Mercurial > hg > sv-dependency-builds
comparison src/fftw-3.3.3/dft/scalar/codelets/t1_3.c @ 10:37bf6b4a2645
Add FFTW3
| author | Chris Cannam |
|---|---|
| date | Wed, 20 Mar 2013 15:35:50 +0000 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| 9:c0fb53affa76 | 10:37bf6b4a2645 |
|---|---|
| 1 /* | |
| 2 * Copyright (c) 2003, 2007-11 Matteo Frigo | |
| 3 * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology | |
| 4 * | |
| 5 * This program is free software; you can redistribute it and/or modify | |
| 6 * it under the terms of the GNU General Public License as published by | |
| 7 * the Free Software Foundation; either version 2 of the License, or | |
| 8 * (at your option) any later version. | |
| 9 * | |
| 10 * This program is distributed in the hope that it will be useful, | |
| 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
| 13 * GNU General Public License for more details. | |
| 14 * | |
| 15 * You should have received a copy of the GNU General Public License | |
| 16 * along with this program; if not, write to the Free Software | |
| 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | |
| 18 * | |
| 19 */ | |
| 20 | |
| 21 /* This file was automatically generated --- DO NOT EDIT */ | |
| 22 /* Generated on Sun Nov 25 07:35:47 EST 2012 */ | |
| 23 | |
| 24 #include "codelet-dft.h" | |
| 25 | |
| 26 #ifdef HAVE_FMA | |
| 27 | |
| 28 /* Generated by: ../../../genfft/gen_twiddle.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 3 -name t1_3 -include t.h */ | |
| 29 | |
| 30 /* | |
| 31 * This function contains 16 FP additions, 14 FP multiplications, | |
| 32 * (or, 6 additions, 4 multiplications, 10 fused multiply/add), | |
| 33 * 21 stack variables, 2 constants, and 12 memory accesses | |
| 34 */ | |
| 35 #include "t.h" | |
| 36 | |
| 37 static void t1_3(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) | |
| 38 { | |
| 39 DK(KP866025403, +0.866025403784438646763723170752936183471402627); | |
| 40 DK(KP500000000, +0.500000000000000000000000000000000000000000000); | |
| 41 { | |
| 42 INT m; | |
| 43 for (m = mb, W = W + (mb * 4); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 4, MAKE_VOLATILE_STRIDE(6, rs)) { | |
| 44 E T1, Tm, T9, Tc, Tb, Th, T7, Ti, Ta, Tj, Td; | |
| 45 T1 = ri[0]; | |
| 46 Tm = ii[0]; | |
| 47 { | |
| 48 E T3, T6, T2, T5, Tg, T4, T8; | |
| 49 T3 = ri[WS(rs, 1)]; | |
| 50 T6 = ii[WS(rs, 1)]; | |
| 51 T2 = W[0]; | |
| 52 T5 = W[1]; | |
| 53 T9 = ri[WS(rs, 2)]; | |
| 54 Tc = ii[WS(rs, 2)]; | |
| 55 Tg = T2 * T6; | |
| 56 T4 = T2 * T3; | |
| 57 T8 = W[2]; | |
| 58 Tb = W[3]; | |
| 59 Th = FNMS(T5, T3, Tg); | |
| 60 T7 = FMA(T5, T6, T4); | |
| 61 Ti = T8 * Tc; | |
| 62 Ta = T8 * T9; | |
| 63 } | |
| 64 Tj = FNMS(Tb, T9, Ti); | |
| 65 Td = FMA(Tb, Tc, Ta); | |
| 66 { | |
| 67 E Tk, Te, To, Tn, Tl, Tf; | |
| 68 Tk = Th - Tj; | |
| 69 Tl = Th + Tj; | |
| 70 Te = T7 + Td; | |
| 71 To = Td - T7; | |
| 72 ii[0] = Tl + Tm; | |
