Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/rdft/simd/common/hc2cbdftv_20.c @ 56:af97cad61ff0

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add updated build of PortAudio for OSX
author Chris Cannam <cannam@all-day-breakfast.com>
date Tue, 03 Jan 2017 15:10:52 +0000
parents 2cd0e3b3e1fd
children
rev   line source
Chris@42 1 /*
Chris@42 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42 4 *
Chris@42 5 * This program is free software; you can redistribute it and/or modify
Chris@42 6 * it under the terms of the GNU General Public License as published by
Chris@42 7 * the Free Software Foundation; either version 2 of the License, or
Chris@42 8 * (at your option) any later version.
Chris@42 9 *
Chris@42 10 * This program is distributed in the hope that it will be useful,
Chris@42 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@42 13 * GNU General Public License for more details.
Chris@42 14 *
Chris@42 15 * You should have received a copy of the GNU General Public License
Chris@42 16 * along with this program; if not, write to the Free Software
Chris@42 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@42 18 *
Chris@42 19 */
Chris@42 20
Chris@42 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@42 22 /* Generated on Sat Jul 30 16:52:46 EDT 2016 */
Chris@42 23
Chris@42 24 #include "codelet-rdft.h"
Chris@42 25
Chris@42 26 #ifdef HAVE_FMA
Chris@42 27
Chris@42 28 /* Generated by: ../../../genfft/gen_hc2cdft_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 20 -dif -sign 1 -name hc2cbdftv_20 -include hc2cbv.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 143 FP additions, 108 FP multiplications,
Chris@42 32 * (or, 77 additions, 42 multiplications, 66 fused multiply/add),
Chris@42 33 * 134 stack variables, 4 constants, and 40 memory accesses
Chris@42 34 */
Chris@42 35 #include "hc2cbv.h"
Chris@42 36
Chris@42 37 static void hc2cbdftv_20(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 38 {
Chris@42 39 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42 40 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42 41 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42 42 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@42 43 {
Chris@42 44 INT m;
Chris@42 45 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@42 46 V T1M, T1T, T4, TF, T12, Te, T16, Ts, Tb, TN, TA, TG, TU, T1Y, T11;
Chris@42 47 V T1e, T29, T21, T15, Th, T13, Tp;
Chris@42 48 {
Chris@42 49 V TS, TT, Tf, T10, T20, T1Z, TX, Tg, Tn, To, T2, T3, TD, TE, T8;
Chris@42 50 V TV, T7, TZ, Tz, T9, Tu, Tv, T5, T6, Tx, Ty, Tc, Td, Tq, Tr;
Chris@42 51 V TY, Ta, TW, Tw;
Chris@42 52 T2 = LD(&(Rp[0]), ms, &(Rp[0]));
Chris@42 53 T3 = LD(&(Rm[WS(rs, 9)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 54 TD = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 55 TE = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Chris@42 56 T5 = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
Chris@42 57 T6 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 58 Tx = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 59 Ty = LD(&(Rm[WS(rs, 8)]), -ms, &(Rm[0]));
