Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/n1bv_14.c @ 79:91c729825bca pa_catalina

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Update build for AUDIO_COMPONENT_FIX
author Chris Cannam
date Wed, 30 Oct 2019 12:40:34 +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 16:39:15 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_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 14 -name n1bv_14 -include n1b.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 74 FP additions, 48 FP multiplications,
Chris@42 32 * (or, 32 additions, 6 multiplications, 42 fused multiply/add),
Chris@42 33 * 63 stack variables, 6 constants, and 28 memory accesses
Chris@42 34 */
Chris@42 35 #include "n1b.h"
Chris@42 36
Chris@42 37 static void n1bv_14(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 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@42 40 DVK(KP801937735, +0.801937735804838252472204639014890102331838324);
Chris@42 41 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@42 42 DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
Chris@42 43 DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
Chris@42 44 DVK(KP356895867, +0.356895867892209443894399510021300583399127187);
Chris@42 45 {
Chris@42 46 INT i;
Chris@42 47 const R *xi;
Chris@42 48 R *xo;
Chris@42 49 xi = ii;
Chris@42 50 xo = io;
Chris@42 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@42 52 V TH, T3, TP, Tn, Ta, Tu, TU, TK, TO, Tk, TM, Tg, TL, Td, T1;
Chris@42 53 V T2;
Chris@42 54 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@42 55 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@42 56 {
Chris@42 57 V Ti, TI, T6, TJ, T9, Tj, Te, Tf, Tb, Tc;
Chris@42 58 {
Chris@42 59 V T4, T5, T7, T8, Tl, Tm;
Chris@42 60 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@42 61 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
Chris@42 62 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
Chris@42 63 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@42 64 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@42 65 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@42 66 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@42 67 TH = VADD(T1, T2);
Chris@42 68 T3 = VSUB(T1, T2);
Chris@42 69 TI = VADD(T4, T5);
Chris@42 70 T6 = VSUB(T4, T5);
Chris@42 71 TJ = VADD(T7, T8);
Chris@42 72 T9 = VSUB(T7, T8);
Chris@42 73 TP = VADD(Tl, Tm);
Chris@42 74 Tn = VSUB(Tl, Tm);
Chris@42 75 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
Chris@42 76 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
Chris@42 77 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@42 78 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@42 79 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
Chris@42 80 }
Chris@42 81 Ta = VADD(T6, T9);
Chris@42 82 Tu = VSUB(T6, T9);
Chris@42 83 TU = VSUB(TI, TJ);
Chris@42 84 TK = VADD(TI, TJ);
Chris@42 85 TO = VADD(Ti, Tj);
Chris@42 86 Tk = VSUB(Ti, Tj);
Chris@42 87 TM = VADD(Te, Tf);
Chris@42 88 Tg = VSUB(Te, Tf);
Chris@42 89 TL = VADD(Tb, Tc);
Chris@42 90 Td = VSUB(Tb, Tc);
Chris@42 91 }
Chris@42 92 {
Chris@42 93 V T13, TG, TY, T18, TB, Tw, TT, Tz, T11, T16, TE, Tr, TV, TQ;
Chris@42 94 TV = VSUB(TP, TO);
Chris@42 95 TQ = VADD(TO, TP);
Chris@42 96 {
Chris@42 97 V Ts, To, TW, TN;
Chris@42 98 Ts = VSUB(Tk, Tn);
