Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/t1bv_15.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 16:44:21 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_twiddle_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 15 -name t1bv_15 -include t1b.h -sign 1 */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 92 FP additions, 77 FP multiplications,
Chris@42 32 * (or, 50 additions, 35 multiplications, 42 fused multiply/add),
Chris@42 33 * 81 stack variables, 8 constants, and 30 memory accesses
Chris@42 34 */
Chris@42 35 #include "t1b.h"
Chris@42 36
Chris@42 37 static void t1bv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 38 {
Chris@42 39 DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
Chris@42 40 DVK(KP910592997, +0.910592997310029334643087372129977886038870291);
Chris@42 41 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42 42 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42 43 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 44 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42 45 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@42 46 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 47 {
Chris@42 48 INT m;
Chris@42 49 R *x;
Chris@42 50 x = ii;
Chris@42 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@42 52 V Tq, Ty, Th, TV, TK, Ts, T1f, T7, Tu, TA, TC, Tj, Tk, T1g, Tf;
Chris@42 53 {
Chris@42 54 V T1, T4, T2, T9, Te;
Chris@42 55 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 56 T4 = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@42 57 T2 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 58 {
Chris@42 59 V T8, Tp, Tx, Tg;
Chris@42 60 T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 61 Tp = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 62 Tx = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@42 63 Tg = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@42 64 {
Chris@42 65 V Tb, Td, Tr, T6, Tt, Tz, TB, Ti;
Chris@42 66 {
Chris@42 67 V T5, T3, Ta, Tc;
Chris@42 68 Ta = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@42 69 Tc = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@42 70 T5 = BYTW(&(W[TWVL * 18]), T4);
Chris@42 71 T3 = BYTW(&(W[TWVL * 8]), T2);
Chris@42 72 T9 = BYTW(&(W[TWVL * 4]), T8);
Chris@42 73 Tq = BYTW(&(W[TWVL * 10]), Tp);
Chris@42 74 Ty = BYTW(&(W[TWVL * 16]), Tx);
Chris@42 75 Th = BYTW(&(W[TWVL * 22]), Tg);
Chris@42 76 Tb = BYTW(&(W[TWVL * 14]), Ta);
Chris@42 77 Td = BYTW(&(W[TWVL * 24]), Tc);
Chris@42 78 Tr = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@42 79 TV = VSUB(T3, T5);
Chris@42 80 T6 = VADD(T3, T5);
Chris@42 81 Tt = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 82 }
