Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/t1sv_8.c @ 84:08ae793730bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +0000
parents d0c2a83c1364
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 08:06:09 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.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 8 -name t1sv_8 -include dft/simd/ts.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 66 FP additions, 36 FP multiplications,
Chris@82 32 * (or, 44 additions, 14 multiplications, 22 fused multiply/add),
Chris@82 33 * 34 stack variables, 1 constants, and 32 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/ts.h"
Chris@82 36
Chris@82 37 static void t1sv_8(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 38 {
Chris@82 39 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 40 {
Chris@82 41 INT m;
Chris@82 42 for (m = mb, W = W + (mb * 14); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 14), MAKE_VOLATILE_STRIDE(16, rs)) {
Chris@82 43 V T1, T1m, T7, T1l, Tk, TS, Te, TQ, TF, T14, TL, T16, T12, T17, Ts;
Chris@82 44 V TX, Ty, TZ, TV, T10;
Chris@82 45 T1 = LD(&(ri[0]), ms, &(ri[0]));
Chris@82 46 T1m = LD(&(ii[0]), ms, &(ii[0]));
Chris@82 47 {
Chris@82 48 V T3, T6, T4, T1k, T2, T5;
Chris@82 49 T3 = LD(&(ri[WS(rs, 4)]), ms, &(ri[0]));
Chris@82 50 T6 = LD(&(ii[WS(rs, 4)]), ms, &(ii[0]));
Chris@82 51 T2 = LDW(&(W[TWVL * 6]));
Chris@82 52 T4 = VMUL(T2, T3);
Chris@82 53 T1k = VMUL(T2, T6);
Chris@82 54 T5 = LDW(&(W[TWVL * 7]));
Chris@82 55 T7 = VFMA(T5, T6, T4);
Chris@82 56 T1l = VFNMS(T5, T3, T1k);
Chris@82 57 }
Chris@82 58 {
Chris@82 59 V Tg, Tj, Th, TR, Tf, Ti;
Chris@82 60 Tg = LD(&(ri[WS(rs, 6)]), ms, &(ri[0]));
Chris@82 61 Tj = LD(&(ii[WS(rs, 6)]), ms, &(ii[0]));
Chris@82 62 Tf = LDW(&(W[TWVL * 10]));
Chris@82 63 Th = VMUL(Tf, Tg);
Chris@82 64 TR = VMUL(Tf, Tj);
Chris@82 65 Ti = LDW(&(W[TWVL * 11]));
Chris@82 66 Tk = VFMA(Ti, Tj, Th);
Chris@82 67 TS = VFNMS(Ti, Tg, TR);
Chris@82 68 }
Chris@82 69 {
Chris@82 70 V Ta, Td, Tb, TP, T9, Tc;
Chris@82 71 Ta = LD(&(ri[WS(rs, 2)]), ms, &(ri[0]));
Chris@82 72 Td = LD(&(ii[WS(rs, 2)]), ms, &(ii[0]));
Chris@82 73 T9 = LDW(&(W[TWVL * 2]));
Chris@82 74 Tb = VMUL(T9, Ta);
Chris@82 75 TP = VMUL(T9, Td);
Chris@82 76 Tc = LDW(&(W[TWVL * 3]));
Chris@82 77 Te = VFMA(Tc, Td, Tb);
Chris@82 78 TQ = VFNMS(Tc, Ta, TP);
Chris@82 79 }
Chris@82 80 {
Chris@82 81 V TB, TE, TC, T13, TH, TK, TI, T15, TA, TG, TD, TJ;
