Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/t1fv_15.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 08:05:28 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_twiddle_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 15 -name t1fv_15 -include dft/simd/t1f.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 92 FP additions, 77 FP multiplications,
Chris@82 32 * (or, 50 additions, 35 multiplications, 42 fused multiply/add),
Chris@82 33 * 50 stack variables, 8 constants, and 30 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/t1f.h"
Chris@82 36
Chris@82 37 static void t1fv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 38 {
Chris@82 39 DVK(KP910592997, +0.910592997310029334643087372129977886038870291);
Chris@82 40 DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
Chris@82 41 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 42 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 43 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82 44 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 45 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 46 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 47 {
Chris@82 48 INT m;
Chris@82 49 R *x;
Chris@82 50 x = ri;
Chris@82 51 for (m = mb, W = W + (mb * ((TWVL / VL) * 28)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 28), MAKE_VOLATILE_STRIDE(15, rs)) {
Chris@82 52 V T1b, T7, TP, T12, T15, Tf, Tn, To, T1c, T1d, T1e, TQ, TR, TS, Tw;
Chris@82 53 V TE, TF, TT, TU, TV;
Chris@82 54 {
Chris@82 55 V T1, T5, T3, T4, T2, T6;
Chris@82 56 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@82 57 T4 = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@82 58 T5 = BYTWJ(&(W[TWVL * 18]), T4);
Chris@82 59 T2 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@82 60 T3 = BYTWJ(&(W[TWVL * 8]), T2);
Chris@82 61 T1b = VSUB(T5, T3);
Chris@82 62 T6 = VADD(T3, T5);
Chris@82 63 T7 = VADD(T1, T6);
Chris@82 64 TP = VFNMS(LDK(KP500000000), T6, T1);
Chris@82 65 }
Chris@82 66 {
Chris@82 67 V T9, Tq, Ty, Th, Te, T10, Tv, T13, TD, T14, Tm, T11;
Chris@82 68 {
Chris@82 69 V T8, Tp, Tx, Tg;
Chris@82 70 T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@82 71 T9 = BYTWJ(&(W[TWVL * 4]), T8);
Chris@82 72 Tp = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@82 73 Tq = BYTWJ(&(W[TWVL * 10]), Tp);
Chris@82 74 Tx = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@82 75 Ty = BYTWJ(&(W[TWVL * 16]), Tx);
Chris@82 76 Tg = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@82 77 Th = BYTWJ(&(W[TWVL * 22]), Tg);
Chris@82 78 }
Chris@82 79 {
Chris@82 80 V Tb, Td, Ta, Tc;
Chris@82 81 Ta = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@82 82 Tb = BYTWJ(&(W[TWVL * 14]), Ta);
Chris@82 83 Tc = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@82 84 Td = BYTWJ(&(W[TWVL * 24]), Tc);
Chris@82 85 Te = VADD(Tb, Td);
