Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/n2bv_20.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 08:05:19 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_notw_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 20 -name n2bv_20 -with-ostride 2 -include dft/simd/n2b.h -store-multiple 2 */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 104 FP additions, 50 FP multiplications,
Chris@82 32 * (or, 58 additions, 4 multiplications, 46 fused multiply/add),
Chris@82 33 * 57 stack variables, 4 constants, and 50 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/n2b.h"
Chris@82 36
Chris@82 37 static void n2bv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 38 {
Chris@82 39 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 40 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82 41 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 42 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 43 {
Chris@82 44 INT i;
Chris@82 45 const R *xi;
Chris@82 46 R *xo;
Chris@82 47 xi = ii;
Chris@82 48 xo = io;
Chris@82 49 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(40, is), MAKE_VOLATILE_STRIDE(40, os)) {
Chris@82 50 V T3, T1r, TE, T13, Ts, TL, TM, Tz, T16, T19, T1a, T1v, T1w, T1x, T1s;
Chris@82 51 V T1t, T1u, T1d, T1g, T1h, Ti, Tk, TH, TJ;
Chris@82 52 {
Chris@82 53 V T1, T2, T11, TC, TD, T12;
Chris@82 54 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 55 T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
Chris@82 56 T11 = VADD(T1, T2);
Chris@82 57 TC = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82 58 TD = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
Chris@82 59 T12 = VADD(TC, TD);
Chris@82 60 T3 = VSUB(T1, T2);
Chris@82 61 T1r = VADD(T11, T12);
Chris@82 62 TE = VSUB(TC, TD);
Chris@82 63 T13 = VSUB(T11, T12);
Chris@82 64 }
Chris@82 65 {
Chris@82 66 V T6, T14, Tv, T1c, Ty, T1f, T9, T17, Td, T1b, To, T15, Tr, T18, Tg;
Chris@82 67 V T1e;
Chris@82 68 {
Chris@82 69 V T4, T5, Tt, Tu;
Chris@82 70 T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82 71 T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
Chris@82 72 T6 = VSUB(T4, T5);
Chris@82 73 T14 = VADD(T4, T5);
Chris@82 74 Tt = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
Chris@82 75 Tu = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 76 Tv = VSUB(Tt, Tu);
Chris@82 77 T1c = VADD(Tt, Tu);
Chris@82 78 }
Chris@82 79 {
Chris@82 80 V Tw, Tx, T7, T8;
Chris@82 81 Tw = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)]));
Chris@82 82 Tx = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@82 83 Ty = VSUB(Tw, Tx);
Chris@82 84 T1f = VADD(Tw, Tx);
Chris@82 85 T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0]));
Chris@82 86 T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82 87 T9 = VSUB(T7, T8);