Chris@82 82 TB = LD(&(ri[WS(rs, 7)]), ms, &(ri[WS(rs, 1)]));
Chris@82 83 TE = LD(&(ii[WS(rs, 7)]), ms, &(ii[WS(rs, 1)]));
Chris@82 84 TA = LDW(&(W[TWVL * 12]));
Chris@82 85 TC = VMUL(TA, TB);
Chris@82 86 T13 = VMUL(TA, TE);
Chris@82 87 TH = LD(&(ri[WS(rs, 3)]), ms, &(ri[WS(rs, 1)]));
Chris@82 88 TK = LD(&(ii[WS(rs, 3)]), ms, &(ii[WS(rs, 1)]));
Chris@82 89 TG = LDW(&(W[TWVL * 4]));
Chris@82 90 TI = VMUL(TG, TH);
Chris@82 91 T15 = VMUL(TG, TK);
Chris@82 92 TD = LDW(&(W[TWVL * 13]));
Chris@82 93 TF = VFMA(TD, TE, TC);
Chris@82 94 T14 = VFNMS(TD, TB, T13);
Chris@82 95 TJ = LDW(&(W[TWVL * 5]));
Chris@82 96 TL = VFMA(TJ, TK, TI);
Chris@82 97 T16 = VFNMS(TJ, TH, T15);
Chris@82 98 T12 = VSUB(TF, TL);
Chris@82 99 T17 = VSUB(T14, T16);
Chris@82 100 }
Chris@82 101 {
Chris@82 102 V To, Tr, Tp, TW, Tu, Tx, Tv, TY, Tn, Tt, Tq, Tw;
Chris@82 103 To = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
Chris@82 104 Tr = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
Chris@82 105 Tn = LDW(&(W[0]));
Chris@82 106 Tp = VMUL(Tn, To);
Chris@82 107 TW = VMUL(Tn, Tr);
Chris@82 108 Tu = LD(&(ri[WS(rs, 5)]), ms, &(ri[WS(rs, 1)]));
Chris@82 109 Tx = LD(&(ii[WS(rs, 5)]), ms, &(ii[WS(rs, 1)]));
Chris@82 110 Tt = LDW(&(W[TWVL * 8]));
Chris@82 111 Tv = VMUL(Tt, Tu);
Chris@82 112 TY = VMUL(Tt, Tx);
Chris@82 113 Tq = LDW(&(W[TWVL * 1]));
Chris@82 114 Ts = VFMA(Tq, Tr, Tp);
Chris@82 115 TX = VFNMS(Tq, To, TW);
Chris@82 116 Tw = LDW(&(W[TWVL * 9]));
Chris@82 117 Ty = VFMA(Tw, Tx, Tv);
Chris@82 118 TZ = VFNMS(Tw, Tu, TY);
Chris@82 119 TV = VSUB(Ts, Ty);
Chris@82 120 T10 = VSUB(TX, TZ);
Chris@82 121 }
Chris@82 122 {
Chris@82 123 V TU, T1a, T1t, T1v, T19, T1w, T1d, T1u;
Chris@82 124 {
Chris@82 125 V TO, TT, T1r, T1s;
Chris@82 126 TO = VSUB(T1, T7);
Chris@82 127 TT = VSUB(TQ, TS);
Chris@82 128 TU = VADD(TO, TT);
Chris@82 129 T1a = VSUB(TO, TT);
Chris@82 130 T1r = VSUB(T1m, T1l);
Chris@82 131 T1s = VSUB(Te, Tk);
Chris@82 132 T1t = VSUB(T1r, T1s);
Chris@82 133 T1v = VADD(T1s, T1r);
Chris@82 134 }
Chris@82 135 {
Chris@82 136 V T11, T18, T1b, T1c;
Chris@82 137 T11 = VADD(TV, T10);
Chris@82 138 T18 = VSUB(T12, T17);
Chris@82 139 T19 = VADD(T11, T18);
Chris@82 140 T1w = VSUB(T18, T11);
Chris@82 141 T1b = VSUB(T10, TV);
Chris@82 142 T1c = VADD(T12, T17);
Chris@82 143 T1d = VSUB(T1b, T1c);
Chris@82 144 T1u = VADD(T1b, T1c);
Chris@82 145 }