Chris@82 86 T10 = VSUB(Td, Tb);
Chris@82 87 }
Chris@82 88 {
Chris@82 89 V Ts, Tu, Tr, Tt;
Chris@82 90 Tr = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@82 91 Ts = BYTWJ(&(W[TWVL * 20]), Tr);
Chris@82 92 Tt = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82 93 Tu = BYTWJ(&(W[0]), Tt);
Chris@82 94 Tv = VADD(Ts, Tu);
Chris@82 95 T13 = VSUB(Tu, Ts);
Chris@82 96 }
Chris@82 97 {
Chris@82 98 V TA, TC, Tz, TB;
Chris@82 99 Tz = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@82 100 TA = BYTWJ(&(W[TWVL * 26]), Tz);
Chris@82 101 TB = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@82 102 TC = BYTWJ(&(W[TWVL * 6]), TB);
Chris@82 103 TD = VADD(TA, TC);
Chris@82 104 T14 = VSUB(TC, TA);
Chris@82 105 }
Chris@82 106 {
Chris@82 107 V Tj, Tl, Ti, Tk;
Chris@82 108 Ti = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@82 109 Tj = BYTWJ(&(W[TWVL * 2]), Ti);
Chris@82 110 Tk = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@82 111 Tl = BYTWJ(&(W[TWVL * 12]), Tk);
Chris@82 112 Tm = VADD(Tj, Tl);
Chris@82 113 T11 = VSUB(Tl, Tj);
Chris@82 114 }
Chris@82 115 T12 = VSUB(T10, T11);
Chris@82 116 T15 = VSUB(T13, T14);
Chris@82 117 Tf = VADD(T9, Te);
Chris@82 118 Tn = VADD(Th, Tm);
Chris@82 119 To = VADD(Tf, Tn);
Chris@82 120 T1c = VADD(T10, T11);
Chris@82 121 T1d = VADD(T13, T14);
Chris@82 122 T1e = VADD(T1c, T1d);
Chris@82 123 TQ = VFNMS(LDK(KP500000000), Te, T9);
Chris@82 124 TR = VFNMS(LDK(KP500000000), Tm, Th);
Chris@82 125 TS = VADD(TQ, TR);
Chris@82 126 Tw = VADD(Tq, Tv);
Chris@82 127 TE = VADD(Ty, TD);
Chris@82 128 TF = VADD(Tw, TE);
Chris@82 129 TT = VFNMS(LDK(KP500000000), Tv, Tq);
Chris@82 130 TU = VFNMS(LDK(KP500000000), TD, Ty);
Chris@82 131 TV = VADD(TT, TU);
Chris@82 132 }
Chris@82 133 {
Chris@82 134 V TI, TG, TH, TM, TO, TK, TL, TN, TJ;
Chris@82 135 TI = VSUB(To, TF);
Chris@82 136 TG = VADD(To, TF);
Chris@82 137 TH = VFNMS(LDK(KP250000000), TG, T7);
Chris@82 138 TK = VSUB(Tw, TE);
Chris@82 139 TL = VSUB(Tf, Tn);
Chris@82 140 TM = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TL, TK));
Chris@82 141 TO = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TK, TL));
Chris@82 142 ST(&(x[0]), VADD(T7, TG), ms, &(x[0]));
Chris@82 143 TN = VFMA(LDK(KP559016994), TI, TH);
Chris@82 144 ST(&(x[WS(rs, 6)]), VFNMSI(TO, TN), ms, &(x[0]));
Chris@82 145 ST(&(x[WS(rs, 9)]), VFMAI(TO, TN), ms, &(x[WS(rs, 1)]));
Chris@82 146 TJ = VFNMS(LDK(KP559016994), TI, TH);
Chris@82 147 ST(&(x[WS(rs, 3)]), VFNMSI(TM, TJ), ms, &(x[WS(rs, 1)]));
Chris@82 148 ST(&(x[WS(rs, 12)]), VFMAI(TM, TJ), ms, &(x[0]));
Chris@82 149 }
Chris@82 150 {
Chris@82 151 V T16, T1m, T1u, T1h, T1p, T1a, T1o, TZ, T1t, T1l, T1f, T1g;
Chris@82 152 T16 = VFMA(LDK(KP618033988), T15, T12);
Chris@82 153 T1m = VFNMS(LDK(KP618033988), T12, T15);
Chris@82 154 T1u = VMUL(LDK(KP866025403), VADD(T1b, T1e));
