Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/t3bv_20.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 08:06:09 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_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 20 -name t3bv_20 -include dft/simd/t3b.h -sign 1 */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 138 FP additions, 118 FP multiplications,
Chris@82 32 * (or, 92 additions, 72 multiplications, 46 fused multiply/add),
Chris@82 33 * 73 stack variables, 4 constants, and 40 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/t3b.h"
Chris@82 36
Chris@82 37 static void t3bv_20(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
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 m;
Chris@82 45 R *x;
Chris@82 46 x = ii;
Chris@82 47 for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(20, rs)) {
Chris@82 48 V T2, T8, T9, TA, T3, Tc, T4, TV, T14, Tl, Tq, Tx, TQ, Td, Te;
Chris@82 49 V T1d, Ti, Tt, T11;
Chris@82 50 T2 = LDW(&(W[0]));
Chris@82 51 T8 = LDW(&(W[TWVL * 2]));
Chris@82 52 T9 = VZMUL(T2, T8);
Chris@82 53 TA = VZMULJ(T2, T8);
Chris@82 54 T3 = LDW(&(W[TWVL * 4]));
Chris@82 55 Tc = VZMULJ(T9, T3);
Chris@82 56 T4 = VZMUL(T2, T3);
Chris@82 57 TV = VZMUL(T9, T3);
Chris@82 58 T14 = VZMULJ(TA, T3);
Chris@82 59 Tl = VZMULJ(T8, T3);
Chris@82 60 Tq = VZMULJ(T2, T3);
Chris@82 61 Tx = VZMUL(T8, T3);
Chris@82 62 TQ = VZMUL(TA, T3);
Chris@82 63 Td = LDW(&(W[TWVL * 6]));
Chris@82 64 Te = VZMULJ(Tc, Td);
Chris@82 65 T1d = VZMULJ(T9, Td);
Chris@82 66 Ti = VZMULJ(T8, Td);
Chris@82 67 Tt = VZMULJ(T2, Td);
Chris@82 68 T11 = VZMULJ(TA, Td);
Chris@82 69 {
Chris@82 70 V T7, T1g, T1F, T23, TU, T1n, T1o, T18, Tp, TE, TF, T27, T28, T29, T1P;
Chris@82 71 V T1S, T1T, T1h, T1i, T1j, T24, T25, T26, T1I, T1L, T1M, T1B, T1C;
Chris@82 72 {
Chris@82 73 V T1, T1f, T6, T1c, T1e, T5, T1b, T1D, T1E;
Chris@82 74 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@82 75 T1e = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
Chris@82 76 T1f = VZMUL(T1d, T1e);
Chris@82 77 T5 = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@82 78 T6 = VZMUL(T4, T5);
Chris@82 79 T1b = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@82 80 T1c = VZMUL(Tc, T1b);
Chris@82 81 T7 = VSUB(T1, T6);
Chris@82 82 T1g = VSUB(T1c, T1f);
Chris@82 83 T1D = VADD(T1, T6);
Chris@82 84 T1E = VADD(T1c, T1f);
Chris@82 85 T1F = VSUB(T1D, T1E);
Chris@82 86 T23 = VADD(T1D, T1E);
Chris@82 87 }
Chris@82 88 {
Chris@82 89 V Th, T1G, T10, T1O, T17, T1R, To, T1J, Tw, T1N, TN, T1H, TT, T1K, TD;
Chris@82 90 V T1Q;
Chris@82 91 {
Chris@82 92 V Tb, Tg, Ta, Tf;
