Mercurial > hg > sv-dependency-builds
comparison src/fftw-3.3.3/dft/scalar/codelets/q1_2.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 07:36:17 EST 2012 */ | |
| 23 | |
| 24 #include "codelet-dft.h" | |
| 25 | |
| 26 #ifdef HAVE_FMA | |
| 27 | |
| 28 /* Generated by: ../../../genfft/gen_twidsq.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -reload-twiddle -dif -n 2 -name q1_2 -include q.h */ | |
| 29 | |
| 30 /* | |
| 31 * This function contains 12 FP additions, 8 FP multiplications, | |
| 32 * (or, 8 additions, 4 multiplications, 4 fused multiply/add), | |
| 33 * 21 stack variables, 0 constants, and 16 memory accesses | |
| 34 */ | |
| 35 #include "q.h" | |
| 36 | |
| 37 static void q1_2(R *rio, R *iio, const R *W, stride rs, stride vs, INT mb, INT me, INT ms) | |
| 38 { | |
| 39 { | |
| 40 INT m; | |
| 41 for (m = mb, W = W + (mb * 2); m < me; m = m + 1, rio = rio + ms, iio = iio + ms, W = W + 2, MAKE_VOLATILE_STRIDE(4, rs), MAKE_VOLATILE_STRIDE(0, vs)) { | |
| 42 E T9, T6, T5; | |
| 43 { | |
| 44 E T1, T2, T7, T8, Tb, T4, Tc, Th, Ti, Te, Tj, Td, Tg; | |
| 45 T1 = rio[0]; | |
| 46 T2 = rio[WS(rs, 1)]; | |
| 47 T7 = iio[0]; | |
| 48 T8 = iio[WS(rs, 1)]; | |
| 49 Tb = rio[WS(vs, 1)]; | |
| 50 T4 = T1 - T2; | |
| 51 Tc = rio[WS(vs, 1) + WS(rs, 1)]; | |
| 52 T9 = T7 - T8; | |
| 53 Th = iio[WS(vs, 1)]; | |
| 54 Ti = iio[WS(vs, 1) + WS(rs, 1)]; | |
| 55 Te = Tb - Tc; | |
| 56 rio[0] = T1 + T2; | |
| 57 iio[0] = T7 + T8; | |
| 58 Tj = Th - Ti; | |
| 59 rio[WS(rs, 1)] = Tb + Tc; | |
| 60 iio[WS(rs, 1)] = Th + Ti; | |
| 61 Td = W[0]; | |
| 62 Tg = W[1]; | |
| 63 { | |
| 64 E T3, Tk, Tf, Ta; | |
| 65 T3 = W[0]; | |
| 66 T6 = W[1]; | |
| 67 Tk = Td * Tj; | |
| 68 Tf = Td * Te; | |
| 69 Ta = T3 * T9; | |
| 70 T5 = T3 * T4; | |
| 71 iio[WS(vs, 1) + WS(rs, 1)] = FNMS(Tg, Te, Tk); | |
| 72 rio[WS(vs, 1) + WS(rs, 1)] = FMA(Tg, Tj, Tf); | |
| 73 iio[WS(vs, 1)] = FNMS(T6, T4, Ta); | |
| 74 } | |
| 75 } | |
| 76 rio[WS(vs, 1)] = FMA(T6, T9, T5); | |
| 77 } | |
| 78 } | |
| 79 } | |
| 80 | |
| 81 static const tw_instr twinstr[] = { | |
| 82 {TW_FULL, 0, 2}, | |
| 83 {TW_NEXT, 1, 0} | |
| 84 }; | |
| 85 | |
| 86 static const ct_desc desc = { 2, "q1_2", twinstr, &GENUS, {8, 4, 4, 0}, 0, 0, 0 }; | |
| 87 | |
| 88 void X(codelet_q1_2) (planner *p) { | |
| 89 X(kdft_difsq_register) (p, q1_2, &desc); | |
| 90 } | |
| 91 #else /* HAVE_FMA */ | |
| 92 | |
| 93 /* Generated by: ../../../genfft/gen_twidsq.native -compact -variables 4 -pipeline-latency 4 -reload-twiddle -dif -n 2 -name q1_2 -include q.h */ | |
| 94 | |
| 95 /* | |
| 96 * This function contains 12 FP additions, 8 FP multiplications, | |
| 97 * (or, 8 additions, 4 multiplications, 4 fused multiply/add), | |
| 98 * 17 stack variables, 0 constants, and 16 memory accesses | |
| 99 */ | |
| 100 #include "q.h" | |
| 101 | |
| 102 static void q1_2(R *rio, R *iio, const R *W, stride rs, stride vs, INT mb, INT me, INT ms) | |
| 103 { | |
| 104 { | |
| 105 INT m; | |
| 106 for (m = mb, W = W + (mb * 2); m < me; m = m + 1, rio = rio + ms, iio = iio + ms, W = W + 2, MAKE_VOLATILE_STRIDE(4, rs), MAKE_VOLATILE_STRIDE(0, vs)) { | |
| 107 E T1, T2, T4, T6, T7, T8, T9, Ta, Tc, Te, Tf, Tg; | |
| 108 T1 = rio[0]; | |
| 109 T2 = rio[WS(rs, 1)]; | |
| 110 T4 = T1 - T2; | |
| 111 T6 = iio[0]; | |
| 112 T7 = iio[WS(rs, 1)]; | |
| 113 T8 = T6 - T7; | |
| 114 T9 = rio[WS(vs, 1)]; | |
| 115 Ta = rio[WS(vs, 1) + WS(rs, 1)]; | |
| 116 Tc = T9 - Ta; | |
| 117 Te = iio[WS(vs, 1)]; | |
| 118 Tf = iio[WS(vs, 1) + WS(rs, 1)]; | |
| 119 Tg = Te - Tf; | |
| 120 rio[0] = T1 + T2; | |
| 121 iio[0] = T6 + T7; | |
| 122 rio[WS(rs, 1)] = T9 + Ta; | |
| 123 iio[WS(rs, 1)] = Te + Tf; | |
| 124 { | |
| 125 E Tb, Td, T3, T5; | |
| 126 Tb = W[0]; | |
| 127 Td = W[1]; | |
| 128 rio[WS(vs, 1) + WS(rs, 1)] = FMA(Tb, Tc, Td * Tg); | |
| 129 iio[WS(vs, 1) + WS(rs, 1)] = FNMS(Td, Tc, Tb * Tg); | |
| 130 T3 = W[0]; | |
| 131 T5 = W[1]; | |
| 132 rio[WS(vs, 1)] = FMA(T3, T4, T5 * T8); | |
| 133 iio[WS(vs, 1)] = FNMS(T5, T4, T3 * T8); | |
| 134 } | |
| 135 } | |
| 136 } | |
| 137 } | |
| 138 | |
| 139 static const tw_instr twinstr[] = { | |
| 140 {TW_FULL, 0, 2}, | |
| 141 {TW_NEXT, 1, 0} | |
| 142 }; | |
| 143 | |
| 144 static const ct_desc desc = { 2, "q1_2", twinstr, &GENUS, {8, 4, 4, 0}, 0, 0, 0 }; | |
| 145 | |
| 146 void X(codelet_q1_2) (planner *p) { | |
| 147 X(kdft_difsq_register) (p, q1_2, &desc); | |
| 148 } | |
| 149 #endif /* HAVE_FMA */ |