Chris@42 60 T8 = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
Chris@42 61 TS = VFMACONJ(T3, T2);
Chris@42 62 T4 = VFNMSCONJ(T3, T2);
Chris@42 63 TT = VFMACONJ(TE, TD);
Chris@42 64 TF = VFNMSCONJ(TE, TD);
Chris@42 65 TV = VFMACONJ(T6, T5);
Chris@42 66 T7 = VFNMSCONJ(T6, T5);
Chris@42 67 TZ = VFMACONJ(Ty, Tx);
Chris@42 68 Tz = VFNMSCONJ(Ty, Tx);
Chris@42 69 T9 = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 70 Tu = LD(&(Rp[WS(rs, 9)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 71 Tv = LD(&(Rm[0]), -ms, &(Rm[0]));
Chris@42 72 Tc = LD(&(Rp[WS(rs, 8)]), ms, &(Rp[0]));
Chris@42 73 Td = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 74 Tq = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 75 Tr = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Chris@42 76 Tf = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Chris@42 77 TY = VFMACONJ(T9, T8);
Chris@42 78 Ta = VFMSCONJ(T9, T8);
Chris@42 79 TW = VFMACONJ(Tv, Tu);
Chris@42 80 Tw = VFNMSCONJ(Tv, Tu);
Chris@42 81 T12 = VFMACONJ(Td, Tc);
Chris@42 82 Te = VFNMSCONJ(Td, Tc);
Chris@42 83 T16 = VFMACONJ(Tr, Tq);
Chris@42 84 Ts = VFMSCONJ(Tr, Tq);
Chris@42 85 T10 = VSUB(TY, TZ);
Chris@42 86 T20 = VADD(TY, TZ);
Chris@42 87 Tb = VADD(T7, Ta);
Chris@42 88 TN = VSUB(T7, Ta);
Chris@42 89 T1Z = VADD(TV, TW);
Chris@42 90 TX = VSUB(TV, TW);
Chris@42 91 TA = VSUB(Tw, Tz);
Chris@42 92 TG = VADD(Tw, Tz);
Chris@42 93 Tg = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 94 Tn = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 95 To = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
Chris@42 96 TU = VSUB(TS, TT);
Chris@42 97 T1Y = VADD(TS, TT);
Chris@42 98 T11 = VADD(TX, T10);
Chris@42 99 T1e = VSUB(TX, T10);
Chris@42 100 T29 = VSUB(T1Z, T20);
Chris@42 101 T21 = VADD(T1Z, T20);
Chris@42 102 T15 = VFMACONJ(Tg, Tf);
Chris@42 103 Th = VFMSCONJ(Tg, Tf);
Chris@42 104 T13 = VFMACONJ(To, Tn);
Chris@42 105 Tp = VFMSCONJ(To, Tn);
Chris@42 106 }
Chris@42 107 {
Chris@42 108 V T1S, T2B, T1W, T1I, T2q, T2w, T2i, T2c, T1C, T1K, T1s, T1g, T1, T2t, T1v;
Chris@42 109 V T1Q, T2A, T1q, T2m, TC, T1w, TP, T1x, T2f, T2r, T2g, T1E, T1D, T2y, T2x;
Chris@42 110 V T1i, T1h, T2D, T2C, T2s, T1t, T1u, T1y, T2u, TQ, T2d, T2e, T1U, T1L, T2j;
Chris@42 111 V T2k;
Chris@42 112 {
Chris@42 113 V T1R, T1F, T1V, T1o, TO, Tl, T1d, T2a, T1l, TB, TK, T1G, Tk, T1b, T19;
Chris@42 114 V T27, T25, T1H, TJ, T17, T23, TM, Ti, T14, T22, Tt, TH, Tj, T18, T24;
Chris@42 115 V TI, T2b, T2p, T1X, T2v, T2h, T2n, T1B, T1f, T28, T2o, T1a, TR, T1J, T1r;
Chris@42 116 V T1z, T26, Tm, TL, T1O, T1m, T1j, T2z, T1N, T1p, T1P, T2l, T1c, T1A, T1n;
Chris@42 117 V T1k;
Chris@42 118 T1R = LDW(&(W[TWVL * 18]));
Chris@42 119 T17 = VSUB(T15, T16);
Chris@42 120 T23 = VADD(T15, T16);
Chris@42 121 TM = VSUB(Te, Th);
Chris@42 122 Ti = VADD(Te, Th);
Chris@42 123 T14 = VSUB(T12, T13);
Chris@42 124 T22 = VADD(T12, T13);
Chris@42 125 Tt = VSUB(Tp, Ts);
Chris@42 126 TH = VADD(Tp, Ts);
Chris@42 127 T1F = LDW(&(W[TWVL * 28]));