Chris@42 83 Tz = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@42 84 TB = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 85 Ti = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 86 Te = VADD(Tb, Td);
Chris@42 87 TK = VSUB(Tb, Td);
Chris@42 88 Ts = BYTW(&(W[TWVL * 20]), Tr);
Chris@42 89 T1f = VADD(T1, T6);
Chris@42 90 T7 = VFNMS(LDK(KP500000000), T6, T1);
Chris@42 91 Tu = BYTW(&(W[0]), Tt);
Chris@42 92 TA = BYTW(&(W[TWVL * 26]), Tz);
Chris@42 93 TC = BYTW(&(W[TWVL * 6]), TB);
Chris@42 94 Tj = BYTW(&(W[TWVL * 2]), Ti);
Chris@42 95 Tk = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 96 }
Chris@42 97 }
Chris@42 98 T1g = VADD(T9, Te);
Chris@42 99 Tf = VFNMS(LDK(KP500000000), Te, T9);
Chris@42 100 }
Chris@42 101 {
Chris@42 102 V Tv, TN, TD, TO, Tl;
Chris@42 103 Tv = VADD(Ts, Tu);
Chris@42 104 TN = VSUB(Ts, Tu);
Chris@42 105 TD = VADD(TA, TC);
Chris@42 106 TO = VSUB(TA, TC);
Chris@42 107 Tl = BYTW(&(W[TWVL * 12]), Tk);
Chris@42 108 {
Chris@42 109 V Tw, T1j, TX, TP, TE, T1k, TL, Tm;
Chris@42 110 Tw = VFNMS(LDK(KP500000000), Tv, Tq);
Chris@42 111 T1j = VADD(Tq, Tv);
Chris@42 112 TX = VADD(TN, TO);
Chris@42 113 TP = VSUB(TN, TO);
Chris@42 114 TE = VFNMS(LDK(KP500000000), TD, Ty);
Chris@42 115 T1k = VADD(Ty, TD);
Chris@42 116 TL = VSUB(Tj, Tl);
Chris@42 117 Tm = VADD(Tj, Tl);
Chris@42 118 {
Chris@42 119 V TT, TF, T1q, T1l, TW, TM, T1h, Tn;
Chris@42 120 TT = VSUB(Tw, TE);
Chris@42 121 TF = VADD(Tw, TE);
Chris@42 122 T1q = VSUB(T1j, T1k);
Chris@42 123 T1l = VADD(T1j, T1k);
Chris@42 124 TW = VADD(TK, TL);
Chris@42 125 TM = VSUB(TK, TL);
Chris@42 126 T1h = VADD(Th, Tm);
Chris@42 127 Tn = VFNMS(LDK(KP500000000), Tm, Th);
Chris@42 128 {
Chris@42 129 V T10, TY, T16, TQ, T1r, T1i, TS, To, TZ, T1e;
Chris@42 130 T10 = VSUB(TW, TX);
Chris@42 131 TY = VADD(TW, TX);
Chris@42 132 T16 = VFNMS(LDK(KP618033988), TM, TP);
Chris@42 133 TQ = VFMA(LDK(KP618033988), TP, TM);
Chris@42 134 T1r = VSUB(T1g, T1h);
Chris@42 135 T1i = VADD(T1g, T1h);
Chris@42 136 TS = VSUB(Tf, Tn);
Chris@42 137 To = VADD(Tf, Tn);
Chris@42 138 TZ = VFNMS(LDK(KP250000000), TY, TV);
Chris@42 139 T1e = VMUL(LDK(KP866025403), VADD(TV, TY));
Chris@42 140 {
Chris@42 141 V T1u, T1s, T1o, T18, TU, TG, TI, T19, T11, T1n, T1m;
Chris@42 142 T1u = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1q, T1r));
Chris@42 143 T1s = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1r, T1q));
Chris@42 144 T1m = VADD(T1i, T1l);
Chris@42 145 T1o = VSUB(T1i, T1l);
Chris@42 146 T18 = VFNMS(LDK(KP618033988), TS, TT);
Chris@42 147 TU = VFMA(LDK(KP618033988), TT, TS);
Chris@42 148 TG = VADD(To, TF);
Chris@42 149 TI = VSUB(To, TF);
Chris@42 150 T19 = VFNMS(LDK(KP559016994), T10, TZ);
Chris@42 151 T11 = VFMA(LDK(KP559016994), T10, TZ);
Chris@42 152 ST(&(x[0]), VADD(T1f, T1m), ms, &(x[0]));
Chris@42 153 T1n = VFNMS(LDK(KP250000000), T1m, T1f);
Chris@42 154 {
Chris@42 155 V T1a, T1c, T14, T12, T1p, T1t, T15, TJ, T1d, TH;