Chris@82 93 Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@82 94 Tb = VZMUL(T9, Ta);
Chris@82 95 Tf = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@82 96 Tg = VZMUL(Te, Tf);
Chris@82 97 Th = VSUB(Tb, Tg);
Chris@82 98 T1G = VADD(Tb, Tg);
Chris@82 99 }
Chris@82 100 {
Chris@82 101 V TX, TZ, TW, TY;
Chris@82 102 TW = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@82 103 TX = VZMUL(TV, TW);
Chris@82 104 TY = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@82 105 TZ = VZMUL(T8, TY);
Chris@82 106 T10 = VSUB(TX, TZ);
Chris@82 107 T1O = VADD(TX, TZ);
Chris@82 108 }
Chris@82 109 {
Chris@82 110 V T13, T16, T12, T15;
Chris@82 111 T12 = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)]));
Chris@82 112 T13 = VZMUL(T11, T12);
Chris@82 113 T15 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@82 114 T16 = VZMUL(T14, T15);
Chris@82 115 T17 = VSUB(T13, T16);
Chris@82 116 T1R = VADD(T13, T16);
Chris@82 117 }
Chris@82 118 {
Chris@82 119 V Tk, Tn, Tj, Tm;
Chris@82 120 Tj = LD(&(x[WS(rs, 16)]), ms, &(x[0]));
Chris@82 121 Tk = VZMUL(Ti, Tj);
Chris@82 122 Tm = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@82 123 Tn = VZMUL(Tl, Tm);
Chris@82 124 To = VSUB(Tk, Tn);
Chris@82 125 T1J = VADD(Tk, Tn);
Chris@82 126 }
Chris@82 127 {
Chris@82 128 V Ts, Tv, Tr, Tu;
Chris@82 129 Tr = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@82 130 Ts = VZMUL(Tq, Tr);
Chris@82 131 Tu = LD(&(x[WS(rs, 18)]), ms, &(x[0]));
Chris@82 132 Tv = VZMUL(Tt, Tu);
Chris@82 133 Tw = VSUB(Ts, Tv);
Chris@82 134 T1N = VADD(Ts, Tv);
Chris@82 135 }
Chris@82 136 {
Chris@82 137 V TK, TM, TJ, TL;
Chris@82 138 TJ = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@82 139 TK = VZMUL(T3, TJ);
Chris@82 140 TL = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)]));
Chris@82 141 TM = VZMUL(Td, TL);
Chris@82 142 TN = VSUB(TK, TM);
Chris@82 143 T1H = VADD(TK, TM);
Chris@82 144 }
Chris@82 145 {
Chris@82 146 V TP, TS, TO, TR;
Chris@82 147 TO = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82 148 TP = VZMUL(T2, TO);
Chris@82 149 TR = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@82 150 TS = VZMUL(TQ, TR);
Chris@82 151 TT = VSUB(TP, TS);
Chris@82 152 T1K = VADD(TP, TS);
Chris@82 153 }
Chris@82 154 {
Chris@82 155 V Tz, TC, Ty, TB;
Chris@82 156 Ty = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@82 157 Tz = VZMUL(Tx, Ty);
Chris@82 158 TB = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@82 159 TC = VZMUL(TA, TB);
Chris@82 160 TD = VSUB(Tz, TC);
Chris@82 161 T1Q = VADD(Tz, TC);
Chris@82 162 }
Chris@82 163 TU = VSUB(TN, TT);
Chris@82 164 T1n = VSUB(Th, To);