Chris@82 88 T17 = VADD(T7, T8);
Chris@82 89 }
Chris@82 90 {
Chris@82 91 V Tb, Tc, Tm, Tn;
Chris@82 92 Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@82 93 Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0]));
Chris@82 94 Td = VSUB(Tb, Tc);
Chris@82 95 T1b = VADD(Tb, Tc);
Chris@82 96 Tm = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
Chris@82 97 Tn = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)]));
Chris@82 98 To = VSUB(Tm, Tn);
Chris@82 99 T15 = VADD(Tm, Tn);
Chris@82 100 }
Chris@82 101 {
Chris@82 102 V Tp, Tq, Te, Tf;
Chris@82 103 Tp = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 104 Tq = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
Chris@82 105 Tr = VSUB(Tp, Tq);
Chris@82 106 T18 = VADD(Tp, Tq);
Chris@82 107 Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
Chris@82 108 Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 109 Tg = VSUB(Te, Tf);
Chris@82 110 T1e = VADD(Te, Tf);
Chris@82 111 }
Chris@82 112 Ts = VSUB(To, Tr);
Chris@82 113 TL = VSUB(T6, T9);
Chris@82 114 TM = VSUB(Td, Tg);
Chris@82 115 Tz = VSUB(Tv, Ty);
Chris@82 116 T16 = VSUB(T14, T15);
Chris@82 117 T19 = VSUB(T17, T18);
Chris@82 118 T1a = VADD(T16, T19);
Chris@82 119 T1v = VADD(T1b, T1c);
Chris@82 120 T1w = VADD(T1e, T1f);
Chris@82 121 T1x = VADD(T1v, T1w);
Chris@82 122 T1s = VADD(T14, T15);
Chris@82 123 T1t = VADD(T17, T18);
Chris@82 124 T1u = VADD(T1s, T1t);
Chris@82 125 T1d = VSUB(T1b, T1c);
Chris@82 126 T1g = VSUB(T1e, T1f);
Chris@82 127 T1h = VADD(T1d, T1g);
Chris@82 128 {
Chris@82 129 V Ta, Th, TF, TG;
Chris@82 130 Ta = VADD(T6, T9);
Chris@82 131 Th = VADD(Td, Tg);
Chris@82 132 Ti = VADD(Ta, Th);
Chris@82 133 Tk = VSUB(Ta, Th);
Chris@82 134 TF = VADD(To, Tr);
Chris@82 135 TG = VADD(Tv, Ty);
Chris@82 136 TH = VADD(TF, TG);
Chris@82 137 TJ = VSUB(TF, TG);
Chris@82 138 }
Chris@82 139 }
Chris@82 140 {
Chris@82 141 V T1H, T1J, T1K, T1L, T1N, T1I, TZ, T10;
Chris@82 142 TZ = VADD(T3, Ti);
Chris@82 143 T10 = VADD(TE, TH);
Chris@82 144 T1H = VFNMSI(T10, TZ);
Chris@82 145 STM2(&(xo[30]), T1H, ovs, &(xo[2]));
Chris@82 146 T1I = VFMAI(T10, TZ);
Chris@82 147 STM2(&(xo[10]), T1I, ovs, &(xo[2]));
Chris@82 148 {
Chris@82 149 V T1A, T1y, T1z, T1E, T1G, T1C, T1D, T1F, T1B, T1M;
Chris@82 150 T1A = VSUB(T1u, T1x);
Chris@82 151 T1y = VADD(T1u, T1x);
Chris@82 152 T1z = VFNMS(LDK(KP250000000), T1y, T1r);
Chris@82 153 T1C = VSUB(T1s, T1t);
Chris@82 154 T1D = VSUB(T1v, T1w);
Chris@82 155 T1E = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1D, T1C));
Chris@82 156 T1G = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1C, T1D));
Chris@82 157 T1J = VADD(T1r, T1y);
Chris@82 158 STM2(&(xo[0]), T1J, ovs, &(xo[0]));
Chris@82 159 T1F = VFNMS(LDK(KP559016994), T1A, T1z);
