Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/rdft/simd/common/hc2cbdftv_16.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 16:52:44 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 16 -dif -sign 1 -name hc2cbdftv_16 -include hc2cbv.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 103 FP additions, 80 FP multiplications,
Chris@42 32 * (or, 53 additions, 30 multiplications, 50 fused multiply/add),
Chris@42 33 * 123 stack variables, 3 constants, and 32 memory accesses
Chris@42 34 */
Chris@42 35 #include "hc2cbv.h"
Chris@42 36
Chris@42 37 static void hc2cbdftv_16(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(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@42 40 DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@42 41 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 42 {
Chris@42 43 INT m;
Chris@42 44 for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 30)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(64, rs)) {
Chris@42 45 V T1D, T1F, TV, TW, T17, T18, T1B, T1A, T1H, T1G;
Chris@42 46 {
Chris@42 47 V T8, Tv, Tb, TF, Tl, TJ, TP, T1w, TE, T1t, T10, T1p, TG, Te, Tg;
Chris@42 48 V Th, T2, T3, Ts, Tt, T5, T6, Tp, Tq, T9, TA, T4, TC, Tu, TN;
Chris@42 49 V T7, TB, Tr, Ta, Tj, Tk, Tc, Td, TY, TD, TO, TZ, T1Q, T19, T1I;
Chris@42 50 V T1d, Tf, T11, TH, TQ, Ti, TI, T1k, T1K, T1S, T1r, T14, T16, TU, Ty;
Chris@42 51 V T1z, TX, T1o, T1, TK, TR, Tm, T12, T1C, Tz, T15;
Chris@42 52 T2 = LD(&(Rp[0]), ms, &(Rp[0]));
Chris@42 53 T3 = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 54 Ts = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
Chris@42 55 Tt = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 56 T5 = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
Chris@42 57 T6 = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 58 Tp = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Chris@42 59 Tq = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 60 T9 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 61 TA = VFNMSCONJ(T3, T2);
Chris@42 62 T4 = VFMACONJ(T3, T2);
Chris@42 63 TC = VFMSCONJ(Tt, Ts);
Chris@42 64 Tu = VFMACONJ(Tt, Ts);
Chris@42 65 TN = VFNMSCONJ(T6, T5);
Chris@42 66 T7 = VFMACONJ(T6, T5);
Chris@42 67 TB = VFNMSCONJ(Tq, Tp);
Chris@42 68 Tr = VFMACONJ(Tq, Tp);
Chris@42 69 Ta = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
Chris@42 70 Tj = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 71 Tk = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Chris@42 72 Tc = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 73 Td = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Chris@42 74 T8 = VSUB(T4, T7);