Chris@42 156 T1d = VADD(T7, TG);
Chris@42 157 TH = VFNMS(LDK(KP250000000), TG, T7);
Chris@42 158 T1a = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), T19, T18));
Chris@42 159 T1c = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), T19, T18));
Chris@42 160 T14 = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), T11, TU));
Chris@42 161 T12 = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), T11, TU));
Chris@42 162 T1p = VFNMS(LDK(KP559016994), T1o, T1n);
Chris@42 163 T1t = VFMA(LDK(KP559016994), T1o, T1n);
Chris@42 164 ST(&(x[WS(rs, 10)]), VFMAI(T1e, T1d), ms, &(x[0]));
Chris@42 165 ST(&(x[WS(rs, 5)]), VFNMSI(T1e, T1d), ms, &(x[WS(rs, 1)]));
Chris@42 166 T15 = VFNMS(LDK(KP559016994), TI, TH);
Chris@42 167 TJ = VFMA(LDK(KP559016994), TI, TH);
Chris@42 168 {
Chris@42 169 V T17, T1b, T13, TR;
Chris@42 170 ST(&(x[WS(rs, 12)]), VFNMSI(T1s, T1p), ms, &(x[0]));
Chris@42 171 ST(&(x[WS(rs, 3)]), VFMAI(T1s, T1p), ms, &(x[WS(rs, 1)]));
Chris@42 172 ST(&(x[WS(rs, 9)]), VFNMSI(T1u, T1t), ms, &(x[WS(rs, 1)]));
Chris@42 173 ST(&(x[WS(rs, 6)]), VFMAI(T1u, T1t), ms, &(x[0]));
Chris@42 174 T17 = VFNMS(LDK(KP823639103), T16, T15);
Chris@42 175 T1b = VFMA(LDK(KP823639103), T16, T15);
Chris@42 176 T13 = VFMA(LDK(KP823639103), TQ, TJ);
Chris@42 177 TR = VFNMS(LDK(KP823639103), TQ, TJ);
Chris@42 178 ST(&(x[WS(rs, 13)]), VFMAI(T1a, T17), ms, &(x[WS(rs, 1)]));
Chris@42 179 ST(&(x[WS(rs, 2)]), VFNMSI(T1a, T17), ms, &(x[0]));
Chris@42 180 ST(&(x[WS(rs, 8)]), VFMAI(T1c, T1b), ms, &(x[0]));
Chris@42 181 ST(&(x[WS(rs, 7)]), VFNMSI(T1c, T1b), ms, &(x[WS(rs, 1)]));
Chris@42 182 ST(&(x[WS(rs, 11)]), VFMAI(T14, T13), ms, &(x[WS(rs, 1)]));
Chris@42 183 ST(&(x[WS(rs, 4)]), VFNMSI(T14, T13), ms, &(x[0]));
Chris@42 184 ST(&(x[WS(rs, 14)]), VFNMSI(T12, TR), ms, &(x[0]));
Chris@42 185 ST(&(x[WS(rs, 1)]), VFMAI(T12, TR), ms, &(x[WS(rs, 1)]));
Chris@42 186 }
Chris@42 187 }
Chris@42 188 }
Chris@42 189 }
Chris@42 190 }
Chris@42 191 }
Chris@42 192 }
Chris@42 193 }
Chris@42 194 }
Chris@42 195 VLEAVE();
Chris@42 196 }
Chris@42 197
Chris@42 198 static const tw_instr twinstr[] = {
Chris@42 199 VTW(0, 1),
Chris@42 200 VTW(0, 2),
Chris@42 201 VTW(0, 3),
Chris@42 202 VTW(0, 4),
Chris@42 203 VTW(0, 5),
Chris@42 204 VTW(0, 6),
Chris@42 205 VTW(0, 7),
Chris@42 206 VTW(0, 8),
Chris@42 207 VTW(0, 9),
Chris@42 208 VTW(0, 10),
Chris@42 209 VTW(0, 11),
Chris@42 210 VTW(0, 12),
Chris@42 211 VTW(0, 13),
Chris@42 212 VTW(0, 14),
Chris@42 213 {TW_NEXT, VL, 0}
Chris@42 214 };
Chris@42 215
Chris@42 216 static const ct_desc desc = { 15, XSIMD_STRING("t1bv_15"), twinstr, &GENUS, {50, 35, 42, 0}, 0, 0, 0 };
Chris@42 217
Chris@42 218 void XSIMD(codelet_t1bv_15) (planner *p) {
Chris@42 219 X(kdft_dit_register) (p, t1bv_15, &desc);
Chris@42 220 }