Chris@42 99 To = VADD(Tk, Tn);
Chris@42 100 TW = VSUB(TM, TL);
Chris@42 101 TN = VADD(TL, TM);
Chris@42 102 {
Chris@42 103 V Tt, Th, TR, T12;
Chris@42 104 Tt = VSUB(Td, Tg);
Chris@42 105 Th = VADD(Td, Tg);
Chris@42 106 TR = VFNMS(LDK(KP356895867), TK, TQ);
Chris@42 107 T12 = VFNMS(LDK(KP554958132), TV, TU);
Chris@42 108 {
Chris@42 109 V Tx, TF, TZ, T14;
Chris@42 110 Tx = VFNMS(LDK(KP356895867), Ta, To);
Chris@42 111 TF = VFMA(LDK(KP554958132), Ts, Tu);
Chris@42 112 ST(&(xo[0]), VADD(TH, VADD(TK, VADD(TN, TQ))), ovs, &(xo[0]));
Chris@42 113 TZ = VFNMS(LDK(KP356895867), TN, TK);
Chris@42 114 T14 = VFNMS(LDK(KP356895867), TQ, TN);
Chris@42 115 {
Chris@42 116 V TX, T17, TC, Tp;
Chris@42 117 TX = VFMA(LDK(KP554958132), TW, TV);
Chris@42 118 T17 = VFMA(LDK(KP554958132), TU, TW);
Chris@42 119 ST(&(xo[WS(os, 7)]), VADD(T3, VADD(Ta, VADD(Th, To))), ovs, &(xo[WS(os, 1)]));
Chris@42 120 TC = VFNMS(LDK(KP356895867), Th, Ta);
Chris@42 121 Tp = VFNMS(LDK(KP356895867), To, Th);
Chris@42 122 {
Chris@42 123 V TA, Tv, TS, Ty;
Chris@42 124 TA = VFMA(LDK(KP554958132), Tt, Ts);
Chris@42 125 Tv = VFNMS(LDK(KP554958132), Tu, Tt);
Chris@42 126 TS = VFNMS(LDK(KP692021471), TR, TN);
Chris@42 127 T13 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T12, TW));
Chris@42 128 Ty = VFNMS(LDK(KP692021471), Tx, Th);
Chris@42 129 TG = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TF, Tt));
Chris@42 130 {
Chris@42 131 V T10, T15, TD, Tq;
Chris@42 132 T10 = VFNMS(LDK(KP692021471), TZ, TQ);
Chris@42 133 T15 = VFNMS(LDK(KP692021471), T14, TK);
Chris@42 134 TY = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TX, TU));
Chris@42 135 T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV));
Chris@42 136 TD = VFNMS(LDK(KP692021471), TC, To);
Chris@42 137 Tq = VFNMS(LDK(KP692021471), Tp, Ta);
Chris@42 138 TB = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TA, Tu));
Chris@42 139 Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts));
Chris@42 140 TT = VFNMS(LDK(KP900968867), TS, TH);
Chris@42 141 Tz = VFNMS(LDK(KP900968867), Ty, T3);
Chris@42 142 T11 = VFNMS(LDK(KP900968867), T10, TH);
Chris@42 143 T16 = VFNMS(LDK(KP900968867), T15, TH);
Chris@42 144 TE = VFNMS(LDK(KP900968867), TD, T3);
Chris@42 145 Tr = VFNMS(LDK(KP900968867), Tq, T3);
Chris@42 146 }
Chris@42 147 }
Chris@42 148 }
Chris@42 149 }
Chris@42 150 }
Chris@42 151 }
Chris@42 152 ST(&(xo[WS(os, 2)]), VFMAI(TY, TT), ovs, &(xo[0]));
Chris@42 153 ST(&(xo[WS(os, 12)]), VFNMSI(TY, TT), ovs, &(xo[0]));
Chris@42 154 ST(&(xo[WS(os, 9)]), VFMAI(TB, Tz), ovs, &(xo[WS(os, 1)]));
Chris@42 155 ST(&(xo[WS(os, 5)]), VFNMSI(TB, Tz), ovs, &(xo[WS(os, 1)]));
Chris@42 156 ST(&(xo[WS(os, 6)]), VFMAI(T13, T11), ovs, &(xo[0]));
Chris@42 157 ST(&(xo[WS(os, 8)]), VFNMSI(T13, T11), ovs, &(xo[0]));
Chris@42 158 ST(&(xo[WS(os, 4)]), VFMAI(T18, T16), ovs, &(xo[0]));
Chris@42 159 ST(&(xo[WS(os, 10)]), VFNMSI(T18, T16), ovs, &(xo[0]));
Chris@42 160 ST(&(xo[WS(os, 13)]), VFNMSI(TG, TE), ovs, &(xo[WS(os, 1)]));
Chris@42 161 ST(&(xo[WS(os, 1)]), VFMAI(TG, TE), ovs, &(xo[WS(os, 1)]));
Chris@42 162 ST(&(xo[WS(os, 11)]), VFNMSI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
Chris@42 163 ST(&(xo[WS(os, 3)]), VFMAI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
Chris@42 164 }
Chris@42 165 }
Chris@42 166 }