Chris@42 128 T1V = LDW(&(W[TWVL * 8]));
Chris@42 129 T1o = VFMA(LDK(KP618033988), TM, TN);
Chris@42 130 TO = VFNMS(LDK(KP618033988), TN, TM);
Chris@42 131 Tj = VADD(Tb, Ti);
Chris@42 132 Tl = VSUB(Tb, Ti);
Chris@42 133 T18 = VADD(T14, T17);
Chris@42 134 T1d = VSUB(T14, T17);
Chris@42 135 T24 = VADD(T22, T23);
Chris@42 136 T2a = VSUB(T22, T23);
Chris@42 137 T1l = VFMA(LDK(KP618033988), Tt, TA);
Chris@42 138 TB = VFNMS(LDK(KP618033988), TA, Tt);
Chris@42 139 TI = VADD(TG, TH);
Chris@42 140 TK = VSUB(TG, TH);
Chris@42 141 T1G = VADD(T4, Tj);
Chris@42 142 Tk = VFNMS(LDK(KP250000000), Tj, T4);
Chris@42 143 T1b = VSUB(T11, T18);
Chris@42 144 T19 = VADD(T11, T18);
Chris@42 145 T27 = VSUB(T21, T24);
Chris@42 146 T25 = VADD(T21, T24);
Chris@42 147 T1H = VADD(TF, TI);
Chris@42 148 TJ = VFNMS(LDK(KP250000000), TI, TF);
Chris@42 149 T2b = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T2a, T29));
Chris@42 150 T2p = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T29, T2a));
Chris@42 151 T1X = LDW(&(W[TWVL * 6]));
Chris@42 152 T1S = VZMUL(T1R, VADD(TU, T19));
Chris@42 153 T2v = LDW(&(W[TWVL * 22]));
Chris@42 154 T2B = VADD(T1Y, T25);
Chris@42 155 T26 = VFNMS(LDK(KP250000000), T25, T1Y);
Chris@42 156 T1W = VZMULI(T1V, VFMAI(T1H, T1G));
Chris@42 157 T1I = VZMULI(T1F, VFNMSI(T1H, T1G));
Chris@42 158 T2h = LDW(&(W[TWVL * 30]));
Chris@42 159 T2n = LDW(&(W[TWVL * 14]));
Chris@42 160 T1B = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1d, T1e));
Chris@42 161 T1f = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1e, T1d));
Chris@42 162 T28 = VFMA(LDK(KP559016994), T27, T26);
Chris@42 163 T2o = VFNMS(LDK(KP559016994), T27, T26);
Chris@42 164 T1a = VFNMS(LDK(KP250000000), T19, TU);
Chris@42 165 TR = LDW(&(W[TWVL * 2]));
Chris@42 166 T1J = LDW(&(W[TWVL * 26]));
Chris@42 167 T1r = LDW(&(W[TWVL * 34]));
Chris@42 168 T1z = LDW(&(W[TWVL * 10]));
Chris@42 169 T1k = VFMA(LDK(KP559016994), Tl, Tk);
Chris@42 170 Tm = VFNMS(LDK(KP559016994), Tl, Tk);
Chris@42 171 T2q = VZMUL(T2n, VFMAI(T2p, T2o));
Chris@42 172 T2w = VZMUL(T2v, VFNMSI(T2p, T2o));
Chris@42 173 T2i = VZMUL(T2h, VFMAI(T2b, T28));
Chris@42 174 T2c = VZMUL(T1X, VFNMSI(T2b, T28));
Chris@42 175 T1c = VFNMS(LDK(KP559016994), T1b, T1a);
Chris@42 176 T1A = VFMA(LDK(KP559016994), T1b, T1a);
Chris@42 177 TL = VFNMS(LDK(KP559016994), TK, TJ);
Chris@42 178 T1n = VFMA(LDK(KP559016994), TK, TJ);
Chris@42 179 T1O = VFMA(LDK(KP951056516), T1l, T1k);
Chris@42 180 T1m = VFNMS(LDK(KP951056516), T1l, T1k);
Chris@42 181 T1j = LDW(&(W[TWVL * 36]));
Chris@42 182 T2z = LDW(&(W[0]));
Chris@42 183 T1N = LDW(&(W[TWVL * 20]));
Chris@42 184 T1C = VZMUL(T1z, VFMAI(T1B, T1A));
Chris@42 185 T1K = VZMUL(T1J, VFNMSI(T1B, T1A));
Chris@42 186 T1s = VZMUL(T1r, VFMAI(T1f, T1c));
Chris@42 187 T1g = VZMUL(TR, VFNMSI(T1f, T1c));
Chris@42 188 T1p = VFMA(LDK(KP951056516), T1o, T1n);
Chris@42 189 T1P = VFNMS(LDK(KP951056516), T1o, T1n);
Chris@42 190 T2l = LDW(&(W[TWVL * 16]));
Chris@42 191 T1 = LDW(&(W[TWVL * 4]));
Chris@42 192 T2t = LDW(&(W[TWVL * 24]));