Chris@82 160 T1K = VFMAI(T1G, T1F);
Chris@82 161 STM2(&(xo[16]), T1K, ovs, &(xo[0]));
Chris@82 162 T1L = VFNMSI(T1G, T1F);
Chris@82 163 STM2(&(xo[24]), T1L, ovs, &(xo[0]));
Chris@82 164 T1B = VFMA(LDK(KP559016994), T1A, T1z);
Chris@82 165 T1M = VFNMSI(T1E, T1B);
Chris@82 166 STM2(&(xo[8]), T1M, ovs, &(xo[0]));
Chris@82 167 STN2(&(xo[8]), T1M, T1I, ovs);
Chris@82 168 T1N = VFMAI(T1E, T1B);
Chris@82 169 STM2(&(xo[32]), T1N, ovs, &(xo[0]));
Chris@82 170 }
Chris@82 171 {
Chris@82 172 V T1O, T1P, T1R, T1S;
Chris@82 173 {
Chris@82 174 V T1k, T1i, T1j, T1o, T1q, T1m, T1n, T1p, T1Q, T1l;
Chris@82 175 T1k = VSUB(T1a, T1h);
Chris@82 176 T1i = VADD(T1a, T1h);
Chris@82 177 T1j = VFNMS(LDK(KP250000000), T1i, T13);
Chris@82 178 T1m = VSUB(T1d, T1g);
Chris@82 179 T1n = VSUB(T16, T19);
Chris@82 180 T1o = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1n, T1m));
Chris@82 181 T1q = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1m, T1n));
Chris@82 182 T1O = VADD(T13, T1i);
Chris@82 183 STM2(&(xo[20]), T1O, ovs, &(xo[0]));
Chris@82 184 T1p = VFMA(LDK(KP559016994), T1k, T1j);
Chris@82 185 T1P = VFMAI(T1q, T1p);
Chris@82 186 STM2(&(xo[12]), T1P, ovs, &(xo[0]));
Chris@82 187 T1Q = VFNMSI(T1q, T1p);
Chris@82 188 STM2(&(xo[28]), T1Q, ovs, &(xo[0]));
Chris@82 189 STN2(&(xo[28]), T1Q, T1H, ovs);
Chris@82 190 T1l = VFNMS(LDK(KP559016994), T1k, T1j);
Chris@82 191 T1R = VFNMSI(T1o, T1l);
Chris@82 192 STM2(&(xo[4]), T1R, ovs, &(xo[0]));
Chris@82 193 T1S = VFMAI(T1o, T1l);
Chris@82 194 STM2(&(xo[36]), T1S, ovs, &(xo[0]));
Chris@82 195 }
Chris@82 196 {
Chris@82 197 V TA, TN, TV, TS, TK, TU, Tl, TR, TI, Tj;
Chris@82 198 TA = VFMA(LDK(KP618033988), Tz, Ts);
Chris@82 199 TN = VFMA(LDK(KP618033988), TM, TL);
Chris@82 200 TV = VFNMS(LDK(KP618033988), TL, TM);
Chris@82 201 TS = VFNMS(LDK(KP618033988), Ts, Tz);
Chris@82 202 TI = VFNMS(LDK(KP250000000), TH, TE);
Chris@82 203 TK = VFMA(LDK(KP559016994), TJ, TI);
Chris@82 204 TU = VFNMS(LDK(KP559016994), TJ, TI);
Chris@82 205 Tj = VFNMS(LDK(KP250000000), Ti, T3);
Chris@82 206 Tl = VFMA(LDK(KP559016994), Tk, Tj);
Chris@82 207 TR = VFNMS(LDK(KP559016994), Tk, Tj);
Chris@82 208 {
Chris@82 209 V TB, TO, T1T, T1U;
Chris@82 210 TB = VFNMS(LDK(KP951056516), TA, Tl);
Chris@82 211 TO = VFMA(LDK(KP951056516), TN, TK);
Chris@82 212 T1T = VFNMSI(TO, TB);
Chris@82 213 STM2(&(xo[38]), T1T, ovs, &(xo[2]));
Chris@82 214 STN2(&(xo[36]), T1S, T1T, ovs);
Chris@82 215 T1U = VFMAI(TO, TB);
Chris@82 216 STM2(&(xo[2]), T1U, ovs, &(xo[2]));
Chris@82 217 STN2(&(xo[0]), T1J, T1U, ovs);
Chris@82 218 }
Chris@82 219 {
Chris@82 220 V TX, TY, T1V, T1W;
Chris@82 221 TX = VFNMS(LDK(KP951056516), TS, TR);
Chris@82 222 TY = VFMA(LDK(KP951056516), TV, TU);
Chris@82 223 T1V = VFNMSI(TY, TX);
Chris@82 224 STM2(&(xo[14]), T1V, ovs, &(xo[2]));