Chris@82 155 T1f = VFNMS(LDK(KP250000000), T1e, T1b);
Chris@82 156 T1g = VSUB(T1c, T1d);
Chris@82 157 T1h = VFMA(LDK(KP559016994), T1g, T1f);
Chris@82 158 T1p = VFNMS(LDK(KP559016994), T1g, T1f);
Chris@82 159 {
Chris@82 160 V T18, T19, TY, TW, TX;
Chris@82 161 T18 = VSUB(TQ, TR);
Chris@82 162 T19 = VSUB(TT, TU);
Chris@82 163 T1a = VFMA(LDK(KP618033988), T19, T18);
Chris@82 164 T1o = VFNMS(LDK(KP618033988), T18, T19);
Chris@82 165 TY = VSUB(TS, TV);
Chris@82 166 TW = VADD(TS, TV);
Chris@82 167 TX = VFNMS(LDK(KP250000000), TW, TP);
Chris@82 168 TZ = VFMA(LDK(KP559016994), TY, TX);
Chris@82 169 T1t = VADD(TP, TW);
Chris@82 170 T1l = VFNMS(LDK(KP559016994), TY, TX);
Chris@82 171 }
Chris@82 172 {
Chris@82 173 V T17, T1i, T1r, T1s;
Chris@82 174 ST(&(x[WS(rs, 5)]), VFNMSI(T1u, T1t), ms, &(x[WS(rs, 1)]));
Chris@82 175 ST(&(x[WS(rs, 10)]), VFMAI(T1u, T1t), ms, &(x[0]));
Chris@82 176 T17 = VFMA(LDK(KP823639103), T16, TZ);
Chris@82 177 T1i = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), T1h, T1a));
Chris@82 178 ST(&(x[WS(rs, 1)]), VFNMSI(T1i, T17), ms, &(x[WS(rs, 1)]));
Chris@82 179 ST(&(x[WS(rs, 14)]), VFMAI(T1i, T17), ms, &(x[0]));
Chris@82 180 T1r = VFNMS(LDK(KP823639103), T1m, T1l);
Chris@82 181 T1s = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), T1p, T1o));
Chris@82 182 ST(&(x[WS(rs, 8)]), VFNMSI(T1s, T1r), ms, &(x[0]));
Chris@82 183 ST(&(x[WS(rs, 7)]), VFMAI(T1s, T1r), ms, &(x[WS(rs, 1)]));
Chris@82 184 {
Chris@82 185 V T1n, T1q, T1j, T1k;
Chris@82 186 T1n = VFMA(LDK(KP823639103), T1m, T1l);
Chris@82 187 T1q = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), T1p, T1o));
Chris@82 188 ST(&(x[WS(rs, 13)]), VFNMSI(T1q, T1n), ms, &(x[WS(rs, 1)]));
Chris@82 189 ST(&(x[WS(rs, 2)]), VFMAI(T1q, T1n), ms, &(x[0]));
Chris@82 190 T1j = VFNMS(LDK(KP823639103), T16, TZ);
Chris@82 191 T1k = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), T1h, T1a));
Chris@82 192 ST(&(x[WS(rs, 11)]), VFNMSI(T1k, T1j), ms, &(x[WS(rs, 1)]));
Chris@82 193 ST(&(x[WS(rs, 4)]), VFMAI(T1k, T1j), ms, &(x[0]));
Chris@82 194 }
Chris@82 195 }
Chris@82 196 }
Chris@82 197 }
Chris@82 198 }
Chris@82 199 VLEAVE();
Chris@82 200 }
Chris@82 201
Chris@82 202 static const tw_instr twinstr[] = {
Chris@82 203 VTW(0, 1),
Chris@82 204 VTW(0, 2),
Chris@82 205 VTW(0, 3),
Chris@82 206 VTW(0, 4),
Chris@82 207 VTW(0, 5),
Chris@82 208 VTW(0, 6),
Chris@82 209 VTW(0, 7),
Chris@82 210 VTW(0, 8),
Chris@82 211 VTW(0, 9),
Chris@82 212 VTW(0, 10),
Chris@82 213 VTW(0, 11),
Chris@82 214 VTW(0, 12),
Chris@82 215 VTW(0, 13),
Chris@82 216 VTW(0, 14),
Chris@82 217 {TW_NEXT, VL, 0}
Chris@82 218 };
Chris@82 219
Chris@82 220 static const ct_desc desc = { 15, XSIMD_STRING("t1fv_15"), twinstr, &GENUS, {50, 35, 42, 0}, 0, 0, 0 };
Chris@82 221
Chris@82 222 void XSIMD(codelet_t1fv_15) (planner *p) {
Chris@82 223 X(kdft_dit_register) (p, t1fv_15, &desc);
