Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/n1fv_7.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 d0c2a83c1364
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 08:04:51 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_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 7 -name n1fv_7 -include dft/simd/n1f.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 30 FP additions, 24 FP multiplications,
Chris@82 32 * (or, 9 additions, 3 multiplications, 21 fused multiply/add),
Chris@82 33 * 33 stack variables, 6 constants, and 14 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/n1f.h"
Chris@82 36
Chris@82 37 static void n1fv_7(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 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@82 40 DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
Chris@82 41 DVK(KP801937735, +0.801937735804838252472204639014890102331838324);
Chris@82 42 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@82 43 DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
Chris@82 44 DVK(KP356895867, +0.356895867892209443894399510021300583399127187);
Chris@82 45 {
Chris@82 46 INT i;
Chris@82 47 const R *xi;
Chris@82 48 R *xo;
Chris@82 49 xi = ri;
Chris@82 50 xo = ro;
Chris@82 51 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(14, is), MAKE_VOLATILE_STRIDE(14, os)) {
Chris@82 52 V T1, T4, Te, Ta, Tf, T7, Tg, Tb, Th, Tr, To, Tm, Tj, T2, T3;
Chris@82 53 V Ts, Tq, Tp;
Chris@82 54 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 55 T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 56 T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82 57 T4 = VADD(T2, T3);
Chris@82 58 Te = VSUB(T3, T2);
Chris@82 59 {
Chris@82 60 V T8, T9, T5, T6;
Chris@82 61 T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 62 T9 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82 63 Ta = VADD(T8, T9);
Chris@82 64 Tf = VSUB(T9, T8);
Chris@82 65 T5 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 66 T6 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82 67 T7 = VADD(T5, T6);
Chris@82 68 Tg = VSUB(T6, T5);
Chris@82 69 }
Chris@82 70 Tb = VFNMS(LDK(KP356895867), T4, Ta);
Chris@82 71 Th = VFMA(LDK(KP554958132), Tg, Tf);
Chris@82 72 Tr = VFNMS(LDK(KP554958132), Te, Tg);
Chris@82 73 To = VFNMS(LDK(KP356895867), Ta, T7);
Chris@82 74 Tm = VFMA(LDK(KP554958132), Tf, Te);
Chris@82 75 Tj = VFNMS(LDK(KP356895867), T7, T4);
Chris@82 76 ST(&(xo[0]), VADD(T1, VADD(T4, VADD(T7, Ta))), ovs, &(xo[0]));
Chris@82 77 Ts = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tr, Tf));
Chris@82 78 Tp = VFNMS(LDK(KP692021471), To, T4);
Chris@82 79 Tq = VFNMS(LDK(KP900968867), Tp, T1);
Chris@82 80 ST(&(xo[WS(os, 4)]), VFNMSI(Ts, Tq), ovs, &(xo[0]));
Chris@82 81 ST(&(xo[WS(os, 3)]), VFMAI(Ts, Tq), ovs, &(xo[WS(os, 1)]));
Chris@82 82 {
Chris@82 83 V Ti, Td, Tc, Tn, Tl, Tk;
Chris@82 84 Ti = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Th, Te));
Chris@82 85 Tc = VFNMS(LDK(KP692021471), Tb, T7);
Chris@82 86 Td = VFNMS(LDK(KP900968867), Tc, T1);
Chris@82 87 ST(&(xo[WS(os, 5)]), VFNMSI(Ti, Td), ovs, &(xo[WS(os, 1)]));
Chris@82 88 ST(&(xo[WS(os, 2)]), VFMAI(Ti, Td), ovs, &(xo[0]));
Chris@82 89 Tn = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), Tm, Tg));
Chris@82 90 Tk = VFNMS(LDK(KP692021471), Tj, Ta);
Chris@82 91 Tl = VFNMS(LDK(KP900968867), Tk, T1);
Chris@82 92 ST(&(xo[WS(os, 6)]), VFNMSI(Tn, Tl), ovs, &(xo[0]));
Chris@82 93 ST(&(xo[WS(os, 1)]), VFMAI(Tn, Tl), ovs, &(xo[WS(os, 1)]));
Chris@82 94 }
Chris@82 95 }
Chris@82 96 }
Chris@82 97 VLEAVE();
Chris@82 98 }
Chris@82 99
Chris@82 100 static const kdft_desc desc = { 7, XSIMD_STRING("n1fv_7"), {9, 3, 21, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 101