Chris@42 167 VLEAVE();
Chris@42 168 }
Chris@42 169
Chris@42 170 static const kdft_desc desc = { 14, XSIMD_STRING("n1bv_14"), {32, 6, 42, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 171
Chris@42 172 void XSIMD(codelet_n1bv_14) (planner *p) {
Chris@42 173 X(kdft_register) (p, n1bv_14, &desc);
Chris@42 174 }
Chris@42 175
Chris@42 176 #else /* HAVE_FMA */
Chris@42 177
Chris@42 178 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 14 -name n1bv_14 -include n1b.h */
Chris@42 179
Chris@42 180 /*
Chris@42 181 * This function contains 74 FP additions, 36 FP multiplications,
Chris@42 182 * (or, 50 additions, 12 multiplications, 24 fused multiply/add),
Chris@42 183 * 33 stack variables, 6 constants, and 28 memory accesses
Chris@42 184 */
Chris@42 185 #include "n1b.h"
Chris@42 186
Chris@42 187 static void n1bv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 188 {
Chris@42 189 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
Chris@42 190 DVK(KP222520933, +0.222520933956314404288902564496794759466355569);
Chris@42 191 DVK(KP623489801, +0.623489801858733530525004884004239810632274731);
Chris@42 192 DVK(KP781831482, +0.781831482468029808708444526674057750232334519);
Chris@42 193 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
Chris@42 194 DVK(KP433883739, +0.433883739117558120475768332848358754609990728);
Chris@42 195 {
Chris@42 196 INT i;
Chris@42 197 const R *xi;
Chris@42 198 R *xo;
Chris@42 199 xi = ii;
Chris@42 200 xo = io;
Chris@42 201 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@42 202 V Tp, Ty, Tl, TL, Tq, TE, T7, TJ, Ts, TB, Te, TK, Tr, TH, Tn;
Chris@42 203 V To;
Chris@42 204 Tn = LD(&(xi[0]), ivs, &(xi[0]));
Chris@42 205 To = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@42 206 Tp = VSUB(Tn, To);
Chris@42 207 Ty = VADD(Tn, To);
Chris@42 208 {
Chris@42 209 V Th, TC, Tk, TD;
Chris@42 210 {
Chris@42 211 V Tf, Tg, Ti, Tj;
Chris@42 212 Tf = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@42 213 Tg = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
Chris@42 214 Th = VSUB(Tf, Tg);
Chris@42 215 TC = VADD(Tf, Tg);
Chris@42 216 Ti = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
Chris@42 217 Tj = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@42 218 Tk = VSUB(Ti, Tj);
Chris@42 219 TD = VADD(Ti, Tj);
Chris@42 220 }
Chris@42 221 Tl = VSUB(Th, Tk);
Chris@42 222 TL = VSUB(TD, TC);
Chris@42 223 Tq = VADD(Th, Tk);
Chris@42 224 TE = VADD(TC, TD);
Chris@42 225 }
Chris@42 226 {
Chris@42 227 V T3, Tz, T6, TA;
Chris@42 228 {
Chris@42 229 V T1, T2, T4, T5;
Chris@42 230 T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@42 231 T2 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
Chris@42 232 T3 = VSUB(T1, T2);
Chris@42 233 Tz = VADD(T1, T2);
Chris@42 234 T4 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
Chris@42 235 T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@42 236 T6 = VSUB(T4, T5);
Chris@42 237 TA = VADD(T4, T5);
Chris@42 238 }
Chris@42 239 T7 = VSUB(T3, T6);
Chris@42 240 TJ = VSUB(Tz, TA);
Chris@42 241 Ts = VADD(T3, T6);
Chris@42 242 TB = VADD(Tz, TA);
Chris@42 243 }
Chris@42 244 {
Chris@42 245 V Ta, TF, Td, TG;
Chris@42 246 {
Chris@42 247 V T8, T9, Tb, Tc;
Chris@42 248 T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@42 249 T9 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
