Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/t1fv_16.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 16 -name t1fv_16 -include dft/simd/t1f.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 87 FP additions, 64 FP multiplications,
Chris@82 32 * (or, 53 additions, 30 multiplications, 34 fused multiply/add),
Chris@82 33 * 36 stack variables, 3 constants, and 32 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/t1f.h"
Chris@82 36
Chris@82 37 static void t1fv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 38 {
Chris@82 39 DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82 40 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 41 DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@82 42 {
Chris@82 43 INT m;
Chris@82 44 R *x;
Chris@82 45 x = ri;
Chris@82 46 for (m = mb, W = W + (mb * ((TWVL / VL) * 30)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(16, rs)) {
Chris@82 47 V T4, TW, T9, T19, TD, TI, TZ, T1a, Tf, Tk, Tl, T13, T1c, Tq, Tv;
Chris@82 48 V Tw, T16, T1d, T1, T3, T2;
Chris@82 49 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@82 50 T2 = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@82 51 T3 = BYTWJ(&(W[TWVL * 14]), T2);
Chris@82 52 T4 = VADD(T1, T3);
Chris@82 53 TW = VSUB(T1, T3);
Chris@82 54 {
Chris@82 55 V T6, T8, T5, T7;
Chris@82 56 T5 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@82 57 T6 = BYTWJ(&(W[TWVL * 6]), T5);
Chris@82 58 T7 = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@82 59 T8 = BYTWJ(&(W[TWVL * 22]), T7);
Chris@82 60 T9 = VADD(T6, T8);
Chris@82 61 T19 = VSUB(T6, T8);
Chris@82 62 }
Chris@82 63 {
Chris@82 64 V TA, TH, TC, TF, TX, TY;
Chris@82 65 {
Chris@82 66 V Tz, TG, TB, TE;
Chris@82 67 Tz = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@82 68 TA = BYTWJ(&(W[TWVL * 26]), Tz);
Chris@82 69 TG = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@82 70 TH = BYTWJ(&(W[TWVL * 18]), TG);
Chris@82 71 TB = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@82 72 TC = BYTWJ(&(W[TWVL * 10]), TB);
Chris@82 73 TE = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@82 74 TF = BYTWJ(&(W[TWVL * 2]), TE);
Chris@82 75 }
Chris@82 76 TD = VADD(TA, TC);
Chris@82 77 TI = VADD(TF, TH);
Chris@82 78 TX = VSUB(TF, TH);
Chris@82 79 TY = VSUB(TA, TC);
Chris@82 80 TZ = VADD(TX, TY);
Chris@82 81 T1a = VSUB(TY, TX);
Chris@82 82 }
Chris@82 83 {
Chris@82 84 V Tc, Tj, Te, Th, T11, T12;
Chris@82 85 {
Chris@82 86 V Tb, Ti, Td, Tg;
Chris@82 87 Tb = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82 88 Tc = BYTWJ(&(W[0]), Tb);
Chris@82 89 Ti = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@82 90 Tj = BYTWJ(&(W[TWVL * 24]), Ti);
Chris@82 91 Td = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@82 92 Te = BYTWJ(&(W[TWVL * 16]), Td);
Chris@82 93 Tg = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@82 94 Th = BYTWJ(&(W[TWVL * 8]), Tg);
Chris@82 95 }
Chris@82 96 Tf = VADD(Tc, Te);
Chris@82 97 Tk = VADD(Th, Tj);
Chris@82 98 Tl = VSUB(Tf, Tk);
Chris@82 99 T11 = VSUB(Tc, Te);