Chris@82 225 STN2(&(xo[12]), T1P, T1V, ovs);
Chris@82 226 T1W = VFMAI(TY, TX);
Chris@82 227 STM2(&(xo[26]), T1W, ovs, &(xo[2]));
Chris@82 228 STN2(&(xo[24]), T1L, T1W, ovs);
Chris@82 229 }
Chris@82 230 {
Chris@82 231 V TP, TQ, T1X, T1Y;
Chris@82 232 TP = VFMA(LDK(KP951056516), TA, Tl);
Chris@82 233 TQ = VFNMS(LDK(KP951056516), TN, TK);
Chris@82 234 T1X = VFNMSI(TQ, TP);
Chris@82 235 STM2(&(xo[22]), T1X, ovs, &(xo[2]));
Chris@82 236 STN2(&(xo[20]), T1O, T1X, ovs);
Chris@82 237 T1Y = VFMAI(TQ, TP);
Chris@82 238 STM2(&(xo[18]), T1Y, ovs, &(xo[2]));
Chris@82 239 STN2(&(xo[16]), T1K, T1Y, ovs);
Chris@82 240 }
Chris@82 241 {
Chris@82 242 V TT, TW, T1Z, T20;
Chris@82 243 TT = VFMA(LDK(KP951056516), TS, TR);
Chris@82 244 TW = VFNMS(LDK(KP951056516), TV, TU);
Chris@82 245 T1Z = VFNMSI(TW, TT);
Chris@82 246 STM2(&(xo[6]), T1Z, ovs, &(xo[2]));
Chris@82 247 STN2(&(xo[4]), T1R, T1Z, ovs);
Chris@82 248 T20 = VFMAI(TW, TT);
Chris@82 249 STM2(&(xo[34]), T20, ovs, &(xo[2]));
Chris@82 250 STN2(&(xo[32]), T1N, T20, ovs);
Chris@82 251 }
Chris@82 252 }
Chris@82 253 }
Chris@82 254 }
Chris@82 255 }
Chris@82 256 }
Chris@82 257 VLEAVE();
Chris@82 258 }
Chris@82 259
Chris@82 260 static const kdft_desc desc = { 20, XSIMD_STRING("n2bv_20"), {58, 4, 46, 0}, &GENUS, 0, 2, 0, 0 };
Chris@82 261
Chris@82 262 void XSIMD(codelet_n2bv_20) (planner *p) {
Chris@82 263 X(kdft_register) (p, n2bv_20, &desc);
Chris@82 264 }
Chris@82 265
Chris@82 266 #else
Chris@82 267
Chris@82 268 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 20 -name n2bv_20 -with-ostride 2 -include dft/simd/n2b.h -store-multiple 2 */
Chris@82 269
Chris@82 270 /*
Chris@82 271 * This function contains 104 FP additions, 24 FP multiplications,
Chris@82 272 * (or, 92 additions, 12 multiplications, 12 fused multiply/add),
Chris@82 273 * 57 stack variables, 4 constants, and 50 memory accesses
Chris@82 274 */
Chris@82 275 #include "dft/simd/n2b.h"
Chris@82 276
Chris@82 277 static void n2bv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82 278 {
Chris@82 279 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@82 280 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 281 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 282 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 283 {
Chris@82 284 INT i;
Chris@82 285 const R *xi;
Chris@82 286 R *xo;
Chris@82 287 xi = ii;
Chris@82 288 xo = io;
Chris@82 289 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(40, is), MAKE_VOLATILE_STRIDE(40, os)) {
Chris@82 290 V T3, T1y, TH, T1i, Ts, TL, TM, Tz, T13, T16, T1j, T1u, T1v, T1w, T1r;
Chris@82 291 V T1s, T1t, T1a, T1d, T1k, Ti, Tk, TE, TI;
Chris@82 292 {
Chris@82 293 V T1, T2, T1g, TF, TG, T1h;
