Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/n1fv_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 08:04:51 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 14 -name n1fv_14 -include dft/simd/n1f.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 74 FP additions, 48 FP multiplications,
Chris@82 32 * (or, 32 additions, 6 multiplications, 42 fused multiply/add),
Chris@82 33 * 51 stack variables, 6 constants, and 28 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/n1f.h"
Chris@82 36
Chris@82 37 static void n1fv_14(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(KP801937735, +0.801937735804838252472204639014890102331838324);
Chris@82 40 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@82 41 DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
Chris@82 42 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@82 43 DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
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(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
Chris@82 52 V T3, TH, Ts, TV, TW, Tt, Tu, TU, Ta, To, Th, Tp, TC, Tx, TK;
Chris@82 53 V TQ, TN, TR, T14, TZ, T1, T2;
Chris@82 54 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 55 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@82 56 T3 = VSUB(T1, T2);
Chris@82 57 TH = VADD(T1, T2);
Chris@82 58 {
Chris@82 59 V T6, TI, T9, TJ, Tn, TP, Tk, TO, Tg, TM, Td, TL;
Chris@82 60 {
Chris@82 61 V T4, T5, Ti, Tj;
Chris@82 62 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 63 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
Chris@82 64 T6 = VSUB(T4, T5);
Chris@82 65 TI = VADD(T4, T5);
Chris@82 66 {
Chris@82 67 V T7, T8, Tl, Tm;
Chris@82 68 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
Chris@82 69 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82 70 T9 = VSUB(T7, T8);
Chris@82 71 TJ = VADD(T7, T8);
Chris@82 72 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@82 73 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 74 Tn = VSUB(Tl, Tm);
Chris@82 75 TP = VADD(Tl, Tm);
Chris@82 76 }
Chris@82 77 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82 78 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
Chris@82 79 Tk = VSUB(Ti, Tj);
Chris@82 80 TO = VADD(Ti, Tj);
Chris@82 81 {
Chris@82 82 V Te, Tf, Tb, Tc;
Chris@82 83 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
Chris@82 84 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 85 Tg = VSUB(Te, Tf);
Chris@82 86 TM = VADD(Te, Tf);
Chris@82 87 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82 88 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
Chris@82 89 Td = VSUB(Tb, Tc);
Chris@82 90 TL = VADD(Tb, Tc);
Chris@82 91 }
Chris@82 92 }
Chris@82 93 Ts = VSUB(T9, T6);
Chris@82 94 TV = VSUB(TL, TM);
Chris@82 95 TW = VSUB(TJ, TI);
Chris@82 96 Tt = VSUB(Tn, Tk);
Chris@82 97 Tu = VSUB(Tg, Td);
Chris@82 98 TU = VSUB(TO, TP);
Chris@82 99 Ta = VADD(T6, T9);
Chris@82 100 To = VADD(Tk, Tn);
Chris@82 101 Th = VADD(Td, Tg);
Chris@82 102 Tp = VFNMS(LDK(KP356895867), Ta, To);
Chris@82 103 TC = VFNMS(LDK(KP356895867), To, Th);
Chris@82 104 Tx = VFNMS(LDK(KP356895867), Th, Ta);
Chris@82 105 TK = VADD(TI, TJ);
Chris@82 106 TQ = VADD(TO, TP);
