Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cb/r2cbIII_9.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 08:07:43 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 9 -name r2cbIII_9 -dft-III -include rdft/scalar/r2cbIII.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 32 FP additions, 24 FP multiplications,
Chris@82 32 * (or, 8 additions, 0 multiplications, 24 fused multiply/add),
Chris@82 33 * 35 stack variables, 12 constants, and 18 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/r2cbIII.h"
Chris@82 36
Chris@82 37 static void r2cbIII_9(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_705737063, +1.705737063904886419256501927880148143872040591);
Chris@82 40 DK(KP1_969615506, +1.969615506024416118733486049179046027341286503);
Chris@82 41 DK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@82 42 DK(KP176326980, +0.176326980708464973471090386868618986121633062);
Chris@82 43 DK(KP1_326827896, +1.326827896337876792410842639271782594433726619);
Chris@82 44 DK(KP1_532088886, +1.532088886237956070404785301110833347871664914);
Chris@82 45 DK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@82 46 DK(KP839099631, +0.839099631177280011763127298123181364687434283);
Chris@82 47 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 48 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 49 DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
Chris@82 50 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 51 {
Chris@82 52 INT i;
Chris@82 53 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(36, rs), MAKE_VOLATILE_STRIDE(36, csr), MAKE_VOLATILE_STRIDE(36, csi)) {
Chris@82 54 E T3, Tr, Th, Td, Tc, T8, Tn, Ts, Tk, Tt, T9, Te;
Chris@82 55 {
Chris@82 56 E Tg, T1, T2, Tf;
Chris@82 57 Tg = Ci[WS(csi, 1)];
Chris@82 58 T1 = Cr[WS(csr, 4)];
Chris@82 59 T2 = Cr[WS(csr, 1)];
Chris@82 60 Tf = T2 - T1;
Chris@82 61 T3 = FMA(KP2_000000000, T2, T1);
Chris@82 62 Tr = FMA(KP1_732050807, Tg, Tf);
Chris@82 63 Th = FNMS(KP1_732050807, Tg, Tf);
Chris@82 64 }
Chris@82 65 {
Chris@82 66 E T4, T7, Tm, Tj, Tl, Ti;
Chris@82 67 T4 = Cr[WS(csr, 3)];
Chris@82 68 Td = Ci[WS(csi, 3)];
Chris@82 69 {
Chris@82 70 E T5, T6, Ta, Tb;
Chris@82 71 T5 = Cr[0];
Chris@82 72 T6 = Cr[WS(csr, 2)];
Chris@82 73 T7 = T5 + T6;
Chris@82 74 Tm = T5 - T6;
Chris@82 75 Ta = Ci[WS(csi, 2)];
Chris@82 76 Tb = Ci[0];
Chris@82 77 Tc = Ta - Tb;
Chris@82 78 Tj = Tb + Ta;
Chris@82 79 }
Chris@82 80 T8 = T4 + T7;
Chris@82 81 Tl = FMA(KP500000000, Tc, Td);
Chris@82 82 Tn = FNMS(KP866025403, Tm, Tl);
Chris@82 83 Ts = FMA(KP866025403, Tm, Tl);
Chris@82 84 Ti = FNMS(KP500000000, T7, T4);
Chris@82 85 Tk = FMA(KP866025403, Tj, Ti);
Chris@82 86 Tt = FNMS(KP866025403, Tj, Ti);
Chris@82 87 }
Chris@82 88 R0[0] = FMA(KP2_000000000, T8, T3);
Chris@82 89 T9 = T8 - T3;
Chris@82 90 Te = Tc - Td;
Chris@82 91 R1[WS(rs, 1)] = FMA(KP1_732050807, Te, T9);
Chris@82 92 R0[WS(rs, 3)] = FMS(KP1_732050807, Te, T9);
Chris@82 93 {
Chris@82 94 E Tq, To, Tp, Tw, Tu, Tv;
Chris@82 95 Tq = FNMS(KP839099631, Tk, Tn);
Chris@82 96 To = FMA(KP839099631, Tn, Tk);
Chris@82 97 Tp = FMA(KP766044443, To, Th);
Chris@82 98 R1[0] = FNMS(KP1_532088886, To, Th);
Chris@82 99 R1[WS(rs, 3)] = FMA(KP1_326827896, Tq, Tp);
Chris@82 100 R0[WS(rs, 2)] = FMS(KP1_326827896, Tq, Tp);
Chris@82 101 Tw = FNMS(KP176326980, Ts, Tt);
Chris@82 102 Tu = FMA(KP176326980, Tt, Ts);
Chris@82 103 Tv = FMA(KP984807753, Tu, Tr);
Chris@82 104 R0[WS(rs, 1)] = FMS(KP1_969615506, Tu, Tr);
Chris@82 105 R1[WS(rs, 2)] = FMA(KP1_705737063, Tw, Tv);
Chris@82 106 R0[WS(rs, 4)] = FMS(KP1_705737063, Tw, Tv);
Chris@82 107 }
Chris@82 108 }
Chris@82 109 }
Chris@82 110 }
Chris@82 111
Chris@82 112 static const kr2c_desc desc = { 9, "r2cbIII_9", {8, 0, 24, 0}, &GENUS };
Chris@82 113
Chris@82 114 void X(codelet_r2cbIII_9) (planner *p) {
Chris@82 115 X(kr2c_register) (p, r2cbIII_9, &desc);
