annotate fft/fftw/fftw-3.3.4/rdft/scalar/r2cb/r2cbIII_7.c @ 40:223f770b5341 kissfft-double tip

Try a double-precision kissfft
author Chris Cannam
date Wed, 07 Sep 2016 10:40:32 +0100
parents 26056e866c29
children
rev   line source
Chris@19 1 /*
Chris@19 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@19 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@19 4 *
Chris@19 5 * This program is free software; you can redistribute it and/or modify
Chris@19 6 * it under the terms of the GNU General Public License as published by
Chris@19 7 * the Free Software Foundation; either version 2 of the License, or
Chris@19 8 * (at your option) any later version.
Chris@19 9 *
Chris@19 10 * This program is distributed in the hope that it will be useful,
Chris@19 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@19 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@19 13 * GNU General Public License for more details.
Chris@19 14 *
Chris@19 15 * You should have received a copy of the GNU General Public License
Chris@19 16 * along with this program; if not, write to the Free Software
Chris@19 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@19 18 *
Chris@19 19 */
Chris@19 20
Chris@19 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@19 22 /* Generated on Tue Mar 4 13:50:31 EST 2014 */
Chris@19 23
Chris@19 24 #include "codelet-rdft.h"
Chris@19 25
Chris@19 26 #ifdef HAVE_FMA
Chris@19 27
Chris@19 28 /* Generated by: ../../../genfft/gen_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 7 -name r2cbIII_7 -dft-III -include r2cbIII.h */
Chris@19 29
Chris@19 30 /*
Chris@19 31 * This function contains 24 FP additions, 22 FP multiplications,
Chris@19 32 * (or, 2 additions, 0 multiplications, 22 fused multiply/add),
Chris@19 33 * 31 stack variables, 7 constants, and 14 memory accesses
Chris@19 34 */
Chris@19 35 #include "r2cbIII.h"
Chris@19 36
Chris@19 37 static void r2cbIII_7(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@19 38 {
Chris@19 39 DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
Chris@19 40 DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
Chris@19 41 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@19 42 DK(KP692021471, +0.692021471630095869627814897002069140197260599);
Chris@19 43 DK(KP801937735, +0.801937735804838252472204639014890102331838324);
Chris@19 44 DK(KP356895867, +0.356895867892209443894399510021300583399127187);
Chris@19 45 DK(KP554958132, +0.554958132087371191422194871006410481067288862);
Chris@19 46 {
Chris@19 47 INT i;
Chris@19 48 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(28, rs), MAKE_VOLATILE_STRIDE(28, csr), MAKE_VOLATILE_STRIDE(28, csi)) {
Chris@19 49 E Tn, Td, Tg, Ti, Tl, T8;
Chris@19 50 {
Chris@19 51 E T1, T9, Tb, Ta, T2, T4, Th, Tm, Tc, T3, Te;
Chris@19 52 T1 = Cr[WS(csr, 3)];
Chris@19 53 T9 = Ci[WS(csi, 1)];
Chris@19 54 Tb = Ci[0];
Chris@19 55 Ta = Ci[WS(csi, 2)];
Chris@19 56 T2 = Cr[WS(csr, 2)];
Chris@19 57 T4 = Cr[0];
Chris@19 58 Th = FMA(KP554958132, T9, Tb);
Chris@19 59 Tm = FNMS(KP554958132, Ta, T9);
Chris@19 60 Tc = FMA(KP554958132, Tb, Ta);
Chris@19 61 T3 = Cr[WS(csr, 1)];
Chris@19 62 Te = FNMS(KP356895867, T2, T4);
Chris@19 63 Tn = FNMS(KP801937735, Tm, Tb);
Chris@19 64 {
Chris@19 65 E Tf, Tk, T7, T5, Tj, T6;
Chris@19 66 Td = FMA(KP801937735, Tc, T9);
Chris@19 67 T5 = T2 + T3 + T4;
Chris@19 68 Tj = FNMS(KP356895867, T4, T3);
Chris@19 69 T6 = FNMS(KP356895867, T3, T2);
Chris@19 70 Tf = FNMS(KP692021471, Te, T3);
Chris@19 71 R0[0] = FMA(KP2_000000000, T5, T1);
Chris@19 72 Tk = FNMS(KP692021471, Tj, T2);