Chris@82 146 ST(&(ri[WS(rs, 5)]), VFNMS(LDK(KP707106781), T19, TU), ms, &(ri[WS(rs, 1)]));
Chris@82 147 ST(&(ii[WS(rs, 5)]), VFNMS(LDK(KP707106781), T1u, T1t), ms, &(ii[WS(rs, 1)]));
Chris@82 148 ST(&(ri[WS(rs, 1)]), VFMA(LDK(KP707106781), T19, TU), ms, &(ri[WS(rs, 1)]));
Chris@82 149 ST(&(ii[WS(rs, 1)]), VFMA(LDK(KP707106781), T1u, T1t), ms, &(ii[WS(rs, 1)]));
Chris@82 150 ST(&(ri[WS(rs, 7)]), VFNMS(LDK(KP707106781), T1d, T1a), ms, &(ri[WS(rs, 1)]));
Chris@82 151 ST(&(ii[WS(rs, 7)]), VFNMS(LDK(KP707106781), T1w, T1v), ms, &(ii[WS(rs, 1)]));
Chris@82 152 ST(&(ri[WS(rs, 3)]), VFMA(LDK(KP707106781), T1d, T1a), ms, &(ri[WS(rs, 1)]));
Chris@82 153 ST(&(ii[WS(rs, 3)]), VFMA(LDK(KP707106781), T1w, T1v), ms, &(ii[WS(rs, 1)]));
Chris@82 154 }
Chris@82 155 {
Chris@82 156 V Tm, T1e, T1o, T1q, TN, T1p, T1h, T1i;
Chris@82 157 {
Chris@82 158 V T8, Tl, T1j, T1n;
Chris@82 159 T8 = VADD(T1, T7);
Chris@82 160 Tl = VADD(Te, Tk);
Chris@82 161 Tm = VADD(T8, Tl);
Chris@82 162 T1e = VSUB(T8, Tl);
Chris@82 163 T1j = VADD(TQ, TS);
Chris@82 164 T1n = VADD(T1l, T1m);
Chris@82 165 T1o = VADD(T1j, T1n);
Chris@82 166 T1q = VSUB(T1n, T1j);
Chris@82 167 }
Chris@82 168 {
Chris@82 169 V Tz, TM, T1f, T1g;
Chris@82 170 Tz = VADD(Ts, Ty);
Chris@82 171 TM = VADD(TF, TL);
Chris@82 172 TN = VADD(Tz, TM);
Chris@82 173 T1p = VSUB(TM, Tz);
Chris@82 174 T1f = VADD(TX, TZ);
Chris@82 175 T1g = VADD(T14, T16);
Chris@82 176 T1h = VSUB(T1f, T1g);
Chris@82 177 T1i = VADD(T1f, T1g);
Chris@82 178 }
Chris@82 179 ST(&(ri[WS(rs, 4)]), VSUB(Tm, TN), ms, &(ri[0]));
Chris@82 180 ST(&(ii[WS(rs, 4)]), VSUB(T1o, T1i), ms, &(ii[0]));
Chris@82 181 ST(&(ri[0]), VADD(Tm, TN), ms, &(ri[0]));
Chris@82 182 ST(&(ii[0]), VADD(T1i, T1o), ms, &(ii[0]));
Chris@82 183 ST(&(ri[WS(rs, 6)]), VSUB(T1e, T1h), ms, &(ri[0]));
Chris@82 184 ST(&(ii[WS(rs, 6)]), VSUB(T1q, T1p), ms, &(ii[0]));
Chris@82 185 ST(&(ri[WS(rs, 2)]), VADD(T1e, T1h), ms, &(ri[0]));
Chris@82 186 ST(&(ii[WS(rs, 2)]), VADD(T1p, T1q), ms, &(ii[0]));
Chris@82 187 }
Chris@82 188 }
Chris@82 189 }
Chris@82 190 VLEAVE();
Chris@82 191 }
Chris@82 192
Chris@82 193 static const tw_instr twinstr[] = {
Chris@82 194 VTW(0, 1),
Chris@82 195 VTW(0, 2),
Chris@82 196 VTW(0, 3),
Chris@82 197 VTW(0, 4),
Chris@82 198 VTW(0, 5),
Chris@82 199 VTW(0, 6),
Chris@82 200 VTW(0, 7),
Chris@82 201 {TW_NEXT, (2 * VL), 0}
Chris@82 202 };
Chris@82 203
Chris@82 204 static const ct_desc desc = { 8, XSIMD_STRING("t1sv_8"), twinstr, &GENUS, {44, 14, 22, 0}, 0, 0, 0 };