Chris@82 294 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 295 T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
Chris@82 296 T1g = VADD(T1, T2);
Chris@82 297 TF = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82 298 TG = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
Chris@82 299 T1h = VADD(TF, TG);
Chris@82 300 T3 = VSUB(T1, T2);
Chris@82 301 T1y = VADD(T1g, T1h);
Chris@82 302 TH = VSUB(TF, TG);
Chris@82 303 T1i = VSUB(T1g, T1h);
Chris@82 304 }
Chris@82 305 {
Chris@82 306 V T6, T11, Tv, T19, Ty, T1c, T9, T14, Td, T18, To, T12, Tr, T15, Tg;
Chris@82 307 V T1b;
Chris@82 308 {
Chris@82 309 V T4, T5, Tt, Tu;
Chris@82 310 T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82 311 T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
Chris@82 312 T6 = VSUB(T4, T5);
Chris@82 313 T11 = VADD(T4, T5);
Chris@82 314 Tt = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
Chris@82 315 Tu = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 316 Tv = VSUB(Tt, Tu);
Chris@82 317 T19 = VADD(Tt, Tu);
Chris@82 318 }
Chris@82 319 {
Chris@82 320 V Tw, Tx, T7, T8;
Chris@82 321 Tw = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)]));
Chris@82 322 Tx = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@82 323 Ty = VSUB(Tw, Tx);
Chris@82 324 T1c = VADD(Tw, Tx);
Chris@82 325 T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0]));
Chris@82 326 T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82 327 T9 = VSUB(T7, T8);
Chris@82 328 T14 = VADD(T7, T8);
Chris@82 329 }
Chris@82 330 {
Chris@82 331 V Tb, Tc, Tm, Tn;
Chris@82 332 Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@82 333 Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0]));
Chris@82 334 Td = VSUB(Tb, Tc);
Chris@82 335 T18 = VADD(Tb, Tc);
Chris@82 336 Tm = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
Chris@82 337 Tn = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)]));
Chris@82 338 To = VSUB(Tm, Tn);
Chris@82 339 T12 = VADD(Tm, Tn);
Chris@82 340 }
Chris@82 341 {
Chris@82 342 V Tp, Tq, Te, Tf;
Chris@82 343 Tp = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 344 Tq = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
Chris@82 345 Tr = VSUB(Tp, Tq);
Chris@82 346 T15 = VADD(Tp, Tq);
Chris@82 347 Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
Chris@82 348 Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 349 Tg = VSUB(Te, Tf);
Chris@82 350 T1b = VADD(Te, Tf);
Chris@82 351 }
Chris@82 352 Ts = VSUB(To, Tr);
Chris@82 353 TL = VSUB(T6, T9);
Chris@82 354 TM = VSUB(Td, Tg);
Chris@82 355 Tz = VSUB(Tv, Ty);
Chris@82 356 T13 = VSUB(T11, T12);
Chris@82 357 T16 = VSUB(T14, T15);
Chris@82 358 T1j = VADD(T13, T16);
Chris@82 359 T1u = VADD(T18, T19);
Chris@82 360 T1v = VADD(T1b, T1c);
Chris@82 361 T1w = VADD(T1u, T1v);
Chris@82 362 T1r = VADD(T11, T12);
