Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/simd/common/hc2cfdftv_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 08:08:11 EDT 2018 */
Chris@82 23
Chris@82 24 #include "rdft/codelet-rdft.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_hc2cdft_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 20 -dit -name hc2cfdftv_20 -include rdft/simd/hc2cfv.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 143 FP additions, 128 FP multiplications,
Chris@82 32 * (or, 77 additions, 62 multiplications, 66 fused multiply/add),
Chris@82 33 * 129 stack variables, 5 constants, and 40 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/simd/hc2cfv.h"
Chris@82 36
Chris@82 37 static void hc2cfdftv_20(R *Rp, R *Ip, R *Rm, R *Im, 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(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 41 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82 42 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 43 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 44 {
Chris@82 45 INT m;
Chris@82 46 for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 38)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 38), MAKE_VOLATILE_STRIDE(80, rs)) {
Chris@82 47 V T1O, T2j, T2c, T2b, T2i, T1X, Tx, TM, TN, T1x, T1y, T1z, T1u, T1v, T1w;
Chris@82 48 V T12, T1d, T1e, T24, T2g, Ti, T1t, T1V, T29, T26, T27, T1W, T25, T1H, T1L;
Chris@82 49 V T1B, T1K, T1E, T1F, T1G, T1D, T1A, T1C, T1N, T1I, T1J, T1M;
Chris@82 50 {
Chris@82 51 V T3, T1Y, TC, T7, Tn, T1P, Tc, Tg, Tw, T1Z, TS, T1S, TL, T21, T17;
Chris@82 52 V T1Q, T11, T22, T1c, T1T, T1, T2, Tz, T5, T6, TB, Ty, TA, T4, Ta;
Chris@82 53 V Tb, Tk, Te, Tf, Tm, Tj, Tl, T9, Td, T20, T23, T8, Th, T1R, T1U;
Chris@82 54 T1 = LD(&(Rp[0]), ms, &(Rp[0]));
Chris@82 55 T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
Chris@82 56 Ty = LDW(&(W[0]));
Chris@82 57 Tz = VZMULIJ(Ty, VFNMSCONJ(T2, T1));
Chris@82 58 T5 = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 59 T6 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 60 TA = LDW(&(W[TWVL * 20]));
Chris@82 61 TB = VZMULIJ(TA, VFNMSCONJ(T6, T5));
Chris@82 62 T3 = VFMACONJ(T2, T1);
Chris@82 63 T1Y = VSUB(TB, Tz);
Chris@82 64 TC = VADD(Tz, TB);
Chris@82 65 T4 = LDW(&(W[TWVL * 18]));
Chris@82 66 T7 = VZMULJ(T4, VFMACONJ(T6, T5));
Chris@82 67 Ta = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Chris@82 68 Tb = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Chris@82 69 Tj = LDW(&(W[TWVL * 6]));
Chris@82 70 Tk = VZMULJ(Tj, VFMACONJ(Tb, Ta));
Chris@82 71 Te = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 72 Tf = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 73 Tl = LDW(&(W[TWVL * 26]));
Chris@82 74 Tm = VZMULJ(Tl, VFMACONJ(Tf, Te));
Chris@82 75 Tn = VADD(Tk, Tm);
Chris@82 76 T1P = VSUB(Tk, Tm);
Chris@82 77 T9 = LDW(&(W[TWVL * 8]));
Chris@82 78 Tc = VZMULIJ(T9, VFNMSCONJ(Tb, Ta));
Chris@82 79 Td = LDW(&(W[TWVL * 28]));
Chris@82 80 Tg = VZMULIJ(Td, VFNMSCONJ(Tf, Te));
Chris@82 81 {
Chris@82 82 V Tr, TP, Tv, TR, Tp, Tq, To, TO, Tt, Tu, Ts, TQ, TG, T14, TK;