Chris@42 75 TY = VADD(T4, T7);
Chris@42 76 TD = VADD(TB, TC);
Chris@42 77 TO = VSUB(TB, TC);
Chris@42 78 Tv = VSUB(Tr, Tu);
Chris@42 79 TZ = VADD(Tr, Tu);
Chris@42 80 Tb = VFMACONJ(Ta, T9);
Chris@42 81 TF = VFNMSCONJ(Ta, T9);
Chris@42 82 Tl = VFMACONJ(Tk, Tj);
Chris@42 83 TJ = VFNMSCONJ(Tk, Tj);
Chris@42 84 TP = VFMA(LDK(KP707106781), TO, TN);
Chris@42 85 T1w = VFNMS(LDK(KP707106781), TO, TN);
Chris@42 86 TE = VFMA(LDK(KP707106781), TD, TA);
Chris@42 87 T1t = VFNMS(LDK(KP707106781), TD, TA);
Chris@42 88 T10 = VADD(TY, TZ);
Chris@42 89 T1p = VSUB(TY, TZ);
Chris@42 90 TG = VFNMSCONJ(Td, Tc);
Chris@42 91 Te = VFMACONJ(Td, Tc);
Chris@42 92 Tg = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 93 Th = LD(&(Rm[0]), -ms, &(Rm[0]));
Chris@42 94 T1Q = LDW(&(W[TWVL * 22]));
Chris@42 95 T19 = LDW(&(W[TWVL * 26]));
Chris@42 96 T1I = LDW(&(W[TWVL * 2]));
Chris@42 97 T1d = LDW(&(W[TWVL * 28]));
Chris@42 98 Tf = VSUB(Tb, Te);
Chris@42 99 T11 = VADD(Tb, Te);
Chris@42 100 TH = VFNMS(LDK(KP414213562), TG, TF);
Chris@42 101 TQ = VFMA(LDK(KP414213562), TF, TG);
Chris@42 102 Ti = VFMACONJ(Th, Tg);
Chris@42 103 TI = VFMSCONJ(Th, Tg);
Chris@42 104 T1k = LDW(&(W[0]));
Chris@42 105 T1K = LDW(&(W[TWVL * 4]));
Chris@42 106 T1S = LDW(&(W[TWVL * 24]));
Chris@42 107 TX = LDW(&(W[TWVL * 14]));
Chris@42 108 T1o = LDW(&(W[TWVL * 6]));
Chris@42 109 T1 = LDW(&(W[TWVL * 10]));
Chris@42 110 TK = VFMA(LDK(KP414213562), TJ, TI);
Chris@42 111 TR = VFNMS(LDK(KP414213562), TI, TJ);
Chris@42 112 Tm = VSUB(Ti, Tl);
Chris@42 113 T12 = VADD(Ti, Tl);
Chris@42 114 T1C = LDW(&(W[TWVL * 18]));
Chris@42 115 Tz = LDW(&(W[TWVL * 12]));
Chris@42 116 T15 = LDW(&(W[TWVL * 16]));
Chris@42 117 {
Chris@42 118 V T1v, T1y, T1N, T1g, T1J, T1c, T1U, T1V, T1m, T1n, T1s, TS, T1u, TL, T1x;
Chris@42 119 V T13, T1q, Tn, Tw, T1L, T1f, TT, T1M, T1e, TM, T1R, T1j, T1b, Tx, T1a;
Chris@42 120 V To, T1T, T1l, T1E, T1O, T1P, T1h, T1i;
Chris@42 121 T1s = LDW(&(W[TWVL * 8]));
Chris@42 122 TS = VADD(TQ, TR);
Chris@42 123 T1u = VSUB(TQ, TR);
Chris@42 124 TL = VADD(TH, TK);
Chris@42 125 T1x = VSUB(TH, TK);
Chris@42 126 T13 = VADD(T11, T12);
Chris@42 127 T1q = VSUB(T11, T12);
Chris@42 128 Tn = VADD(Tf, Tm);
Chris@42 129 Tw = VSUB(Tf, Tm);
Chris@42 130 T1L = VFMA(LDK(KP923879532), T1u, T1t);
Chris@42 131 T1v = VFNMS(LDK(KP923879532), T1u, T1t);
Chris@42 132 T1f = VFMA(LDK(KP923879532), TS, TP);
Chris@42 133 TT = VFNMS(LDK(KP923879532), TS, TP);