Chris@82 100 T12 = VSUB(Th, Tj);
Chris@82 101 T13 = VFNMS(LDK(KP414213562), T12, T11);
Chris@82 102 T1c = VFMA(LDK(KP414213562), T11, T12);
Chris@82 103 }
Chris@82 104 {
Chris@82 105 V Tn, Tu, Tp, Ts, T14, T15;
Chris@82 106 {
Chris@82 107 V Tm, Tt, To, Tr;
Chris@82 108 Tm = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
Chris@82 109 Tn = BYTWJ(&(W[TWVL * 28]), Tm);
Chris@82 110 Tt = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@82 111 Tu = BYTWJ(&(W[TWVL * 20]), Tt);
Chris@82 112 To = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@82 113 Tp = BYTWJ(&(W[TWVL * 12]), To);
Chris@82 114 Tr = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@82 115 Ts = BYTWJ(&(W[TWVL * 4]), Tr);
Chris@82 116 }
Chris@82 117 Tq = VADD(Tn, Tp);
Chris@82 118 Tv = VADD(Ts, Tu);
Chris@82 119 Tw = VSUB(Tq, Tv);
Chris@82 120 T14 = VSUB(Tn, Tp);
Chris@82 121 T15 = VSUB(Tu, Ts);
Chris@82 122 T16 = VFNMS(LDK(KP414213562), T15, T14);
Chris@82 123 T1d = VFMA(LDK(KP414213562), T14, T15);
Chris@82 124 }
Chris@82 125 {
Chris@82 126 V Ty, TM, TL, TN;
Chris@82 127 {
Chris@82 128 V Ta, Tx, TJ, TK;
Chris@82 129 Ta = VSUB(T4, T9);
Chris@82 130 Tx = VADD(Tl, Tw);
Chris@82 131 Ty = VFNMS(LDK(KP707106781), Tx, Ta);
Chris@82 132 TM = VFMA(LDK(KP707106781), Tx, Ta);
Chris@82 133 TJ = VSUB(TD, TI);
Chris@82 134 TK = VSUB(Tw, Tl);
Chris@82 135 TL = VFNMS(LDK(KP707106781), TK, TJ);
Chris@82 136 TN = VFMA(LDK(KP707106781), TK, TJ);
Chris@82 137 }
Chris@82 138 ST(&(x[WS(rs, 6)]), VFNMSI(TL, Ty), ms, &(x[0]));
Chris@82 139 ST(&(x[WS(rs, 2)]), VFMAI(TN, TM), ms, &(x[0]));
Chris@82 140 ST(&(x[WS(rs, 10)]), VFMAI(TL, Ty), ms, &(x[0]));
Chris@82 141 ST(&(x[WS(rs, 14)]), VFNMSI(TN, TM), ms, &(x[0]));
Chris@82 142 }
Chris@82 143 {
Chris@82 144 V T1k, T1o, T1n, T1p;
Chris@82 145 {
Chris@82 146 V T1i, T1j, T1l, T1m;
Chris@82 147 T1i = VFNMS(LDK(KP707106781), TZ, TW);
Chris@82 148 T1j = VADD(T1c, T1d);
Chris@82 149 T1k = VFNMS(LDK(KP923879532), T1j, T1i);
Chris@82 150 T1o = VFMA(LDK(KP923879532), T1j, T1i);
Chris@82 151 T1l = VFMA(LDK(KP707106781), T1a, T19);
Chris@82 152 T1m = VSUB(T16, T13);
Chris@82 153 T1n = VFNMS(LDK(KP923879532), T1m, T1l);
Chris@82 154 T1p = VFMA(LDK(KP923879532), T1m, T1l);
Chris@82 155 }
Chris@82 156 ST(&(x[WS(rs, 5)]), VFNMSI(T1n, T1k), ms, &(x[WS(rs, 1)]));
Chris@82 157 ST(&(x[WS(rs, 13)]), VFNMSI(T1p, T1o), ms, &(x[WS(rs, 1)]));
Chris@82 158 ST(&(x[WS(rs, 11)]), VFMAI(T1n, T1k), ms, &(x[WS(rs, 1)]));
Chris@82 159 ST(&(x[WS(rs, 3)]), VFMAI(T1p, T1o), ms, &(x[WS(rs, 1)]));
Chris@82 160 }
Chris@82 161 {
Chris@82 162 V TQ, TU, TT, TV;
Chris@82 163 {
Chris@82 164 V TO, TP, TR, TS;
Chris@82 165 TO = VADD(T4, T9);
Chris@82 166 TP = VADD(TI, TD);
Chris@82 167 TQ = VADD(TO, TP);
Chris@82 168 TU = VSUB(TO, TP);
Chris@82 169 TR = VADD(Tf, Tk);
Chris@82 170 TS = VADD(Tq, Tv);
Chris@82 171 TT = VADD(TR, TS);
Chris@82 172 TV = VSUB(TS, TR);
Chris@82 173 }
Chris@82 174 ST(&(x[WS(rs, 8)]), VSUB(TQ, TT), ms, &(x[0]));
Chris@82 175 ST(&(x[WS(rs, 4)]), VFMAI(TV, TU), ms, &(x[0]));