Chris@82 83 V T16, TE, TF, TD, T13, TI, TJ, TH, T15, TW, T19, T10, T1b, TU, TV;
Chris@82 84 V TT, T18, TY, TZ, TX, T1a;
Chris@82 85 Tp = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
Chris@82 86 Tq = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Chris@82 87 To = LDW(&(W[TWVL * 16]));
Chris@82 88 Tr = VZMULIJ(To, VFNMSCONJ(Tq, Tp));
Chris@82 89 TO = LDW(&(W[TWVL * 14]));
Chris@82 90 TP = VZMULJ(TO, VFMACONJ(Tq, Tp));
Chris@82 91 Tt = LD(&(Rp[WS(rs, 9)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 92 Tu = LD(&(Rm[WS(rs, 9)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 93 Ts = LDW(&(W[TWVL * 36]));
Chris@82 94 Tv = VZMULIJ(Ts, VFNMSCONJ(Tu, Tt));
Chris@82 95 TQ = LDW(&(W[TWVL * 34]));
Chris@82 96 TR = VZMULJ(TQ, VFMACONJ(Tu, Tt));
Chris@82 97 Tw = VADD(Tr, Tv);
Chris@82 98 T1Z = VSUB(Tv, Tr);
Chris@82 99 TS = VADD(TP, TR);
Chris@82 100 T1S = VSUB(TP, TR);
Chris@82 101 TE = LD(&(Rp[WS(rs, 8)]), ms, &(Rp[0]));
Chris@82 102 TF = LD(&(Rm[WS(rs, 8)]), -ms, &(Rm[0]));
Chris@82 103 TD = LDW(&(W[TWVL * 30]));
Chris@82 104 TG = VZMULJ(TD, VFMACONJ(TF, TE));
Chris@82 105 T13 = LDW(&(W[TWVL * 32]));
Chris@82 106 T14 = VZMULIJ(T13, VFNMSCONJ(TF, TE));
Chris@82 107 TI = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 108 TJ = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 109 TH = LDW(&(W[TWVL * 10]));
Chris@82 110 TK = VZMULJ(TH, VFMACONJ(TJ, TI));
Chris@82 111 T15 = LDW(&(W[TWVL * 12]));
Chris@82 112 T16 = VZMULIJ(T15, VFNMSCONJ(TJ, TI));
Chris@82 113 TL = VADD(TG, TK);
Chris@82 114 T21 = VSUB(T16, T14);
Chris@82 115 T17 = VADD(T14, T16);
Chris@82 116 T1Q = VSUB(TK, TG);
Chris@82 117 TU = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
Chris@82 118 TV = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
Chris@82 119 TT = LDW(&(W[TWVL * 24]));
Chris@82 120 TW = VZMULIJ(TT, VFNMSCONJ(TV, TU));
Chris@82 121 T18 = LDW(&(W[TWVL * 22]));
Chris@82 122 T19 = VZMULJ(T18, VFMACONJ(TV, TU));
Chris@82 123 TY = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 124 TZ = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 125 TX = LDW(&(W[TWVL * 4]));
Chris@82 126 T10 = VZMULIJ(TX, VFNMSCONJ(TZ, TY));
Chris@82 127 T1a = LDW(&(W[TWVL * 2]));
Chris@82 128 T1b = VZMULJ(T1a, VFMACONJ(TZ, TY));
Chris@82 129 T11 = VADD(TW, T10);
Chris@82 130 T22 = VSUB(T10, TW);
Chris@82 131 T1c = VADD(T19, T1b);
Chris@82 132 T1T = VSUB(T1b, T19);
Chris@82 133 }
Chris@82 134 T1O = VSUB(T3, T7);
Chris@82 135 T2j = VADD(T1S, T1T);
Chris@82 136 T2c = VSUB(T21, T22);
Chris@82 137 T2b = VSUB(T1Y, T1Z);
Chris@82 138 T2i = VADD(T1P, T1Q);
Chris@82 139 T1X = VSUB(Tg, Tc);
Chris@82 140 Tx = VSUB(Tn, Tw);
Chris@82 141 TM = VSUB(TC, TL);
Chris@82 142 TN = VSUB(Tx, TM);
Chris@82 143 T1x = VADD(TS, T11);
Chris@82 144 T1y = VADD(T17, T1c);
Chris@82 145 T1z = VADD(T1x, T1y);
Chris@82 146 T1u = VADD(Tn, Tw);
Chris@82 147 T1v = VADD(TC, TL);
Chris@82 148 T1w = VADD(T1u, T1v);
Chris@82 149 T12 = VSUB(TS, T11);
Chris@82 150 T1d = VSUB(T17, T1c);
Chris@82 151 T1e = VSUB(T12, T1d);
Chris@82 152 T20 = VADD(T1Y, T1Z);
Chris@82 153 T23 = VADD(T21, T22);
