_000000000, T6, T9);
Chris@82 76 }
Chris@82 77 {
Chris@82 78 E To, Th, Ti, Tn;
Chris@82 79 To = Ci[WS(csi, 4)];
Chris@82 80 Th = Ci[0];
Chris@82 81 Ti = Ci[WS(csi, 3)];
Chris@82 82 Tn = Ti - Th;
Chris@82 83 Tj = Th + Ti;
Chris@82 84 Tz = FMA(KP2_000000000, To, Tn);
Chris@82 85 Tp = Tn - To;
Chris@82 86 }
Chris@82 87 R0[0] = KP2_000000000 * (T5 + Ta);
Chris@82 88 R0[WS(rs, 3)] = KP2_000000000 * (Ts + Tp);
Chris@82 89 Tt = Tp - Ts;
Chris@82 90 Tu = T5 - Ta;
Chris@82 91 R1[WS(rs, 1)] = KP1_414213562 * (Tt - Tu);
Chris@82 92 R1[WS(rs, 4)] = KP1_414213562 * (Tu + Tt);
Chris@82 93 {
Chris@82 94 E Tf, Tk, Tv, Ty, TB, TC;
Chris@82 95 Tf = FMA(KP1_732050807, Te, Tb);
Chris@82 96 Tk = FNMS(KP1_732050807, Tj, Tg);
Chris@82 97 Tv = Tk - Tf;
Chris@82 98 Ty = FMA(KP1_732050807, Tx, Tw);
Chris@82 99 TB = FNMS(KP1_732050807, TA, Tz);
Chris@82 100 TC = Ty + TB;
Chris@82 101 R0[WS(rs, 2)] = Tf + Tk;
Chris@82 102 R0[WS(rs, 5)] = TB - Ty;
Chris@82 103 R1[0] = KP707106781 * (Tv - TC);
Chris@82 104 R1[WS(rs, 3)] = KP707106781 * (Tv + TC);
Chris@82 105 }
Chris@82 106 {
Chris@82 107 E Tl, Tm, TF, TD, TE, TG;
Chris@82 108 Tl = FNMS(KP1_732050807, Te, Tb);
Chris@82 109 Tm = FMA(KP1_732050807, Tj, Tg);
Chris@82 110 TF = Tl - Tm;
Chris@82 111 TD = FMA(KP1_732050807, TA, Tz);
Chris@82 112 TE = FNMS(KP1_732050807, Tx, Tw);
Chris@82 113 TG = TE + TD;
Chris@82 114 R0[WS(rs, 4)] = -(Tl + Tm);
Chris@82 115 R1[WS(rs, 2)] = KP707106781 * (TF + TG);
Chris@82 116 R0[WS(rs, 1)] = TD - TE;
Chris@82 117 R1[WS(rs, 5)] = KP707106781 * (TF - TG);
Chris@82 118 }
Chris@82 119 }
Chris@82 120 }
Chris@82 121 }
Chris@82 122
Chris@82 123 static const kr2c_desc desc = { 12, "r2cbIII_12", {30, 8, 12, 0}, &GENUS };
Chris@82 124
Chris@82 125 void X(codelet_r2cbIII_12) (planner *p) {
Chris@82 126 X(kr2c_register) (p, r2cbIII_12, &desc);
Chris@82 127 }
Chris@82 128
Chris@82 129 #else
Chris@82 130
Chris@82 131 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 12 -name r2cbIII_12 -dft-III -include rdft/scalar/r2cbIII.h */
Chris@82 132
Chris@82 133 /*
Chris@82 134 * This function contains 42 FP additions, 20 FP multiplications,
Chris@82 135 * (or, 38 additions, 16 multiplications, 4 fused multiply/add),
Chris@82 136 * 25 stack variables, 4 constants, and 24 memory accesses
Chris@82 137 */
Chris@82 138 #include "rdft/scalar/r2cbIII.h"
Chris@82 139
Chris@82 140 static void r2cbIII_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 141 {
Chris@82 142 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
Chris@82 143 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 144 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 145 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 146 {
Chris@82 147 INT i;
Chris@82 148 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) {