Chris@42 134 T1M = VFNMS(LDK(KP923879532), T1x, T1w);
Chris@42 135 T1y = VFMA(LDK(KP923879532), T1x, T1w);
Chris@42 136 T1e = VFMA(LDK(KP923879532), TL, TE);
Chris@42 137 TM = VFNMS(LDK(KP923879532), TL, TE);
Chris@42 138 T1r = VZMUL(T1o, VFMAI(T1q, T1p));
Chris@42 139 T1R = VZMUL(T1Q, VFNMSI(T1q, T1p));
Chris@42 140 T14 = VZMUL(TX, VSUB(T10, T13));
Chris@42 141 T1j = VADD(T10, T13);
Chris@42 142 T1b = VFMA(LDK(KP707106781), Tw, Tv);
Chris@42 143 Tx = VFNMS(LDK(KP707106781), Tw, Tv);
Chris@42 144 T1a = VFMA(LDK(KP707106781), Tn, T8);
Chris@42 145 To = VFNMS(LDK(KP707106781), Tn, T8);
Chris@42 146 T1T = VZMULI(T1S, VFMAI(T1M, T1L));
Chris@42 147 T1N = VZMULI(T1K, VFNMSI(T1M, T1L));
Chris@42 148 T16 = VZMULI(T15, VFMAI(TT, TM));
Chris@42 149 TU = VZMULI(Tz, VFNMSI(TT, TM));
Chris@42 150 T1l = VZMULI(T1k, VFMAI(T1f, T1e));
Chris@42 151 T1g = VZMULI(T1d, VFNMSI(T1f, T1e));
Chris@42 152 T1D = VZMUL(T1C, VFMAI(Tx, To));
Chris@42 153 Ty = VZMUL(T1, VFNMSI(Tx, To));
Chris@42 154 T1J = VZMUL(T1I, VFMAI(T1b, T1a));
Chris@42 155 T1c = VZMUL(T19, VFNMSI(T1b, T1a));
Chris@42 156 T1U = VCONJ(VSUB(T1R, T1T));
Chris@42 157 T1V = VADD(T1R, T1T);
Chris@42 158 T1m = VCONJ(VSUB(T1j, T1l));
Chris@42 159 T1n = VADD(T1j, T1l);
Chris@42 160 T1z = VZMULI(T1s, VFMAI(T1y, T1v));
Chris@42 161 T1E = LDW(&(W[TWVL * 20]));
Chris@42 162 T1O = VCONJ(VSUB(T1J, T1N));
Chris@42 163 T1P = VADD(T1J, T1N);
Chris@42 164 T1h = VCONJ(VSUB(T1c, T1g));
Chris@42 165 T1i = VADD(T1c, T1g);
Chris@42 166 ST(&(Rp[WS(rs, 6)]), T1V, ms, &(Rp[0]));
Chris@42 167 ST(&(Rm[WS(rs, 6)]), T1U, -ms, &(Rm[0]));
Chris@42 168 ST(&(Rp[0]), T1n, ms, &(Rp[0]));
Chris@42 169 ST(&(Rm[0]), T1m, -ms, &(Rm[0]));
Chris@42 170 ST(&(Rp[WS(rs, 1)]), T1P, ms, &(Rp[WS(rs, 1)]));
Chris@42 171 ST(&(Rm[WS(rs, 1)]), T1O, -ms, &(Rm[WS(rs, 1)]));
Chris@42 172 ST(&(Rp[WS(rs, 7)]), T1i, ms, &(Rp[WS(rs, 1)]));
Chris@42 173 ST(&(Rm[WS(rs, 7)]), T1h, -ms, &(Rm[WS(rs, 1)]));
Chris@42 174 T1F = VZMULI(T1E, VFNMSI(T1y, T1v));
Chris@42 175 }
Chris@42 176 TV = VCONJ(VSUB(Ty, TU));
Chris@42 177 TW = VADD(Ty, TU);
Chris@42 178 T17 = VCONJ(VSUB(T14, T16));
Chris@42 179 T18 = VADD(T14, T16);
Chris@42 180 T1B = VADD(T1r, T1z);
Chris@42 181 T1A = VCONJ(VSUB(T1r, T1z));
Chris@42 182 }
Chris@42 183 T1H = VADD(T1D, T1F);
Chris@42 184 T1G = VCONJ(VSUB(T1D, T1F));
Chris@42 185 ST(&(Rm[WS(rs, 3)]), TV, -ms, &(Rm[WS(rs, 1)]));
Chris@42 186 ST(&(Rp[WS(rs, 3)]), TW, ms, &(Rp[WS(rs, 1)]));
Chris@42 187 ST(&(Rm[WS(rs, 4)]), T17, -ms, &(Rm[0]));