Chris@82 154 T24 = VADD(T20, T23);
Chris@82 155 T2g = VSUB(T23, T20);
Chris@82 156 T8 = VADD(T3, T7);
Chris@82 157 Th = VADD(Tc, Tg);
Chris@82 158 Ti = VSUB(T8, Th);
Chris@82 159 T1t = VADD(T8, Th);
Chris@82 160 T1R = VSUB(T1P, T1Q);
Chris@82 161 T1U = VSUB(T1S, T1T);
Chris@82 162 T1V = VADD(T1R, T1U);
Chris@82 163 T29 = VSUB(T1R, T1U);
Chris@82 164 }
Chris@82 165 T1W = VADD(T1O, T1V);
Chris@82 166 T25 = VADD(T1X, T24);
Chris@82 167 T26 = VMUL(LDK(KP500000000), VFNMSI(T25, T1W));
Chris@82 168 T27 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T25, T1W)));
Chris@82 169 ST(&(Rp[WS(rs, 5)]), T26, ms, &(Rp[WS(rs, 1)]));
Chris@82 170 ST(&(Rm[WS(rs, 4)]), T27, -ms, &(Rm[0]));
Chris@82 171 T1F = VSUB(T1x, T1y);
Chris@82 172 T1G = VSUB(T1u, T1v);
Chris@82 173 T1H = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1G, T1F));
Chris@82 174 T1L = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1F, T1G));
Chris@82 175 T1D = VSUB(T1w, T1z);
Chris@82 176 T1A = VADD(T1w, T1z);
Chris@82 177 T1C = VFNMS(LDK(KP250000000), T1A, T1t);
Chris@82 178 T1B = VCONJ(VMUL(LDK(KP500000000), VADD(T1t, T1A)));
Chris@82 179 T1K = VFMA(LDK(KP559016994), T1D, T1C);
Chris@82 180 T1E = VFNMS(LDK(KP559016994), T1D, T1C);
Chris@82 181 ST(&(Rm[WS(rs, 9)]), T1B, -ms, &(Rm[WS(rs, 1)]));
Chris@82 182 T1N = VCONJ(VMUL(LDK(KP500000000), VFMAI(T1L, T1K)));
Chris@82 183 ST(&(Rm[WS(rs, 5)]), T1N, -ms, &(Rm[WS(rs, 1)]));
Chris@82 184 T1I = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T1H, T1E)));
Chris@82 185 ST(&(Rm[WS(rs, 1)]), T1I, -ms, &(Rm[WS(rs, 1)]));
Chris@82 186 T1J = VMUL(LDK(KP500000000), VFMAI(T1H, T1E));
Chris@82 187 ST(&(Rp[WS(rs, 2)]), T1J, ms, &(Rp[0]));
Chris@82 188 T1M = VMUL(LDK(KP500000000), VFNMSI(T1L, T1K));
Chris@82 189 ST(&(Rp[WS(rs, 6)]), T1M, ms, &(Rp[0]));
Chris@82 190 {
Chris@82 191 V T1m, T1q, T1g, T1p, T1j, T1k, T1l, T1i, T1f, T1h, T1s, T1n, T1o, T1r, T2e;
Chris@82 192 V T2A, T2o, T2u, T2l, T2B, T2p, T2x, T2d, T2t, T2a, T2s, T28, T2k, T2w, T2h;
Chris@82 193 V T2v, T2f, T2m, T2C, T2D, T2n, T2q, T2y, T2z, T2r;
Chris@82 194 T1k = VADD(Tx, TM);
Chris@82 195 T1l = VADD(T12, T1d);
Chris@82 196 T1m = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1l, T1k));
Chris@82 197 T1q = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1k, T1l));
Chris@82 198 T1i = VSUB(TN, T1e);
Chris@82 199 T1f = VADD(TN, T1e);
Chris@82 200 T1h = VFNMS(LDK(KP250000000), T1f, Ti);
Chris@82 201 T1g = VMUL(LDK(KP500000000), VADD(Ti, T1f));
Chris@82 202 T1p = VFNMS(LDK(KP559016994), T1i, T1h);
Chris@82 203 T1j = VFMA(LDK(KP559016994), T1i, T1h);
Chris@82 204 ST(&(Rp[0]), T1g, ms, &(Rp[0]));
Chris@82 205 T1s = VCONJ(VMUL(LDK(KP500000000), VFMAI(T1q, T1p)));
Chris@82 206 ST(&(Rm[WS(rs, 7)]), T1s, -ms, &(Rm[WS(rs, 1)]));
Chris@82 207 T1n = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T1m, T1j)));
Chris@82 208 ST(&(Rm[WS(rs, 3)]), T1n, -ms, &(Rm[WS(rs, 1)]));
Chris@82 209 T1o = VMUL(LDK(KP500000000), VFMAI(T1m, T1j));
Chris@82 210 ST(&(Rp[WS(rs, 4)]), T1o, ms, &(Rp[0]));
Chris@82 211 T1r = VMUL(LDK(KP500000000), VFNMSI(T1q, T1p));
Chris@82 212 ST(&(Rp[WS(rs, 8)]), T1r, ms, &(Rp[0]));
