Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/scalar/codelets/n1_3.c @ 84:08ae793730bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +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 16:35:50 EDT 2016 */
Chris@42 23
Chris@42 24 #include "codelet-dft.h"
Chris@42 25
Chris@42 26 #ifdef HAVE_FMA
Chris@42 27
Chris@42 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@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 12 FP additions, 6 FP multiplications,
Chris@42 32 * (or, 6 additions, 0 multiplications, 6 fused multiply/add),
Chris@42 33 * 15 stack variables, 2 constants, and 12 memory accesses
Chris@42 34 */
Chris@42 35 #include "n.h"
Chris@42 36
Chris@42 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@42 38 {
Chris@42 39 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 40 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 41 {
Chris@42 42 INT i;
Chris@42 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@42 44 E T1, T9, T2, T3, T6, T7;
Chris@42 45 T1 = ri[0];
Chris@42 46 T9 = ii[0];
Chris@42 47 T2 = ri[WS(is, 1)];
Chris@42 48 T3 = ri[WS(is, 2)];
Chris@42 49 T6 = ii[WS(is, 1)];
Chris@42 50 T7 = ii[WS(is, 2)];
Chris@42 51 {
Chris@42 52 E T4, Tc, T8, Ta, T5, Tb;
Chris@42 53 T4 = T2 + T3;
Chris@42 54 Tc = T3 - T2;
Chris@42 55 T8 = T6 - T7;
Chris@42 56 Ta = T6 + T7;
Chris@42 57 T5 = FNMS(KP500000000, T4, T1);
Chris@42 58 ro[0] = T1 + T4;
Chris@42 59 Tb = FNMS(KP500000000, Ta, T9);
Chris@42 60 io[0] = T9 + Ta;
Chris@42 61 ro[WS(os, 1)] = FMA(KP866025403, T8, T5);
Chris@42 62 ro[WS(os, 2)] = FNMS(KP866025403, T8, T5);
Chris@42 63 io[WS(os, 2)] = FNMS(KP866025403, Tc, Tb);
Chris@42 64 io[WS(os, 1)] = FMA(KP866025403, Tc, Tb);
Chris@42 65 }
Chris@42 66 }
Chris@42 67 }
Chris@42 68 }
Chris@42 69
Chris@42 70 static const kdft_desc desc = { 3, "n1_3", {6, 0, 6, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 71
Chris@42 72 void X(codelet_n1_3) (planner *p) {
Chris@42 73 X(kdft_register) (p, n1_3, &desc);
Chris@42 74 }
Chris@42 75
Chris@42 76 #else /* HAVE_FMA */
Chris@42 77
Chris@42 78 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 3 -name n1_3 -include n.h */
Chris@42 79
Chris@42 80 /*
Chris@42 81 * This function contains 12 FP additions, 4 FP multiplications,
Chris@42 82 * (or, 10 additions, 2 multiplications, 2 fused multiply/add),
Chris@42 83 * 15 stack variables, 2 constants, and 12 memory accesses
Chris@42 84 */
Chris@42 85 #include "n.h"
Chris@42 86
Chris@42 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@42 88 {
Chris@42 89 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 90 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 91 {
Chris@42 92 INT i;
Chris@42 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@42 94 E T1, Ta, T4, T9, T8, Tb, T5, Tc;
Chris@42 95 T1 = ri[0];
Chris@42 96 Ta = ii[0];
Chris@42 97 {
Chris@42 98 E T2, T3, T6, T7;
Chris@42 99 T2 = ri[WS(is, 1)];
Chris@42 100 T3 = ri[WS(is, 2)];
Chris@42 101 T4 = T2 + T3;
Chris@42 102 T9 = KP866025403 * (T3 - T2);
Chris@42 103 T6 = ii[WS(is, 1)];
Chris@42 104 T7 = ii[WS(is, 2)];
Chris@42 105 T8 = KP866025403 * (T6 - T7);
Chris@42 106 Tb = T6 + T7;
Chris@42 107 }
Chris@42 108 ro[0] = T1 + T4;
Chris@42 109 io[0] = Ta + Tb;
Chris@42 110 T5 = FNMS(KP500000000, T4, T1);
Chris@42 111 ro[WS(os, 2)] = T5 - T8;
Chris@42 112 ro[WS(os, 1)] = T5 + T8;
Chris@42 113 Tc = FNMS(KP500000000, Tb, Ta);
Chris@42 114 io[WS(os, 1)] = T9 + Tc;
Chris@42 115 io[WS(os, 2)] = Tc - T9;
Chris@42 116 }
Chris@42 117 }
Chris@42 118 }
Chris@42 119
Chris@42 120 static const kdft_desc desc = { 3, "n1_3", {10, 2, 2, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 121
Chris@42 122 void X(codelet_n1_3) (planner *p) {
Chris@42 123 X(kdft_register) (p, n1_3, &desc);
Chris@42 124 }
Chris@42 125
Chris@42 126 #endif /* HAVE_FMA */