Chris@42 193 T1v = LDW(&(W[TWVL * 12]));
Chris@42 194 T1Q = VZMULI(T1N, VFNMSI(T1P, T1O));
Chris@42 195 T2A = VZMULI(T2z, VFMAI(T1p, T1m));
Chris@42 196 T1q = VZMULI(T1j, VFNMSI(T1p, T1m));
Chris@42 197 T2m = VZMULI(T2l, VFMAI(T1P, T1O));
Chris@42 198 TC = VFMA(LDK(KP951056516), TB, Tm);
Chris@42 199 T1w = VFNMS(LDK(KP951056516), TB, Tm);
Chris@42 200 TP = VFNMS(LDK(KP951056516), TO, TL);
Chris@42 201 T1x = VFMA(LDK(KP951056516), TO, TL);
Chris@42 202 T2f = LDW(&(W[TWVL * 32]));
Chris@42 203 }
Chris@42 204 T2D = VCONJ(VSUB(T2B, T2A));
Chris@42 205 T2C = VADD(T2A, T2B);
Chris@42 206 T2s = VCONJ(VSUB(T2q, T2m));
Chris@42 207 T2r = VADD(T2m, T2q);
Chris@42 208 T1t = VADD(T1q, T1s);
Chris@42 209 T1u = VCONJ(VSUB(T1s, T1q));
Chris@42 210 T1y = VZMULI(T1v, VFNMSI(T1x, T1w));
Chris@42 211 T2u = VZMULI(T2t, VFMAI(T1x, T1w));
Chris@42 212 TQ = VZMULI(T1, VFNMSI(TP, TC));
Chris@42 213 T2g = VZMULI(T2f, VFMAI(TP, TC));
Chris@42 214 ST(&(Rm[0]), T2D, -ms, &(Rm[0]));
Chris@42 215 ST(&(Rp[0]), T2C, ms, &(Rp[0]));
Chris@42 216 ST(&(Rm[WS(rs, 4)]), T2s, -ms, &(Rm[0]));
Chris@42 217 ST(&(Rm[WS(rs, 9)]), T1u, -ms, &(Rm[WS(rs, 1)]));
Chris@42 218 T1E = VCONJ(VSUB(T1C, T1y));
Chris@42 219 T1D = VADD(T1y, T1C);
Chris@42 220 T2y = VCONJ(VSUB(T2w, T2u));
Chris@42 221 T2x = VADD(T2u, T2w);
Chris@42 222 T1i = VCONJ(VSUB(T1g, TQ));
Chris@42 223 T1h = VADD(TQ, T1g);
Chris@42 224 ST(&(Rp[WS(rs, 9)]), T1t, ms, &(Rp[WS(rs, 1)]));
Chris@42 225 T1L = VADD(T1I, T1K);
Chris@42 226 T1M = VCONJ(VSUB(T1K, T1I));
Chris@42 227 ST(&(Rp[WS(rs, 3)]), T1D, ms, &(Rp[WS(rs, 1)]));
Chris@42 228 ST(&(Rm[WS(rs, 6)]), T2y, -ms, &(Rm[0]));
Chris@42 229 ST(&(Rp[WS(rs, 6)]), T2x, ms, &(Rp[0]));
Chris@42 230 ST(&(Rm[WS(rs, 1)]), T1i, -ms, &(Rm[WS(rs, 1)]));
Chris@42 231 ST(&(Rp[WS(rs, 1)]), T1h, ms, &(Rp[WS(rs, 1)]));
Chris@42 232 T2d = VADD(T1W, T2c);
Chris@42 233 T2e = VCONJ(VSUB(T2c, T1W));
Chris@42 234 ST(&(Rm[WS(rs, 3)]), T1E, -ms, &(Rm[WS(rs, 1)]));
Chris@42 235 ST(&(Rp[WS(rs, 7)]), T1L, ms, &(Rp[WS(rs, 1)]));
Chris@42 236 T1U = VCONJ(VSUB(T1S, T1Q));
Chris@42 237 T1T = VADD(T1Q, T1S);
Chris@42 238 T2j = VADD(T2g, T2i);
Chris@42 239 T2k = VCONJ(VSUB(T2i, T2g));
Chris@42 240 ST(&(Rp[WS(rs, 2)]), T2d, ms, &(Rp[0]));
Chris@42 241 ST(&(Rp[WS(rs, 4)]), T2r, ms, &(Rp[0]));
Chris@42 242 ST(&(Rm[WS(rs, 5)]), T1U, -ms, &(Rm[WS(rs, 1)]));
Chris@42 243 ST(&(Rm[WS(rs, 2)]), T2e, -ms, &(Rm[0]));
Chris@42 244 ST(&(Rp[WS(rs, 8)]), T2j, ms, &(Rp[0]));
Chris@42 245 ST(&(Rm[WS(rs, 8)]), T2k, -ms, &(Rm[0]));
Chris@42 246 }
Chris@42 247 ST(&(Rp[WS(rs, 5)]), T1T, ms, &(Rp[WS(rs, 1)]));
Chris@42 248 ST(&(Rm[WS(rs, 7)]), T1M, -ms, &(Rm[WS(rs, 1)]));
Chris@42 249 }
Chris@42 250 }
Chris@42 251 VLEAVE();
Chris@42 252 }
Chris@42 253
Chris@42 254 static const tw_instr twinstr[] = {
Chris@42 255 VTW(1, 1),
Chris@42 256 VTW(1, 2),
Chris@42 257 VTW(1, 3),
Chris@42 258 VTW(1, 4),
Chris@42 259 VTW(1, 5),
Chris@42 260 VTW(1, 6),
Chris@42 261 VTW(1, 7),
Chris@42 262 VTW(1, 8),
Chris@42 263 VTW(1, 9),
Chris@42 264 VTW(1, 10),
Chris@42 265 VTW(1, 11),
