Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cb/r2cb_14.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 08:07:28 EDT 2018 */
Chris@82 23
Chris@82 24 #include "rdft/codelet-rdft.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_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 14 -name r2cb_14 -include rdft/scalar/r2cb.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 62 FP additions, 44 FP multiplications,
Chris@82 32 * (or, 18 additions, 0 multiplications, 44 fused multiply/add),
Chris@82 33 * 46 stack variables, 7 constants, and 28 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/r2cb.h"
Chris@82 36
Chris@82 37 static void r2cb_14(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 38 {
Chris@82 39 DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
Chris@82 40 DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
Chris@82 41 DK(KP692021471, +0.692021471630095869627814897002069140197260599);
Chris@82 42 DK(KP801937735, +0.801937735804838252472204639014890102331838324);
Chris@82 43 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 44 DK(KP356895867, +0.356895867892209443894399510021300583399127187);
Chris@82 45 DK(KP554958132, +0.554958132087371191422194871006410481067288862);
Chris@82 46 {
Chris@82 47 INT i;
Chris@82 48 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
Chris@82 49 E T3, Te, To, TK, Tu, TM, Tr, TL, Tv, TA, TX, TS, TN, TF, T6;
Chris@82 50 E Tf, Tc, Th, T9, Tg, Tj, Tx, TU, TP, TH, TC, T1, T2, Td, Ti;
Chris@82 51 T1 = Cr[0];
Chris@82 52 T2 = Cr[WS(csr, 7)];
Chris@82 53 T3 = T1 - T2;
Chris@82 54 Te = T1 + T2;
Chris@82 55 {
Chris@82 56 E Tm, Tn, T4, T5;
Chris@82 57 Tm = Ci[WS(csi, 4)];
Chris@82 58 Tn = Ci[WS(csi, 3)];
Chris@82 59 To = Tm - Tn;
Chris@82 60 TK = Tm + Tn;
Chris@82 61 {
Chris@82 62 E Ts, Tt, Tp, Tq;
Chris@82 63 Ts = Ci[WS(csi, 6)];
Chris@82 64 Tt = Ci[WS(csi, 1)];
Chris@82 65 Tu = Ts - Tt;
Chris@82 66 TM = Ts + Tt;
Chris@82 67 Tp = Ci[WS(csi, 2)];
Chris@82 68 Tq = Ci[WS(csi, 5)];
Chris@82 69 Tr = Tp - Tq;
Chris@82 70 TL = Tp + Tq;
Chris@82 71 }
Chris@82 72 Tv = FMA(KP554958132, Tu, Tr);
Chris@82 73 TA = FMA(KP554958132, To, Tu);
Chris@82 74 TX = FNMS(KP554958132, TL, TK);
Chris@82 75 TS = FMA(KP554958132, TK, TM);
Chris@82 76 TN = FMA(KP554958132, TM, TL);
Chris@82 77 TF = FNMS(KP554958132, Tr, To);
Chris@82 78 T4 = Cr[WS(csr, 2)];
Chris@82 79 T5 = Cr[WS(csr, 5)];
Chris@82 80 T6 = T4 - T5;
Chris@82 81 Tf = T4 + T5;
Chris@82 82 {
Chris@82 83 E Ta, Tb, T7, T8;
Chris@82 84 Ta = Cr[WS(csr, 6)];
Chris@82 85 Tb = Cr[WS(csr, 1)];
Chris@82 86 Tc = Ta - Tb;
Chris@82 87 Th = Ta + Tb;
Chris@82 88 T7 = Cr[WS(csr, 4)];
Chris@82 89 T8 = Cr[WS(csr, 3)];
Chris@82 90 T9 = T7 - T8;
Chris@82 91 Tg = T7 + T8;
Chris@82 92 }
Chris@82 93 Tj = FNMS(KP356895867, Tg, Tf);
Chris@82 94 Tx = FNMS(KP356895867, Tf, Th);
Chris@82 95 TU = FNMS(KP356895867, Tc, T9);
Chris@82 96 TP = FNMS(KP356895867, T6, Tc);
Chris@82 97 TH = FNMS(KP356895867, T9, T6);
Chris@82 98 TC = FNMS(KP356895867, Th, Tg);
Chris@82 99 }
Chris@82 100 Td = T6 + T9 + Tc;
Chris@82 101 R1[WS(rs, 3)] = FMA(KP2_000000000, Td, T3);
Chris@82 102 Ti = Tf + Tg + Th;
Chris@82 103 R0[0] = FMA(KP2_000000000, Ti, Te);
Chris@82 104 {