Chris@42 188 ST(&(Rm[WS(rs, 2)]), T1A, -ms, &(Rm[0]));
Chris@42 189 ST(&(Rp[WS(rs, 2)]), T1B, ms, &(Rp[0]));
Chris@42 190 ST(&(Rp[WS(rs, 4)]), T18, ms, &(Rp[0]));
Chris@42 191 ST(&(Rp[WS(rs, 5)]), T1H, ms, &(Rp[WS(rs, 1)]));
Chris@42 192 ST(&(Rm[WS(rs, 5)]), T1G, -ms, &(Rm[WS(rs, 1)]));
Chris@42 193 }
Chris@42 194 }
Chris@42 195 VLEAVE();
Chris@42 196 }
Chris@42 197
Chris@42 198 static const tw_instr twinstr[] = {
Chris@42 199 VTW(1, 1),
Chris@42 200 VTW(1, 2),
Chris@42 201 VTW(1, 3),
Chris@42 202 VTW(1, 4),
Chris@42 203 VTW(1, 5),
Chris@42 204 VTW(1, 6),
Chris@42 205 VTW(1, 7),
Chris@42 206 VTW(1, 8),
Chris@42 207 VTW(1, 9),
Chris@42 208 VTW(1, 10),
Chris@42 209 VTW(1, 11),
Chris@42 210 VTW(1, 12),
Chris@42 211 VTW(1, 13),
Chris@42 212 VTW(1, 14),
Chris@42 213 VTW(1, 15),
Chris@42 214 {TW_NEXT, VL, 0}
Chris@42 215 };
Chris@42 216
Chris@42 217 static const hc2c_desc desc = { 16, XSIMD_STRING("hc2cbdftv_16"), twinstr, &GENUS, {53, 30, 50, 0} };
Chris@42 218
Chris@42 219 void XSIMD(codelet_hc2cbdftv_16) (planner *p) {
Chris@42 220 X(khc2c_register) (p, hc2cbdftv_16, &desc, HC2C_VIA_DFT);
Chris@42 221 }
Chris@42 222 #else /* HAVE_FMA */
Chris@42 223
Chris@42 224 /* Generated by: ../../../genfft/gen_hc2cdft_c.native -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 16 -dif -sign 1 -name hc2cbdftv_16 -include hc2cbv.h */
Chris@42 225
Chris@42 226 /*
Chris@42 227 * This function contains 103 FP additions, 42 FP multiplications,
Chris@42 228 * (or, 99 additions, 38 multiplications, 4 fused multiply/add),
Chris@42 229 * 83 stack variables, 3 constants, and 32 memory accesses
Chris@42 230 */
Chris@42 231 #include "hc2cbv.h"
Chris@42 232
Chris@42 233 static void hc2cbdftv_16(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 234 {
Chris@42 235 DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
Chris@42 236 DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@42 237 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 238 {
Chris@42 239 INT m;
Chris@42 240 for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 30)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(64, rs)) {
Chris@42 241 V Tf, T16, TZ, T1C, TI, T1a, TV, T1D, T1F, T1G, Ty, T19, TC, T17, TS;
Chris@42 242 V T10;
Chris@42 243 {
Chris@42 244 V T2, TD, T4, TF, Tc, Tb, Td, T6, T8, T9, T3, TE, Ta, T7, T5;
Chris@42 245 V Te, TX, TY, TG, TH, TT, TU, Tj, TM, Tw, TQ, Tn, TN, Ts, TP;
Chris@42 246 V Tg, Ti, Th, Tt, Tv, Tu, Tk, Tm, Tl, Tr, Tq, Tp, To, Tx, TA;
Chris@42 247 V TB, TO, TR;
Chris@42 248 T2 = LD(&(Rp[0]), ms, &(Rp[0]));