Chris@82 107 TN = VADD(TL, TM);
Chris@82 108 TR = VFNMS(LDK(KP356895867), TQ, TN);
Chris@82 109 T14 = VFNMS(LDK(KP356895867), TN, TK);
Chris@82 110 TZ = VFNMS(LDK(KP356895867), TK, TQ);
Chris@82 111 }
Chris@82 112 ST(&(xo[WS(os, 7)]), VADD(T3, VADD(Ta, VADD(Th, To))), ovs, &(xo[WS(os, 1)]));
Chris@82 113 ST(&(xo[0]), VADD(TH, VADD(TK, VADD(TN, TQ))), ovs, &(xo[0]));
Chris@82 114 {
Chris@82 115 V Tr, Tw, Tq, Tv;
Chris@82 116 Tq = VFNMS(LDK(KP692021471), Tp, Th);
Chris@82 117 Tr = VFNMS(LDK(KP900968867), Tq, T3);
Chris@82 118 Tv = VFMA(LDK(KP554958132), Tu, Tt);
Chris@82 119 Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts));
Chris@82 120 ST(&(xo[WS(os, 5)]), VFNMSI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
Chris@82 121 ST(&(xo[WS(os, 9)]), VFMAI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
Chris@82 122 }
Chris@82 123 {
Chris@82 124 V T16, T18, T15, T17;
Chris@82 125 T15 = VFNMS(LDK(KP692021471), T14, TQ);
Chris@82 126 T16 = VFNMS(LDK(KP900968867), T15, TH);
Chris@82 127 T17 = VFNMS(LDK(KP554958132), TU, TW);
Chris@82 128 T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV));
Chris@82 129 ST(&(xo[WS(os, 6)]), VFMAI(T18, T16), ovs, &(xo[0]));
Chris@82 130 ST(&(xo[WS(os, 8)]), VFNMSI(T18, T16), ovs, &(xo[0]));
Chris@82 131 }
Chris@82 132 {
Chris@82 133 V Tz, TB, Ty, TA;
Chris@82 134 Ty = VFNMS(LDK(KP692021471), Tx, To);
Chris@82 135 Tz = VFNMS(LDK(KP900968867), Ty, T3);
Chris@82 136 TA = VFMA(LDK(KP554958132), Tt, Ts);
Chris@82 137 TB = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TA, Tu));
Chris@82 138 ST(&(xo[WS(os, 13)]), VFNMSI(TB, Tz), ovs, &(xo[WS(os, 1)]));
Chris@82 139 ST(&(xo[WS(os, 1)]), VFMAI(TB, Tz), ovs, &(xo[WS(os, 1)]));
Chris@82 140 }
Chris@82 141 {
Chris@82 142 V TT, TY, TS, TX;
Chris@82 143 TS = VFNMS(LDK(KP692021471), TR, TK);
Chris@82 144 TT = VFNMS(LDK(KP900968867), TS, TH);
Chris@82 145 TX = VFMA(LDK(KP554958132), TW, TV);
Chris@82 146 TY = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TX, TU));
Chris@82 147 ST(&(xo[WS(os, 4)]), VFMAI(TY, TT), ovs, &(xo[0]));
Chris@82 148 ST(&(xo[WS(os, 10)]), VFNMSI(TY, TT), ovs, &(xo[0]));
Chris@82 149 }
Chris@82 150 {
Chris@82 151 V T11, T13, T10, T12;
Chris@82 152 T10 = VFNMS(LDK(KP692021471), TZ, TN);
Chris@82 153 T11 = VFNMS(LDK(KP900968867), T10, TH);
Chris@82 154 T12 = VFMA(LDK(KP554958132), TV, TU);
Chris@82 155 T13 = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), T12, TW));
Chris@82 156 ST(&(xo[WS(os, 2)]), VFMAI(T13, T11), ovs, &(xo[0]));
Chris@82 157 ST(&(xo[WS(os, 12)]), VFNMSI(T13, T11), ovs, &(xo[0]));
Chris@82 158 }
Chris@82 159 {
Chris@82 160 V TE, TG, TD, TF;
Chris@82 161 TD = VFNMS(LDK(KP692021471), TC, Ta);
Chris@82 162 TE = VFNMS(LDK(KP900968867), TD, T3);
Chris@82 163 TF = VFNMS(LDK(KP554958132), Ts, Tu);
Chris@82 164 TG = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TF, Tt));
Chris@82 165 ST(&(xo[WS(os, 11)]), VFNMSI(TG, TE), ovs, &(xo[WS(os, 1)]));
Chris@82 166 ST(&(xo[WS(os, 3)]), VFMAI(TG, TE), ovs, &(xo[WS(os, 1)]));
Chris@82 167 }
Chris@82 168 }
Chris@82 169 }
Chris@82 170 VLEAVE();
Chris@82 171 }
Chris@82 172
Chris@82 173 static const kdft_desc desc = { 14, XSIMD_STRING("n1fv_14"), {32, 6, 42, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 174
