Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/rdft/scalar/r2cb/r2cb_9.c @ 56:af97cad61ff0

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add updated build of PortAudio for OSX
author Chris Cannam <cannam@all-day-breakfast.com>
date Tue, 03 Jan 2017 15:10:52 +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 16:49:26 EDT 2016 */
Chris@42 23
Chris@42 24 #include "codelet-rdft.h"
Chris@42 25
Chris@42 26 #ifdef HAVE_FMA
Chris@42 27
Chris@42 28 /* Generated by: ../../../genfft/gen_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 9 -name r2cb_9 -include r2cb.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 32 FP additions, 24 FP multiplications,
Chris@42 32 * (or, 8 additions, 0 multiplications, 24 fused multiply/add),
Chris@42 33 * 40 stack variables, 12 constants, and 18 memory accesses
Chris@42 34 */
Chris@42 35 #include "r2cb.h"
Chris@42 36
Chris@42 37 static void r2cb_9(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@42 38 {
Chris@42 39 DK(KP1_326827896, +1.326827896337876792410842639271782594433726619);
Chris@42 40 DK(KP1_705737063, +1.705737063904886419256501927880148143872040591);
Chris@42 41 DK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@42 42 DK(KP1_532088886, +1.532088886237956070404785301110833347871664914);
Chris@42 43 DK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@42 44 DK(KP1_969615506, +1.969615506024416118733486049179046027341286503);
Chris@42 45 DK(KP839099631, +0.839099631177280011763127298123181364687434283);
Chris@42 46 DK(KP176326980, +0.176326980708464973471090386868618986121633062);
Chris@42 47 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 48 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 49 DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
Chris@42 50 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@42 51 {
Chris@42 52 INT i;
Chris@42 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@42 54 E T4, Th, T3, Tb, Tp, Tk, T7, Tf, Ti, Ta, T1, T2;
Chris@42 55 Ta = Ci[WS(csi, 3)];
Chris@42 56 T1 = Cr[0];
Chris@42 57 T2 = Cr[WS(csr, 3)];
Chris@42 58 T4 = Cr[WS(csr, 1)];
Chris@42 59 Th = Ci[WS(csi, 1)];
Chris@42 60 {
Chris@42 61 E T5, T9, T6, Td, Te;
Chris@42 62 T5 = Cr[WS(csr, 4)];
Chris@42 63 T9 = T1 - T2;
Chris@42 64 T3 = FMA(KP2_000000000, T2, T1);
Chris@42 65 T6 = Cr[WS(csr, 2)];
Chris@42 66 Td = Ci[WS(csi, 4)];
Chris@42 67 Te = Ci[WS(csi, 2)];
Chris@42 68 Tb = FNMS(KP1_732050807, Ta, T9);
Chris@42 69 Tp = FMA(KP1_732050807, Ta, T9);
Chris@42 70 Tk = T6 - T5;
Chris@42 71 T7 = T5 + T6;
Chris@42 72 Tf = Td + Te;
Chris@42 73 Ti = Td - Te;
Chris@42 74 }
Chris@42 75 {
Chris@42 76 E Tu, To, Tt, Tn, Tc, T8;
Chris@42 77 Tc = FNMS(KP500000000, T7, T4);
Chris@42 78 T8 = T4 + T7;
Chris@42 79 {
Chris@42 80 E Tw, Tj, Tr, Tg, Tv;
Chris@42 81 Tw = Ti + Th;
Chris@42 82 Tj = FNMS(KP500000000, Ti, Th);
Chris@42 83 Tr = FMA(KP866025403, Tf, Tc);
Chris@42 84 Tg = FNMS(KP866025403, Tf, Tc);
Chris@42 85 Tv = T3 - T8;
Chris@42 86 R0[0] = FMA(KP2_000000000, T8, T3);
Chris@42 87 {
Chris@42 88 E Tq, Tl, Ts, Tm;
Chris@42 89 Tq = FMA(KP866025403, Tk, Tj);
Chris@42 90 Tl = FNMS(KP866025403, Tk, Tj);
Chris@42 91 R0[WS(rs, 3)] = FMA(KP1_732050807, Tw, Tv);
Chris@42 92 R1[WS(rs, 1)] = FNMS(KP1_732050807, Tw, Tv);
Chris@42 93 Ts = FNMS(KP176326980, Tr, Tq);
Chris@42 94 Tu = FMA(KP176326980, Tq, Tr);
Chris@42 95 Tm = FNMS(KP839099631, Tl, Tg);
Chris@42 96 To = FMA(KP839099631, Tg, Tl);
Chris@42 97 R0[WS(rs, 1)] = FNMS(KP1_969615506, Ts, Tp);
Chris@42 98 Tt = FMA(KP984807753, Ts, Tp);
Chris@42 99 R1[0] = FMA(KP1_532088886, Tm, Tb);
Chris@42 100 Tn = FNMS(KP766044443, Tm, Tb);
Chris@42 101 }
Chris@42 102 }
Chris@42 103 R1[WS(rs, 2)] = FNMS(KP1_705737063, Tu, Tt);
Chris@42 104 R0[WS(rs, 4)] = FMA(KP1_705737063, Tu, Tt);
Chris@42 105 R0[WS(rs, 2)] = FNMS(KP1_326827896, To, Tn);
Chris@42 106 R1[WS(rs, 3)] = FMA(KP1_326827896, To, Tn);
Chris@42 107 }
Chris@42 108 }
Chris@42 109 }
Chris@42 110 }
Chris@42 111