Chris@82 116 }
Chris@82 117
Chris@82 118 #else
Chris@82 119
Chris@82 120 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 9 -name r2cbIII_9 -dft-III -include rdft/scalar/r2cbIII.h */
Chris@82 121
Chris@82 122 /*
Chris@82 123 * This function contains 32 FP additions, 18 FP multiplications,
Chris@82 124 * (or, 22 additions, 8 multiplications, 10 fused multiply/add),
Chris@82 125 * 35 stack variables, 12 constants, and 18 memory accesses
Chris@82 126 */
Chris@82 127 #include "rdft/scalar/r2cbIII.h"
Chris@82 128
Chris@82 129 static void r2cbIII_9(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 130 {
Chris@82 131 DK(KP642787609, +0.642787609686539326322643409907263432907559884);
Chris@82 132 DK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@82 133 DK(KP1_326827896, +1.326827896337876792410842639271782594433726619);
Chris@82 134 DK(KP1_113340798, +1.113340798452838732905825904094046265936583811);
Chris@82 135 DK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@82 136 DK(KP173648177, +0.173648177666930348851716626769314796000375677);
Chris@82 137 DK(KP1_705737063, +1.705737063904886419256501927880148143872040591);
Chris@82 138 DK(KP300767466, +0.300767466360870593278543795225003852144476517);
Chris@82 139 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 140 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 141 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 142 DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
Chris@82 143 {
Chris@82 144 INT i;
Chris@82 145 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(36, rs), MAKE_VOLATILE_STRIDE(36, csr), MAKE_VOLATILE_STRIDE(36, csi)) {
Chris@82 146 E T3, Ts, Ti, Td, Tc, T8, To, Tu, Tl, Tt, T9, Te;
Chris@82 147 {
Chris@82 148 E Th, T1, T2, Tf, Tg;
Chris@82 149 Tg = Ci[WS(csi, 1)];
Chris@82 150 Th = KP1_732050807 * Tg;
Chris@82 151 T1 = Cr[WS(csr, 4)];
Chris@82 152 T2 = Cr[WS(csr, 1)];
Chris@82 153 Tf = T2 - T1;
Chris@82 154 T3 = FMA(KP2_000000000, T2, T1);
Chris@82 155 Ts = Tf - Th;
Chris@82 156 Ti = Tf + Th;
Chris@82 157 }
Chris@82 158 {
Chris@82 159 E T4, T7, Tm, Tk, Tn, Tj;
Chris@82 160 T4 = Cr[WS(csr, 3)];
Chris@82 161 Td = Ci[WS(csi, 3)];
Chris@82 162 {
Chris@82 163 E T5, T6, Ta, Tb;
Chris@82 164 T5 = Cr[0];
Chris@82 165 T6 = Cr[WS(csr, 2)];
Chris@82 166 T7 = T5 + T6;
Chris@82 167 Tm = KP866025403 * (T6 - T5);
Chris@82 168 Ta = Ci[WS(csi, 2)];
Chris@82 169 Tb = Ci[0];
Chris@82 170 Tc = Ta - Tb;
Chris@82 171 Tk = KP866025403 * (Tb + Ta);
Chris@82 172 }
Chris@82 173 T8 = T4 + T7;
Chris@82 174 Tn = FMA(KP500000000, Tc, Td);
Chris@82 175 To = Tm - Tn;
Chris@82 176 Tu = Tm + Tn;
Chris@82 177 Tj = FMS(KP500000000, T7, T4);
Chris@82 178 Tl = Tj + Tk;
Chris@82 179 Tt = Tj - Tk;
Chris@82 180 }
Chris@82 181 R0[0] = FMA(KP2_000000000, T8, T3);
Chris@82 182 T9 = T8 - T3;
Chris@82 183 Te = KP1_732050807 * (Tc - Td);
Chris@82 184 R1[WS(rs, 1)] = T9 + Te;
Chris@82 185 R0[WS(rs, 3)] = Te - T9;
Chris@82 186 {
Chris@82 187 E Tr, Tp, Tq, Tx, Tv, Tw;
Chris@82 188 Tr = FNMS(KP1_705737063, Tl, KP300767466 * To);
Chris@82 189 Tp = FMA(KP173648177, Tl, KP984807753 * To);
Chris@82 190 Tq = Ti - Tp;
Chris@82 191 R0[WS(rs, 1)] = -(FMA(KP2_000000000, Tp, Ti));
Chris@82 192 R0[WS(rs, 4)] = Tr - Tq;
Chris@82 193 R1[WS(rs, 2)] = Tq + Tr;
Chris@82 194 Tx = FMA(KP1_113340798, Tt, KP1_326827896 * Tu);
Chris@82 195 Tv = FNMS(KP642787609, Tu, KP766044443 * Tt);
Chris@82 196 Tw = Tv - Ts;
Chris@82 197 R1[0] = FMA(KP2_000000000, Tv, Ts);
Chris@82 198 R1[WS(rs, 3)] = Tx - Tw;
Chris@82 199 R0[WS(rs, 2)] = Tw + Tx;
Chris@82 200 }
Chris@82 201 }
Chris@82 202 }
Chris@82 203 }
Chris@82 204
Chris@82 205 static const kr2c_desc desc = { 9, "r2cbIII_9", {22, 8, 10, 0}, &GENUS };
Chris@82 206
Chris@82 207 void X(codelet_r2cbIII_9) (planner *p) {
Chris@82 208 X(kr2c_register) (p, r2cbIII_9, &desc);
Chris@82 209 }
Chris@82 210
Chris@82 211 #endif