Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/rdft/scalar/r2cb/r2cbIII_12.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 16:50:41 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 12 -name r2cbIII_12 -dft-III -include r2cbIII.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 42 FP additions, 20 FP multiplications,
Chris@42 32 * (or, 30 additions, 8 multiplications, 12 fused multiply/add),
Chris@42 33 * 37 stack variables, 4 constants, and 24 memory accesses
Chris@42 34 */
Chris@42 35 #include "r2cbIII.h"
Chris@42 36
Chris@42 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@42 38 {
Chris@42 39 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 40 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
Chris@42 41 DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
Chris@42 42 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@42 43 {
Chris@42 44 INT i;
Chris@42 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@42 46 E TE, TD, TF, TG;
Chris@42 47 {
Chris@42 48 E Tx, T6, Te, Tb, T5, Tw, Ts, To, Th, Ti, T9, TA;
Chris@42 49 {
Chris@42 50 E T1, Tq, Tc, Td, T4, T2, T3, T7, T8, Tr;
Chris@42 51 T1 = Cr[WS(csr, 1)];
Chris@42 52 T2 = Cr[WS(csr, 5)];
Chris@42 53 T3 = Cr[WS(csr, 2)];
Chris@42 54 Tq = Ci[WS(csi, 1)];
Chris@42 55 Tc = Ci[WS(csi, 5)];
Chris@42 56 Td = Ci[WS(csi, 2)];
Chris@42 57 T4 = T2 + T3;
Chris@42 58 Tx = T2 - T3;
Chris@42 59 T6 = Cr[WS(csr, 4)];
Chris@42 60 Te = Tc + Td;
Chris@42 61 Tr = Td - Tc;
Chris@42 62 Tb = FNMS(KP2_000000000, T1, T4);
Chris@42 63 T5 = T1 + T4;
Chris@42 64 T7 = Cr[0];
Chris@42 65 Tw = FMA(KP2_000000000, Tq, Tr);
Chris@42 66 Ts = Tq - Tr;
Chris@42 67 T8 = Cr[WS(csr, 3)];
Chris@42 68 To = Ci[WS(csi, 4)];
Chris@42 69 Th = Ci[0];
Chris@42 70 Ti = Ci[WS(csi, 3)];
Chris@42 71 T9 = T7 + T8;
Chris@42 72 TA = T7 - T8;
Chris@42 73 }
Chris@42 74 {
Chris@42 75 E Tl, Tm, Tv, TC;
Chris@42 76 {
Chris@42 77 E Tf, Ty, Tk, TB;
Chris@42 78 {
Chris@42 79 E Tj, Tn, Tg, Ta;
Chris@42 80 Tl = FNMS(KP1_732050807, Te, Tb);
Chris@42 81 Tf = FMA(KP1_732050807, Te, Tb);
Chris@42 82 Tj = Th + Ti;
Chris@42 83 Tn = Ti - Th;
Chris@42 84 Tg = FNMS(KP2_000000000, T6, T9);
Chris@42 85 Ta = T6 + T9;
Chris@42 86 {
Chris@42 87 E Tu, Tt, Tz, Tp;
Chris@42 88 Ty = FMA(KP1_732050807, Tx, Tw);
Chris@42 89 TE = FNMS(KP1_732050807, Tx, Tw);
Chris@42 90 Tz = FMA(KP2_000000000, To, Tn);
Chris@42 91 Tp = Tn - To;
Chris@42 92 Tm = FMA(KP1_732050807, Tj, Tg);
Chris@42 93 Tk = FNMS(KP1_732050807, Tj, Tg);
Chris@42 94 Tu = T5 - Ta;
Chris@42 95 R0[0] = KP2_000000000 * (T5 + Ta);
Chris@42 96 Tt = Tp - Ts;
Chris@42 97 R0[WS(rs, 3)] = KP2_000000000 * (Ts + Tp);
Chris@42 98 Tv = Tk - Tf;
Chris@42 99 TD = FMA(KP1_732050807, TA, Tz);
Chris@42 100 TB = FNMS(KP1_732050807, TA, Tz);
Chris@42 101 R1[WS(rs, 4)] = KP1_414213562 * (Tu + Tt);
Chris@42 102 R1[WS(rs, 1)] = KP1_414213562 * (Tt - Tu);
Chris@42 103 }
Chris@42 104 }
Chris@42 105 R0[WS(rs, 2)] = Tf + Tk;
Chris@42 106 TC = Ty + TB;
Chris@42 107 R0[WS(rs, 5)] = TB - Ty;
Chris@42 108 }
Chris@42 109 R1[WS(rs, 3)] = KP707106781 * (Tv + TC);
Chris@42 110 R1[0] = KP707106781 * (Tv - TC);
Chris@42 111 TF = Tl - Tm;
Chris@42 112 R0[WS(rs, 4)] = -(Tl + Tm);
Chris@42 113 }
Chris@42 114 }
Chris@42 115 R0[WS(rs, 1)] = TD - TE;
Chris@42 116 TG = TE + TD;
Chris@42 117 R1[WS(rs, 5)] = KP707106781 * (TF - TG);
Chris@42 118 R1[WS(rs, 2)] = KP707106781 * (TF + TG);
Chris@42 119 }
Chris@42 120 }
Chris@42 121 }
Chris@42 122
Chris@42 123 static const kr2c_desc desc = { 12, "r2cbIII_12", {30, 8, 12, 0}, &GENUS };
Chris@42 124
Chris@42 125 void X(codelet_r2cbIII_12) (planner *p) {
Chris@42 126 X(kr2c_register) (p, r2cbIII_12, &desc);
Chris@42 127 }
Chris@42 128