Chris@82 176 ST(&(x[0]), VADD(TQ, TT), ms, &(x[0]));
Chris@82 177 ST(&(x[WS(rs, 12)]), VFNMSI(TV, TU), ms, &(x[0]));
Chris@82 178 }
Chris@82 179 {
Chris@82 180 V T18, T1g, T1f, T1h;
Chris@82 181 {
Chris@82 182 V T10, T17, T1b, T1e;
Chris@82 183 T10 = VFMA(LDK(KP707106781), TZ, TW);
Chris@82 184 T17 = VADD(T13, T16);
Chris@82 185 T18 = VFNMS(LDK(KP923879532), T17, T10);
Chris@82 186 T1g = VFMA(LDK(KP923879532), T17, T10);
Chris@82 187 T1b = VFNMS(LDK(KP707106781), T1a, T19);
Chris@82 188 T1e = VSUB(T1c, T1d);
Chris@82 189 T1f = VFNMS(LDK(KP923879532), T1e, T1b);
Chris@82 190 T1h = VFMA(LDK(KP923879532), T1e, T1b);
Chris@82 191 }
Chris@82 192 ST(&(x[WS(rs, 9)]), VFNMSI(T1f, T18), ms, &(x[WS(rs, 1)]));
Chris@82 193 ST(&(x[WS(rs, 15)]), VFMAI(T1h, T1g), ms, &(x[WS(rs, 1)]));
Chris@82 194 ST(&(x[WS(rs, 7)]), VFMAI(T1f, T18), ms, &(x[WS(rs, 1)]));
Chris@82 195 ST(&(x[WS(rs, 1)]), VFNMSI(T1h, T1g), ms, &(x[WS(rs, 1)]));
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 VTW(0, 15),
Chris@82 218 {TW_NEXT, VL, 0}
Chris@82 219 };
Chris@82 220
Chris@82 221 static const ct_desc desc = { 16, XSIMD_STRING("t1fv_16"), twinstr, &GENUS, {53, 30, 34, 0}, 0, 0, 0 };
Chris@82 222
Chris@82 223 void XSIMD(codelet_t1fv_16) (planner *p) {
Chris@82 224 X(kdft_dit_register) (p, t1fv_16, &desc);
Chris@82 225 }
Chris@82 226 #else
Chris@82 227
Chris@82 228 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 16 -name t1fv_16 -include dft/simd/t1f.h */
Chris@82 229
Chris@82 230 /*
Chris@82 231 * This function contains 87 FP additions, 42 FP multiplications,
Chris@82 232 * (or, 83 additions, 38 multiplications, 4 fused multiply/add),
Chris@82 233 * 36 stack variables, 3 constants, and 32 memory accesses
Chris@82 234 */
Chris@82 235 #include "dft/simd/t1f.h"
Chris@82 236
Chris@82 237 static void t1fv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 238 {
Chris@82 239 DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82 240 DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
Chris@82 241 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 242 {
Chris@82 243 INT m;
Chris@82 244 R *x;
Chris@82 245 x = ri;
Chris@82 246 for (m = mb, W = W + (mb * ((TWVL / VL) * 30)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(16, rs)) {
Chris@82 247 V TJ, T10, TD, T11, T1b, T1c, Ty, TK, T16, T17, T18, Tb, TN, T13, T14;
Chris@82 248 V T15, Tm, TM, TG, TI, TH;
Chris@82 249 TG = LD(&(x[0]), ms, &(x[0]));
Chris@82 250 TH = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@82 251 TI = BYTWJ(&(W[TWVL * 14]), TH);
Chris@82 252 TJ = VSUB(TG, TI);
Chris@82 253 T10 = VADD(TG, TI);
Chris@82 254 {
Chris@82 255 V TA, TC, Tz, TB;
Chris@82 256 Tz = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@82 257 TA = BYTWJ(&(W[TWVL * 6]), Tz);
Chris@82 258 TB = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@82 259 TC = BYTWJ(&(W[TWVL * 22]), TB);
Chris@82 260 TD = VSUB(TA, TC);
Chris@82 261 T11 = VADD(TA, TC);
Chris@82 262 }
Chris@82 263 {
Chris@82 264 V Tp, Tw, Tr, Tu, Ts, Tx;
Chris@82 265 {
Chris@82 266 V To, Tv, Tq, Tt;