Chris@82 175 void XSIMD(codelet_n1fv_14) (planner *p) {
Chris@82 176 X(kdft_register) (p, n1fv_14, &desc);
Chris@82 177 }
Chris@82 178
Chris@82 179 #else
Chris@82 180
Chris@82 181 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 14 -name n1fv_14 -include dft/simd/n1f.h */
Chris@82 182
Chris@82 183 /*
Chris@82 184 * This function contains 74 FP additions, 36 FP multiplications,
Chris@82 185 * (or, 50 additions, 12 multiplications, 24 fused multiply/add),
Chris@82 186 * 33 stack variables, 6 constants, and 28 memory accesses
Chris@82 187 */
Chris@82 188 #include "dft/simd/n1f.h"
Chris@82 189
Chris@82 190 static void n1fv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 191 {
Chris@82 192 DVK(KP222520933, +0.222520933956314404288902564496794759466355569);
Chris@82 193 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@82 194 DVK(KP623489801, +0.623489801858733530525004884004239810632274731);
Chris@82 195 DVK(KP433883739, +0.433883739117558120475768332848358754609990728);
Chris@82 196 DVK(KP781831482, +0.781831482468029808708444526674057750232334519);
Chris@82 197 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@82 198 {
Chris@82 199 INT i;
Chris@82 200 const R *xi;
Chris@82 201 R *xo;
Chris@82 202 xi = ri;
Chris@82 203 xo = ro;
Chris@82 204 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
Chris@82 205 V T3, Ty, To, TK, Tr, TE, Ta, TJ, Tq, TB, Th, TL, Ts, TH, T1;
Chris@82 206 V T2;
Chris@82 207 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 208 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@82 209 T3 = VSUB(T1, T2);
Chris@82 210 Ty = VADD(T1, T2);
Chris@82 211 {
Chris@82 212 V Tk, TC, Tn, TD;
Chris@82 213 {
Chris@82 214 V Ti, Tj, Tl, Tm;
Chris@82 215 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82 216 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
Chris@82 217 Tk = VSUB(Ti, Tj);
Chris@82 218 TC = VADD(Ti, Tj);
Chris@82 219 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@82 220 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 221 Tn = VSUB(Tl, Tm);
Chris@82 222 TD = VADD(Tl, Tm);
Chris@82 223 }
Chris@82 224 To = VADD(Tk, Tn);
Chris@82 225 TK = VSUB(TC, TD);
Chris@82 226 Tr = VSUB(Tn, Tk);
Chris@82 227 TE = VADD(TC, TD);
Chris@82 228 }
Chris@82 229 {
Chris@82 230 V T6, Tz, T9, TA;
Chris@82 231 {
Chris@82 232 V T4, T5, T7, T8;
Chris@82 233 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 234 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
Chris@82 235 T6 = VSUB(T4, T5);
Chris@82 236 Tz = VADD(T4, T5);
Chris@82 237 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
Chris@82 238 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82 239 T9 = VSUB(T7, T8);
Chris@82 240 TA = VADD(T7, T8);
Chris@82 241 }
Chris@82 242 Ta = VADD(T6, T9);
Chris@82 243 TJ = VSUB(TA, Tz);
Chris@82 244 Tq = VSUB(T9, T6);
Chris@82 245 TB = VADD(Tz, TA);
Chris@82 246 }
Chris@82 247 {
Chris@82 248 V Td, TF, Tg, TG;
Chris@82 249 {
Chris@82 250 V Tb, Tc, Te, Tf;
Chris@82 251 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82 252 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
Chris@82 253 Td = VSUB(Tb, Tc);
Chris@82 254 TF = VADD(Tb, Tc);