Chris@42 249 TD = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
Chris@42 250 T3 = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 251 T4 = VCONJ(T3);
Chris@42 252 TE = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 253 TF = VCONJ(TE);
Chris@42 254 Tc = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
Chris@42 255 Ta = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 256 Tb = VCONJ(Ta);
Chris@42 257 Td = VSUB(Tb, Tc);
Chris@42 258 T6 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Chris@42 259 T7 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42 260 T8 = VCONJ(T7);
Chris@42 261 T9 = VSUB(T6, T8);
Chris@42 262 T5 = VSUB(T2, T4);
Chris@42 263 Te = VMUL(LDK(KP707106781), VADD(T9, Td));
Chris@42 264 Tf = VADD(T5, Te);
Chris@42 265 T16 = VSUB(T5, Te);
Chris@42 266 TX = VADD(T2, T4);
Chris@42 267 TY = VADD(TD, TF);
Chris@42 268 TZ = VSUB(TX, TY);
Chris@42 269 T1C = VADD(TX, TY);
Chris@42 270 TG = VSUB(TD, TF);
Chris@42 271 TH = VMUL(LDK(KP707106781), VSUB(T9, Td));
Chris@42 272 TI = VADD(TG, TH);
Chris@42 273 T1a = VSUB(TH, TG);
Chris@42 274 TT = VADD(T6, T8);
Chris@42 275 TU = VADD(Tb, Tc);
Chris@42 276 TV = VSUB(TT, TU);
Chris@42 277 T1D = VADD(TT, TU);
Chris@42 278 Tg = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 279 Th = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
Chris@42 280 Ti = VCONJ(Th);
Chris@42 281 Tj = VSUB(Tg, Ti);
Chris@42 282 TM = VADD(Tg, Ti);
Chris@42 283 Tt = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 284 Tu = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Chris@42 285 Tv = VCONJ(Tu);
Chris@42 286 Tw = VSUB(Tt, Tv);
Chris@42 287 TQ = VADD(Tt, Tv);
Chris@42 288 Tk = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 289 Tl = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Chris@42 290 Tm = VCONJ(Tl);
Chris@42 291 Tn = VSUB(Tk, Tm);
Chris@42 292 TN = VADD(Tk, Tm);
Chris@42 293 Tr = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
Chris@42 294 Tp = LD(&(Rm[0]), -ms, &(Rm[0]));
Chris@42 295 Tq = VCONJ(Tp);
Chris@42 296 Ts = VSUB(Tq, Tr);
Chris@42 297 TP = VADD(Tq, Tr);
Chris@42 298 T1F = VADD(TM, TN);
Chris@42 299 T1G = VADD(TP, TQ);
Chris@42 300 To = VFNMS(LDK(KP382683432), Tn, VMUL(LDK(KP923879532), Tj));
Chris@42 301 Tx = VFMA(LDK(KP923879532), Ts, VMUL(LDK(KP382683432), Tw));
Chris@42 302 Ty = VADD(To, Tx);
Chris@42 303 T19 = VSUB(To, Tx);
Chris@42 304 TA = VFMA(LDK(KP382683432), Tj, VMUL(LDK(KP923879532), Tn));
Chris@42 305 TB = VFNMS(LDK(KP382683432), Ts, VMUL(LDK(KP923879532), Tw));