Chris@42 221 #else /* HAVE_FMA */
Chris@42 222
Chris@42 223 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 15 -name t1bv_15 -include t1b.h -sign 1 */
Chris@42 224
Chris@42 225 /*
Chris@42 226 * This function contains 92 FP additions, 53 FP multiplications,
Chris@42 227 * (or, 78 additions, 39 multiplications, 14 fused multiply/add),
Chris@42 228 * 52 stack variables, 10 constants, and 30 memory accesses
Chris@42 229 */
Chris@42 230 #include "t1b.h"
Chris@42 231
Chris@42 232 static void t1bv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 233 {
Chris@42 234 DVK(KP216506350, +0.216506350946109661690930792688234045867850657);
Chris@42 235 DVK(KP484122918, +0.484122918275927110647408174972799951354115213);
Chris@42 236 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 237 DVK(KP509036960, +0.509036960455127183450980863393907648510733164);
Chris@42 238 DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
Chris@42 239 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42 240 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@42 241 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42 242 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42 243 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 244 {
Chris@42 245 INT m;
Chris@42 246 R *x;
Chris@42 247 x = ii;
Chris@42 248 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@42 249 V Ts, TV, T1f, TZ, T10, Tb, Tm, Tt, T1j, T1k, T1l, TI, TM, TR, Tz;
Chris@42 250 V TD, TQ, T1g, T1h, T1i;
Chris@42 251 {
Chris@42 252 V TT, Tr, Tp, Tq, To, TU;
Chris@42 253 TT = LD(&(x[0]), ms, &(x[0]));
Chris@42 254 Tq = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@42 255 Tr = BYTW(&(W[TWVL * 18]), Tq);
Chris@42 256 To = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 257 Tp = BYTW(&(W[TWVL * 8]), To);
Chris@42 258 Ts = VSUB(Tp, Tr);
Chris@42 259 TU = VADD(Tp, Tr);
Chris@42 260 TV = VFNMS(LDK(KP500000000), TU, TT);
Chris@42 261 T1f = VADD(TT, TU);
Chris@42 262 }
Chris@42 263 {
Chris@42 264 V Tx, TG, TK, TB, T5, Ty, Tg, TH, Tl, TL, Ta, TC;
Chris@42 265 {
Chris@42 266 V Tw, TF, TJ, TA;
Chris@42 267 Tw = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 268 Tx = BYTW(&(W[TWVL * 4]), Tw);
Chris@42 269 TF = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 270 TG = BYTW(&(W[TWVL * 10]), TF);
Chris@42 271 TJ = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@42 272 TK = BYTW(&(W[TWVL * 16]), TJ);
Chris@42 273 TA = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@42 274 TB = BYTW(&(W[TWVL * 22]), TA);
Chris@42 275 }
Chris@42 276 {
Chris@42 277 V T2, T4, T1, T3;
Chris@42 278 T1 = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@42 279 T2 = BYTW(&(W[TWVL * 14]), T1);
Chris@42 280 T3 = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@42 281 T4 = BYTW(&(W[TWVL * 24]), T3);