Chris@82 255 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
Chris@82 256 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 257 Tg = VSUB(Te, Tf);
Chris@82 258 TG = VADD(Te, Tf);
Chris@82 259 }
Chris@82 260 Th = VADD(Td, Tg);
Chris@82 261 TL = VSUB(TF, TG);
Chris@82 262 Ts = VSUB(Tg, Td);
Chris@82 263 TH = VADD(TF, TG);
Chris@82 264 }
Chris@82 265 ST(&(xo[WS(os, 7)]), VADD(T3, VADD(Ta, VADD(Th, To))), ovs, &(xo[WS(os, 1)]));
Chris@82 266 ST(&(xo[0]), VADD(Ty, VADD(TB, VADD(TH, TE))), ovs, &(xo[0]));
Chris@82 267 {
Chris@82 268 V Tt, Tp, TP, TQ;
Chris@82 269 Tt = VBYI(VFNMS(LDK(KP781831482), Tr, VFNMS(LDK(KP433883739), Ts, VMUL(LDK(KP974927912), Tq))));
Chris@82 270 Tp = VFMA(LDK(KP623489801), To, VFNMS(LDK(KP900968867), Th, VFNMS(LDK(KP222520933), Ta, T3)));
Chris@82 271 ST(&(xo[WS(os, 5)]), VSUB(Tp, Tt), ovs, &(xo[WS(os, 1)]));
Chris@82 272 ST(&(xo[WS(os, 9)]), VADD(Tp, Tt), ovs, &(xo[WS(os, 1)]));
Chris@82 273 TP = VBYI(VFMA(LDK(KP974927912), TJ, VFMA(LDK(KP433883739), TL, VMUL(LDK(KP781831482), TK))));
Chris@82 274 TQ = VFMA(LDK(KP623489801), TE, VFNMS(LDK(KP900968867), TH, VFNMS(LDK(KP222520933), TB, Ty)));
Chris@82 275 ST(&(xo[WS(os, 2)]), VADD(TP, TQ), ovs, &(xo[0]));
Chris@82 276 ST(&(xo[WS(os, 12)]), VSUB(TQ, TP), ovs, &(xo[0]));
Chris@82 277 }
Chris@82 278 {
Chris@82 279 V Tv, Tu, TM, TI;
Chris@82 280 Tv = VBYI(VFMA(LDK(KP781831482), Tq, VFMA(LDK(KP974927912), Ts, VMUL(LDK(KP433883739), Tr))));
Chris@82 281 Tu = VFMA(LDK(KP623489801), Ta, VFNMS(LDK(KP900968867), To, VFNMS(LDK(KP222520933), Th, T3)));
Chris@82 282 ST(&(xo[WS(os, 13)]), VSUB(Tu, Tv), ovs, &(xo[WS(os, 1)]));
Chris@82 283 ST(&(xo[WS(os, 1)]), VADD(Tu, Tv), ovs, &(xo[WS(os, 1)]));
Chris@82 284 TM = VBYI(VFNMS(LDK(KP433883739), TK, VFNMS(LDK(KP974927912), TL, VMUL(LDK(KP781831482), TJ))));
Chris@82 285 TI = VFMA(LDK(KP623489801), TB, VFNMS(LDK(KP900968867), TE, VFNMS(LDK(KP222520933), TH, Ty)));
Chris@82 286 ST(&(xo[WS(os, 6)]), VSUB(TI, TM), ovs, &(xo[0]));
Chris@82 287 ST(&(xo[WS(os, 8)]), VADD(TM, TI), ovs, &(xo[0]));
Chris@82 288 }
Chris@82 289 {
Chris@82 290 V TO, TN, Tx, Tw;
Chris@82 291 TO = VBYI(VFMA(LDK(KP433883739), TJ, VFNMS(LDK(KP974927912), TK, VMUL(LDK(KP781831482), TL))));
Chris@82 292 TN = VFMA(LDK(KP623489801), TH, VFNMS(LDK(KP222520933), TE, VFNMS(LDK(KP900968867), TB, Ty)));
Chris@82 293 ST(&(xo[WS(os, 4)]), VSUB(TN, TO), ovs, &(xo[0]));
Chris@82 294 ST(&(xo[WS(os, 10)]), VADD(TO, TN), ovs, &(xo[0]));
Chris@82 295 Tx = VBYI(VFMA(LDK(KP433883739), Tq, VFNMS(LDK(KP781831482), Ts, VMUL(LDK(KP974927912), Tr))));
Chris@82 296 Tw = VFMA(LDK(KP623489801), Th, VFNMS(LDK(KP222520933), To, VFNMS(LDK(KP900968867), Ta, T3)));
Chris@82 297 ST(&(xo[WS(os, 11)]), VSUB(Tw, Tx), ovs, &(xo[WS(os, 1)]));
Chris@82 298 ST(&(xo[WS(os, 3)]), VADD(Tw, Tx), ovs, &(xo[WS(os, 1)]));
Chris@82 299 }
Chris@82 300 }
Chris@82 301 }
Chris@82 302 VLEAVE();
Chris@82 303 }
Chris@82 304
Chris@82 305 static const kdft_desc desc = { 14, XSIMD_STRING("n1fv_14"), {50, 12, 24, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 306
Chris@82 307 void XSIMD(codelet_n1fv_14) (planner *p) {
Chris@82 308 X(kdft_register) (p, n1fv_14, &desc);
Chris@82 309 }
Chris@82 310
Chris@82 311 #endif