Chris@82 363 T1s = VADD(T14, T15);
Chris@82 364 T1t = VADD(T1r, T1s);
Chris@82 365 T1a = VSUB(T18, T19);
Chris@82 366 T1d = VSUB(T1b, T1c);
Chris@82 367 T1k = VADD(T1a, T1d);
Chris@82 368 {
Chris@82 369 V Ta, Th, TC, TD;
Chris@82 370 Ta = VADD(T6, T9);
Chris@82 371 Th = VADD(Td, Tg);
Chris@82 372 Ti = VADD(Ta, Th);
Chris@82 373 Tk = VMUL(LDK(KP559016994), VSUB(Ta, Th));
Chris@82 374 TC = VADD(To, Tr);
Chris@82 375 TD = VADD(Tv, Ty);
Chris@82 376 TE = VMUL(LDK(KP559016994), VSUB(TC, TD));
Chris@82 377 TI = VADD(TC, TD);
Chris@82 378 }
Chris@82 379 }
Chris@82 380 {
Chris@82 381 V T1H, T1J, T1K, T1L, T1N, T1I, TZ, T10;
Chris@82 382 TZ = VADD(T3, Ti);
Chris@82 383 T10 = VBYI(VADD(TH, TI));
Chris@82 384 T1H = VSUB(TZ, T10);
Chris@82 385 STM2(&(xo[30]), T1H, ovs, &(xo[2]));
Chris@82 386 T1I = VADD(TZ, T10);
Chris@82 387 STM2(&(xo[10]), T1I, ovs, &(xo[2]));
Chris@82 388 {
Chris@82 389 V T1x, T1z, T1A, T1E, T1G, T1C, T1D, T1F, T1B, T1M;
Chris@82 390 T1x = VMUL(LDK(KP559016994), VSUB(T1t, T1w));
Chris@82 391 T1z = VADD(T1t, T1w);
Chris@82 392 T1A = VFNMS(LDK(KP250000000), T1z, T1y);
Chris@82 393 T1C = VSUB(T1r, T1s);
Chris@82 394 T1D = VSUB(T1u, T1v);
Chris@82 395 T1E = VBYI(VFMA(LDK(KP951056516), T1C, VMUL(LDK(KP587785252), T1D)));
Chris@82 396 T1G = VBYI(VFNMS(LDK(KP951056516), T1D, VMUL(LDK(KP587785252), T1C)));
Chris@82 397 T1J = VADD(T1y, T1z);
Chris@82 398 STM2(&(xo[0]), T1J, ovs, &(xo[0]));
Chris@82 399 T1F = VSUB(T1A, T1x);
Chris@82 400 T1K = VSUB(T1F, T1G);
Chris@82 401 STM2(&(xo[16]), T1K, ovs, &(xo[0]));
Chris@82 402 T1L = VADD(T1G, T1F);
Chris@82 403 STM2(&(xo[24]), T1L, ovs, &(xo[0]));
Chris@82 404 T1B = VADD(T1x, T1A);
Chris@82 405 T1M = VSUB(T1B, T1E);
Chris@82 406 STM2(&(xo[8]), T1M, ovs, &(xo[0]));
Chris@82 407 STN2(&(xo[8]), T1M, T1I, ovs);
Chris@82 408 T1N = VADD(T1E, T1B);
Chris@82 409 STM2(&(xo[32]), T1N, ovs, &(xo[0]));
Chris@82 410 }
Chris@82 411 {
Chris@82 412 V T1O, T1P, T1R, T1S;
Chris@82 413 {
Chris@82 414 V T1n, T1l, T1m, T1f, T1p, T17, T1e, T1q, T1Q, T1o;
Chris@82 415 T1n = VMUL(LDK(KP559016994), VSUB(T1j, T1k));
Chris@82 416 T1l = VADD(T1j, T1k);
Chris@82 417 T1m = VFNMS(LDK(KP250000000), T1l, T1i);
Chris@82 418 T17 = VSUB(T13, T16);
Chris@82 419 T1e = VSUB(T1a, T1d);
Chris@82 420 T1f = VBYI(VFNMS(LDK(KP951056516), T1e, VMUL(LDK(KP587785252), T17)));
Chris@82 421 T1p = VBYI(VFMA(LDK(KP951056516), T17, VMUL(LDK(KP587785252), T1e)));
Chris@82 422 T1O = VADD(T1i, T1l);
Chris@82 423 STM2(&(xo[20]), T1O, ovs, &(xo[0]));
Chris@82 424 T1q = VADD(T1n, T1m);
Chris@82 425 T1P = VADD(T1p, T1q);
Chris@82 426 STM2(&(xo[12]), T1P, ovs, &(xo[0]));
Chris@82 427 T1Q = VSUB(T1q, T1p);
Chris@82 428 STM2(&(xo[28]), T1Q, ovs, &(xo[0]));
Chris@82 429 STN2(&(xo[28]), T1Q, T1H, ovs);
Chris@82 430 T1o = VSUB(T1m, T1n);