Chris@82 213 T2d = VFMA(LDK(KP618033988), T2c, T2b);
Chris@82 214 T2t = VFNMS(LDK(KP618033988), T2b, T2c);
Chris@82 215 T28 = VFNMS(LDK(KP250000000), T1V, T1O);
Chris@82 216 T2a = VFMA(LDK(KP559016994), T29, T28);
Chris@82 217 T2s = VFNMS(LDK(KP559016994), T29, T28);
Chris@82 218 T2e = VFNMS(LDK(KP951056516), T2d, T2a);
Chris@82 219 T2A = VFMA(LDK(KP951056516), T2t, T2s);
Chris@82 220 T2o = VFMA(LDK(KP951056516), T2d, T2a);
Chris@82 221 T2u = VFNMS(LDK(KP951056516), T2t, T2s);
Chris@82 222 T2k = VFMA(LDK(KP618033988), T2j, T2i);
Chris@82 223 T2w = VFNMS(LDK(KP618033988), T2i, T2j);
Chris@82 224 T2f = VFNMS(LDK(KP250000000), T24, T1X);
Chris@82 225 T2h = VFNMS(LDK(KP559016994), T2g, T2f);
Chris@82 226 T2v = VFMA(LDK(KP559016994), T2g, T2f);
Chris@82 227 T2l = VFNMS(LDK(KP951056516), T2k, T2h);
Chris@82 228 T2B = VFMA(LDK(KP951056516), T2w, T2v);
Chris@82 229 T2p = VFMA(LDK(KP951056516), T2k, T2h);
Chris@82 230 T2x = VFNMS(LDK(KP951056516), T2w, T2v);
Chris@82 231 T2m = VMUL(LDK(KP500000000), VFNMSI(T2l, T2e));
Chris@82 232 ST(&(Rp[WS(rs, 9)]), T2m, ms, &(Rp[WS(rs, 1)]));
Chris@82 233 T2C = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T2B, T2A)));
Chris@82 234 ST(&(Rm[WS(rs, 6)]), T2C, -ms, &(Rm[0]));
Chris@82 235 T2D = VMUL(LDK(KP500000000), VFMAI(T2B, T2A));
Chris@82 236 ST(&(Rp[WS(rs, 7)]), T2D, ms, &(Rp[WS(rs, 1)]));
Chris@82 237 T2n = VCONJ(VMUL(LDK(KP500000000), VFMAI(T2l, T2e)));
Chris@82 238 ST(&(Rm[WS(rs, 8)]), T2n, -ms, &(Rm[0]));
Chris@82 239 T2q = VMUL(LDK(KP500000000), VFNMSI(T2p, T2o));
Chris@82 240 ST(&(Rp[WS(rs, 1)]), T2q, ms, &(Rp[WS(rs, 1)]));
Chris@82 241 T2y = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T2x, T2u)));
Chris@82 242 ST(&(Rm[WS(rs, 2)]), T2y, -ms, &(Rm[0]));
Chris@82 243 T2z = VMUL(LDK(KP500000000), VFMAI(T2x, T2u));
Chris@82 244 ST(&(Rp[WS(rs, 3)]), T2z, ms, &(Rp[WS(rs, 1)]));
Chris@82 245 T2r = VCONJ(VMUL(LDK(KP500000000), VFMAI(T2p, T2o)));
Chris@82 246 ST(&(Rm[0]), T2r, -ms, &(Rm[0]));
Chris@82 247 }
Chris@82 248 }
Chris@82 249 }
Chris@82 250 VLEAVE();
Chris@82 251 }
Chris@82 252
Chris@82 253 static const tw_instr twinstr[] = {
Chris@82 254 VTW(1, 1),
Chris@82 255 VTW(1, 2),
Chris@82 256 VTW(1, 3),
Chris@82 257 VTW(1, 4),
Chris@82 258 VTW(1, 5),
Chris@82 259 VTW(1, 6),
Chris@82 260 VTW(1, 7),
Chris@82 261 VTW(1, 8),
Chris@82 262 VTW(1, 9),
Chris@82 263 VTW(1, 10),
Chris@82 264 VTW(1, 11),
Chris@82 265 VTW(1, 12),
Chris@82 266 VTW(1, 13),
Chris@82 267 VTW(1, 14),
Chris@82 268 VTW(1, 15),
Chris@82 269 VTW(1, 16),
Chris@82 270 VTW(1, 17),
Chris@82 271 VTW(1, 18),
Chris@82 272 VTW(1, 19),
Chris@82 273 {TW_NEXT, VL, 0}
Chris@82 274 };
Chris@82 275
Chris@82 276 static const hc2c_desc desc = { 20, XSIMD_STRING("hc2cfdftv_20"), twinstr, &GENUS, {77, 62, 66, 0} };
Chris@82 277
Chris@82 278 void XSIMD(codelet_hc2cfdftv_20) (planner *p) {
Chris@82 279 X(khc2c_register) (p, hc2cfdftv_20, &desc, HC2C_VIA_DFT);
Chris@82 280 }
Chris@82 281 #else
Chris@82 282
Chris@82 283 /* Generated by: ../../../genfft/gen_hc2cdft_c.native -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 20 -dit -name hc2cfdftv_20 -include rdft/simd/hc2cfv.h */