Chris@82 149 E T5, Tw, Tb, Te, Tx, Ts, Ta, TA, Tg, Tj, Tz, Tp, Tt, Tu;
Chris@82 150 {
Chris@82 151 E T1, T2, T3, T4;
Chris@82 152 T1 = Cr[WS(csr, 1)];
Chris@82 153 T2 = Cr[WS(csr, 5)];
Chris@82 154 T3 = Cr[WS(csr, 2)];
Chris@82 155 T4 = T2 + T3;
Chris@82 156 T5 = T1 + T4;
Chris@82 157 Tw = KP866025403 * (T2 - T3);
Chris@82 158 Tb = FNMS(KP500000000, T4, T1);
Chris@82 159 }
Chris@82 160 {
Chris@82 161 E Tq, Tc, Td, Tr;
Chris@82 162 Tq = Ci[WS(csi, 1)];
Chris@82 163 Tc = Ci[WS(csi, 5)];
Chris@82 164 Td = Ci[WS(csi, 2)];
Chris@82 165 Tr = Td - Tc;
Chris@82 166 Te = KP866025403 * (Tc + Td);
Chris@82 167 Tx = FMA(KP500000000, Tr, Tq);
Chris@82 168 Ts = Tq - Tr;
Chris@82 169 }
Chris@82 170 {
Chris@82 171 E T6, T7, T8, T9;
Chris@82 172 T6 = Cr[WS(csr, 4)];
Chris@82 173 T7 = Cr[0];
Chris@82 174 T8 = Cr[WS(csr, 3)];
Chris@82 175 T9 = T7 + T8;
Chris@82 176 Ta = T6 + T9;
Chris@82 177 TA = KP866025403 * (T7 - T8);
Chris@82 178 Tg = FNMS(KP500000000, T9, T6);
Chris@82 179 }
Chris@82 180 {
Chris@82 181 E To, Th, Ti, Tn;
Chris@82 182 To = Ci[WS(csi, 4)];
Chris@82 183 Th = Ci[0];
Chris@82 184 Ti = Ci[WS(csi, 3)];
Chris@82 185 Tn = Ti - Th;
Chris@82 186 Tj = KP866025403 * (Th + Ti);
Chris@82 187 Tz = FMA(KP500000000, Tn, To);
Chris@82 188 Tp = Tn - To;
Chris@82 189 }
Chris@82 190 R0[0] = KP2_000000000 * (T5 + Ta);
Chris@82 191 R0[WS(rs, 3)] = KP2_000000000 * (Ts + Tp);
Chris@82 192 Tt = Tp - Ts;
Chris@82 193 Tu = T5 - Ta;
Chris@82 194 R1[WS(rs, 1)] = KP1_414213562 * (Tt - Tu);
Chris@82 195 R1[WS(rs, 4)] = KP1_414213562 * (Tu + Tt);
Chris@82 196 {
Chris@82 197 E Tf, Tk, Tv, Ty, TB, TC;
Chris@82 198 Tf = Tb - Te;
Chris@82 199 Tk = Tg + Tj;
Chris@82 200 Tv = Tf - Tk;
Chris@82 201 Ty = Tw + Tx;
Chris@82 202 TB = Tz - TA;
Chris@82 203 TC = Ty + TB;
Chris@82 204 R0[WS(rs, 2)] = -(KP2_000000000 * (Tf + Tk));
Chris@82 205 R0[WS(rs, 5)] = KP2_000000000 * (TB - Ty);
Chris@82 206 R1[0] = KP1_414213562 * (Tv - TC);
Chris@82 207 R1[WS(rs, 3)] = KP1_414213562 * (Tv + TC);
Chris@82 208 }
Chris@82 209 {
Chris@82 210 E Tl, Tm, TF, TD, TE, TG;
Chris@82 211 Tl = Tb + Te;
Chris@82 212 Tm = Tg - Tj;
Chris@82 213 TF = Tm - Tl;
Chris@82 214 TD = TA + Tz;
Chris@82 215 TE = Tx - Tw;
Chris@82 216 TG = TE + TD;
Chris@82 217 R0[WS(rs, 4)] = KP2_000000000 * (Tl + Tm);
Chris@82 218 R1[WS(rs, 2)] = KP1_414213562 * (TF + TG);
Chris@82 219 R0[WS(rs, 1)] = KP2_000000000 * (TD - TE);
Chris@82 220 R1[WS(rs, 5)] = KP1_414213562 * (TF - TG);
Chris@82 221 }
Chris@82 222 }
Chris@82 223 }
Chris@82 224 }
Chris@82 225
Chris@82 226 static const kr2c_desc desc = { 12, "r2cbIII_12", {38, 16, 4, 0}, &GENUS };
Chris@82 227
Chris@82 228 void X(codelet_r2cbIII_12) (planner *p) {
Chris@82 229 X(kr2c_register) (p, r2cbIII_12, &desc);
Chris@82 230 }
Chris@82 231
Chris@82 232 #endif