Chris@82 165 T1o = VSUB(Tw, TD);
Chris@82 166 T18 = VSUB(T10, T17);
Chris@82 167 Tp = VADD(Th, To);
Chris@82 168 TE = VADD(Tw, TD);
Chris@82 169 TF = VADD(Tp, TE);
Chris@82 170 T27 = VADD(T1N, T1O);
Chris@82 171 T28 = VADD(T1Q, T1R);
Chris@82 172 T29 = VADD(T27, T28);
Chris@82 173 T1P = VSUB(T1N, T1O);
Chris@82 174 T1S = VSUB(T1Q, T1R);
Chris@82 175 T1T = VADD(T1P, T1S);
Chris@82 176 T1h = VADD(TN, TT);
Chris@82 177 T1i = VADD(T10, T17);
Chris@82 178 T1j = VADD(T1h, T1i);
Chris@82 179 T24 = VADD(T1G, T1H);
Chris@82 180 T25 = VADD(T1J, T1K);
Chris@82 181 T26 = VADD(T24, T25);
Chris@82 182 T1I = VSUB(T1G, T1H);
Chris@82 183 T1L = VSUB(T1J, T1K);
Chris@82 184 T1M = VADD(T1I, T1L);
Chris@82 185 }
Chris@82 186 T1B = VADD(T7, TF);
Chris@82 187 T1C = VADD(T1g, T1j);
Chris@82 188 ST(&(x[WS(rs, 15)]), VFNMSI(T1C, T1B), ms, &(x[WS(rs, 1)]));
Chris@82 189 ST(&(x[WS(rs, 5)]), VFMAI(T1C, T1B), ms, &(x[WS(rs, 1)]));
Chris@82 190 {
Chris@82 191 V T2c, T2a, T2b, T2g, T2i, T2e, T2f, T2h, T2d;
Chris@82 192 T2c = VSUB(T26, T29);
Chris@82 193 T2a = VADD(T26, T29);
Chris@82 194 T2b = VFNMS(LDK(KP250000000), T2a, T23);
Chris@82 195 T2e = VSUB(T24, T25);
Chris@82 196 T2f = VSUB(T27, T28);
Chris@82 197 T2g = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T2f, T2e));
Chris@82 198 T2i = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T2e, T2f));
Chris@82 199 ST(&(x[0]), VADD(T23, T2a), ms, &(x[0]));
Chris@82 200 T2h = VFNMS(LDK(KP559016994), T2c, T2b);
Chris@82 201 ST(&(x[WS(rs, 8)]), VFMAI(T2i, T2h), ms, &(x[0]));
Chris@82 202 ST(&(x[WS(rs, 12)]), VFNMSI(T2i, T2h), ms, &(x[0]));
Chris@82 203 T2d = VFMA(LDK(KP559016994), T2c, T2b);
Chris@82 204 ST(&(x[WS(rs, 4)]), VFNMSI(T2g, T2d), ms, &(x[0]));
Chris@82 205 ST(&(x[WS(rs, 16)]), VFMAI(T2g, T2d), ms, &(x[0]));
Chris@82 206 }
Chris@82 207 {
Chris@82 208 V T1W, T1U, T1V, T20, T22, T1Y, T1Z, T21, T1X;
Chris@82 209 T1W = VSUB(T1M, T1T);
Chris@82 210 T1U = VADD(T1M, T1T);
Chris@82 211 T1V = VFNMS(LDK(KP250000000), T1U, T1F);
Chris@82 212 T1Y = VSUB(T1P, T1S);
Chris@82 213 T1Z = VSUB(T1I, T1L);
Chris@82 214 T20 = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1Z, T1Y));
Chris@82 215 T22 = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1Y, T1Z));
Chris@82 216 ST(&(x[WS(rs, 10)]), VADD(T1F, T1U), ms, &(x[0]));
Chris@82 217 T21 = VFMA(LDK(KP559016994), T1W, T1V);
Chris@82 218 ST(&(x[WS(rs, 6)]), VFMAI(T22, T21), ms, &(x[0]));
Chris@82 219 ST(&(x[WS(rs, 14)]), VFNMSI(T22, T21), ms, &(x[0]));
Chris@82 220 T1X = VFNMS(LDK(KP559016994), T1W, T1V);
Chris@82 221 ST(&(x[WS(rs, 2)]), VFNMSI(T20, T1X), ms, &(x[0]));