Chris@42 266 VTW(1, 12),
Chris@42 267 VTW(1, 13),
Chris@42 268 VTW(1, 14),
Chris@42 269 VTW(1, 15),
Chris@42 270 VTW(1, 16),
Chris@42 271 VTW(1, 17),
Chris@42 272 VTW(1, 18),
Chris@42 273 VTW(1, 19),
Chris@42 274 {TW_NEXT, VL, 0}
Chris@42 275 };
Chris@42 276
Chris@42 277 static const hc2c_desc desc = { 20, XSIMD_STRING("hc2cbdftv_20"), twinstr, &GENUS, {77, 42, 66, 0} };
Chris@42 278
Chris@42 279 void XSIMD(codelet_hc2cbdftv_20) (planner *p) {
Chris@42 280 X(khc2c_register) (p, hc2cbdftv_20, &desc, HC2C_VIA_DFT);
Chris@42 281 }
Chris@42 282 #else /* HAVE_FMA */
Chris@42 283
Chris@42 284 /* Generated by: ../../../genfft/gen_hc2cdft_c.native -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 20 -dif -sign 1 -name hc2cbdftv_20 -include hc2cbv.h */
Chris@42 285
Chris@42 286 /*
Chris@42 287 * This function contains 143 FP additions, 62 FP multiplications,
Chris@42 288 * (or, 131 additions, 50 multiplications, 12 fused multiply/add),
Chris@42 289 * 114 stack variables, 4 constants, and 40 memory accesses
Chris@42 290 */
Chris@42 291 #include "hc2cbv.h"
Chris@42 292
Chris@42 293 static void hc2cbdftv_20(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 294 {
Chris@42 295 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42 296 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42 297 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42 298 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@42 299 {
Chris@42 300 INT m;
Chris@42 301 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@42 302 V TK, T1v, TY, T1x, T1j, T2f, TS, TT, TO, TU, T5, To, Tp, Tq, T2a;
Chris@42 303 V T2d, T2g, T2k, T2j, T1k, T1l, T18, T1m, T1f;
Chris@42 304 {
Chris@42 305 V T2, TP, T4, TR, TI, T1d, T9, T12, Td, T15, TE, T1a, Tv, T13, Tm;
Chris@42 306 V T1c, Tz, T16, Ti, T19, T3, TQ, TH, TG, TF, T6, T8, T7, Tc, Tb;
Chris@42 307 V Ta, TD, TC, TB, Ts, Tu, Tt, Tl, Tk, Tj, Tw, Ty, Tx, Tf, Th;
Chris@42 308 V Tg, TA, TJ, TW, TX, T1h, T1i, TM, TN, Te, Tn, T28, T29, T2b, T2c;
Chris@42 309 V T14, T17, T1b, T1e;
Chris@42 310 T2 = LD(&(Rp[0]), ms, &(Rp[0]));
Chris@42 311 TP = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 312 T3 = LD(&(Rm[WS(rs, 9)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 313 T4 = VCONJ(T3);
Chris@42 314 TQ = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Chris@42 315 TR = VCONJ(TQ);
Chris@42 316 TH = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 317 TF = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Chris@42 318 TG = VCONJ(TF);
Chris@42 319 TI = VSUB(TG, TH);
Chris@42 320 T1d = VADD(TG, TH);
Chris@42 321 T6 = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
Chris@42 322 T7 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 323 T8 = VCONJ(T7);
Chris@42 324 T9 = VSUB(T6, T8);
Chris@42 325 T12 = VADD(T6, T8);
Chris@42 326 Tc = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
Chris@42 327 Ta = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 328 Tb = VCONJ(Ta);
Chris@42 329 Td = VSUB(Tb, Tc);
Chris@42 330 T15 = VADD(Tb, Tc);