Chris@82 102 void XSIMD(codelet_n1fv_7) (planner *p) {
Chris@82 103 X(kdft_register) (p, n1fv_7, &desc);
Chris@82 104 }
Chris@82 105
Chris@82 106 #else
Chris@82 107
Chris@82 108 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 7 -name n1fv_7 -include dft/simd/n1f.h */
Chris@82 109
Chris@82 110 /*
Chris@82 111 * This function contains 30 FP additions, 18 FP multiplications,
Chris@82 112 * (or, 18 additions, 6 multiplications, 12 fused multiply/add),
Chris@82 113 * 24 stack variables, 6 constants, and 14 memory accesses
Chris@82 114 */
Chris@82 115 #include "dft/simd/n1f.h"
Chris@82 116
Chris@82 117 static void n1fv_7(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 118 {
Chris@82 119 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@82 120 DVK(KP222520933, +0.222520933956314404288902564496794759466355569);
Chris@82 121 DVK(KP623489801, +0.623489801858733530525004884004239810632274731);
Chris@82 122 DVK(KP781831482, +0.781831482468029808708444526674057750232334519);
Chris@82 123 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@82 124 DVK(KP433883739, +0.433883739117558120475768332848358754609990728);
Chris@82 125 {
Chris@82 126 INT i;
Chris@82 127 const R *xi;
Chris@82 128 R *xo;
Chris@82 129 xi = ri;
Chris@82 130 xo = ro;
Chris@82 131 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(14, is), MAKE_VOLATILE_STRIDE(14, os)) {
Chris@82 132 V T1, Ta, Td, T4, Tc, T7, Te, T8, T9, Tj, Ti;
Chris@82 133 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 134 T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 135 T9 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82 136 Ta = VADD(T8, T9);
Chris@82 137 Td = VSUB(T9, T8);
Chris@82 138 {
Chris@82 139 V T2, T3, T5, T6;
Chris@82 140 T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 141 T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82 142 T4 = VADD(T2, T3);
Chris@82 143 Tc = VSUB(T3, T2);
Chris@82 144 T5 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 145 T6 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82 146 T7 = VADD(T5, T6);
Chris@82 147 Te = VSUB(T6, T5);
Chris@82 148 }
Chris@82 149 ST(&(xo[0]), VADD(T1, VADD(T4, VADD(T7, Ta))), ovs, &(xo[0]));
Chris@82 150 Tj = VBYI(VFMA(LDK(KP433883739), Tc, VFNMS(LDK(KP781831482), Te, VMUL(LDK(KP974927912), Td))));
Chris@82 151 Ti = VFMA(LDK(KP623489801), T7, VFNMS(LDK(KP222520933), Ta, VFNMS(LDK(KP900968867), T4, T1)));
Chris@82 152 ST(&(xo[WS(os, 4)]), VSUB(Ti, Tj), ovs, &(xo[0]));
Chris@82 153 ST(&(xo[WS(os, 3)]), VADD(Ti, Tj), ovs, &(xo[WS(os, 1)]));
Chris@82 154 {
Chris@82 155 V Tf, Tb, Th, Tg;
Chris@82 156 Tf = VBYI(VFNMS(LDK(KP781831482), Td, VFNMS(LDK(KP433883739), Te, VMUL(LDK(KP974927912), Tc))));
Chris@82 157 Tb = VFMA(LDK(KP623489801), Ta, VFNMS(LDK(KP900968867), T7, VFNMS(LDK(KP222520933), T4, T1)));
Chris@82 158 ST(&(xo[WS(os, 5)]), VSUB(Tb, Tf), ovs, &(xo[WS(os, 1)]));
Chris@82 159 ST(&(xo[WS(os, 2)]), VADD(Tb, Tf), ovs, &(xo[0]));
Chris@82 160 Th = VBYI(VFMA(LDK(KP781831482), Tc, VFMA(LDK(KP974927912), Te, VMUL(LDK(KP433883739), Td))));
Chris@82 161 Tg = VFMA(LDK(KP623489801), T4, VFNMS(LDK(KP900968867), Ta, VFNMS(LDK(KP222520933), T7, T1)));
Chris@82 162 ST(&(xo[WS(os, 6)]), VSUB(Tg, Th), ovs, &(xo[0]));
Chris@82 163 ST(&(xo[WS(os, 1)]), VADD(Tg, Th), ovs, &(xo[WS(os, 1)]));
Chris@82 164 }
Chris@82 165 }
Chris@82 166 }
Chris@82 167 VLEAVE();
Chris@82 168 }
Chris@82 169
Chris@82 170 static const kdft_desc desc = { 7, XSIMD_STRING("n1fv_7"), {18, 6, 12, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 171
Chris@82 172 void XSIMD(codelet_n1fv_7) (planner *p) {
Chris@82 173 X(kdft_register) (p, n1fv_7, &desc);
Chris@82 174 }
Chris@82 175
Chris@82 176 #endif