Chris@82 205
Chris@82 206 void XSIMD(codelet_t1sv_8) (planner *p) {
Chris@82 207 X(kdft_dit_register) (p, t1sv_8, &desc);
Chris@82 208 }
Chris@82 209 #else
Chris@82 210
Chris@82 211 /* Generated by: ../../../genfft/gen_twiddle.native -simd -compact -variables 4 -pipeline-latency 8 -n 8 -name t1sv_8 -include dft/simd/ts.h */
Chris@82 212
Chris@82 213 /*
Chris@82 214 * This function contains 66 FP additions, 32 FP multiplications,
Chris@82 215 * (or, 52 additions, 18 multiplications, 14 fused multiply/add),
Chris@82 216 * 28 stack variables, 1 constants, and 32 memory accesses
Chris@82 217 */
Chris@82 218 #include "dft/simd/ts.h"
Chris@82 219
Chris@82 220 static void t1sv_8(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 221 {
Chris@82 222 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 223 {
Chris@82 224 INT m;
Chris@82 225 for (m = mb, W = W + (mb * 14); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 14), MAKE_VOLATILE_STRIDE(16, rs)) {
Chris@82 226 V T7, T1e, TH, T19, TF, T13, TR, TU, Ti, T1f, TK, T16, Tu, T12, TM;
Chris@82 227 V TP;
Chris@82 228 {
Chris@82 229 V T1, T18, T6, T17;
Chris@82 230 T1 = LD(&(ri[0]), ms, &(ri[0]));
Chris@82 231 T18 = LD(&(ii[0]), ms, &(ii[0]));
Chris@82 232 {
Chris@82 233 V T3, T5, T2, T4;
Chris@82 234 T3 = LD(&(ri[WS(rs, 4)]), ms, &(ri[0]));
Chris@82 235 T5 = LD(&(ii[WS(rs, 4)]), ms, &(ii[0]));
Chris@82 236 T2 = LDW(&(W[TWVL * 6]));
Chris@82 237 T4 = LDW(&(W[TWVL * 7]));
Chris@82 238 T6 = VFMA(T2, T3, VMUL(T4, T5));
Chris@82 239 T17 = VFNMS(T4, T3, VMUL(T2, T5));
Chris@82 240 }
Chris@82 241 T7 = VADD(T1, T6);
Chris@82 242 T1e = VSUB(T18, T17);
Chris@82 243 TH = VSUB(T1, T6);
Chris@82 244 T19 = VADD(T17, T18);
Chris@82 245 }
Chris@82 246 {
Chris@82 247 V Tz, TS, TE, TT;
Chris@82 248 {
Chris@82 249 V Tw, Ty, Tv, Tx;
Chris@82 250 Tw = LD(&(ri[WS(rs, 7)]), ms, &(ri[WS(rs, 1)]));
Chris@82 251 Ty = LD(&(ii[WS(rs, 7)]), ms, &(ii[WS(rs, 1)]));
Chris@82 252 Tv = LDW(&(W[TWVL * 12]));
Chris@82 253 Tx = LDW(&(W[TWVL * 13]));
Chris@82 254 Tz = VFMA(Tv, Tw, VMUL(Tx, Ty));
Chris@82 255 TS = VFNMS(Tx, Tw, VMUL(Tv, Ty));
Chris@82 256 }
Chris@82 257 {
Chris@82 258 V TB, TD, TA, TC;
Chris@82 259 TB = LD(&(ri[WS(rs, 3)]), ms, &(ri[WS(rs, 1)]));
Chris@82 260 TD = LD(&(ii[WS(rs, 3)]), ms, &(ii[WS(rs, 1)]));
Chris@82 261 TA = LDW(&(W[TWVL * 4]));
Chris@82 262 TC = LDW(&(W[TWVL * 5]));