Chris@42 331 TD = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 332 TB = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
Chris@42 333 TC = VCONJ(TB);
Chris@42 334 TE = VSUB(TC, TD);
Chris@42 335 T1a = VADD(TC, TD);
Chris@42 336 Ts = LD(&(Rp[WS(rs, 9)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 337 Tt = LD(&(Rm[0]), -ms, &(Rm[0]));
Chris@42 338 Tu = VCONJ(Tt);
Chris@42 339 Tv = VSUB(Ts, Tu);
Chris@42 340 T13 = VADD(Ts, Tu);
Chris@42 341 Tl = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Chris@42 342 Tj = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 343 Tk = VCONJ(Tj);
Chris@42 344 Tm = VSUB(Tk, Tl);
Chris@42 345 T1c = VADD(Tk, Tl);
Chris@42 346 Tw = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 347 Tx = LD(&(Rm[WS(rs, 8)]), -ms, &(Rm[0]));
Chris@42 348 Ty = VCONJ(Tx);
Chris@42 349 Tz = VSUB(Tw, Ty);
Chris@42 350 T16 = VADD(Tw, Ty);
Chris@42 351 Tf = LD(&(Rp[WS(rs, 8)]), ms, &(Rp[0]));
Chris@42 352 Tg = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 353 Th = VCONJ(Tg);
Chris@42 354 Ti = VSUB(Tf, Th);
Chris@42 355 T19 = VADD(Tf, Th);
Chris@42 356 TA = VSUB(Tv, Tz);
Chris@42 357 TJ = VSUB(TE, TI);
Chris@42 358 TK = VFNMS(LDK(KP951056516), TJ, VMUL(LDK(KP587785252), TA));
Chris@42 359 T1v = VFMA(LDK(KP951056516), TA, VMUL(LDK(KP587785252), TJ));
Chris@42 360 TW = VSUB(T9, Td);
Chris@42 361 TX = VSUB(Ti, Tm);
Chris@42 362 TY = VFNMS(LDK(KP951056516), TX, VMUL(LDK(KP587785252), TW));
Chris@42 363 T1x = VFMA(LDK(KP951056516), TW, VMUL(LDK(KP587785252), TX));
Chris@42 364 T1h = VADD(T2, T4);
Chris@42 365 T1i = VADD(TP, TR);
Chris@42 366 T1j = VSUB(T1h, T1i);
Chris@42 367 T2f = VADD(T1h, T1i);
Chris@42 368 TS = VSUB(TP, TR);
Chris@42 369 TM = VADD(Tv, Tz);
Chris@42 370 TN = VADD(TE, TI);
Chris@42 371 TT = VADD(TM, TN);
Chris@42 372 TO = VMUL(LDK(KP559016994), VSUB(TM, TN));
Chris@42 373 TU = VFNMS(LDK(KP250000000), TT, TS);
Chris@42 374 T5 = VSUB(T2, T4);
Chris@42 375 Te = VADD(T9, Td);
Chris@42 376 Tn = VADD(Ti, Tm);
Chris@42 377 To = VADD(Te, Tn);
Chris@42 378 Tp = VFNMS(LDK(KP250000000), To, T5);
Chris@42 379 Tq = VMUL(LDK(KP559016994), VSUB(Te, Tn));
Chris@42 380 T28 = VADD(T12, T13);
Chris@42 381 T29 = VADD(T15, T16);
Chris@42 382 T2a = VADD(T28, T29);
Chris@42 383 T2b = VADD(T19, T1a);
Chris@42 384 T2c = VADD(T1c, T1d);
Chris@42 385 T2d = VADD(T2b, T2c);
Chris@42 386 T2g = VADD(T2a, T2d);
Chris@42 387 T2k = VSUB(T2b, T2c);
Chris@42 388 T2j = VSUB(T28, T29);
Chris@42 389 T14 = VSUB(T12, T13);
Chris@42 390 T17 = VSUB(T15, T16);
Chris@42 391 T1k = VADD(T14, T17);
Chris@42 392 T1b = VSUB(T19, T1a);
Chris@42 393 T1e = VSUB(T1c, T1d);
Chris@42 394 T1l = VADD(T1b, T1e);
Chris@42 395 T18 = VSUB(T14, T17);
Chris@42 396 T1m = VADD(T1k, T1l);
Chris@42 397 T1f = VSUB(T1b, T1e);
Chris@42 398 }
Chris@42 399 {
Chris@42 400 V T2L, T22, T1S, T26, T2m, T2G, T2s, T2A, T1q, T1U, T1C, T1M, T10, T2E, T1I;
Chris@42 401 V T2q, T1A, T2K, T20, T2w, T21, T1Q, T1R, T1P, T25, T1r, T1s, T2C, T2N, T1N;
Chris@42 402 V T2H, T2I, T2M, T1E, T1D, T1O, T1V, T2n, T2B, T24, T2o, T2t, T2u, T23, T1W;