Chris@42 129 #else /* HAVE_FMA */
Chris@42 130
Chris@42 131 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 12 -name r2cbIII_12 -dft-III -include r2cbIII.h */
Chris@42 132
Chris@42 133 /*
Chris@42 134 * This function contains 42 FP additions, 20 FP multiplications,
Chris@42 135 * (or, 38 additions, 16 multiplications, 4 fused multiply/add),
Chris@42 136 * 25 stack variables, 4 constants, and 24 memory accesses
Chris@42 137 */
Chris@42 138 #include "r2cbIII.h"
Chris@42 139
Chris@42 140 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@42 141 {
Chris@42 142 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
Chris@42 143 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@42 144 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 145 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 146 {
Chris@42 147 INT i;
Chris@42 148 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@42 149 E T5, Tw, Tb, Te, Tx, Ts, Ta, TA, Tg, Tj, Tz, Tp, Tt, Tu;
Chris@42 150 {
Chris@42 151 E T1, T2, T3, T4;
Chris@42 152 T1 = Cr[WS(csr, 1)];
Chris@42 153 T2 = Cr[WS(csr, 5)];
Chris@42 154 T3 = Cr[WS(csr, 2)];
Chris@42 155 T4 = T2 + T3;
Chris@42 156 T5 = T1 + T4;
Chris@42 157 Tw = KP866025403 * (T2 - T3);
Chris@42 158 Tb = FNMS(KP500000000, T4, T1);
Chris@42 159 }
Chris@42 160 {
Chris@42 161 E Tq, Tc, Td, Tr;
Chris@42 162 Tq = Ci[WS(csi, 1)];
Chris@42 163 Tc = Ci[WS(csi, 5)];
Chris@42 164 Td = Ci[WS(csi, 2)];
Chris@42 165 Tr = Td - Tc;
Chris@42 166 Te = KP866025403 * (Tc + Td);
Chris@42 167 Tx = FMA(KP500000000, Tr, Tq);
Chris@42 168 Ts = Tq - Tr;
Chris@42 169 }
Chris@42 170 {
Chris@42 171 E T6, T7, T8, T9;
Chris@42 172 T6 = Cr[WS(csr, 4)];
Chris@42 173 T7 = Cr[0];
Chris@42 174 T8 = Cr[WS(csr, 3)];
Chris@42 175 T9 = T7 + T8;
Chris@42 176 Ta = T6 + T9;
Chris@42 177 TA = KP866025403 * (T7 - T8);
Chris@42 178 Tg = FNMS(KP500000000, T9, T6);
Chris@42 179 }
Chris@42 180 {
Chris@42 181 E To, Th, Ti, Tn;
Chris@42 182 To = Ci[WS(csi, 4)];
Chris@42 183 Th = Ci[0];
Chris@42 184 Ti = Ci[WS(csi, 3)];
Chris@42 185 Tn = Ti - Th;
Chris@42 186 Tj = KP866025403 * (Th + Ti);
Chris@42 187 Tz = FMA(KP500000000, Tn, To);
Chris@42 188 Tp = Tn - To;
Chris@42 189 }
Chris@42 190 R0[0] = KP2_000000000 * (T5 + Ta);
Chris@42 191 R0[WS(rs, 3)] = KP2_000000000 * (Ts + Tp);
Chris@42 192 Tt = Tp - Ts;
Chris@42 193 Tu = T5 - Ta;
Chris@42 194 R1[WS(rs, 1)] = KP1_414213562 * (Tt - Tu);
Chris@42 195 R1[WS(rs, 4)] = KP1_414213562 * (Tu + Tt);
Chris@42 196 {
Chris@42 197 E Tf, Tk, Tv, Ty, TB, TC;
Chris@42 198 Tf = Tb - Te;
Chris@42 199 Tk = Tg + Tj;
Chris@42 200 Tv = Tf - Tk;
Chris@42 201 Ty = Tw + Tx;
Chris@42 202 TB = Tz - TA;
Chris@42 203 TC = Ty + TB;
Chris@42 204 R0[WS(rs, 2)] = -(KP2_000000000 * (Tf + Tk));
Chris@42 205 R0[WS(rs, 5)] = KP2_000000000 * (TB - Ty);
Chris@42 206 R1[0] = KP1_414213562 * (Tv - TC);
Chris@42 207 R1[WS(rs, 3)] = KP1_414213562 * (Tv + TC);
Chris@42 208 }
Chris@42 209 {
Chris@42 210 E Tl, Tm, TF, TD, TE, TG;
Chris@42 211 Tl = Tb + Te;
Chris@42 212 Tm = Tg - Tj;
Chris@42 213 TF = Tm - Tl;
Chris@42 214 TD = TA + Tz;
Chris@42 215 TE = Tx - Tw;
Chris@42 216 TG = TE + TD;
Chris@42 217 R0[WS(rs, 4)] = KP2_000000000 * (Tl + Tm);
Chris@42 218 R1[WS(rs, 2)] = KP1_414213562 * (TF + TG);
Chris@42 219 R0[WS(rs, 1)] = KP2_000000000 * (TD - TE);
Chris@42 220 R1[WS(rs, 5)] = KP1_414213562 * (TF - TG);
Chris@42 221 }
Chris@42 222 }
Chris@42 223 }
Chris@42 224 }
Chris@42 225
Chris@42 226 static const kr2c_desc desc = { 12, "r2cbIII_12", {38, 16, 4, 0}, &GENUS };
Chris@42 227
Chris@42 228 void X(codelet_r2cbIII_12) (planner *p) {
Chris@42 229 X(kr2c_register) (p, r2cbIII_12, &desc);
Chris@42 230 }
Chris@42 231
Chris@42 232 #endif /* HAVE_FMA */