Chris@82 263 TE = VFMA(TA, TB, VMUL(TC, TD));
Chris@82 264 TT = VFNMS(TC, TB, VMUL(TA, TD));
Chris@82 265 }
Chris@82 266 TF = VADD(Tz, TE);
Chris@82 267 T13 = VADD(TS, TT);
Chris@82 268 TR = VSUB(Tz, TE);
Chris@82 269 TU = VSUB(TS, TT);
Chris@82 270 }
Chris@82 271 {
Chris@82 272 V Tc, TI, Th, TJ;
Chris@82 273 {
Chris@82 274 V T9, Tb, T8, Ta;
Chris@82 275 T9 = LD(&(ri[WS(rs, 2)]), ms, &(ri[0]));
Chris@82 276 Tb = LD(&(ii[WS(rs, 2)]), ms, &(ii[0]));
Chris@82 277 T8 = LDW(&(W[TWVL * 2]));
Chris@82 278 Ta = LDW(&(W[TWVL * 3]));
Chris@82 279 Tc = VFMA(T8, T9, VMUL(Ta, Tb));
Chris@82 280 TI = VFNMS(Ta, T9, VMUL(T8, Tb));
Chris@82 281 }
Chris@82 282 {
Chris@82 283 V Te, Tg, Td, Tf;
Chris@82 284 Te = LD(&(ri[WS(rs, 6)]), ms, &(ri[0]));
Chris@82 285 Tg = LD(&(ii[WS(rs, 6)]), ms, &(ii[0]));
Chris@82 286 Td = LDW(&(W[TWVL * 10]));
Chris@82 287 Tf = LDW(&(W[TWVL * 11]));
Chris@82 288 Th = VFMA(Td, Te, VMUL(Tf, Tg));
Chris@82 289 TJ = VFNMS(Tf, Te, VMUL(Td, Tg));
Chris@82 290 }
Chris@82 291 Ti = VADD(Tc, Th);
Chris@82 292 T1f = VSUB(Tc, Th);
Chris@82 293 TK = VSUB(TI, TJ);
Chris@82 294 T16 = VADD(TI, TJ);
Chris@82 295 }
Chris@82 296 {
Chris@82 297 V To, TN, Tt, TO;
Chris@82 298 {
Chris@82 299 V Tl, Tn, Tk, Tm;
Chris@82 300 Tl = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
Chris@82 301 Tn = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
Chris@82 302 Tk = LDW(&(W[0]));
Chris@82 303 Tm = LDW(&(W[TWVL * 1]));
Chris@82 304 To = VFMA(Tk, Tl, VMUL(Tm, Tn));
Chris@82 305 TN = VFNMS(Tm, Tl, VMUL(Tk, Tn));
Chris@82 306 }
Chris@82 307 {
Chris@82 308 V Tq, Ts, Tp, Tr;
Chris@82 309 Tq = LD(&(ri[WS(rs, 5)]), ms, &(ri[WS(rs, 1)]));
Chris@82 310 Ts = LD(&(ii[WS(rs, 5)]), ms, &(ii[WS(rs, 1)]));
Chris@82 311 Tp = LDW(&(W[TWVL * 8]));
Chris@82 312 Tr = LDW(&(W[TWVL * 9]));
Chris@82 313 Tt = VFMA(Tp, Tq, VMUL(Tr, Ts));
Chris@82 314 TO = VFNMS(Tr, Tq, VMUL(Tp, Ts));
Chris@82 315 }
Chris@82 316 Tu = VADD(To, Tt);
Chris@82 317 T12 = VADD(TN, TO);
Chris@82 318 TM = VSUB(To, Tt);
Chris@82 319 TP = VSUB(TN, TO);
Chris@82 320 }
Chris@82 321 {
Chris@82 322 V Tj, TG, T1b, T1c;
Chris@82 323 Tj = VADD(T7, Ti);
Chris@82 324 TG = VADD(Tu, TF);
Chris@82 325 ST(&(ri[WS(rs, 4)]), VSUB(Tj, TG), ms, &(ri[0]));
Chris@82 326 ST(&(ri[0]), VADD(Tj, TG), ms, &(ri[0]));
Chris@82 327 {
Chris@82 328 V T15, T1a, T11, T14;
Chris@82 329 T15 = VADD(T12, T13);
Chris@82 330 T1a = VADD(T16, T19);