Chris@82 267 To = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@82 268 Tp = BYTWJ(&(W[TWVL * 26]), To);
Chris@82 269 Tv = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@82 270 Tw = BYTWJ(&(W[TWVL * 18]), Tv);
Chris@82 271 Tq = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@82 272 Tr = BYTWJ(&(W[TWVL * 10]), Tq);
Chris@82 273 Tt = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@82 274 Tu = BYTWJ(&(W[TWVL * 2]), Tt);
Chris@82 275 }
Chris@82 276 T1b = VADD(Tp, Tr);
Chris@82 277 T1c = VADD(Tu, Tw);
Chris@82 278 Ts = VSUB(Tp, Tr);
Chris@82 279 Tx = VSUB(Tu, Tw);
Chris@82 280 Ty = VMUL(LDK(KP707106781), VSUB(Ts, Tx));
Chris@82 281 TK = VMUL(LDK(KP707106781), VADD(Tx, Ts));
Chris@82 282 }
Chris@82 283 {
Chris@82 284 V T2, T9, T4, T7, T5, Ta;
Chris@82 285 {
Chris@82 286 V T1, T8, T3, T6;
Chris@82 287 T1 = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
Chris@82 288 T2 = BYTWJ(&(W[TWVL * 28]), T1);
Chris@82 289 T8 = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@82 290 T9 = BYTWJ(&(W[TWVL * 20]), T8);
Chris@82 291 T3 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@82 292 T4 = BYTWJ(&(W[TWVL * 12]), T3);
Chris@82 293 T6 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@82 294 T7 = BYTWJ(&(W[TWVL * 4]), T6);
Chris@82 295 }
Chris@82 296 T16 = VADD(T2, T4);
Chris@82 297 T17 = VADD(T7, T9);
Chris@82 298 T18 = VSUB(T16, T17);
Chris@82 299 T5 = VSUB(T2, T4);
Chris@82 300 Ta = VSUB(T7, T9);
Chris@82 301 Tb = VFNMS(LDK(KP923879532), Ta, VMUL(LDK(KP382683432), T5));
Chris@82 302 TN = VFMA(LDK(KP923879532), T5, VMUL(LDK(KP382683432), Ta));
Chris@82 303 }
Chris@82 304 {
Chris@82 305 V Td, Tk, Tf, Ti, Tg, Tl;
Chris@82 306 {
Chris@82 307 V Tc, Tj, Te, Th;
Chris@82 308 Tc = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82 309 Td = BYTWJ(&(W[0]), Tc);
Chris@82 310 Tj = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@82 311 Tk = BYTWJ(&(W[TWVL * 24]), Tj);
Chris@82 312 Te = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@82 313 Tf = BYTWJ(&(W[TWVL * 16]), Te);
Chris@82 314 Th = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@82 315 Ti = BYTWJ(&(W[TWVL * 8]), Th);
Chris@82 316 }
Chris@82 317 T13 = VADD(Td, Tf);
Chris@82 318 T14 = VADD(Ti, Tk);
Chris@82 319 T15 = VSUB(T13, T14);
Chris@82 320 Tg = VSUB(Td, Tf);
Chris@82 321 Tl = VSUB(Ti, Tk);
Chris@82 322 Tm = VFMA(LDK(KP382683432), Tg, VMUL(LDK(KP923879532), Tl));
Chris@82 323 TM = VFNMS(LDK(KP382683432), Tl, VMUL(LDK(KP923879532), Tg));
Chris@82 324 }
Chris@82 325 {
Chris@82 326 V T1a, T1g, T1f, T1h;
Chris@82 327 {
Chris@82 328 V T12, T19, T1d, T1e;
Chris@82 329 T12 = VSUB(T10, T11);
Chris@82 330 T19 = VMUL(LDK(KP707106781), VADD(T15, T18));
Chris@82 331 T1a = VADD(T12, T19);
Chris@82 332 T1g = VSUB(T12, T19);
Chris@82 333 T1d = VSUB(T1b, T1c);
Chris@82 334 T1e = VMUL(LDK(KP707106781), VSUB(T18, T15));
Chris@82 335 T1f = VBYI(VADD(T1d, T1e));
Chris@82 336 T1h = VBYI(VSUB(T1e, T1d));
Chris@82 337 }
Chris@82 338 ST(&(x[WS(rs, 14)]), VSUB(T1a, T1f), ms, &(x[0]));