Chris@82 222 ST(&(x[WS(rs, 18)]), VFMAI(T20, T1X), ms, &(x[0]));
Chris@82 223 }
Chris@82 224 {
Chris@82 225 V T19, T1p, T1x, T1u, T1m, T1w, TI, T1t;
Chris@82 226 T19 = VFMA(LDK(KP618033988), T18, TU);
Chris@82 227 T1p = VFMA(LDK(KP618033988), T1o, T1n);
Chris@82 228 T1x = VFNMS(LDK(KP618033988), T1n, T1o);
Chris@82 229 T1u = VFNMS(LDK(KP618033988), TU, T18);
Chris@82 230 {
Chris@82 231 V T1k, T1l, TG, TH;
Chris@82 232 T1k = VFNMS(LDK(KP250000000), T1j, T1g);
Chris@82 233 T1l = VSUB(T1h, T1i);
Chris@82 234 T1m = VFMA(LDK(KP559016994), T1l, T1k);
Chris@82 235 T1w = VFNMS(LDK(KP559016994), T1l, T1k);
Chris@82 236 TG = VFNMS(LDK(KP250000000), TF, T7);
Chris@82 237 TH = VSUB(Tp, TE);
Chris@82 238 TI = VFMA(LDK(KP559016994), TH, TG);
Chris@82 239 T1t = VFNMS(LDK(KP559016994), TH, TG);
Chris@82 240 }
Chris@82 241 {
Chris@82 242 V T1a, T1q, T1z, T1A;
Chris@82 243 T1a = VFNMS(LDK(KP951056516), T19, TI);
Chris@82 244 T1q = VFMA(LDK(KP951056516), T1p, T1m);
Chris@82 245 ST(&(x[WS(rs, 19)]), VFNMSI(T1q, T1a), ms, &(x[WS(rs, 1)]));
Chris@82 246 ST(&(x[WS(rs, 1)]), VFMAI(T1q, T1a), ms, &(x[WS(rs, 1)]));
Chris@82 247 T1z = VFNMS(LDK(KP951056516), T1u, T1t);
Chris@82 248 T1A = VFMA(LDK(KP951056516), T1x, T1w);
Chris@82 249 ST(&(x[WS(rs, 7)]), VFNMSI(T1A, T1z), ms, &(x[WS(rs, 1)]));
Chris@82 250 ST(&(x[WS(rs, 13)]), VFMAI(T1A, T1z), ms, &(x[WS(rs, 1)]));
Chris@82 251 }
Chris@82 252 {
Chris@82 253 V T1r, T1s, T1v, T1y;
Chris@82 254 T1r = VFMA(LDK(KP951056516), T19, TI);
Chris@82 255 T1s = VFNMS(LDK(KP951056516), T1p, T1m);
Chris@82 256 ST(&(x[WS(rs, 11)]), VFNMSI(T1s, T1r), ms, &(x[WS(rs, 1)]));
Chris@82 257 ST(&(x[WS(rs, 9)]), VFMAI(T1s, T1r), ms, &(x[WS(rs, 1)]));
Chris@82 258 T1v = VFMA(LDK(KP951056516), T1u, T1t);
Chris@82 259 T1y = VFNMS(LDK(KP951056516), T1x, T1w);
Chris@82 260 ST(&(x[WS(rs, 3)]), VFNMSI(T1y, T1v), ms, &(x[WS(rs, 1)]));
Chris@82 261 ST(&(x[WS(rs, 17)]), VFMAI(T1y, T1v), ms, &(x[WS(rs, 1)]));
Chris@82 262 }
Chris@82 263 }
Chris@82 264 }
Chris@82 265 }
Chris@82 266 }
Chris@82 267 VLEAVE();
Chris@82 268 }
Chris@82 269
Chris@82 270 static const tw_instr twinstr[] = {
Chris@82 271 VTW(0, 1),
Chris@82 272 VTW(0, 3),
Chris@82 273 VTW(0, 9),
Chris@82 274 VTW(0, 19),
Chris@82 275 {TW_NEXT, VL, 0}
Chris@82 276 };
Chris@82 277
Chris@82 278 static const ct_desc desc = { 20, XSIMD_STRING("t3bv_20"), twinstr, &GENUS, {92, 72, 46, 0}, 0, 0, 0 };
Chris@82 279
Chris@82 280 void XSIMD(codelet_t3bv_20) (planner *p) {
Chris@82 281 X(kdft_dit_register) (p, t3bv_20, &desc);
Chris@82 282 }
Chris@82 283 #else
Chris@82 284