Chris@42 282 T5 = VSUB(T2, T4);
Chris@42 283 Ty = VADD(T2, T4);
Chris@42 284 }
Chris@42 285 {
Chris@42 286 V Td, Tf, Tc, Te;
Chris@42 287 Tc = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@42 288 Td = BYTW(&(W[TWVL * 20]), Tc);
Chris@42 289 Te = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 290 Tf = BYTW(&(W[0]), Te);
Chris@42 291 Tg = VSUB(Td, Tf);
Chris@42 292 TH = VADD(Td, Tf);
Chris@42 293 }
Chris@42 294 {
Chris@42 295 V Ti, Tk, Th, Tj;
Chris@42 296 Th = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@42 297 Ti = BYTW(&(W[TWVL * 26]), Th);
Chris@42 298 Tj = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 299 Tk = BYTW(&(W[TWVL * 6]), Tj);
Chris@42 300 Tl = VSUB(Ti, Tk);
Chris@42 301 TL = VADD(Ti, Tk);
Chris@42 302 }
Chris@42 303 {
Chris@42 304 V T7, T9, T6, T8;
Chris@42 305 T6 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 306 T7 = BYTW(&(W[TWVL * 2]), T6);
Chris@42 307 T8 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 308 T9 = BYTW(&(W[TWVL * 12]), T8);
Chris@42 309 Ta = VSUB(T7, T9);
Chris@42 310 TC = VADD(T7, T9);
Chris@42 311 }
Chris@42 312 TZ = VSUB(T5, Ta);
Chris@42 313 T10 = VSUB(Tg, Tl);
Chris@42 314 Tb = VADD(T5, Ta);
Chris@42 315 Tm = VADD(Tg, Tl);
Chris@42 316 Tt = VADD(Tb, Tm);
Chris@42 317 T1j = VADD(TG, TH);
Chris@42 318 T1k = VADD(TK, TL);
Chris@42 319 T1l = VADD(T1j, T1k);
Chris@42 320 TI = VFNMS(LDK(KP500000000), TH, TG);
Chris@42 321 TM = VFNMS(LDK(KP500000000), TL, TK);
Chris@42 322 TR = VADD(TI, TM);
Chris@42 323 Tz = VFNMS(LDK(KP500000000), Ty, Tx);
Chris@42 324 TD = VFNMS(LDK(KP500000000), TC, TB);
Chris@42 325 TQ = VADD(Tz, TD);
Chris@42 326 T1g = VADD(Tx, Ty);
Chris@42 327 T1h = VADD(TB, TC);
Chris@42 328 T1i = VADD(T1g, T1h);
Chris@42 329 }
Chris@42 330 {
Chris@42 331 V T1o, T1m, T1n, T1s, T1t, T1q, T1r, T1u, T1p;
Chris@42 332 T1o = VMUL(LDK(KP559016994), VSUB(T1i, T1l));
Chris@42 333 T1m = VADD(T1i, T1l);
Chris@42 334 T1n = VFNMS(LDK(KP250000000), T1m, T1f);
Chris@42 335 T1q = VSUB(T1g, T1h);
Chris@42 336 T1r = VSUB(T1j, T1k);
Chris@42 337 T1s = VBYI(VFNMS(LDK(KP951056516), T1r, VMUL(LDK(KP587785252), T1q)));
Chris@42 338 T1t = VBYI(VFMA(LDK(KP951056516), T1q, VMUL(LDK(KP587785252), T1r)));
Chris@42 339 ST(&(x[0]), VADD(T1f, T1m), ms, &(x[0]));
Chris@42 340 T1u = VADD(T1o, T1n);
Chris@42 341 ST(&(x[WS(rs, 6)]), VADD(T1t, T1u), ms, &(x[0]));
Chris@42 342 ST(&(x[WS(rs, 9)]), VSUB(T1u, T1t), ms, &(x[WS(rs, 1)]));
Chris@42 343 T1p = VSUB(T1n, T1o);
Chris@42 344 ST(&(x[WS(rs, 3)]), VSUB(T1p, T1s), ms, &(x[WS(rs, 1)]));
Chris@42 345 ST(&(x[WS(rs, 12)]), VADD(T1s, T1p), ms, &(x[0]));
Chris@42 346 }
Chris@42 347 {
Chris@42 348 V T11, T18, T1e, TO, T16, Tv, T15, TY, T1d, T19, TE, TN;
Chris@42 349 T11 = VFMA(LDK(KP823639103), TZ, VMUL(LDK(KP509036960), T10));