Chris@82 224 }
Chris@82 225 #else
Chris@82 226
Chris@82 227 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 15 -name t1fv_15 -include dft/simd/t1f.h */
Chris@82 228
Chris@82 229 /*
Chris@82 230 * This function contains 92 FP additions, 53 FP multiplications,
Chris@82 231 * (or, 78 additions, 39 multiplications, 14 fused multiply/add),
Chris@82 232 * 52 stack variables, 10 constants, and 30 memory accesses
Chris@82 233 */
Chris@82 234 #include "dft/simd/t1f.h"
Chris@82 235
Chris@82 236 static void t1fv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 237 {
Chris@82 238 DVK(KP216506350, +0.216506350946109661690930792688234045867850657);
Chris@82 239 DVK(KP484122918, +0.484122918275927110647408174972799951354115213);
Chris@82 240 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 241 DVK(KP509036960, +0.509036960455127183450980863393907648510733164);
Chris@82 242 DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
Chris@82 243 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@82 244 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 245 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 246 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 247 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 248 {
Chris@82 249 INT m;
Chris@82 250 R *x;
Chris@82 251 x = ri;
Chris@82 252 for (m = mb, W = W + (mb * ((TWVL / VL) * 28)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 28), MAKE_VOLATILE_STRIDE(15, rs)) {
Chris@82 253 V T1e, T7, TP, T12, T15, Tf, Tn, To, T1b, T1c, T1f, TQ, TR, TS, Tw;
Chris@82 254 V TE, TF, TT, TU, TV;
Chris@82 255 {
Chris@82 256 V T1, T5, T3, T4, T2, T6;
Chris@82 257 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@82 258 T4 = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@82 259 T5 = BYTWJ(&(W[TWVL * 18]), T4);
Chris@82 260 T2 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@82 261 T3 = BYTWJ(&(W[TWVL * 8]), T2);
Chris@82 262 T1e = VSUB(T5, T3);
Chris@82 263 T6 = VADD(T3, T5);
Chris@82 264 T7 = VADD(T1, T6);
Chris@82 265 TP = VFNMS(LDK(KP500000000), T6, T1);
Chris@82 266 }
Chris@82 267 {
Chris@82 268 V T9, Tq, Ty, Th, Te, T13, Tv, T10, TD, T11, Tm, T14;
Chris@82 269 {
Chris@82 270 V T8, Tp, Tx, Tg;
Chris@82 271 T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@82 272 T9 = BYTWJ(&(W[TWVL * 4]), T8);
Chris@82 273 Tp = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@82 274 Tq = BYTWJ(&(W[TWVL * 10]), Tp);
Chris@82 275 Tx = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@82 276 Ty = BYTWJ(&(W[TWVL * 16]), Tx);
Chris@82 277 Tg = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@82 278 Th = BYTWJ(&(W[TWVL * 22]), Tg);
Chris@82 279 }
Chris@82 280 {
Chris@82 281 V Tb, Td, Ta, Tc;
Chris@82 282 Ta = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@82 283 Tb = BYTWJ(&(W[TWVL * 14]), Ta);