Chris@82 105 E Tw, Tl, Tk, TY, TW, TV;
Chris@82 106 Tw = FMA(KP801937735, Tv, To);
Chris@82 107 Tk = FNMS(KP692021471, Tj, Th);
Chris@82 108 Tl = FNMS(KP1_801937735, Tk, Te);
Chris@82 109 R0[WS(rs, 4)] = FNMS(KP1_949855824, Tw, Tl);
Chris@82 110 R0[WS(rs, 3)] = FMA(KP1_949855824, Tw, Tl);
Chris@82 111 TY = FNMS(KP801937735, TX, TM);
Chris@82 112 TV = FNMS(KP692021471, TU, T6);
Chris@82 113 TW = FNMS(KP1_801937735, TV, T3);
Chris@82 114 R1[WS(rs, 1)] = FNMS(KP1_949855824, TY, TW);
Chris@82 115 R1[WS(rs, 5)] = FMA(KP1_949855824, TY, TW);
Chris@82 116 }
Chris@82 117 {
Chris@82 118 E TB, Tz, Ty, TO, TJ, TI;
Chris@82 119 TB = FNMS(KP801937735, TA, Tr);
Chris@82 120 Ty = FNMS(KP692021471, Tx, Tg);
Chris@82 121 Tz = FNMS(KP1_801937735, Ty, Te);
Chris@82 122 R0[WS(rs, 1)] = FNMS(KP1_949855824, TB, Tz);
Chris@82 123 R0[WS(rs, 6)] = FMA(KP1_949855824, TB, Tz);
Chris@82 124 TO = FMA(KP801937735, TN, TK);
Chris@82 125 TI = FNMS(KP692021471, TH, Tc);
Chris@82 126 TJ = FNMS(KP1_801937735, TI, T3);
Chris@82 127 R1[0] = FNMS(KP1_949855824, TO, TJ);
Chris@82 128 R1[WS(rs, 6)] = FMA(KP1_949855824, TO, TJ);
Chris@82 129 }
Chris@82 130 {
Chris@82 131 E TT, TR, TQ, TG, TE, TD;
Chris@82 132 TT = FNMS(KP801937735, TS, TL);
Chris@82 133 TQ = FNMS(KP692021471, TP, T9);
Chris@82 134 TR = FNMS(KP1_801937735, TQ, T3);
Chris@82 135 R1[WS(rs, 4)] = FNMS(KP1_949855824, TT, TR);
Chris@82 136 R1[WS(rs, 2)] = FMA(KP1_949855824, TT, TR);
Chris@82 137 TG = FNMS(KP801937735, TF, Tu);
Chris@82 138 TD = FNMS(KP692021471, TC, Tf);
Chris@82 139 TE = FNMS(KP1_801937735, TD, Te);
Chris@82 140 R0[WS(rs, 5)] = FNMS(KP1_949855824, TG, TE);
Chris@82 141 R0[WS(rs, 2)] = FMA(KP1_949855824, TG, TE);
Chris@82 142 }
Chris@82 143 }
Chris@82 144 }
Chris@82 145 }
Chris@82 146
Chris@82 147 static const kr2c_desc desc = { 14, "r2cb_14", {18, 0, 44, 0}, &GENUS };
Chris@82 148
Chris@82 149 void X(codelet_r2cb_14) (planner *p) {
Chris@82 150 X(kr2c_register) (p, r2cb_14, &desc);
Chris@82 151 }
Chris@82 152
Chris@82 153 #else
Chris@82 154
Chris@82 155 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 14 -name r2cb_14 -include rdft/scalar/r2cb.h */
Chris@82 156
Chris@82 157 /*
Chris@82 158 * This function contains 62 FP additions, 38 FP multiplications,
Chris@82 159 * (or, 36 additions, 12 multiplications, 26 fused multiply/add),
Chris@82 160 * 28 stack variables, 7 constants, and 28 memory accesses
Chris@82 161 */
Chris@82 162 #include "rdft/scalar/r2cb.h"
Chris@82 163
Chris@82 164 static void r2cb_14(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 165 {
Chris@82 166 DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
Chris@82 167 DK(KP445041867, +0.445041867912628808577805128993589518932711138);
Chris@82 168 DK(KP1_246979603, +1.246979603717467061050009768008479621264549462);
Chris@82 169 DK(KP867767478, +0.867767478235116240951536665696717509219981456);
Chris@82 170 DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
Chris@82 171 DK(KP1_563662964, +1.563662964936059617416889053348115500464669037);
Chris@82 172 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 173 {
Chris@82 174 INT i;
Chris@82 175 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