| 73 Tn = FNMS(KP500000000, Tl, Tm); | |
| 74 ri[0] = T1 + Te; | |
| 75 Tf = FNMS(KP500000000, Te, T1); | |
| 76 ii[WS(rs, 1)] = FMA(KP866025403, To, Tn); | |
| 77 ii[WS(rs, 2)] = FNMS(KP866025403, To, Tn); | |
| 78 ri[WS(rs, 2)] = FNMS(KP866025403, Tk, Tf); | |
| 79 ri[WS(rs, 1)] = FMA(KP866025403, Tk, Tf); | |
| 80 } | |
| 81 } | |
| 82 } | |
| 83 } | |
| 84 | |
| 85 static const tw_instr twinstr[] = { | |
| 86 {TW_FULL, 0, 3}, | |
| 87 {TW_NEXT, 1, 0} | |
| 88 }; | |
| 89 | |
| 90 static const ct_desc desc = { 3, "t1_3", twinstr, &GENUS, {6, 4, 10, 0}, 0, 0, 0 }; | |
| 91 | |
| 92 void X(codelet_t1_3) (planner *p) { | |
| 93 X(kdft_dit_register) (p, t1_3, &desc); | |
| 94 } | |
| 95 #else /* HAVE_FMA */ | |
| 96 | |
| 97 /* Generated by: ../../../genfft/gen_twiddle.native -compact -variables 4 -pipeline-latency 4 -n 3 -name t1_3 -include t.h */ | |
| 98 | |
| 99 /* | |
| 100 * This function contains 16 FP additions, 12 FP multiplications, | |
| 101 * (or, 10 additions, 6 multiplications, 6 fused multiply/add), | |
| 102 * 15 stack variables, 2 constants, and 12 memory accesses | |
| 103 */ | |
| 104 #include "t.h" | |
| 105 | |
| 106 static void t1_3(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) | |
| 107 { | |
| 108 DK(KP866025403, +0.866025403784438646763723170752936183471402627); | |
| 109 DK(KP500000000, +0.500000000000000000000000000000000000000000000); | |
| 110 { | |
| 111 INT m; | |
| 112 for (m = mb, W = W + (mb * 4); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 4, MAKE_VOLATILE_STRIDE(6, rs)) { | |
| 113 E T1, Ti, T6, Te, Tb, Tf, Tc, Th; | |
| 114 T1 = ri[0]; | |
| 115 Ti = ii[0]; | |
| 116 { | |
| 117 E T3, T5, T2, T4; | |
| 118 T3 = ri[WS(rs, 1)]; | |
| 119 T5 = ii[WS(rs, 1)]; | |
| 120 T2 = W[0]; | |
| 121 T4 = W[1]; | |
| 122 T6 = FMA(T2, T3, T4 * T5); | |
| 123 Te = FNMS(T4, T3, T2 * T5); | |
| 124 } | |
| 125 { | |
| 126 E T8, Ta, T7, T9; | |
| 127 T8 = ri[WS(rs, 2)]; | |
| 128 Ta = ii[WS(rs, 2)]; | |
| 129 T7 = W[2]; | |
| 130 T9 = W[3]; | |
| 131 Tb = FMA(T7, T8, T9 * Ta); | |
| 132 Tf = FNMS(T9, T8, T7 * Ta); | |
| 133 } | |
| 134 Tc = T6 + Tb; | |
| 135 Th = Te + Tf; | |
| 136 ri[0] = T1 + Tc; | |
| 137 ii[0] = Th + Ti; | |
| 138 { | |
| 139 E Td, Tg, Tj, Tk; | |
| 140 Td = FNMS(KP500000000, Tc, T1); | |
| 141 Tg = KP866025403 * (Te - Tf); | |
| 142 ri[WS(rs, 2)] = Td - Tg; | |
| 143 ri[WS(rs, 1)] = Td + Tg; | |
| 144 Tj = KP866025403 * (Tb - T6); | |
| 145 Tk = FNMS(KP500000000, Th, Ti); | |
| 146 ii[WS(rs, 1)] = Tj + Tk; | |
| 147 ii[WS(rs, 2)] = Tk - Tj; | |
| 148 } | |
| 149 } | |
| 150 } | |
| 151 } | |
| 152 | |
| 153 static const tw_instr twinstr[] = { | |
| 154 {TW_FULL, 0, 3}, | |
| 155 {TW_NEXT, 1, 0} | |
| 156 }; | |
| 157 | |
| 158 static const ct_desc desc = { 3, "t1_3", twinstr, &GENUS, {10, 6, 6, 0}, 0, 0, 0 }; | |
| 159 | |
| 160 void X(codelet_t1_3) (planner *p) { | |
| 161 X(kdft_dit_register) (p, t1_3, &desc); | |
| 162 } | |
| 163 #endif /* HAVE_FMA */ |