Chris@82 284 Tc = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@82 285 Td = BYTWJ(&(W[TWVL * 24]), Tc);
Chris@82 286 Te = VADD(Tb, Td);
Chris@82 287 T13 = VSUB(Td, Tb);
Chris@82 288 }
Chris@82 289 {
Chris@82 290 V Ts, Tu, Tr, Tt;
Chris@82 291 Tr = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@82 292 Ts = BYTWJ(&(W[TWVL * 20]), Tr);
Chris@82 293 Tt = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82 294 Tu = BYTWJ(&(W[0]), Tt);
Chris@82 295 Tv = VADD(Ts, Tu);
Chris@82 296 T10 = VSUB(Tu, Ts);
Chris@82 297 }
Chris@82 298 {
Chris@82 299 V TA, TC, Tz, TB;
Chris@82 300 Tz = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@82 301 TA = BYTWJ(&(W[TWVL * 26]), Tz);
Chris@82 302 TB = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@82 303 TC = BYTWJ(&(W[TWVL * 6]), TB);
Chris@82 304 TD = VADD(TA, TC);
Chris@82 305 T11 = VSUB(TC, TA);
Chris@82 306 }
Chris@82 307 {
Chris@82 308 V Tj, Tl, Ti, Tk;
Chris@82 309 Ti = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@82 310 Tj = BYTWJ(&(W[TWVL * 2]), Ti);
Chris@82 311 Tk = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@82 312 Tl = BYTWJ(&(W[TWVL * 12]), Tk);
Chris@82 313 Tm = VADD(Tj, Tl);
Chris@82 314 T14 = VSUB(Tl, Tj);
Chris@82 315 }
Chris@82 316 T12 = VSUB(T10, T11);
Chris@82 317 T15 = VSUB(T13, T14);
Chris@82 318 Tf = VADD(T9, Te);
Chris@82 319 Tn = VADD(Th, Tm);
Chris@82 320 To = VADD(Tf, Tn);
Chris@82 321 T1b = VADD(T13, T14);
Chris@82 322 T1c = VADD(T10, T11);
Chris@82 323 T1f = VADD(T1b, T1c);
Chris@82 324 TQ = VFNMS(LDK(KP500000000), Te, T9);
Chris@82 325 TR = VFNMS(LDK(KP500000000), Tm, Th);
Chris@82 326 TS = VADD(TQ, TR);
Chris@82 327 Tw = VADD(Tq, Tv);
Chris@82 328 TE = VADD(Ty, TD);
Chris@82 329 TF = VADD(Tw, TE);
Chris@82 330 TT = VFNMS(LDK(KP500000000), Tv, Tq);
Chris@82 331 TU = VFNMS(LDK(KP500000000), TD, Ty);
Chris@82 332 TV = VADD(TT, TU);
Chris@82 333 }
Chris@82 334 {
Chris@82 335 V TI, TG, TH, TM, TO, TK, TL, TN, TJ;
Chris@82 336 TI = VMUL(LDK(KP559016994), VSUB(To, TF));
Chris@82 337 TG = VADD(To, TF);
Chris@82 338 TH = VFNMS(LDK(KP250000000), TG, T7);
Chris@82 339 TK = VSUB(Tw, TE);
Chris@82 340 TL = VSUB(Tf, Tn);
Chris@82 341 TM = VBYI(VFNMS(LDK(KP587785252), TL, VMUL(LDK(KP951056516), TK)));
Chris@82 342 TO = VBYI(VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TK)));
Chris@82 343 ST(&(x[0]), VADD(T7, TG), ms, &(x[0]));
Chris@82 344 TN = VADD(TI, TH);
Chris@82 345 ST(&(x[WS(rs, 6)]), VSUB(TN, TO), ms, &(x[0]));
Chris@82 346 ST(&(x[WS(rs, 9)]), VADD(TO, TN), ms, &(x[WS(rs, 1)]));
Chris@82 347 TJ = VSUB(TH, TI);
Chris@82 348 ST(&(x[WS(rs, 3)]), VSUB(TJ, TM), ms, &(x[WS(rs, 1)]));
Chris@82 349 ST(&(x[WS(rs, 12)]), VADD(TM, TJ), ms, &(x[0]));
Chris@82 350 }
Chris@82 351 {
Chris@82 352 V T16, T1m, T1u, T1h, T1o, T1a, T1p, TZ, T1t, T1l, T1d, T1g;
Chris@82 353 T16 = VFNMS(LDK(KP509036960), T15, VMUL(LDK(KP823639103), T12));