Chris@82 176 E T3, Td, T6, Te, Tq, Tz, Tn, Ty, Tc, Tg, Tk, Tx, T9, Tf, T1;
Chris@82 177 E T2;
Chris@82 178 T1 = Cr[0];
Chris@82 179 T2 = Cr[WS(csr, 7)];
Chris@82 180 T3 = T1 - T2;
Chris@82 181 Td = T1 + T2;
Chris@82 182 {
Chris@82 183 E T4, T5, To, Tp;
Chris@82 184 T4 = Cr[WS(csr, 2)];
Chris@82 185 T5 = Cr[WS(csr, 5)];
Chris@82 186 T6 = T4 - T5;
Chris@82 187 Te = T4 + T5;
Chris@82 188 To = Ci[WS(csi, 2)];
Chris@82 189 Tp = Ci[WS(csi, 5)];
Chris@82 190 Tq = To - Tp;
Chris@82 191 Tz = To + Tp;
Chris@82 192 }
Chris@82 193 {
Chris@82 194 E Tl, Tm, Ta, Tb;
Chris@82 195 Tl = Ci[WS(csi, 6)];
Chris@82 196 Tm = Ci[WS(csi, 1)];
Chris@82 197 Tn = Tl - Tm;
Chris@82 198 Ty = Tl + Tm;
Chris@82 199 Ta = Cr[WS(csr, 6)];
Chris@82 200 Tb = Cr[WS(csr, 1)];
Chris@82 201 Tc = Ta - Tb;
Chris@82 202 Tg = Ta + Tb;
Chris@82 203 }
Chris@82 204 {
Chris@82 205 E Ti, Tj, T7, T8;
Chris@82 206 Ti = Ci[WS(csi, 4)];
Chris@82 207 Tj = Ci[WS(csi, 3)];
Chris@82 208 Tk = Ti - Tj;
Chris@82 209 Tx = Ti + Tj;
Chris@82 210 T7 = Cr[WS(csr, 4)];
Chris@82 211 T8 = Cr[WS(csr, 3)];
Chris@82 212 T9 = T7 - T8;
Chris@82 213 Tf = T7 + T8;
Chris@82 214 }
Chris@82 215 R1[WS(rs, 3)] = FMA(KP2_000000000, T6 + T9 + Tc, T3);
Chris@82 216 R0[0] = FMA(KP2_000000000, Te + Tf + Tg, Td);
Chris@82 217 {
Chris@82 218 E Tr, Th, TE, TD;
Chris@82 219 Tr = FNMS(KP1_949855824, Tn, KP1_563662964 * Tk) - (KP867767478 * Tq);
Chris@82 220 Th = FMA(KP1_246979603, Tf, Td) + FNMA(KP445041867, Tg, KP1_801937735 * Te);
Chris@82 221 R0[WS(rs, 2)] = Th - Tr;
Chris@82 222 R0[WS(rs, 5)] = Th + Tr;
Chris@82 223 TE = FMA(KP867767478, Tx, KP1_563662964 * Ty) - (KP1_949855824 * Tz);
Chris@82 224 TD = FMA(KP1_246979603, Tc, T3) + FNMA(KP1_801937735, T9, KP445041867 * T6);
Chris@82 225 R1[WS(rs, 2)] = TD - TE;
Chris@82 226 R1[WS(rs, 4)] = TD + TE;
Chris@82 227 }
Chris@82 228 {
Chris@82 229 E Tt, Ts, TA, Tw;
Chris@82 230 Tt = FMA(KP867767478, Tk, KP1_563662964 * Tn) - (KP1_949855824 * Tq);
Chris@82 231 Ts = FMA(KP1_246979603, Tg, Td) + FNMA(KP1_801937735, Tf, KP445041867 * Te);
Chris@82 232 R0[WS(rs, 6)] = Ts - Tt;
Chris@82 233 R0[WS(rs, 1)] = Ts + Tt;
Chris@82 234 TA = FNMS(KP1_949855824, Ty, KP1_563662964 * Tx) - (KP867767478 * Tz);
Chris@82 235 Tw = FMA(KP1_246979603, T9, T3) + FNMA(KP445041867, Tc, KP1_801937735 * T6);
Chris@82 236 R1[WS(rs, 5)] = Tw - TA;
Chris@82 237 R1[WS(rs, 1)] = Tw + TA;
Chris@82 238 }
Chris@82 239 {
Chris@82 240 E TC, TB, Tv, Tu;
Chris@82 241 TC = FMA(KP1_563662964, Tz, KP1_949855824 * Tx) + (KP867767478 * Ty);
Chris@82 242 TB = FMA(KP1_246979603, T6, T3) + FNMA(KP1_801937735, Tc, KP445041867 * T9);
Chris@82 243 R1[0] = TB - TC;
Chris@82 244 R1[WS(rs, 6)] = TB + TC;
Chris@82 245 Tv = FMA(KP1_563662964, Tq, KP1_949855824 * Tk) + (KP867767478 * Tn);
Chris@82 246 Tu = FMA(KP1_246979603, Te, Td) + FNMA(KP1_801937735, Tg, KP445041867 * Tf);
Chris@82 247 R0[WS(rs, 4)] = Tu - Tv;
Chris@82 248 R0[WS(rs, 3)] = Tu + Tv;
Chris@82 249 }
Chris@82 250 }
Chris@82 251 }
Chris@82 252 }
Chris@82 253
Chris@82 254 static const kr2c_desc desc = { 14, "r2cb_14", {36, 12, 26, 0}, &GENUS };
Chris@82 255
Chris@82 256 void X(codelet_r2cb_14) (planner *p) {
Chris@82 257 X(kr2c_register) (p, r2cb_14, &desc);
Chris@82 258 }
Chris@82 259
Chris@82 260 #endif