Chris@82 285 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 20 -name t3bv_20 -include dft/simd/t3b.h -sign 1 */
Chris@82 286
Chris@82 287 /*
Chris@82 288 * This function contains 138 FP additions, 92 FP multiplications,
Chris@82 289 * (or, 126 additions, 80 multiplications, 12 fused multiply/add),
Chris@82 290 * 73 stack variables, 4 constants, and 40 memory accesses
Chris@82 291 */
Chris@82 292 #include "dft/simd/t3b.h"
Chris@82 293
Chris@82 294 static void t3bv_20(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 295 {
Chris@82 296 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@82 297 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 298 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 299 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 300 {
Chris@82 301 INT m;
Chris@82 302 R *x;
Chris@82 303 x = ii;
Chris@82 304 for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(20, rs)) {
Chris@82 305 V T2, T8, T9, TA, T3, Tc, T4, TV, T14, Tl, Tq, Tx, TQ, Td, Te;
Chris@82 306 V T1g, Ti, Tt, T11;
Chris@82 307 T2 = LDW(&(W[0]));
Chris@82 308 T8 = LDW(&(W[TWVL * 2]));
Chris@82 309 T9 = VZMUL(T2, T8);
Chris@82 310 TA = VZMULJ(T2, T8);
Chris@82 311 T3 = LDW(&(W[TWVL * 4]));
Chris@82 312 Tc = VZMULJ(T9, T3);
Chris@82 313 T4 = VZMUL(T2, T3);
Chris@82 314 TV = VZMUL(T9, T3);
Chris@82 315 T14 = VZMULJ(TA, T3);
Chris@82 316 Tl = VZMULJ(T8, T3);
Chris@82 317 Tq = VZMULJ(T2, T3);
Chris@82 318 Tx = VZMUL(T8, T3);
Chris@82 319 TQ = VZMUL(TA, T3);
Chris@82 320 Td = LDW(&(W[TWVL * 6]));
Chris@82 321 Te = VZMULJ(Tc, Td);
Chris@82 322 T1g = VZMULJ(T9, Td);
Chris@82 323 Ti = VZMULJ(T8, Td);
Chris@82 324 Tt = VZMULJ(T2, Td);
Chris@82 325 T11 = VZMULJ(TA, Td);
Chris@82 326 {
Chris@82 327 V T7, T1j, T1U, T2a, TU, T1n, T1o, T18, Tp, TE, TF, T26, T27, T28, T1M;
Chris@82 328 V T1P, T1W, T1b, T1c, T1k, T23, T24, T25, T1F, T1I, T1V, T1B, T1C;
Chris@82 329 {
Chris@82 330 V T1, T1i, T6, T1f, T1h, T5, T1e, T1S, T1T;
Chris@82 331 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@82 332 T1h = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
Chris@82 333 T1i = VZMUL(T1g, T1h);
Chris@82 334 T5 = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@82 335 T6 = VZMUL(T4, T5);
Chris@82 336 T1e = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@82 337 T1f = VZMUL(Tc, T1e);
Chris@82 338 T7 = VSUB(T1, T6);
Chris@82 339 T1j = VSUB(T1f, T1i);
Chris@82 340 T1S = VADD(T1, T6);
Chris@82 341 T1T = VADD(T1f, T1i);
Chris@82 342 T1U = VSUB(T1S, T1T);
Chris@82 343 T2a = VADD(T1S, T1T);
Chris@82 344 }
Chris@82 345 {
Chris@82 346 V Th, T1D, T10, T1L, T17, T1O, To, T1G, Tw, T1K, TN, T1E, TT, T1H, TD;
