annotate src/fftw-3.3.3/dft/scalar/codelets/n1_3.c @ 10:37bf6b4a2645

Add FFTW3
author Chris Cannam
date Wed, 20 Mar 2013 15:35:50 +0000
parents
children
rev   line source
Chris@10 1 /*
Chris@10 2 * Copyright (c) 2003, 2007-11 Matteo Frigo
Chris@10 3 * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology
Chris@10 4 *
Chris@10 5 * This program is free software; you can redistribute it and/or modify
Chris@10 6 * it under the terms of the GNU General Public License as published by
Chris@10 7 * the Free Software Foundation; either version 2 of the License, or
Chris@10 8 * (at your option) any later version.
Chris@10 9 *
Chris@10 10 * This program is distributed in the hope that it will be useful,
Chris@10 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@10 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@10 13 * GNU General Public License for more details.
Chris@10 14 *
Chris@10 15 * You should have received a copy of the GNU General Public License
Chris@10 16 * along with this program; if not, write to the Free Software
Chris@10 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@10 18 *
Chris@10 19 */
Chris@10 20
Chris@10 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@10 22 /* Generated on Sun Nov 25 07:35:42 EST 2012 */
Chris@10 23
Chris@10 24 #include "codelet-dft.h"
Chris@10 25
Chris@10 26 #ifdef HAVE_FMA
Chris@10 27
Chris@10 28 /* Generated by: ../../../genfft/gen_notw.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 3 -name n1_3 -include n.h */
Chris@10 29
Chris@10 30 /*
Chris@10 31 * This function contains 12 FP additions, 6 FP multiplications,
Chris@10 32 * (or, 6 additions, 0 multiplications, 6 fused multiply/add),
Chris@10 33 * 15 stack variables, 2 constants, and 12 memory accesses
Chris@10 34 */
Chris@10 35 #include "n.h"
Chris@10 36
Chris@10 37 static void n1_3(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@10 38 {
Chris@10 39 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@10 40 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10 41 {
Chris@10 42 INT i;
Chris@10 43 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(12, is), MAKE_VOLATILE_STRIDE(12, os)) {
Chris@10 44 E T1, T9, T2, T3, T6, T7;
Chris@10 45 T1 = ri[0];
Chris@10 46 T9 = ii[0];
Chris@10 47 T2 = ri[WS(is, 1)];
Chris@10 48 T3 = ri[WS(is, 2)];
Chris@10 49 T6 = ii[WS(is, 1)];
Chris@10 50 T7 = ii[WS(is, 2)];
Chris@10 51 {
Chris@10 52 E T4, Tc, T8, Ta, T5, Tb;
Chris@10 53 T4 = T2 + T3;
Chris@10 54 Tc = T3 - T2;
Chris@10 55 T8 = T6 - T7;
Chris@10 56 Ta = T6 + T7;
Chris@10 57 T5 = FNMS(KP500000000, T4, T1);
Chris@10 58 ro[0] = T1 + T4;
Chris@10 59 Tb = FNMS(KP500000000, Ta, T9);
Chris@10 60 io[0] = T9 + Ta;
Chris@10 61 ro[WS(os, 1)] = FMA(KP866025403, T8, T5);
Chris@10 62 ro[WS(os, 2)] = FNMS(KP866025403, T8, T5);
Chris@10 63 io[WS(os, 2)] = FNMS(KP866025403, Tc, Tb);
Chris@10 64 io[WS(os, 1)] = FMA(KP866025403, Tc, Tb);
Chris@10 65 }
Chris@10 66 }
Chris@10 67 }
Chris@10 68 }
Chris@10 69
Chris@10 70 static const kdft_desc desc = { 3, "n1_3", {6, 0, 6, 0}, &GENUS, 0, 0, 0, 0 };
Chris@10 71
Chris@10 72 void X(codelet_n1_3) (planner *p) {
Chris@10 73 X(kdft_register) (p, n1_3, &desc);
Chris@10 74 }
Chris@10 75
Chris@10 76 #else /* HAVE_FMA */
Chris@10 77
Chris@10 78 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 3 -name n1_3 -include n.h */
Chris@10 79
Chris@10 80 /*
Chris@10 81 * This function contains 12 FP additions, 4 FP multiplications,
Chris@10 82 * (or, 10 additions, 2 multiplications, 2 fused multiply/add),
Chris@10 83 * 15 stack variables, 2 constants, and 12 memory accesses
Chris@10 84 */
Chris@10 85 #include "n.h"
Chris@10 86
Chris@10 87 static void n1_3(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@10 88 {
Chris@10 89 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10 90 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@10 91 {
Chris@10 92 INT i;
Chris@10 93 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(12, is), MAKE_VOLATILE_STRIDE(12, os)) {
Chris@10 94 E T1, Ta, T4, T9, T8, Tb, T5, Tc;
Chris@10 95 T1 = ri[0];
Chris@10 96 Ta = ii[0];
Chris@10 97 {
Chris@10 98 E T2, T3, T6, T7;
Chris@10 99 T2 = ri[WS(is, 1)];
Chris@10 100 T3 = ri[WS(is, 2)];
Chris@10 101 T4 = T2 + T3;
Chris@10 102 T9 = KP866025403 * (T3 - T2);
Chris@10 103 T6 = ii[WS(is, 1)];
Chris@10 104 T7 = ii[WS(is, 2)];
Chris@10 105 T8 = KP866025403 * (T6 - T7);
Chris@10 106 Tb = T6 + T7;
Chris@10 107 }
Chris@10 108 ro[0] = T1 + T4;
Chris@10 109 io[0] = Ta + Tb;
Chris@10 110 T5 = FNMS(KP500000000, T4, T1);
Chris@10 111 ro[WS(os, 2)] = T5 - T8;
Chris@10 112 ro[WS(os, 1)] = T5 + T8;
Chris@10 113 Tc = FNMS(KP500000000, Tb, Ta);
Chris@10 114 io[WS(os, 1)] = T9 + Tc;
Chris@10 115 io[WS(os, 2)] = Tc - T9;
Chris@10 116 }
Chris@10 117 }
Chris@10 118 }
Chris@10 119
Chris@10 120 static const kdft_desc desc = { 3, "n1_3", {10, 2, 2, 0}, &GENUS, 0, 0, 0, 0 };
Chris@10 121
Chris@10 122 void X(codelet_n1_3) (planner *p) {
Chris@10 123 X(kdft_register) (p, n1_3, &desc);
Chris@10 124 }
Chris@10 125
Chris@10 126 #endif /* HAVE_FMA */