Chris@42 403 T2L = VADD(T2f, T2g);
Chris@42 404 T21 = LDW(&(W[TWVL * 18]));
Chris@42 405 T22 = VZMUL(T21, VADD(T1j, T1m));
Chris@42 406 T1Q = VADD(T5, To);
Chris@42 407 T1R = VBYI(VADD(TS, TT));
Chris@42 408 T1P = LDW(&(W[TWVL * 28]));
Chris@42 409 T1S = VZMULI(T1P, VSUB(T1Q, T1R));
Chris@42 410 T25 = LDW(&(W[TWVL * 8]));
Chris@42 411 T26 = VZMULI(T25, VADD(T1Q, T1R));
Chris@42 412 {
Chris@42 413 V T2l, T2z, T2i, T2y, T2e, T2h, T27, T2F, T2r, T2x, T1g, T1K, T1p, T1L, T1n;
Chris@42 414 V T1o, T11, T1T, T1B, T1J, TL, T1G, TZ, T1H, Tr, TV, T1, T2D, T1F, T2p;
Chris@42 415 V T1w, T1Y, T1z, T1Z, T1u, T1y, T1t, T2J, T1X, T2v;
Chris@42 416 T2l = VBYI(VFMA(LDK(KP951056516), T2j, VMUL(LDK(KP587785252), T2k)));
Chris@42 417 T2z = VBYI(VFNMS(LDK(KP951056516), T2k, VMUL(LDK(KP587785252), T2j)));
Chris@42 418 T2e = VMUL(LDK(KP559016994), VSUB(T2a, T2d));
Chris@42 419 T2h = VFNMS(LDK(KP250000000), T2g, T2f);
Chris@42 420 T2i = VADD(T2e, T2h);
Chris@42 421 T2y = VSUB(T2h, T2e);
Chris@42 422 T27 = LDW(&(W[TWVL * 6]));
Chris@42 423 T2m = VZMUL(T27, VSUB(T2i, T2l));
Chris@42 424 T2F = LDW(&(W[TWVL * 22]));
Chris@42 425 T2G = VZMUL(T2F, VADD(T2z, T2y));
Chris@42 426 T2r = LDW(&(W[TWVL * 30]));
Chris@42 427 T2s = VZMUL(T2r, VADD(T2l, T2i));
Chris@42 428 T2x = LDW(&(W[TWVL * 14]));
Chris@42 429 T2A = VZMUL(T2x, VSUB(T2y, T2z));
Chris@42 430 T1g = VBYI(VFNMS(LDK(KP951056516), T1f, VMUL(LDK(KP587785252), T18)));
Chris@42 431 T1K = VBYI(VFMA(LDK(KP951056516), T18, VMUL(LDK(KP587785252), T1f)));
Chris@42 432 T1n = VFNMS(LDK(KP250000000), T1m, T1j);
Chris@42 433 T1o = VMUL(LDK(KP559016994), VSUB(T1k, T1l));
Chris@42 434 T1p = VSUB(T1n, T1o);
Chris@42 435 T1L = VADD(T1o, T1n);
Chris@42 436 T11 = LDW(&(W[TWVL * 2]));
Chris@42 437 T1q = VZMUL(T11, VADD(T1g, T1p));
Chris@42 438 T1T = LDW(&(W[TWVL * 26]));
Chris@42 439 T1U = VZMUL(T1T, VSUB(T1L, T1K));
Chris@42 440 T1B = LDW(&(W[TWVL * 34]));
Chris@42 441 T1C = VZMUL(T1B, VSUB(T1p, T1g));
Chris@42 442 T1J = LDW(&(W[TWVL * 10]));
Chris@42 443 T1M = VZMUL(T1J, VADD(T1K, T1L));
Chris@42 444 Tr = VSUB(Tp, Tq);
Chris@42 445 TL = VSUB(Tr, TK);
Chris@42 446 T1G = VADD(Tr, TK);
Chris@42 447 TV = VSUB(TO, TU);
Chris@42 448 TZ = VBYI(VSUB(TV, TY));
Chris@42 449 T1H = VBYI(VADD(TY, TV));
Chris@42 450 T1 = LDW(&(W[TWVL * 4]));
Chris@42 451 T10 = VZMULI(T1, VADD(TL, TZ));
Chris@42 452 T2D = LDW(&(W[TWVL * 24]));
Chris@42 453 T2E = VZMULI(T2D, VSUB(T1G, T1H));
Chris@42 454 T1F = LDW(&(W[TWVL * 12]));
Chris@42 455 T1I = VZMULI(T1F, VADD(T1G, T1H));
Chris@42 456 T2p = LDW(&(W[TWVL * 32]));
Chris@42 457 T2q = VZMULI(T2p, VSUB(TL, TZ));
Chris@42 458 T1u = VADD(Tq, Tp);
Chris@42 459 T1w = VSUB(T1u, T1v);
Chris@42 460 T1Y = VADD(T1u, T1v);
Chris@42 461 T1y = VADD(TO, TU);
Chris@42 462 T1z = VBYI(VADD(T1x, T1y));
Chris@42 463 T1Z = VBYI(VSUB(T1y, T1x));
Chris@42 464 T1t = LDW(&(W[TWVL * 36]));
Chris@42 465 T1A = VZMULI(T1t, VSUB(T1w, T1z));
Chris@42 466 T2J = LDW(&(W[0]));
Chris@42 467 T2K = VZMULI(T2J, VADD(T1w, T1z));
Chris@42 468 T1X = LDW(&(W[TWVL * 20]));