Chris@82 347 V T1N;
Chris@82 348 {
Chris@82 349 V Tb, Tg, Ta, Tf;
Chris@82 350 Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@82 351 Tb = VZMUL(T9, Ta);
Chris@82 352 Tf = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@82 353 Tg = VZMUL(Te, Tf);
Chris@82 354 Th = VSUB(Tb, Tg);
Chris@82 355 T1D = VADD(Tb, Tg);
Chris@82 356 }
Chris@82 357 {
Chris@82 358 V TX, TZ, TW, TY;
Chris@82 359 TW = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@82 360 TX = VZMUL(TV, TW);
Chris@82 361 TY = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@82 362 TZ = VZMUL(T8, TY);
Chris@82 363 T10 = VSUB(TX, TZ);
Chris@82 364 T1L = VADD(TX, TZ);
Chris@82 365 }
Chris@82 366 {
Chris@82 367 V T13, T16, T12, T15;
Chris@82 368 T12 = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)]));
Chris@82 369 T13 = VZMUL(T11, T12);
Chris@82 370 T15 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@82 371 T16 = VZMUL(T14, T15);
Chris@82 372 T17 = VSUB(T13, T16);
Chris@82 373 T1O = VADD(T13, T16);
Chris@82 374 }
Chris@82 375 {
Chris@82 376 V Tk, Tn, Tj, Tm;
Chris@82 377 Tj = LD(&(x[WS(rs, 16)]), ms, &(x[0]));
Chris@82 378 Tk = VZMUL(Ti, Tj);
Chris@82 379 Tm = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@82 380 Tn = VZMUL(Tl, Tm);
Chris@82 381 To = VSUB(Tk, Tn);
Chris@82 382 T1G = VADD(Tk, Tn);
Chris@82 383 }
Chris@82 384 {
Chris@82 385 V Ts, Tv, Tr, Tu;
Chris@82 386 Tr = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@82 387 Ts = VZMUL(Tq, Tr);
Chris@82 388 Tu = LD(&(x[WS(rs, 18)]), ms, &(x[0]));
Chris@82 389 Tv = VZMUL(Tt, Tu);
Chris@82 390 Tw = VSUB(Ts, Tv);
Chris@82 391 T1K = VADD(Ts, Tv);
Chris@82 392 }
Chris@82 393 {
Chris@82 394 V TK, TM, TJ, TL;
Chris@82 395 TJ = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@82 396 TK = VZMUL(T3, TJ);
Chris@82 397 TL = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)]));
Chris@82 398 TM = VZMUL(Td, TL);
Chris@82 399 TN = VSUB(TK, TM);
Chris@82 400 T1E = VADD(TK, TM);
Chris@82 401 }
Chris@82 402 {
Chris@82 403 V TP, TS, TO, TR;
Chris@82 404 TO = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82 405 TP = VZMUL(T2, TO);
Chris@82 406 TR = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@82 407 TS = VZMUL(TQ, TR);
Chris@82 408 TT = VSUB(TP, TS);
Chris@82 409 T1H = VADD(TP, TS);
Chris@82 410 }
Chris@82 411 {
Chris@82 412 V Tz, TC, Ty, TB;
Chris@82 413 Ty = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@82 414 Tz = VZMUL(Tx, Ty);
Chris@82 415 TB = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@82 416 TC = VZMUL(TA, TB);