Chris@42 112 static const kr2c_desc desc = { 9, "r2cb_9", {8, 0, 24, 0}, &GENUS };
Chris@42 113
Chris@42 114 void X(codelet_r2cb_9) (planner *p) {
Chris@42 115 X(kr2c_register) (p, r2cb_9, &desc);
Chris@42 116 }
Chris@42 117
Chris@42 118 #else /* HAVE_FMA */
Chris@42 119
Chris@42 120 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 9 -name r2cb_9 -include r2cb.h */
Chris@42 121
Chris@42 122 /*
Chris@42 123 * This function contains 32 FP additions, 18 FP multiplications,
Chris@42 124 * (or, 22 additions, 8 multiplications, 10 fused multiply/add),
Chris@42 125 * 35 stack variables, 12 constants, and 18 memory accesses
Chris@42 126 */
Chris@42 127 #include "r2cb.h"
Chris@42 128
Chris@42 129 static void r2cb_9(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@42 130 {
Chris@42 131 DK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@42 132 DK(KP173648177, +0.173648177666930348851716626769314796000375677);
Chris@42 133 DK(KP300767466, +0.300767466360870593278543795225003852144476517);
Chris@42 134 DK(KP1_705737063, +1.705737063904886419256501927880148143872040591);
Chris@42 135 DK(KP642787609, +0.642787609686539326322643409907263432907559884);
Chris@42 136 DK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@42 137 DK(KP1_326827896, +1.326827896337876792410842639271782594433726619);
Chris@42 138 DK(KP1_113340798, +1.113340798452838732905825904094046265936583811);
Chris@42 139 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 140 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 141 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@42 142 DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
Chris@42 143 {
Chris@42 144 INT i;
Chris@42 145 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@42 146 E T3, Tq, Tc, Tk, Tj, T8, Tm, Ts, Th, Tr, Tw, Tx;
Chris@42 147 {
Chris@42 148 E Tb, T1, T2, T9, Ta;
Chris@42 149 Ta = Ci[WS(csi, 3)];
Chris@42 150 Tb = KP1_732050807 * Ta;
Chris@42 151 T1 = Cr[0];
Chris@42 152 T2 = Cr[WS(csr, 3)];
Chris@42 153 T9 = T1 - T2;
Chris@42 154 T3 = FMA(KP2_000000000, T2, T1);
Chris@42 155 Tq = T9 + Tb;
Chris@42 156 Tc = T9 - Tb;
Chris@42 157 }
Chris@42 158 {
Chris@42 159 E T4, T7, Ti, Tg, Tl, Td;
Chris@42 160 T4 = Cr[WS(csr, 1)];
Chris@42 161 Tk = Ci[WS(csi, 1)];
Chris@42 162 {
Chris@42 163 E T5, T6, Te, Tf;
Chris@42 164 T5 = Cr[WS(csr, 4)];
Chris@42 165 T6 = Cr[WS(csr, 2)];
Chris@42 166 T7 = T5 + T6;
Chris@42 167 Ti = KP866025403 * (T5 - T6);
Chris@42 168 Te = Ci[WS(csi, 4)];
Chris@42 169 Tf = Ci[WS(csi, 2)];
Chris@42 170 Tg = KP866025403 * (Te + Tf);
Chris@42 171 Tj = Tf - Te;
Chris@42 172 }
Chris@42 173 T8 = T4 + T7;
Chris@42 174 Tl = FMA(KP500000000, Tj, Tk);
Chris@42 175 Tm = Ti + Tl;
Chris@42 176 Ts = Tl - Ti;
Chris@42 177 Td = FNMS(KP500000000, T7, T4);
Chris@42 178 Th = Td - Tg;
Chris@42 179 Tr = Td + Tg;
Chris@42 180 }
Chris@42 181 R0[0] = FMA(KP2_000000000, T8, T3);
Chris@42 182 Tw = T3 - T8;
Chris@42 183 Tx = KP1_732050807 * (Tk - Tj);
Chris@42 184 R1[WS(rs, 1)] = Tw - Tx;
Chris@42 185 R0[WS(rs, 3)] = Tw + Tx;
Chris@42 186 {
Chris@42 187 E Tp, Tn, To, Tv, Tt, Tu;
Chris@42 188 Tp = FMA(KP1_113340798, Th, KP1_326827896 * Tm);
Chris@42 189 Tn = FNMS(KP642787609, Tm, KP766044443 * Th);
Chris@42 190 To = Tc - Tn;
Chris@42 191 R1[0] = FMA(KP2_000000000, Tn, Tc);
Chris@42 192 R1[WS(rs, 3)] = To + Tp;
Chris@42 193 R0[WS(rs, 2)] = To - Tp;
Chris@42 194 Tv = FMA(KP1_705737063, Tr, KP300767466 * Ts);
Chris@42 195 Tt = FNMS(KP984807753, Ts, KP173648177 * Tr);
Chris@42 196 Tu = Tq - Tt;
Chris@42 197 R0[WS(rs, 1)] = FMA(KP2_000000000, Tt, Tq);
Chris@42 198 R0[WS(rs, 4)] = Tu + Tv;
Chris@42 199 R1[WS(rs, 2)] = Tu - Tv;
Chris@42 200 }
Chris@42 201 }
Chris@42 202 }
Chris@42 203 }
Chris@42 204
Chris@42 205 static const kr2c_desc desc = { 9, "r2cb_9", {22, 8, 10, 0}, &GENUS };
Chris@42 206
Chris@42 207 void X(codelet_r2cb_9) (planner *p) {
Chris@42 208 X(kr2c_register) (p, r2cb_9, &desc);
Chris@42 209 }
Chris@42 210
Chris@42 211 #endif /* HAVE_FMA */