Chris@82 284
Chris@82 285 /*
Chris@82 286 * This function contains 143 FP additions, 77 FP multiplications,
Chris@82 287 * (or, 131 additions, 65 multiplications, 12 fused multiply/add),
Chris@82 288 * 141 stack variables, 9 constants, and 40 memory accesses
Chris@82 289 */
Chris@82 290 #include "rdft/simd/hc2cfv.h"
Chris@82 291
Chris@82 292 static void hc2cfdftv_20(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 293 {
Chris@82 294 DVK(KP293892626, +0.293892626146236564584352977319536384298826219);
Chris@82 295 DVK(KP475528258, +0.475528258147576786058219666689691071702849317);
Chris@82 296 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82 297 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82 298 DVK(KP125000000, +0.125000000000000000000000000000000000000000000);
Chris@82 299 DVK(KP279508497, +0.279508497187473712051146708591409529430077295);
Chris@82 300 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@82 301 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82 302 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 303 {
Chris@82 304 INT m;
Chris@82 305 for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 38)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 38), MAKE_VOLATILE_STRIDE(80, rs)) {
Chris@82 306 V TW, T1x, T2i, T2A, T1r, T1s, T1a, T1y, T1l, Tn, TK, TL, T1p, T1o, T27;
Chris@82 307 V T2t, T2a, T2u, T2e, T2C, T20, T2w, T23, T2x, T2d, T2B, T1W, T1X, T1U, T1V;
Chris@82 308 V T2z, T2K, T2G, T2N, T2J, T2v, T2y, T2F, T2D, T2E, T2M, T2H, T2I, T2L;
Chris@82 309 {
Chris@82 310 V T1u, T5, Tg, T1c, TV, T13, Ta, T1w, TQ, T11, TI, T1j, Tx, T18, Tl;
Chris@82 311 V T1e, TD, T1h, Ts, T16, T2g, T2h, T14, T19, T1f, T1k, Tb, Tm, Ty, TJ;
Chris@82 312 V T25, T26, T28, T29, T1Y, T1Z, T21, T22;
Chris@82 313 {
Chris@82 314 V T4, T3, T2, T1, Tf, Te, Td, Tc, T1b, TU, TT, TS, TR, T12, T9;
Chris@82 315 V T8, T7, T6, T1v, TP, TO, TN, TM, T10, TH, TG, TF, TE, T1i, Tw;
Chris@82 316 V Tv, Tu, Tt, T17, Tk, Tj, Ti, Th, T1d, TC, TB, TA, Tz, T1g, Tr;
Chris@82 317 V Tq, Tp, To, T15;
Chris@82 318 T4 = LD(&(Rp[0]), ms, &(Rp[0]));
Chris@82 319 T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
Chris@82 320 T3 = VCONJ(T2);
Chris@82 321 T1u = VADD(T4, T3);
Chris@82 322 T1 = LDW(&(W[0]));
Chris@82 323 T5 = VZMULIJ(T1, VSUB(T3, T4));
Chris@82 324 Tf = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
Chris@82 325 Td = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Chris@82 326 Te = VCONJ(Td);
Chris@82 327 Tc = LDW(&(W[TWVL * 16]));
Chris@82 328 Tg = VZMULIJ(Tc, VSUB(Te, Tf));
Chris@82 329 T1b = LDW(&(W[TWVL * 14]));
Chris@82 330 T1c = VZMULJ(T1b, VADD(Te, Tf));
Chris@82 331 TU = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 332 TS = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 333 TT = VCONJ(TS);
Chris@82 334 TR = LDW(&(W[TWVL * 28]));
Chris@82 335 TV = VZMULIJ(TR, VSUB(TT, TU));
Chris@82 336 T12 = LDW(&(W[TWVL * 26]));
Chris@82 337 T13 = VZMULJ(T12, VADD(TT, TU));