Chris@82 417 TD = VSUB(Tz, TC);
Chris@82 418 T1N = VADD(Tz, TC);
Chris@82 419 }
Chris@82 420 TU = VSUB(TN, TT);
Chris@82 421 T1n = VSUB(Th, To);
Chris@82 422 T1o = VSUB(Tw, TD);
Chris@82 423 T18 = VSUB(T10, T17);
Chris@82 424 Tp = VADD(Th, To);
Chris@82 425 TE = VADD(Tw, TD);
Chris@82 426 TF = VADD(Tp, TE);
Chris@82 427 T26 = VADD(T1K, T1L);
Chris@82 428 T27 = VADD(T1N, T1O);
Chris@82 429 T28 = VADD(T26, T27);
Chris@82 430 T1M = VSUB(T1K, T1L);
Chris@82 431 T1P = VSUB(T1N, T1O);
Chris@82 432 T1W = VADD(T1M, T1P);
Chris@82 433 T1b = VADD(TN, TT);
Chris@82 434 T1c = VADD(T10, T17);
Chris@82 435 T1k = VADD(T1b, T1c);
Chris@82 436 T23 = VADD(T1D, T1E);
Chris@82 437 T24 = VADD(T1G, T1H);
Chris@82 438 T25 = VADD(T23, T24);
Chris@82 439 T1F = VSUB(T1D, T1E);
Chris@82 440 T1I = VSUB(T1G, T1H);
Chris@82 441 T1V = VADD(T1F, T1I);
Chris@82 442 }
Chris@82 443 T1B = VADD(T7, TF);
Chris@82 444 T1C = VBYI(VADD(T1j, T1k));
Chris@82 445 ST(&(x[WS(rs, 15)]), VSUB(T1B, T1C), ms, &(x[WS(rs, 1)]));
Chris@82 446 ST(&(x[WS(rs, 5)]), VADD(T1B, T1C), ms, &(x[WS(rs, 1)]));
Chris@82 447 {
Chris@82 448 V T29, T2b, T2c, T2g, T2i, T2e, T2f, T2h, T2d;
Chris@82 449 T29 = VMUL(LDK(KP559016994), VSUB(T25, T28));
Chris@82 450 T2b = VADD(T25, T28);
Chris@82 451 T2c = VFNMS(LDK(KP250000000), T2b, T2a);
Chris@82 452 T2e = VSUB(T23, T24);
Chris@82 453 T2f = VSUB(T26, T27);
Chris@82 454 T2g = VBYI(VFMA(LDK(KP951056516), T2e, VMUL(LDK(KP587785252), T2f)));
Chris@82 455 T2i = VBYI(VFNMS(LDK(KP951056516), T2f, VMUL(LDK(KP587785252), T2e)));
Chris@82 456 ST(&(x[0]), VADD(T2a, T2b), ms, &(x[0]));
Chris@82 457 T2h = VSUB(T2c, T29);
Chris@82 458 ST(&(x[WS(rs, 8)]), VSUB(T2h, T2i), ms, &(x[0]));
Chris@82 459 ST(&(x[WS(rs, 12)]), VADD(T2i, T2h), ms, &(x[0]));
Chris@82 460 T2d = VADD(T29, T2c);
Chris@82 461 ST(&(x[WS(rs, 4)]), VSUB(T2d, T2g), ms, &(x[0]));
Chris@82 462 ST(&(x[WS(rs, 16)]), VADD(T2g, T2d), ms, &(x[0]));
Chris@82 463 }
Chris@82 464 {
Chris@82 465 V T1Z, T1X, T1Y, T1R, T21, T1J, T1Q, T22, T20;
Chris@82 466 T1Z = VMUL(LDK(KP559016994), VSUB(T1V, T1W));
Chris@82 467 T1X = VADD(T1V, T1W);
Chris@82 468 T1Y = VFNMS(LDK(KP250000000), T1X, T1U);
Chris@82 469 T1J = VSUB(T1F, T1I);
Chris@82 470 T1Q = VSUB(T1M, T1P);
Chris@82 471 T1R = VBYI(VFNMS(LDK(KP951056516), T1Q, VMUL(LDK(KP587785252), T1J)));
Chris@82 472 T21 = VBYI(VFMA(LDK(KP951056516), T1J, VMUL(LDK(KP587785252), T1Q)));
Chris@82 473 ST(&(x[WS(rs, 10)]), VADD(T1U, T1X), ms, &(x[0]));
Chris@82 474 T22 = VADD(T1Z, T1Y);
Chris@82 475 ST(&(x[WS(rs, 6)]), VADD(T21, T22), ms, &(x[0]));
Chris@82 476 ST(&(x[WS(rs, 14)]), VSUB(T22, T21), ms, &(x[0]));