Chris@42 306 TC = VADD(TA, TB);
Chris@42 307 T17 = VSUB(TA, TB);
Chris@42 308 TO = VSUB(TM, TN);
Chris@42 309 TR = VSUB(TP, TQ);
Chris@42 310 TS = VMUL(LDK(KP707106781), VSUB(TO, TR));
Chris@42 311 T10 = VMUL(LDK(KP707106781), VADD(TO, TR));
Chris@42 312 }
Chris@42 313 {
Chris@42 314 V T21, T1W, T1u, T20, T1I, T1O, TK, T1S, T12, T1e, T1k, T1A, T1o, T1w, T1c;
Chris@42 315 V T1M, T1U, T1V, T1T, T1s, T1t, T1r, T1Z, T1E, T1H, T1B, T1N, Tz, TJ, T1;
Chris@42 316 V T1R, TW, T11, TL, T1d, T1i, T1j, T1h, T1z, T1m, T1n, T1l, T1v, T18, T1b;
Chris@42 317 V T15, T1L, T13, T1g, T1X, T23, T14, T1f, T1Y, T22, T1p, T1y, T1J, T1Q, T1q;
Chris@42 318 V T1x, T1K, T1P;
Chris@42 319 T1U = VADD(T1C, T1D);
Chris@42 320 T1V = VADD(T1F, T1G);
Chris@42 321 T21 = VADD(T1U, T1V);
Chris@42 322 T1T = LDW(&(W[TWVL * 14]));
Chris@42 323 T1W = VZMUL(T1T, VSUB(T1U, T1V));
Chris@42 324 T1s = VADD(Tf, Ty);
Chris@42 325 T1t = VBYI(VADD(TI, TC));
Chris@42 326 T1r = LDW(&(W[TWVL * 28]));
Chris@42 327 T1u = VZMULI(T1r, VSUB(T1s, T1t));
Chris@42 328 T1Z = LDW(&(W[0]));
Chris@42 329 T20 = VZMULI(T1Z, VADD(T1s, T1t));
Chris@42 330 T1E = VSUB(T1C, T1D);
Chris@42 331 T1H = VBYI(VSUB(T1F, T1G));
Chris@42 332 T1B = LDW(&(W[TWVL * 22]));
Chris@42 333 T1I = VZMUL(T1B, VSUB(T1E, T1H));
Chris@42 334 T1N = LDW(&(W[TWVL * 6]));
Chris@42 335 T1O = VZMUL(T1N, VADD(T1E, T1H));
Chris@42 336 Tz = VSUB(Tf, Ty);
Chris@42 337 TJ = VBYI(VSUB(TC, TI));
Chris@42 338 T1 = LDW(&(W[TWVL * 12]));
Chris@42 339 TK = VZMULI(T1, VADD(Tz, TJ));
Chris@42 340 T1R = LDW(&(W[TWVL * 16]));
Chris@42 341 T1S = VZMULI(T1R, VSUB(Tz, TJ));
Chris@42 342 TW = VBYI(VSUB(TS, TV));
Chris@42 343 T11 = VSUB(TZ, T10);
Chris@42 344 TL = LDW(&(W[TWVL * 10]));
Chris@42 345 T12 = VZMUL(TL, VADD(TW, T11));
Chris@42 346 T1d = LDW(&(W[TWVL * 18]));
Chris@42 347 T1e = VZMUL(T1d, VSUB(T11, TW));
Chris@42 348 T1i = VBYI(VADD(T1a, T19));
Chris@42 349 T1j = VADD(T16, T17);
Chris@42 350 T1h = LDW(&(W[TWVL * 4]));
Chris@42 351 T1k = VZMULI(T1h, VADD(T1i, T1j));
Chris@42 352 T1z = LDW(&(W[TWVL * 24]));
Chris@42 353 T1A = VZMULI(T1z, VSUB(T1j, T1i));
Chris@42 354 T1m = VBYI(VADD(TV, TS));
Chris@42 355 T1n = VADD(TZ, T10);
Chris@42 356 T1l = LDW(&(W[TWVL * 2]));
Chris@42 357 T1o = VZMUL(T1l, VADD(T1m, T1n));
Chris@42 358 T1v = LDW(&(W[TWVL * 26]));
Chris@42 359 T1w = VZMUL(T1v, VSUB(T1n, T1m));
Chris@42 360 T18 = VSUB(T16, T17);
Chris@42 361 T1b = VBYI(VSUB(T19, T1a));
Chris@42 362 T15 = LDW(&(W[TWVL * 20]));
Chris@42 363 T1c = VZMULI(T15, VSUB(T18, T1b));