Chris@82 331 ST(&(ii[0]), VADD(T15, T1a), ms, &(ii[0]));
Chris@82 332 ST(&(ii[WS(rs, 4)]), VSUB(T1a, T15), ms, &(ii[0]));
Chris@82 333 T11 = VSUB(T7, Ti);
Chris@82 334 T14 = VSUB(T12, T13);
Chris@82 335 ST(&(ri[WS(rs, 6)]), VSUB(T11, T14), ms, &(ri[0]));
Chris@82 336 ST(&(ri[WS(rs, 2)]), VADD(T11, T14), ms, &(ri[0]));
Chris@82 337 }
Chris@82 338 T1b = VSUB(TF, Tu);
Chris@82 339 T1c = VSUB(T19, T16);
Chris@82 340 ST(&(ii[WS(rs, 2)]), VADD(T1b, T1c), ms, &(ii[0]));
Chris@82 341 ST(&(ii[WS(rs, 6)]), VSUB(T1c, T1b), ms, &(ii[0]));
Chris@82 342 {
Chris@82 343 V TX, T1g, T10, T1d, TY, TZ;
Chris@82 344 TX = VSUB(TH, TK);
Chris@82 345 T1g = VSUB(T1e, T1f);
Chris@82 346 TY = VSUB(TP, TM);
Chris@82 347 TZ = VADD(TR, TU);
Chris@82 348 T10 = VMUL(LDK(KP707106781), VSUB(TY, TZ));
Chris@82 349 T1d = VMUL(LDK(KP707106781), VADD(TY, TZ));
Chris@82 350 ST(&(ri[WS(rs, 7)]), VSUB(TX, T10), ms, &(ri[WS(rs, 1)]));
Chris@82 351 ST(&(ii[WS(rs, 5)]), VSUB(T1g, T1d), ms, &(ii[WS(rs, 1)]));
Chris@82 352 ST(&(ri[WS(rs, 3)]), VADD(TX, T10), ms, &(ri[WS(rs, 1)]));
Chris@82 353 ST(&(ii[WS(rs, 1)]), VADD(T1d, T1g), ms, &(ii[WS(rs, 1)]));
Chris@82 354 }
Chris@82 355 {
Chris@82 356 V TL, T1i, TW, T1h, TQ, TV;
Chris@82 357 TL = VADD(TH, TK);
Chris@82 358 T1i = VADD(T1f, T1e);
Chris@82 359 TQ = VADD(TM, TP);
Chris@82 360 TV = VSUB(TR, TU);
Chris@82 361 TW = VMUL(LDK(KP707106781), VADD(TQ, TV));
Chris@82 362 T1h = VMUL(LDK(KP707106781), VSUB(TV, TQ));
Chris@82 363 ST(&(ri[WS(rs, 5)]), VSUB(TL, TW), ms, &(ri[WS(rs, 1)]));
Chris@82 364 ST(&(ii[WS(rs, 7)]), VSUB(T1i, T1h), ms, &(ii[WS(rs, 1)]));
Chris@82 365 ST(&(ri[WS(rs, 1)]), VADD(TL, TW), ms, &(ri[WS(rs, 1)]));
Chris@82 366 ST(&(ii[WS(rs, 3)]), VADD(T1h, T1i), ms, &(ii[WS(rs, 1)]));
Chris@82 367 }
Chris@82 368 }
Chris@82 369 }
Chris@82 370 }
Chris@82 371 VLEAVE();
Chris@82 372 }
Chris@82 373
Chris@82 374 static const tw_instr twinstr[] = {
Chris@82 375 VTW(0, 1),
Chris@82 376 VTW(0, 2),
Chris@82 377 VTW(0, 3),
Chris@82 378 VTW(0, 4),
Chris@82 379 VTW(0, 5),
Chris@82 380 VTW(0, 6),
Chris@82 381 VTW(0, 7),
Chris@82 382 {TW_NEXT, (2 * VL), 0}
Chris@82 383 };
Chris@82 384
Chris@82 385 static const ct_desc desc = { 8, XSIMD_STRING("t1sv_8"), twinstr, &GENUS, {52, 18, 14, 0}, 0, 0, 0 };
Chris@82 386
Chris@82 387 void XSIMD(codelet_t1sv_8) (planner *p) {
Chris@82 388 X(kdft_dit_register) (p, t1sv_8, &desc);
Chris@82 389 }
Chris@82 390 #endif