Chris@82 338 T9 = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 339 T7 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 340 T8 = VCONJ(T7);
Chris@82 341 T6 = LDW(&(W[TWVL * 20]));
Chris@82 342 Ta = VZMULIJ(T6, VSUB(T8, T9));
Chris@82 343 T1v = LDW(&(W[TWVL * 18]));
Chris@82 344 T1w = VZMULJ(T1v, VADD(T9, T8));
Chris@82 345 TP = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Chris@82 346 TN = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Chris@82 347 TO = VCONJ(TN);
Chris@82 348 TM = LDW(&(W[TWVL * 8]));
Chris@82 349 TQ = VZMULIJ(TM, VSUB(TO, TP));
Chris@82 350 T10 = LDW(&(W[TWVL * 6]));
Chris@82 351 T11 = VZMULJ(T10, VADD(TO, TP));
Chris@82 352 TH = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 353 TF = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 354 TG = VCONJ(TF);
Chris@82 355 TE = LDW(&(W[TWVL * 4]));
Chris@82 356 TI = VZMULIJ(TE, VSUB(TG, TH));
Chris@82 357 T1i = LDW(&(W[TWVL * 2]));
Chris@82 358 T1j = VZMULJ(T1i, VADD(TG, TH));
Chris@82 359 Tw = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 360 Tu = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 361 Tv = VCONJ(Tu);
Chris@82 362 Tt = LDW(&(W[TWVL * 12]));
Chris@82 363 Tx = VZMULIJ(Tt, VSUB(Tv, Tw));
Chris@82 364 T17 = LDW(&(W[TWVL * 10]));
Chris@82 365 T18 = VZMULJ(T17, VADD(Tw, Tv));
Chris@82 366 Tk = LD(&(Rp[WS(rs, 9)]), ms, &(Rp[WS(rs, 1)]));
Chris@82 367 Ti = LD(&(Rm[WS(rs, 9)]), -ms, &(Rm[WS(rs, 1)]));
Chris@82 368 Tj = VCONJ(Ti);
Chris@82 369 Th = LDW(&(W[TWVL * 36]));
Chris@82 370 Tl = VZMULIJ(Th, VSUB(Tj, Tk));
Chris@82 371 T1d = LDW(&(W[TWVL * 34]));
Chris@82 372 T1e = VZMULJ(T1d, VADD(Tj, Tk));
Chris@82 373 TC = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
Chris@82 374 TA = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
Chris@82 375 TB = VCONJ(TA);
Chris@82 376 Tz = LDW(&(W[TWVL * 24]));
Chris@82 377 TD = VZMULIJ(Tz, VSUB(TB, TC));
Chris@82 378 T1g = LDW(&(W[TWVL * 22]));
Chris@82 379 T1h = VZMULJ(T1g, VADD(TB, TC));
Chris@82 380 Tr = LD(&(Rp[WS(rs, 8)]), ms, &(Rp[0]));
Chris@82 381 Tp = LD(&(Rm[WS(rs, 8)]), -ms, &(Rm[0]));
Chris@82 382 Tq = VCONJ(Tp);
Chris@82 383 To = LDW(&(W[TWVL * 32]));
Chris@82 384 Ts = VZMULIJ(To, VSUB(Tq, Tr));
Chris@82 385 T15 = LDW(&(W[TWVL * 30]));
Chris@82 386 T16 = VZMULJ(T15, VADD(Tr, Tq));
Chris@82 387 }
Chris@82 388 TW = VSUB(TQ, TV);
Chris@82 389 T1x = VSUB(T1u, T1w);
Chris@82 390 T2g = VADD(T1u, T1w);
Chris@82 391 T2h = VADD(TQ, TV);
Chris@82 392 T2i = VADD(T2g, T2h);
Chris@82 393 T2A = VSUB(T2g, T2h);
Chris@82 394 T14 = VSUB(T11, T13);
Chris@82 395 T19 = VSUB(T16, T18);
Chris@82 396 T1r = VADD(T14, T19);
Chris@82 397 T1f = VSUB(T1c, T1e);
Chris@82 398 T1k = VSUB(T1h, T1j);
Chris@82 399 T1s = VADD(T1f, T1k);
Chris@82 400 T1a = VSUB(T14, T19);
Chris@82 401 T1y = VADD(T1r, T1s);
Chris@82 402 T1l = VSUB(T1f, T1k);
Chris@82 403 Tb = VSUB(T5, Ta);
Chris@82 404 Tm = VSUB(Tg, Tl);
Chris@82 405 Tn = VADD(Tb, Tm);
Chris@82 406 Ty = VSUB(Ts, Tx);
Chris@82 407 TJ = VSUB(TD, TI);
Chris@82 408 TK = VADD(Ty, TJ);
Chris@82 409 TL = VADD(Tn, TK);
Chris@82 410 T1p = VSUB(Ty, TJ);
Chris@82 411 T1o = VSUB(Tb, Tm);
Chris@82 412 T25 = VADD(T1c, T1e);