Chris@82 339 ST(&(x[WS(rs, 6)]), VADD(T1g, T1h), ms, &(x[0]));
Chris@82 340 ST(&(x[WS(rs, 2)]), VADD(T1a, T1f), ms, &(x[0]));
Chris@82 341 ST(&(x[WS(rs, 10)]), VSUB(T1g, T1h), ms, &(x[0]));
Chris@82 342 }
Chris@82 343 {
Chris@82 344 V T1k, T1o, T1n, T1p;
Chris@82 345 {
Chris@82 346 V T1i, T1j, T1l, T1m;
Chris@82 347 T1i = VADD(T10, T11);
Chris@82 348 T1j = VADD(T1c, T1b);
Chris@82 349 T1k = VADD(T1i, T1j);
Chris@82 350 T1o = VSUB(T1i, T1j);
Chris@82 351 T1l = VADD(T13, T14);
Chris@82 352 T1m = VADD(T16, T17);
Chris@82 353 T1n = VADD(T1l, T1m);
Chris@82 354 T1p = VBYI(VSUB(T1m, T1l));
Chris@82 355 }
Chris@82 356 ST(&(x[WS(rs, 8)]), VSUB(T1k, T1n), ms, &(x[0]));
Chris@82 357 ST(&(x[WS(rs, 4)]), VADD(T1o, T1p), ms, &(x[0]));
Chris@82 358 ST(&(x[0]), VADD(T1k, T1n), ms, &(x[0]));
Chris@82 359 ST(&(x[WS(rs, 12)]), VSUB(T1o, T1p), ms, &(x[0]));
Chris@82 360 }
Chris@82 361 {
Chris@82 362 V TF, TQ, TP, TR;
Chris@82 363 {
Chris@82 364 V Tn, TE, TL, TO;
Chris@82 365 Tn = VSUB(Tb, Tm);
Chris@82 366 TE = VSUB(Ty, TD);
Chris@82 367 TF = VBYI(VSUB(Tn, TE));
Chris@82 368 TQ = VBYI(VADD(TE, Tn));
Chris@82 369 TL = VADD(TJ, TK);
Chris@82 370 TO = VADD(TM, TN);
Chris@82 371 TP = VSUB(TL, TO);
Chris@82 372 TR = VADD(TL, TO);
Chris@82 373 }
Chris@82 374 ST(&(x[WS(rs, 7)]), VADD(TF, TP), ms, &(x[WS(rs, 1)]));
Chris@82 375 ST(&(x[WS(rs, 15)]), VSUB(TR, TQ), ms, &(x[WS(rs, 1)]));
Chris@82 376 ST(&(x[WS(rs, 9)]), VSUB(TP, TF), ms, &(x[WS(rs, 1)]));
Chris@82 377 ST(&(x[WS(rs, 1)]), VADD(TQ, TR), ms, &(x[WS(rs, 1)]));
Chris@82 378 }
Chris@82 379 {
Chris@82 380 V TU, TY, TX, TZ;
Chris@82 381 {
Chris@82 382 V TS, TT, TV, TW;
Chris@82 383 TS = VSUB(TJ, TK);
Chris@82 384 TT = VADD(Tm, Tb);
Chris@82 385 TU = VADD(TS, TT);
Chris@82 386 TY = VSUB(TS, TT);
Chris@82 387 TV = VADD(TD, Ty);
Chris@82 388 TW = VSUB(TN, TM);
Chris@82 389 TX = VBYI(VADD(TV, TW));
Chris@82 390 TZ = VBYI(VSUB(TW, TV));
Chris@82 391 }
Chris@82 392 ST(&(x[WS(rs, 13)]), VSUB(TU, TX), ms, &(x[WS(rs, 1)]));
Chris@82 393 ST(&(x[WS(rs, 5)]), VADD(TY, TZ), ms, &(x[WS(rs, 1)]));
Chris@82 394 ST(&(x[WS(rs, 3)]), VADD(TU, TX), ms, &(x[WS(rs, 1)]));
Chris@82 395 ST(&(x[WS(rs, 11)]), VSUB(TY, TZ), ms, &(x[WS(rs, 1)]));
Chris@82 396 }
Chris@82 397 }
Chris@82 398 }
Chris@82 399 VLEAVE();
Chris@82 400 }
Chris@82 401
Chris@82 402 static const tw_instr twinstr[] = {
Chris@82 403 VTW(0, 1),
Chris@82 404 VTW(0, 2),
Chris@82 405 VTW(0, 3),
Chris@82 406 VTW(0, 4),
Chris@82 407 VTW(0, 5),
Chris@82 408 VTW(0, 6),
Chris@82 409 VTW(0, 7),
Chris@82 410 VTW(0, 8),
Chris@82 411 VTW(0, 9),
Chris@82 412 VTW(0, 10),
Chris@82 413 VTW(0, 11),
Chris@82 414 VTW(0, 12),
Chris@82 415 VTW(0, 13),
Chris@82 416 VTW(0, 14),
Chris@82 417 VTW(0, 15),
Chris@82 418 {TW_NEXT, VL, 0}
Chris@82 419 };
Chris@82 420
Chris@82 421 static const ct_desc desc = { 16, XSIMD_STRING("t1fv_16"), twinstr, &GENUS, {83, 38, 4, 0}, 0, 0, 0 };
Chris@82 422
Chris@82 423 void XSIMD(codelet_t1fv_16) (planner *p) {
Chris@82 424 X(kdft_dit_register) (p, t1fv_16, &desc);
Chris@82 425 }
Chris@82 426 #endif