Chris@82 477 T20 = VSUB(T1Y, T1Z);
Chris@82 478 ST(&(x[WS(rs, 2)]), VADD(T1R, T20), ms, &(x[0]));
Chris@82 479 ST(&(x[WS(rs, 18)]), VSUB(T20, T1R), ms, &(x[0]));
Chris@82 480 }
Chris@82 481 {
Chris@82 482 V T19, T1p, T1w, T1u, T1m, T1x, TI, T1t;
Chris@82 483 T19 = VFNMS(LDK(KP951056516), T18, VMUL(LDK(KP587785252), TU));
Chris@82 484 T1p = VFNMS(LDK(KP951056516), T1o, VMUL(LDK(KP587785252), T1n));
Chris@82 485 T1w = VFMA(LDK(KP951056516), T1n, VMUL(LDK(KP587785252), T1o));
Chris@82 486 T1u = VFMA(LDK(KP951056516), TU, VMUL(LDK(KP587785252), T18));
Chris@82 487 {
Chris@82 488 V T1d, T1l, TG, TH;
Chris@82 489 T1d = VMUL(LDK(KP559016994), VSUB(T1b, T1c));
Chris@82 490 T1l = VFNMS(LDK(KP250000000), T1k, T1j);
Chris@82 491 T1m = VSUB(T1d, T1l);
Chris@82 492 T1x = VADD(T1d, T1l);
Chris@82 493 TG = VFNMS(LDK(KP250000000), TF, T7);
Chris@82 494 TH = VMUL(LDK(KP559016994), VSUB(Tp, TE));
Chris@82 495 TI = VSUB(TG, TH);
Chris@82 496 T1t = VADD(TH, TG);
Chris@82 497 }
Chris@82 498 {
Chris@82 499 V T1a, T1q, T1z, T1A;
Chris@82 500 T1a = VSUB(TI, T19);
Chris@82 501 T1q = VBYI(VSUB(T1m, T1p));
Chris@82 502 ST(&(x[WS(rs, 17)]), VSUB(T1a, T1q), ms, &(x[WS(rs, 1)]));
Chris@82 503 ST(&(x[WS(rs, 3)]), VADD(T1a, T1q), ms, &(x[WS(rs, 1)]));
Chris@82 504 T1z = VADD(T1t, T1u);
Chris@82 505 T1A = VBYI(VSUB(T1x, T1w));
Chris@82 506 ST(&(x[WS(rs, 11)]), VSUB(T1z, T1A), ms, &(x[WS(rs, 1)]));
Chris@82 507 ST(&(x[WS(rs, 9)]), VADD(T1z, T1A), ms, &(x[WS(rs, 1)]));
Chris@82 508 }
Chris@82 509 {
Chris@82 510 V T1r, T1s, T1v, T1y;
Chris@82 511 T1r = VADD(TI, T19);
Chris@82 512 T1s = VBYI(VADD(T1p, T1m));
Chris@82 513 ST(&(x[WS(rs, 13)]), VSUB(T1r, T1s), ms, &(x[WS(rs, 1)]));
Chris@82 514 ST(&(x[WS(rs, 7)]), VADD(T1r, T1s), ms, &(x[WS(rs, 1)]));
Chris@82 515 T1v = VSUB(T1t, T1u);
Chris@82 516 T1y = VBYI(VADD(T1w, T1x));
Chris@82 517 ST(&(x[WS(rs, 19)]), VSUB(T1v, T1y), ms, &(x[WS(rs, 1)]));
Chris@82 518 ST(&(x[WS(rs, 1)]), VADD(T1v, T1y), ms, &(x[WS(rs, 1)]));
Chris@82 519 }
Chris@82 520 }
Chris@82 521 }
Chris@82 522 }
Chris@82 523 }
Chris@82 524 VLEAVE();
Chris@82 525 }
Chris@82 526
Chris@82 527 static const tw_instr twinstr[] = {
Chris@82 528 VTW(0, 1),
Chris@82 529 VTW(0, 3),
Chris@82 530 VTW(0, 9),
Chris@82 531 VTW(0, 19),
Chris@82 532 {TW_NEXT, VL, 0}
Chris@82 533 };
Chris@82 534
Chris@82 535 static const ct_desc desc = { 20, XSIMD_STRING("t3bv_20"), twinstr, &GENUS, {126, 80, 12, 0}, 0, 0, 0 };
Chris@82 536
Chris@82 537 void XSIMD(codelet_t3bv_20) (planner *p) {
Chris@82 538 X(kdft_dit_register) (p, t3bv_20, &desc);
Chris@82 539 }
Chris@82 540 #endif