Chris@82 413 T26 = VADD(TD, TI);
Chris@82 414 T27 = VADD(T25, T26);
Chris@82 415 T2t = VSUB(T25, T26);
Chris@82 416 T28 = VADD(Ts, Tx);
Chris@82 417 T29 = VADD(T1h, T1j);
Chris@82 418 T2a = VADD(T28, T29);
Chris@82 419 T2u = VSUB(T29, T28);
Chris@82 420 T2e = VADD(T27, T2a);
Chris@82 421 T2C = VADD(T2t, T2u);
Chris@82 422 T1Y = VADD(T11, T13);
Chris@82 423 T1Z = VADD(Tg, Tl);
Chris@82 424 T20 = VADD(T1Y, T1Z);
Chris@82 425 T2w = VSUB(T1Y, T1Z);
Chris@82 426 T21 = VADD(T5, Ta);
Chris@82 427 T22 = VADD(T16, T18);
Chris@82 428 T23 = VADD(T21, T22);
Chris@82 429 T2x = VSUB(T22, T21);
Chris@82 430 T2d = VADD(T20, T23);
Chris@82 431 T2B = VADD(T2w, T2x);
Chris@82 432 }
Chris@82 433 T1U = VADD(T1x, T1y);
Chris@82 434 T1V = VBYI(VADD(TW, TL));
Chris@82 435 T1W = VMUL(LDK(KP500000000), VSUB(T1U, T1V));
Chris@82 436 T1X = VCONJ(VMUL(LDK(KP500000000), VADD(T1V, T1U)));
Chris@82 437 ST(&(Rp[WS(rs, 5)]), T1W, ms, &(Rp[WS(rs, 1)]));
Chris@82 438 ST(&(Rm[WS(rs, 4)]), T1X, -ms, &(Rm[0]));
Chris@82 439 T2v = VSUB(T2t, T2u);
Chris@82 440 T2y = VSUB(T2w, T2x);
Chris@82 441 T2z = VMUL(LDK(KP500000000), VBYI(VFNMS(LDK(KP587785252), T2y, VMUL(LDK(KP951056516), T2v))));
Chris@82 442 T2K = VMUL(LDK(KP500000000), VBYI(VFMA(LDK(KP951056516), T2y, VMUL(LDK(KP587785252), T2v))));
Chris@82 443 T2F = VMUL(LDK(KP279508497), VSUB(T2B, T2C));
Chris@82 444 T2D = VADD(T2B, T2C);
Chris@82 445 T2E = VFNMS(LDK(KP125000000), T2D, VMUL(LDK(KP500000000), T2A));
Chris@82 446 T2G = VSUB(T2E, T2F);
Chris@82 447 T2N = VCONJ(VMUL(LDK(KP500000000), VADD(T2A, T2D)));
Chris@82 448 T2J = VADD(T2F, T2E);
Chris@82 449 ST(&(Rm[WS(rs, 9)]), T2N, -ms, &(Rm[WS(rs, 1)]));
Chris@82 450 T2M = VCONJ(VADD(T2K, T2J));
Chris@82 451 ST(&(Rm[WS(rs, 5)]), T2M, -ms, &(Rm[WS(rs, 1)]));
Chris@82 452 T2H = VADD(T2z, T2G);
Chris@82 453 ST(&(Rp[WS(rs, 2)]), T2H, ms, &(Rp[0]));
Chris@82 454 T2I = VCONJ(VSUB(T2G, T2z));
Chris@82 455 ST(&(Rm[WS(rs, 1)]), T2I, -ms, &(Rm[WS(rs, 1)]));
Chris@82 456 T2L = VSUB(T2J, T2K);
Chris@82 457 ST(&(Rp[WS(rs, 6)]), T2L, ms, &(Rp[0]));
Chris@82 458 {
Chris@82 459 V T2c, T2p, T2l, T2s, T2o, T24, T2b, T2f, T2j, T2k, T2r, T2m, T2n, T2q, T1n;
Chris@82 460 V T1Q, T1E, T1K, T1B, T1R, T1F, T1N, T1m, T1J, TZ, T1I, TX, TY, T1q, T1M;
Chris@82 461 V T1A, T1L, T1t, T1z, T1C, T1S, T1T, T1D, T1G, T1O, T1P, T1H;
Chris@82 462 T24 = VSUB(T20, T23);
Chris@82 463 T2b = VSUB(T27, T2a);
Chris@82 464 T2c = VMUL(LDK(KP500000000), VBYI(VFMA(LDK(KP951056516), T24, VMUL(LDK(KP587785252), T2b))));
Chris@82 465 T2p = VMUL(LDK(KP500000000), VBYI(VFNMS(LDK(KP587785252), T24, VMUL(LDK(KP951056516), T2b))));
Chris@82 466 T2f = VMUL(LDK(KP279508497), VSUB(T2d, T2e));
Chris@82 467 T2j = VADD(T2d, T2e);
Chris@82 468 T2k = VFNMS(LDK(KP125000000), T2j, VMUL(LDK(KP500000000), T2i));
Chris@82 469 T2l = VADD(T2f, T2k);
Chris@82 470 T2s = VMUL(LDK(KP500000000), VADD(T2i, T2j));
Chris@82 471 T2o = VSUB(T2k, T2f);
Chris@82 472 ST(&(Rp[0]), T2s, ms, &(Rp[0]));
Chris@82 473 T2r = VCONJ(VADD(T2p, T2o));
Chris@82 474 ST(&(Rm[WS(rs, 7)]), T2r, -ms, &(Rm[WS(rs, 1)]));
Chris@82 475 T2m = VADD(T2c, T2l);