Chris@42 469 T20 = VZMULI(T1X, VSUB(T1Y, T1Z));
Chris@42 470 T2v = LDW(&(W[TWVL * 16]));
Chris@42 471 T2w = VZMULI(T2v, VADD(T1Y, T1Z));
Chris@42 472 }
Chris@42 473 T1r = VADD(T10, T1q);
Chris@42 474 ST(&(Rp[WS(rs, 1)]), T1r, ms, &(Rp[WS(rs, 1)]));
Chris@42 475 T1s = VCONJ(VSUB(T1q, T10));
Chris@42 476 ST(&(Rm[WS(rs, 1)]), T1s, -ms, &(Rm[WS(rs, 1)]));
Chris@42 477 T2C = VCONJ(VSUB(T2A, T2w));
Chris@42 478 ST(&(Rm[WS(rs, 4)]), T2C, -ms, &(Rm[0]));
Chris@42 479 T2N = VCONJ(VSUB(T2L, T2K));
Chris@42 480 ST(&(Rm[0]), T2N, -ms, &(Rm[0]));
Chris@42 481 T1N = VADD(T1I, T1M);
Chris@42 482 ST(&(Rp[WS(rs, 3)]), T1N, ms, &(Rp[WS(rs, 1)]));
Chris@42 483 T2H = VADD(T2E, T2G);
Chris@42 484 ST(&(Rp[WS(rs, 6)]), T2H, ms, &(Rp[0]));
Chris@42 485 T2I = VCONJ(VSUB(T2G, T2E));
Chris@42 486 ST(&(Rm[WS(rs, 6)]), T2I, -ms, &(Rm[0]));
Chris@42 487 T2M = VADD(T2K, T2L);
Chris@42 488 ST(&(Rp[0]), T2M, ms, &(Rp[0]));
Chris@42 489 T1E = VCONJ(VSUB(T1C, T1A));
Chris@42 490 ST(&(Rm[WS(rs, 9)]), T1E, -ms, &(Rm[WS(rs, 1)]));
Chris@42 491 T1D = VADD(T1A, T1C);
Chris@42 492 ST(&(Rp[WS(rs, 9)]), T1D, ms, &(Rp[WS(rs, 1)]));
Chris@42 493 T1O = VCONJ(VSUB(T1M, T1I));
Chris@42 494 ST(&(Rm[WS(rs, 3)]), T1O, -ms, &(Rm[WS(rs, 1)]));
Chris@42 495 T1V = VADD(T1S, T1U);
Chris@42 496 ST(&(Rp[WS(rs, 7)]), T1V, ms, &(Rp[WS(rs, 1)]));
Chris@42 497 T2n = VADD(T26, T2m);
Chris@42 498 ST(&(Rp[WS(rs, 2)]), T2n, ms, &(Rp[0]));
Chris@42 499 T2B = VADD(T2w, T2A);
Chris@42 500 ST(&(Rp[WS(rs, 4)]), T2B, ms, &(Rp[0]));
Chris@42 501 T24 = VCONJ(VSUB(T22, T20));
Chris@42 502 ST(&(Rm[WS(rs, 5)]), T24, -ms, &(Rm[WS(rs, 1)]));
Chris@42 503 T2o = VCONJ(VSUB(T2m, T26));
Chris@42 504 ST(&(Rm[WS(rs, 2)]), T2o, -ms, &(Rm[0]));
Chris@42 505 T2t = VADD(T2q, T2s);
Chris@42 506 ST(&(Rp[WS(rs, 8)]), T2t, ms, &(Rp[0]));
Chris@42 507 T2u = VCONJ(VSUB(T2s, T2q));
Chris@42 508 ST(&(Rm[WS(rs, 8)]), T2u, -ms, &(Rm[0]));
Chris@42 509 T23 = VADD(T20, T22);
Chris@42 510 ST(&(Rp[WS(rs, 5)]), T23, ms, &(Rp[WS(rs, 1)]));
Chris@42 511 T1W = VCONJ(VSUB(T1U, T1S));
Chris@42 512 ST(&(Rm[WS(rs, 7)]), T1W, -ms, &(Rm[WS(rs, 1)]));
Chris@42 513 }
Chris@42 514 }
Chris@42 515 }
Chris@42 516 VLEAVE();
Chris@42 517 }
Chris@42 518
Chris@42 519 static const tw_instr twinstr[] = {
Chris@42 520 VTW(1, 1),
Chris@42 521 VTW(1, 2),
Chris@42 522 VTW(1, 3),
Chris@42 523 VTW(1, 4),
Chris@42 524 VTW(1, 5),
Chris@42 525 VTW(1, 6),
Chris@42 526 VTW(1, 7),
Chris@42 527 VTW(1, 8),
Chris@42 528 VTW(1, 9),
Chris@42 529 VTW(1, 10),
Chris@42 530 VTW(1, 11),
Chris@42 531 VTW(1, 12),
Chris@42 532 VTW(1, 13),
Chris@42 533 VTW(1, 14),
Chris@42 534 VTW(1, 15),
Chris@42 535 VTW(1, 16),
Chris@42 536 VTW(1, 17),
Chris@42 537 VTW(1, 18),
Chris@42 538 VTW(1, 19),
Chris@42 539 {TW_NEXT, VL, 0}
Chris@42 540 };
Chris@42 541
Chris@42 542 static const hc2c_desc desc = { 20, XSIMD_STRING("hc2cbdftv_20"), twinstr, &GENUS, {131, 50, 12, 0} };
Chris@42 543
Chris@42 544 void XSIMD(codelet_hc2cbdftv_20) (planner *p) {
Chris@42 545 X(khc2c_register) (p, hc2cbdftv_20, &desc, HC2C_VIA_DFT);
Chris@42 546 }
Chris@42 547 #endif /* HAVE_FMA */