Chris@82 354 T1m = VFMA(LDK(KP823639103), T15, VMUL(LDK(KP509036960), T12));
Chris@82 355 T1u = VBYI(VMUL(LDK(KP866025403), VADD(T1e, T1f)));
Chris@82 356 T1d = VMUL(LDK(KP484122918), VSUB(T1b, T1c));
Chris@82 357 T1g = VFNMS(LDK(KP216506350), T1f, VMUL(LDK(KP866025403), T1e));
Chris@82 358 T1h = VSUB(T1d, T1g);
Chris@82 359 T1o = VADD(T1d, T1g);
Chris@82 360 {
Chris@82 361 V T18, T19, TY, TW, TX;
Chris@82 362 T18 = VSUB(TT, TU);
Chris@82 363 T19 = VSUB(TQ, TR);
Chris@82 364 T1a = VFNMS(LDK(KP587785252), T19, VMUL(LDK(KP951056516), T18));
Chris@82 365 T1p = VFMA(LDK(KP951056516), T19, VMUL(LDK(KP587785252), T18));
Chris@82 366 TY = VMUL(LDK(KP559016994), VSUB(TS, TV));
Chris@82 367 TW = VADD(TS, TV);
Chris@82 368 TX = VFNMS(LDK(KP250000000), TW, TP);
Chris@82 369 TZ = VSUB(TX, TY);
Chris@82 370 T1t = VADD(TP, TW);
Chris@82 371 T1l = VADD(TY, TX);
Chris@82 372 }
Chris@82 373 {
Chris@82 374 V T17, T1i, T1r, T1s;
Chris@82 375 ST(&(x[WS(rs, 5)]), VSUB(T1t, T1u), ms, &(x[WS(rs, 1)]));
Chris@82 376 ST(&(x[WS(rs, 10)]), VADD(T1t, T1u), ms, &(x[0]));
Chris@82 377 T17 = VSUB(TZ, T16);
Chris@82 378 T1i = VBYI(VSUB(T1a, T1h));
Chris@82 379 ST(&(x[WS(rs, 8)]), VSUB(T17, T1i), ms, &(x[0]));
Chris@82 380 ST(&(x[WS(rs, 7)]), VADD(T17, T1i), ms, &(x[WS(rs, 1)]));
Chris@82 381 T1r = VSUB(T1l, T1m);
Chris@82 382 T1s = VBYI(VADD(T1p, T1o));
Chris@82 383 ST(&(x[WS(rs, 11)]), VSUB(T1r, T1s), ms, &(x[WS(rs, 1)]));
Chris@82 384 ST(&(x[WS(rs, 4)]), VADD(T1r, T1s), ms, &(x[0]));
Chris@82 385 {
Chris@82 386 V T1n, T1q, T1j, T1k;
Chris@82 387 T1n = VADD(T1l, T1m);
Chris@82 388 T1q = VBYI(VSUB(T1o, T1p));
Chris@82 389 ST(&(x[WS(rs, 14)]), VSUB(T1n, T1q), ms, &(x[0]));
Chris@82 390 ST(&(x[WS(rs, 1)]), VADD(T1n, T1q), ms, &(x[WS(rs, 1)]));
Chris@82 391 T1j = VADD(TZ, T16);
Chris@82 392 T1k = VBYI(VADD(T1a, T1h));
Chris@82 393 ST(&(x[WS(rs, 13)]), VSUB(T1j, T1k), ms, &(x[WS(rs, 1)]));
Chris@82 394 ST(&(x[WS(rs, 2)]), VADD(T1j, T1k), ms, &(x[0]));
Chris@82 395 }
Chris@82 396 }
Chris@82 397 }
Chris@82 398 }
Chris@82 399 }
Chris@82 400 VLEAVE();
Chris@82 401 }
Chris@82 402
Chris@82 403 static const tw_instr twinstr[] = {
Chris@82 404 VTW(0, 1),
Chris@82 405 VTW(0, 2),
Chris@82 406 VTW(0, 3),
Chris@82 407 VTW(0, 4),
Chris@82 408 VTW(0, 5),
Chris@82 409 VTW(0, 6),
Chris@82 410 VTW(0, 7),
Chris@82 411 VTW(0, 8),
Chris@82 412 VTW(0, 9),
Chris@82 413 VTW(0, 10),
Chris@82 414 VTW(0, 11),
Chris@82 415 VTW(0, 12),
Chris@82 416 VTW(0, 13),
Chris@82 417 VTW(0, 14),
Chris@82 418 {TW_NEXT, VL, 0}
Chris@82 419 };
Chris@82 420
Chris@82 421 static const ct_desc desc = { 15, XSIMD_STRING("t1fv_15"), twinstr, &GENUS, {78, 39, 14, 0}, 0, 0, 0 };
Chris@82 422
Chris@82 423 void XSIMD(codelet_t1fv_15) (planner *p) {
Chris@82 424 X(kdft_dit_register) (p, t1fv_15, &desc);
Chris@82 425 }
Chris@82 426 #endif