Chris@82 476 ST(&(Rp[WS(rs, 4)]), T2m, ms, &(Rp[0]));
Chris@82 477 T2n = VCONJ(VSUB(T2l, T2c));
Chris@82 478 ST(&(Rm[WS(rs, 3)]), T2n, -ms, &(Rm[WS(rs, 1)]));
Chris@82 479 T2q = VSUB(T2o, T2p);
Chris@82 480 ST(&(Rp[WS(rs, 8)]), T2q, ms, &(Rp[0]));
Chris@82 481 T1m = VFMA(LDK(KP951056516), T1a, VMUL(LDK(KP587785252), T1l));
Chris@82 482 T1J = VFNMS(LDK(KP587785252), T1a, VMUL(LDK(KP951056516), T1l));
Chris@82 483 TX = VFMS(LDK(KP250000000), TL, TW);
Chris@82 484 TY = VMUL(LDK(KP559016994), VSUB(TK, Tn));
Chris@82 485 TZ = VADD(TX, TY);
Chris@82 486 T1I = VSUB(TY, TX);
Chris@82 487 T1n = VMUL(LDK(KP500000000), VBYI(VSUB(TZ, T1m)));
Chris@82 488 T1Q = VMUL(LDK(KP500000000), VBYI(VADD(T1I, T1J)));
Chris@82 489 T1E = VMUL(LDK(KP500000000), VBYI(VADD(TZ, T1m)));
Chris@82 490 T1K = VMUL(LDK(KP500000000), VBYI(VSUB(T1I, T1J)));
Chris@82 491 T1q = VFMA(LDK(KP475528258), T1o, VMUL(LDK(KP293892626), T1p));
Chris@82 492 T1M = VFNMS(LDK(KP293892626), T1o, VMUL(LDK(KP475528258), T1p));
Chris@82 493 T1t = VMUL(LDK(KP279508497), VSUB(T1r, T1s));
Chris@82 494 T1z = VFNMS(LDK(KP125000000), T1y, VMUL(LDK(KP500000000), T1x));
Chris@82 495 T1A = VADD(T1t, T1z);
Chris@82 496 T1L = VSUB(T1z, T1t);
Chris@82 497 T1B = VADD(T1q, T1A);
Chris@82 498 T1R = VADD(T1M, T1L);
Chris@82 499 T1F = VSUB(T1A, T1q);
Chris@82 500 T1N = VSUB(T1L, T1M);
Chris@82 501 T1C = VADD(T1n, T1B);
Chris@82 502 ST(&(Rp[WS(rs, 1)]), T1C, ms, &(Rp[WS(rs, 1)]));
Chris@82 503 T1S = VADD(T1Q, T1R);
Chris@82 504 ST(&(Rp[WS(rs, 7)]), T1S, ms, &(Rp[WS(rs, 1)]));
Chris@82 505 T1T = VCONJ(VSUB(T1R, T1Q));
Chris@82 506 ST(&(Rm[WS(rs, 6)]), T1T, -ms, &(Rm[0]));
Chris@82 507 T1D = VCONJ(VSUB(T1B, T1n));
Chris@82 508 ST(&(Rm[0]), T1D, -ms, &(Rm[0]));
Chris@82 509 T1G = VADD(T1E, T1F);
Chris@82 510 ST(&(Rp[WS(rs, 9)]), T1G, ms, &(Rp[WS(rs, 1)]));
Chris@82 511 T1O = VADD(T1K, T1N);
Chris@82 512 ST(&(Rp[WS(rs, 3)]), T1O, ms, &(Rp[WS(rs, 1)]));
Chris@82 513 T1P = VCONJ(VSUB(T1N, T1K));
Chris@82 514 ST(&(Rm[WS(rs, 2)]), T1P, -ms, &(Rm[0]));
Chris@82 515 T1H = VCONJ(VSUB(T1F, T1E));
Chris@82 516 ST(&(Rm[WS(rs, 8)]), T1H, -ms, &(Rm[0]));
Chris@82 517 }
Chris@82 518 }
Chris@82 519 }
Chris@82 520 VLEAVE();
Chris@82 521 }
Chris@82 522
Chris@82 523 static const tw_instr twinstr[] = {
Chris@82 524 VTW(1, 1),
Chris@82 525 VTW(1, 2),
Chris@82 526 VTW(1, 3),
Chris@82 527 VTW(1, 4),
Chris@82 528 VTW(1, 5),
Chris@82 529 VTW(1, 6),
Chris@82 530 VTW(1, 7),
Chris@82 531 VTW(1, 8),
Chris@82 532 VTW(1, 9),
Chris@82 533 VTW(1, 10),
Chris@82 534 VTW(1, 11),
Chris@82 535 VTW(1, 12),
Chris@82 536 VTW(1, 13),
Chris@82 537 VTW(1, 14),
Chris@82 538 VTW(1, 15),
Chris@82 539 VTW(1, 16),
Chris@82 540 VTW(1, 17),
Chris@82 541 VTW(1, 18),
Chris@82 542 VTW(1, 19),
Chris@82 543 {TW_NEXT, VL, 0}
Chris@82 544 };
Chris@82 545
Chris@82 546 static const hc2c_desc desc = { 20, XSIMD_STRING("hc2cfdftv_20"), twinstr, &GENUS, {131, 65, 12, 0} };
Chris@82 547
Chris@82 548 void XSIMD(codelet_hc2cfdftv_20) (planner *p) {
Chris@82 549 X(khc2c_register) (p, hc2cfdftv_20, &desc, HC2C_VIA_DFT);
Chris@82 550 }
Chris@82 551 #endif