Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/scalar/codelets/n1_4.c @ 82:d0c2a83c1364

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW 3.3.8 source, and a Linux build
author Chris Cannam
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
rev   line source
Chris@82 1 /*
Chris@82 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@82 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@82 4 *
Chris@82 5 * This program is free software; you can redistribute it and/or modify
Chris@82 6 * it under the terms of the GNU General Public License as published by
Chris@82 7 * the Free Software Foundation; either version 2 of the License, or
Chris@82 8 * (at your option) any later version.
Chris@82 9 *
Chris@82 10 * This program is distributed in the hope that it will be useful,
Chris@82 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@82 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@82 13 * GNU General Public License for more details.
Chris@82 14 *
Chris@82 15 * You should have received a copy of the GNU General Public License
Chris@82 16 * along with this program; if not, write to the Free Software
Chris@82 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@82 18 *
Chris@82 19 */
Chris@82 20
Chris@82 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@82 22 /* Generated on Thu May 24 08:04:10 EDT 2018 */
Chris@82 23
Chris@82 24 #include "dft/codelet-dft.h"
Chris@82 25
Chris@82 26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
Chris@82 27
Chris@82 28 /* Generated by: ../../../genfft/gen_notw.native -fma -compact -variables 4 -pipeline-latency 4 -n 4 -name n1_4 -include dft/scalar/n.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 16 FP additions, 0 FP multiplications,
Chris@82 32 * (or, 16 additions, 0 multiplications, 0 fused multiply/add),
Chris@82 33 * 13 stack variables, 0 constants, and 16 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/scalar/n.h"
Chris@82 36
Chris@82 37 static void n1_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 38 {
Chris@82 39 {
Chris@82 40 INT i;
Chris@82 41 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(16, is), MAKE_VOLATILE_STRIDE(16, os)) {
Chris@82 42 E T3, Tb, T9, Tf, T6, Ta, Te, Tg;
Chris@82 43 {
Chris@82 44 E T1, T2, T7, T8;
Chris@82 45 T1 = ri[0];
Chris@82 46 T2 = ri[WS(is, 2)];
Chris@82 47 T3 = T1 + T2;
Chris@82 48 Tb = T1 - T2;
Chris@82 49 T7 = ii[0];
Chris@82 50 T8 = ii[WS(is, 2)];
Chris@82 51 T9 = T7 - T8;
Chris@82 52 Tf = T7 + T8;
Chris@82 53 }
Chris@82 54 {
Chris@82 55 E T4, T5, Tc, Td;
Chris@82 56 T4 = ri[WS(is, 1)];
Chris@82 57 T5 = ri[WS(is, 3)];
Chris@82 58 T6 = T4 + T5;
Chris@82 59 Ta = T4 - T5;
Chris@82 60 Tc = ii[WS(is, 1)];
Chris@82 61 Td = ii[WS(is, 3)];
Chris@82 62 Te = Tc - Td;
Chris@82 63 Tg = Tc + Td;
Chris@82 64 }
Chris@82 65 ro[WS(os, 2)] = T3 - T6;
Chris@82 66 io[WS(os, 2)] = Tf - Tg;
Chris@82 67 ro[0] = T3 + T6;
Chris@82 68 io[0] = Tf + Tg;
Chris@82 69 io[WS(os, 1)] = T9 - Ta;
Chris@82 70 ro[WS(os, 1)] = Tb + Te;
Chris@82 71 io[WS(os, 3)] = Ta + T9;
Chris@82 72 ro[WS(os, 3)] = Tb - Te;
Chris@82 73 }
Chris@82 74 }
Chris@82 75 }
Chris@82 76
Chris@82 77 static const kdft_desc desc = { 4, "n1_4", {16, 0, 0, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 78
Chris@82 79 void X(codelet_n1_4) (planner *p) {
Chris@82 80 X(kdft_register) (p, n1_4, &desc);
Chris@82 81 }
Chris@82 82
Chris@82 83 #else
Chris@82 84
Chris@82 85 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 4 -name n1_4 -include dft/scalar/n.h */
Chris@82 86
Chris@82 87 /*
Chris@82 88 * This function contains 16 FP additions, 0 FP multiplications,
Chris@82 89 * (or, 16 additions, 0 multiplications, 0 fused multiply/add),
Chris@82 90 * 13 stack variables, 0 constants, and 16 memory accesses
Chris@82 91 */
Chris@82 92 #include "dft/scalar/n.h"
Chris@82 93
Chris@82 94 static void n1_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 95 {
Chris@82 96 {
Chris@82 97 INT i;
Chris@82 98 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(16, is), MAKE_VOLATILE_STRIDE(16, os)) {
Chris@82 99 E T3, Tb, T9, Tf, T6, Ta, Te, Tg;
Chris@82 100 {
Chris@82 101 E T1, T2, T7, T8;
Chris@82 102 T1 = ri[0];
Chris@82 103 T2 = ri[WS(is, 2)];
Chris@82 104 T3 = T1 + T2;
Chris@82 105 Tb = T1 - T2;
Chris@82 106 T7 = ii[0];
Chris@82 107 T8 = ii[WS(is, 2)];
Chris@82 108 T9 = T7 - T8;
Chris@82 109 Tf = T7 + T8;
Chris@82 110 }
Chris@82 111 {
Chris@82 112 E T4, T5, Tc, Td;
Chris@82 113 T4 = ri[WS(is, 1)];
Chris@82 114 T5 = ri[WS(is, 3)];
Chris@82 115 T6 = T4 + T5;
Chris@82 116 Ta = T4 - T5;
Chris@82 117 Tc = ii[WS(is, 1)];
Chris@82 118 Td = ii[WS(is, 3)];
Chris@82 119 Te = Tc - Td;
Chris@82 120 Tg = Tc + Td;
Chris@82 121 }
Chris@82 122 ro[WS(os, 2)] = T3 - T6;
Chris@82 123 io[WS(os, 2)] = Tf - Tg;
Chris@82 124 ro[0] = T3 + T6;
Chris@82 125 io[0] = Tf + Tg;
Chris@82 126 io[WS(os, 1)] = T9 - Ta;
Chris@82 127 ro[WS(os, 1)] = Tb + Te;
Chris@82 128 io[WS(os, 3)] = Ta + T9;
Chris@82 129 ro[WS(os, 3)] = Tb - Te;
Chris@82 130 }
Chris@82 131 }
Chris@82 132 }
Chris@82 133
Chris@82 134 static const kdft_desc desc = { 4, "n1_4", {16, 0, 0, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 135
Chris@82 136 void X(codelet_n1_4) (planner *p) {
Chris@82 137 X(kdft_register) (p, n1_4, &desc);
Chris@82 138 }
Chris@82 139
Chris@82 140 #endif