Chris@19 73 T7 = FNMS(KP692021471, T6, T4);
Chris@19 74 Tg = FNMS(KP1_801937735, Tf, T1);
Chris@19 75 Ti = FNMS(KP801937735, Th, Ta);
Chris@19 76 Tl = FNMS(KP1_801937735, Tk, T1);
Chris@19 77 T8 = FNMS(KP1_801937735, T7, T1);
Chris@19 78 }
Chris@19 79 }
Chris@19 80 R1[WS(rs, 2)] = FMS(KP1_949855824, Ti, Tg);
Chris@19 81 R0[WS(rs, 1)] = FMA(KP1_949855824, Ti, Tg);
Chris@19 82 R0[WS(rs, 2)] = FNMS(KP1_949855824, Tn, Tl);
Chris@19 83 R1[WS(rs, 1)] = -(FMA(KP1_949855824, Tn, Tl));
Chris@19 84 R0[WS(rs, 3)] = FNMS(KP1_949855824, Td, T8);
Chris@19 85 R1[0] = -(FMA(KP1_949855824, Td, T8));
Chris@19 86 }
Chris@19 87 }
Chris@19 88 }
Chris@19 89
Chris@19 90 static const kr2c_desc desc = { 7, "r2cbIII_7", {2, 0, 22, 0}, &GENUS };
Chris@19 91
Chris@19 92 void X(codelet_r2cbIII_7) (planner *p) {
Chris@19 93 X(kr2c_register) (p, r2cbIII_7, &desc);
Chris@19 94 }
Chris@19 95
Chris@19 96 #else /* HAVE_FMA */
Chris@19 97
Chris@19 98 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 7 -name r2cbIII_7 -dft-III -include r2cbIII.h */
Chris@19 99
Chris@19 100 /*
Chris@19 101 * This function contains 24 FP additions, 19 FP multiplications,
Chris@19 102 * (or, 9 additions, 4 multiplications, 15 fused multiply/add),
Chris@19 103 * 21 stack variables, 7 constants, and 14 memory accesses
Chris@19 104 */
Chris@19 105 #include "r2cbIII.h"
Chris@19 106
Chris@19 107 static void r2cbIII_7(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@19 108 {
Chris@19 109 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@19 110 DK(KP1_246979603, +1.246979603717467061050009768008479621264549462);
Chris@19 111 DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
Chris@19 112 DK(KP445041867, +0.445041867912628808577805128993589518932711138);
Chris@19 113 DK(KP867767478, +0.867767478235116240951536665696717509219981456);
Chris@19 114 DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
Chris@19 115 DK(KP1_563662964, +1.563662964936059617416889053348115500464669037);
Chris@19 116 {
Chris@19 117 INT i;
Chris@19 118 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(28, rs), MAKE_VOLATILE_STRIDE(28, csr), MAKE_VOLATILE_STRIDE(28, csi)) {
Chris@19 119 E T9, Td, Tb, T1, T4, T2, T3, T5, Tc, Ta, T6, T8, T7;
Chris@19 120 T6 = Ci[WS(csi, 2)];
Chris@19 121 T8 = Ci[0];
Chris@19 122 T7 = Ci[WS(csi, 1)];
Chris@19 123 T9 = FMA(KP1_563662964, T6, KP1_949855824 * T7) + (KP867767478 * T8);
Chris@19 124 Td = FNMS(KP1_949855824, T8, KP1_563662964 * T7) - (KP867767478 * T6);
Chris@19 125 Tb = FNMS(KP1_563662964, T8, KP1_949855824 * T6) - (KP867767478 * T7);
Chris@19 126 T1 = Cr[WS(csr, 3)];
Chris@19 127 T4 = Cr[0];
Chris@19 128 T2 = Cr[WS(csr, 2)];
Chris@19 129 T3 = Cr[WS(csr, 1)];
Chris@19 130 T5 = FMA(KP445041867, T3, KP1_801937735 * T4) + FNMA(KP1_246979603, T2, T1);
Chris@19 131 Tc = FMA(KP1_801937735, T2, KP445041867 * T4) + FNMA(KP1_246979603, T3, T1);
Chris@19 132 Ta = FMA(KP1_246979603, T4, T1) + FNMA(KP1_801937735, T3, KP445041867 * T2);
Chris@19 133 R1[0] = T5 - T9;
Chris@19 134 R0[WS(rs, 3)] = -(T5 + T9);
Chris@19 135 R0[WS(rs, 2)] = Td - Tc;
Chris@19 136 R1[WS(rs, 1)] = Tc + Td;
Chris@19 137 R1[WS(rs, 2)] = Tb - Ta;
Chris@19 138 R0[WS(rs, 1)] = Ta + Tb;
Chris@19 139 R0[0] = FMA(KP2_000000000, T2 + T3 + T4, T1);
Chris@19 140 }
Chris@19 141 }
Chris@19 142 }
Chris@19 143
Chris@19 144 static const kr2c_desc desc = { 7, "r2cbIII_7", {9, 4, 15, 0}, &GENUS };
Chris@19 145
Chris@19 146 void X(codelet_r2cbIII_7) (planner *p) {
Chris@19 147 X(kr2c_register) (p, r2cbIII_7, &desc);
Chris@19 148 }
Chris@19 149
Chris@19 150 #endif /* HAVE_FMA */