Chris@42 350 T18 = VFNMS(LDK(KP823639103), T10, VMUL(LDK(KP509036960), TZ));
Chris@42 351 T1e = VBYI(VMUL(LDK(KP866025403), VADD(Ts, Tt)));
Chris@42 352 TE = VSUB(Tz, TD);
Chris@42 353 TN = VSUB(TI, TM);
Chris@42 354 TO = VFMA(LDK(KP951056516), TE, VMUL(LDK(KP587785252), TN));
Chris@42 355 T16 = VFNMS(LDK(KP951056516), TN, VMUL(LDK(KP587785252), TE));
Chris@42 356 {
Chris@42 357 V Tn, Tu, TS, TW, TX;
Chris@42 358 Tn = VMUL(LDK(KP484122918), VSUB(Tb, Tm));
Chris@42 359 Tu = VFNMS(LDK(KP216506350), Tt, VMUL(LDK(KP866025403), Ts));
Chris@42 360 Tv = VADD(Tn, Tu);
Chris@42 361 T15 = VSUB(Tn, Tu);
Chris@42 362 TS = VMUL(LDK(KP559016994), VSUB(TQ, TR));
Chris@42 363 TW = VADD(TQ, TR);
Chris@42 364 TX = VFNMS(LDK(KP250000000), TW, TV);
Chris@42 365 TY = VADD(TS, TX);
Chris@42 366 T1d = VADD(TV, TW);
Chris@42 367 T19 = VSUB(TX, TS);
Chris@42 368 }
Chris@42 369 {
Chris@42 370 V TP, T12, T1b, T1c;
Chris@42 371 ST(&(x[WS(rs, 5)]), VSUB(T1d, T1e), ms, &(x[WS(rs, 1)]));
Chris@42 372 ST(&(x[WS(rs, 10)]), VADD(T1e, T1d), ms, &(x[0]));
Chris@42 373 TP = VBYI(VADD(Tv, TO));
Chris@42 374 T12 = VSUB(TY, T11);
Chris@42 375 ST(&(x[WS(rs, 1)]), VADD(TP, T12), ms, &(x[WS(rs, 1)]));
Chris@42 376 ST(&(x[WS(rs, 14)]), VSUB(T12, TP), ms, &(x[0]));
Chris@42 377 T1b = VBYI(VSUB(T16, T15));
Chris@42 378 T1c = VSUB(T19, T18);
Chris@42 379 ST(&(x[WS(rs, 7)]), VADD(T1b, T1c), ms, &(x[WS(rs, 1)]));
Chris@42 380 ST(&(x[WS(rs, 8)]), VSUB(T1c, T1b), ms, &(x[0]));
Chris@42 381 {
Chris@42 382 V T17, T1a, T13, T14;
Chris@42 383 T17 = VBYI(VADD(T15, T16));
Chris@42 384 T1a = VADD(T18, T19);
Chris@42 385 ST(&(x[WS(rs, 2)]), VADD(T17, T1a), ms, &(x[0]));
Chris@42 386 ST(&(x[WS(rs, 13)]), VSUB(T1a, T17), ms, &(x[WS(rs, 1)]));
Chris@42 387 T13 = VBYI(VSUB(Tv, TO));
Chris@42 388 T14 = VADD(T11, TY);
Chris@42 389 ST(&(x[WS(rs, 4)]), VADD(T13, T14), ms, &(x[0]));
Chris@42 390 ST(&(x[WS(rs, 11)]), VSUB(T14, T13), ms, &(x[WS(rs, 1)]));
Chris@42 391 }
Chris@42 392 }
Chris@42 393 }
Chris@42 394 }
Chris@42 395 }
Chris@42 396 VLEAVE();
Chris@42 397 }
Chris@42 398
Chris@42 399 static const tw_instr twinstr[] = {
Chris@42 400 VTW(0, 1),
Chris@42 401 VTW(0, 2),
Chris@42 402 VTW(0, 3),
Chris@42 403 VTW(0, 4),
Chris@42 404 VTW(0, 5),
Chris@42 405 VTW(0, 6),
Chris@42 406 VTW(0, 7),
Chris@42 407 VTW(0, 8),
Chris@42 408 VTW(0, 9),
Chris@42 409 VTW(0, 10),
Chris@42 410 VTW(0, 11),
Chris@42 411 VTW(0, 12),
Chris@42 412 VTW(0, 13),
Chris@42 413 VTW(0, 14),
Chris@42 414 {TW_NEXT, VL, 0}
Chris@42 415 };
Chris@42 416
Chris@42 417 static const ct_desc desc = { 15, XSIMD_STRING("t1bv_15"), twinstr, &GENUS, {78, 39, 14, 0}, 0, 0, 0 };
Chris@42 418
Chris@42 419 void XSIMD(codelet_t1bv_15) (planner *p) {
Chris@42 420 X(kdft_dit_register) (p, t1bv_15, &desc);
Chris@42 421 }
Chris@42 422 #endif /* HAVE_FMA */