Chris@82 431 T1R = VADD(T1f, T1o);
Chris@82 432 STM2(&(xo[4]), T1R, ovs, &(xo[0]));
Chris@82 433 T1S = VSUB(T1o, T1f);
Chris@82 434 STM2(&(xo[36]), T1S, ovs, &(xo[0]));
Chris@82 435 }
Chris@82 436 {
Chris@82 437 V TA, TN, TU, TS, TK, TV, Tl, TR, TJ, Tj;
Chris@82 438 TA = VFNMS(LDK(KP951056516), Tz, VMUL(LDK(KP587785252), Ts));
Chris@82 439 TN = VFNMS(LDK(KP951056516), TM, VMUL(LDK(KP587785252), TL));
Chris@82 440 TU = VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TM));
Chris@82 441 TS = VFMA(LDK(KP951056516), Ts, VMUL(LDK(KP587785252), Tz));
Chris@82 442 TJ = VFNMS(LDK(KP250000000), TI, TH);
Chris@82 443 TK = VSUB(TE, TJ);
Chris@82 444 TV = VADD(TE, TJ);
Chris@82 445 Tj = VFNMS(LDK(KP250000000), Ti, T3);
Chris@82 446 Tl = VSUB(Tj, Tk);
Chris@82 447 TR = VADD(Tk, Tj);
Chris@82 448 {
Chris@82 449 V TB, TO, T1T, T1U;
Chris@82 450 TB = VSUB(Tl, TA);
Chris@82 451 TO = VBYI(VSUB(TK, TN));
Chris@82 452 T1T = VSUB(TB, TO);
Chris@82 453 STM2(&(xo[34]), T1T, ovs, &(xo[2]));
Chris@82 454 STN2(&(xo[32]), T1N, T1T, ovs);
Chris@82 455 T1U = VADD(TB, TO);
Chris@82 456 STM2(&(xo[6]), T1U, ovs, &(xo[2]));
Chris@82 457 STN2(&(xo[4]), T1R, T1U, ovs);
Chris@82 458 }
Chris@82 459 {
Chris@82 460 V TX, TY, T1V, T1W;
Chris@82 461 TX = VADD(TR, TS);
Chris@82 462 TY = VBYI(VSUB(TV, TU));
Chris@82 463 T1V = VSUB(TX, TY);
Chris@82 464 STM2(&(xo[22]), T1V, ovs, &(xo[2]));
Chris@82 465 STN2(&(xo[20]), T1O, T1V, ovs);
Chris@82 466 T1W = VADD(TX, TY);
Chris@82 467 STM2(&(xo[18]), T1W, ovs, &(xo[2]));
Chris@82 468 STN2(&(xo[16]), T1K, T1W, ovs);
Chris@82 469 }
Chris@82 470 {
Chris@82 471 V TP, TQ, T1X, T1Y;
Chris@82 472 TP = VADD(Tl, TA);
Chris@82 473 TQ = VBYI(VADD(TN, TK));
Chris@82 474 T1X = VSUB(TP, TQ);
Chris@82 475 STM2(&(xo[26]), T1X, ovs, &(xo[2]));
Chris@82 476 STN2(&(xo[24]), T1L, T1X, ovs);
Chris@82 477 T1Y = VADD(TP, TQ);
Chris@82 478 STM2(&(xo[14]), T1Y, ovs, &(xo[2]));
Chris@82 479 STN2(&(xo[12]), T1P, T1Y, ovs);
Chris@82 480 }
Chris@82 481 {
Chris@82 482 V TT, TW, T1Z, T20;
Chris@82 483 TT = VSUB(TR, TS);
Chris@82 484 TW = VBYI(VADD(TU, TV));
Chris@82 485 T1Z = VSUB(TT, TW);
Chris@82 486 STM2(&(xo[38]), T1Z, ovs, &(xo[2]));
Chris@82 487 STN2(&(xo[36]), T1S, T1Z, ovs);
Chris@82 488 T20 = VADD(TT, TW);
Chris@82 489 STM2(&(xo[2]), T20, ovs, &(xo[2]));
Chris@82 490 STN2(&(xo[0]), T1J, T20, ovs);
Chris@82 491 }
Chris@82 492 }
Chris@82 493 }
Chris@82 494 }
Chris@82 495 }
Chris@82 496 }
Chris@82 497 VLEAVE();
Chris@82 498 }
Chris@82 499
Chris@82 500 static const kdft_desc desc = { 20, XSIMD_STRING("n2bv_20"), {92, 12, 12, 0}, &GENUS, 0, 2, 0, 0 };
Chris@82 501
Chris@82 502 void XSIMD(codelet_n2bv_20) (planner *p) {
Chris@82 503 X(kdft_register) (p, n2bv_20, &desc);
Chris@82 504 }
Chris@82 505
Chris@82 506 #endif