Chris@42 250 Ta = VSUB(T8, T9);
Chris@42 251 TF = VADD(T8, T9);
Chris@42 252 Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@42 253 Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@42 254 Td = VSUB(Tb, Tc);
Chris@42 255 TG = VADD(Tb, Tc);
Chris@42 256 }
Chris@42 257 Te = VSUB(Ta, Td);
Chris@42 258 TK = VSUB(TG, TF);
Chris@42 259 Tr = VADD(Ta, Td);
Chris@42 260 TH = VADD(TF, TG);
Chris@42 261 }
Chris@42 262 ST(&(xo[WS(os, 7)]), VADD(Tp, VADD(Ts, VADD(Tq, Tr))), ovs, &(xo[WS(os, 1)]));
Chris@42 263 ST(&(xo[0]), VADD(Ty, VADD(TB, VADD(TE, TH))), ovs, &(xo[0]));
Chris@42 264 {
Chris@42 265 V Tm, Tt, TQ, TP;
Chris@42 266 Tm = VBYI(VFMA(LDK(KP433883739), T7, VFNMS(LDK(KP781831482), Tl, VMUL(LDK(KP974927912), Te))));
Chris@42 267 Tt = VFMA(LDK(KP623489801), Tq, VFNMS(LDK(KP222520933), Tr, VFNMS(LDK(KP900968867), Ts, Tp)));
Chris@42 268 ST(&(xo[WS(os, 3)]), VADD(Tm, Tt), ovs, &(xo[WS(os, 1)]));
Chris@42 269 ST(&(xo[WS(os, 11)]), VSUB(Tt, Tm), ovs, &(xo[WS(os, 1)]));
Chris@42 270 TQ = VBYI(VFMA(LDK(KP974927912), TJ, VFMA(LDK(KP433883739), TL, VMUL(LDK(KP781831482), TK))));
Chris@42 271 TP = VFMA(LDK(KP623489801), TH, VFNMS(LDK(KP900968867), TE, VFNMS(LDK(KP222520933), TB, Ty)));
Chris@42 272 ST(&(xo[WS(os, 12)]), VSUB(TP, TQ), ovs, &(xo[0]));
Chris@42 273 ST(&(xo[WS(os, 2)]), VADD(TP, TQ), ovs, &(xo[0]));
Chris@42 274 }
Chris@42 275 {
Chris@42 276 V Tu, Tv, TM, TI;
Chris@42 277 Tu = VBYI(VFMA(LDK(KP781831482), T7, VFMA(LDK(KP974927912), Tl, VMUL(LDK(KP433883739), Te))));
Chris@42 278 Tv = VFMA(LDK(KP623489801), Ts, VFNMS(LDK(KP900968867), Tr, VFNMS(LDK(KP222520933), Tq, Tp)));
Chris@42 279 ST(&(xo[WS(os, 1)]), VADD(Tu, Tv), ovs, &(xo[WS(os, 1)]));
Chris@42 280 ST(&(xo[WS(os, 13)]), VSUB(Tv, Tu), ovs, &(xo[WS(os, 1)]));
Chris@42 281 TM = VBYI(VFNMS(LDK(KP433883739), TK, VFNMS(LDK(KP974927912), TL, VMUL(LDK(KP781831482), TJ))));
Chris@42 282 TI = VFMA(LDK(KP623489801), TB, VFNMS(LDK(KP900968867), TH, VFNMS(LDK(KP222520933), TE, Ty)));
Chris@42 283 ST(&(xo[WS(os, 6)]), VSUB(TI, TM), ovs, &(xo[0]));
Chris@42 284 ST(&(xo[WS(os, 8)]), VADD(TI, TM), ovs, &(xo[0]));
Chris@42 285 }
Chris@42 286 {
Chris@42 287 V TO, TN, Tx, Tw;
Chris@42 288 TO = VBYI(VFMA(LDK(KP433883739), TJ, VFNMS(LDK(KP974927912), TK, VMUL(LDK(KP781831482), TL))));
Chris@42 289 TN = VFMA(LDK(KP623489801), TE, VFNMS(LDK(KP222520933), TH, VFNMS(LDK(KP900968867), TB, Ty)));
Chris@42 290 ST(&(xo[WS(os, 4)]), VSUB(TN, TO), ovs, &(xo[0]));
Chris@42 291 ST(&(xo[WS(os, 10)]), VADD(TN, TO), ovs, &(xo[0]));
Chris@42 292 Tx = VBYI(VFNMS(LDK(KP781831482), Te, VFNMS(LDK(KP433883739), Tl, VMUL(LDK(KP974927912), T7))));
Chris@42 293 Tw = VFMA(LDK(KP623489801), Tr, VFNMS(LDK(KP900968867), Tq, VFNMS(LDK(KP222520933), Ts, Tp)));
Chris@42 294 ST(&(xo[WS(os, 5)]), VSUB(Tw, Tx), ovs, &(xo[WS(os, 1)]));
Chris@42 295 ST(&(xo[WS(os, 9)]), VADD(Tx, Tw), ovs, &(xo[WS(os, 1)]));
Chris@42 296 }
Chris@42 297 }
Chris@42 298 }
Chris@42 299 VLEAVE();
Chris@42 300 }
Chris@42 301
Chris@42 302 static const kdft_desc desc = { 14, XSIMD_STRING("n1bv_14"), {50, 12, 24, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 303
Chris@42 304 void XSIMD(codelet_n1bv_14) (planner *p) {
Chris@42 305 X(kdft_register) (p, n1bv_14, &desc);
Chris@42 306 }
Chris@42 307
Chris@42 308 #endif /* HAVE_FMA */