Chris@42 364 T1L = LDW(&(W[TWVL * 8]));
Chris@42 365 T1M = VZMULI(T1L, VADD(T1b, T18));
Chris@42 366 T13 = VADD(TK, T12);
Chris@42 367 ST(&(Rp[WS(rs, 3)]), T13, ms, &(Rp[WS(rs, 1)]));
Chris@42 368 T1g = VCONJ(VSUB(T1e, T1c));
Chris@42 369 ST(&(Rm[WS(rs, 5)]), T1g, -ms, &(Rm[WS(rs, 1)]));
Chris@42 370 T1X = VADD(T1S, T1W);
Chris@42 371 ST(&(Rp[WS(rs, 4)]), T1X, ms, &(Rp[0]));
Chris@42 372 T23 = VCONJ(VSUB(T21, T20));
Chris@42 373 ST(&(Rm[0]), T23, -ms, &(Rm[0]));
Chris@42 374 T14 = VCONJ(VSUB(T12, TK));
Chris@42 375 ST(&(Rm[WS(rs, 3)]), T14, -ms, &(Rm[WS(rs, 1)]));
Chris@42 376 T1f = VADD(T1c, T1e);
Chris@42 377 ST(&(Rp[WS(rs, 5)]), T1f, ms, &(Rp[WS(rs, 1)]));
Chris@42 378 T1Y = VCONJ(VSUB(T1W, T1S));
Chris@42 379 ST(&(Rm[WS(rs, 4)]), T1Y, -ms, &(Rm[0]));
Chris@42 380 T22 = VADD(T20, T21);
Chris@42 381 ST(&(Rp[0]), T22, ms, &(Rp[0]));
Chris@42 382 T1p = VADD(T1k, T1o);
Chris@42 383 ST(&(Rp[WS(rs, 1)]), T1p, ms, &(Rp[WS(rs, 1)]));
Chris@42 384 T1y = VCONJ(VSUB(T1w, T1u));
Chris@42 385 ST(&(Rm[WS(rs, 7)]), T1y, -ms, &(Rm[WS(rs, 1)]));
Chris@42 386 T1J = VADD(T1A, T1I);
Chris@42 387 ST(&(Rp[WS(rs, 6)]), T1J, ms, &(Rp[0]));
Chris@42 388 T1Q = VCONJ(VSUB(T1O, T1M));
Chris@42 389 ST(&(Rm[WS(rs, 2)]), T1Q, -ms, &(Rm[0]));
Chris@42 390 T1q = VCONJ(VSUB(T1o, T1k));
Chris@42 391 ST(&(Rm[WS(rs, 1)]), T1q, -ms, &(Rm[WS(rs, 1)]));
Chris@42 392 T1x = VADD(T1u, T1w);
Chris@42 393 ST(&(Rp[WS(rs, 7)]), T1x, ms, &(Rp[WS(rs, 1)]));
Chris@42 394 T1K = VCONJ(VSUB(T1I, T1A));
Chris@42 395 ST(&(Rm[WS(rs, 6)]), T1K, -ms, &(Rm[0]));
Chris@42 396 T1P = VADD(T1M, T1O);
Chris@42 397 ST(&(Rp[WS(rs, 2)]), T1P, ms, &(Rp[0]));
Chris@42 398 }
Chris@42 399 }
Chris@42 400 }
Chris@42 401 VLEAVE();
Chris@42 402 }
Chris@42 403
Chris@42 404 static const tw_instr twinstr[] = {
Chris@42 405 VTW(1, 1),
Chris@42 406 VTW(1, 2),
Chris@42 407 VTW(1, 3),
Chris@42 408 VTW(1, 4),
Chris@42 409 VTW(1, 5),
Chris@42 410 VTW(1, 6),
Chris@42 411 VTW(1, 7),
Chris@42 412 VTW(1, 8),
Chris@42 413 VTW(1, 9),
Chris@42 414 VTW(1, 10),
Chris@42 415 VTW(1, 11),
Chris@42 416 VTW(1, 12),
Chris@42 417 VTW(1, 13),
Chris@42 418 VTW(1, 14),
Chris@42 419 VTW(1, 15),
Chris@42 420 {TW_NEXT, VL, 0}
Chris@42 421 };
Chris@42 422
Chris@42 423 static const hc2c_desc desc = { 16, XSIMD_STRING("hc2cbdftv_16"), twinstr, &GENUS, {99, 38, 4, 0} };
Chris@42 424
Chris@42 425 void XSIMD(codelet_hc2cbdftv_16) (planner *p) {
Chris@42 426 X(khc2c_register) (p, hc2cbdftv_16, &desc, HC2C_VIA_DFT);
Chris@42 427 }
Chris@42 428 #endif /* HAVE_FMA */