Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/scalar/codelets/t1_8.c @ 84:08ae793730bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +0000
parents d0c2a83c1364
children
rev   line source
Chris@82 1 /*
Chris@82 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@82 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@82 4 *
Chris@82 5 * This program is free software; you can redistribute it and/or modify
Chris@82 6 * it under the terms of the GNU General Public License as published by
Chris@82 7 * the Free Software Foundation; either version 2 of the License, or
Chris@82 8 * (at your option) any later version.
Chris@82 9 *
Chris@82 10 * This program is distributed in the hope that it will be useful,
Chris@82 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@82 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@82 13 * GNU General Public License for more details.
Chris@82 14 *
Chris@82 15 * You should have received a copy of the GNU General Public License
Chris@82 16 * along with this program; if not, write to the Free Software
Chris@82 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@82 18 *
Chris@82 19 */
Chris@82 20
Chris@82 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@82 22 /* Generated on Thu May 24 08:04:13 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.native -fma -compact -variables 4 -pipeline-latency 4 -n 8 -name t1_8 -include dft/scalar/t.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 66 FP additions, 36 FP multiplications,
Chris@82 32 * (or, 44 additions, 14 multiplications, 22 fused multiply/add),
Chris@82 33 * 34 stack variables, 1 constants, and 32 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/scalar/t.h"
Chris@82 36
Chris@82 37 static void t1_8(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 38 {
Chris@82 39 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 40 {
Chris@82 41 INT m;
Chris@82 42 for (m = mb, W = W + (mb * 14); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 14, MAKE_VOLATILE_STRIDE(16, rs)) {
Chris@82 43 E T1, T1m, T7, T1l, Tk, TS, Te, TQ, TF, T14, TL, T16, T12, T17, Ts;
Chris@82 44 E TX, Ty, TZ, TV, T10;
Chris@82 45 T1 = ri[0];
Chris@82 46 T1m = ii[0];
Chris@82 47 {
Chris@82 48 E T3, T6, T4, T1k, T2, T5;
Chris@82 49 T3 = ri[WS(rs, 4)];
Chris@82 50 T6 = ii[WS(rs, 4)];
Chris@82 51 T2 = W[6];
Chris@82 52 T4 = T2 * T3;
Chris@82 53 T1k = T2 * T6;
Chris@82 54 T5 = W[7];
Chris@82 55 T7 = FMA(T5, T6, T4);
Chris@82 56 T1l = FNMS(T5, T3, T1k);
Chris@82 57 }
Chris@82 58 {
Chris@82 59 E Tg, Tj, Th, TR, Tf, Ti;
Chris@82 60 Tg = ri[WS(rs, 6)];
Chris@82 61 Tj = ii[WS(rs, 6)];
Chris@82 62 Tf = W[10];
Chris@82 63 Th = Tf * Tg;
Chris@82 64 TR = Tf * Tj;
Chris@82 65 Ti = W[11];
Chris@82 66 Tk = FMA(Ti, Tj, Th);
Chris@82 67 TS = FNMS(Ti, Tg, TR);
Chris@82 68 }
Chris@82 69 {
Chris@82 70 E Ta, Td, Tb, TP, T9, Tc;
Chris@82 71 Ta = ri[WS(rs, 2)];
Chris@82 72 Td = ii[WS(rs, 2)];
Chris@82 73 T9 = W[2];
Chris@82 74 Tb = T9 * Ta;
Chris@82 75 TP = T9 * Td;
Chris@82 76 Tc = W[3];
Chris@82 77 Te = FMA(Tc, Td, Tb);
Chris@82 78 TQ = FNMS(Tc, Ta, TP);
Chris@82 79 }
Chris@82 80 {
Chris@82 81 E TB, TE, TC, T13, TH, TK, TI, T15, TA, TG, TD, TJ;
Chris@82 82 TB = ri[WS(rs, 7)];
Chris@82 83 TE = ii[WS(rs, 7)];
Chris@82 84 TA = W[12];
Chris@82 85 TC = TA * TB;
Chris@82 86 T13 = TA * TE;
Chris@82 87 TH = ri[WS(rs, 3)];
Chris@82 88 TK = ii[WS(rs, 3)];
Chris@82 89 TG = W[4];
Chris@82 90 TI = TG * TH;
Chris@82 91 T15 = TG * TK;
Chris@82 92 TD = W[13];
Chris@82 93 TF = FMA(TD, TE, TC);
Chris@82 94 T14 = FNMS(TD, TB, T13);
Chris@82 95 TJ = W[5];
Chris@82 96 TL = FMA(TJ, TK, TI);
Chris@82 97 T16 = FNMS(TJ, TH, T15);
Chris@82 98 T12 = TF - TL;
Chris@82 99 T17 = T14 - T16;
Chris@82 100 }
Chris@82 101 {
Chris@82 102 E To, Tr, Tp, TW, Tu, Tx, Tv, TY, Tn, Tt, Tq, Tw;
Chris@82 103 To = ri[WS(rs, 1)];
Chris@82 104 Tr = ii[WS(rs, 1)];
Chris@82 105 Tn = W[0];
Chris@82 106 Tp = Tn * To;
Chris@82 107 TW = Tn * Tr;
Chris@82 108 Tu = ri[WS(rs, 5)];
Chris@82 109 Tx = ii[WS(rs, 5)];
Chris@82 110 Tt = W[8];
Chris@82 111 Tv = Tt * Tu;
Chris@82 112 TY = Tt * Tx;
Chris@82 113 Tq = W[1];
Chris@82 114 Ts = FMA(Tq, Tr, Tp);
Chris@82 115 TX = FNMS(Tq, To, TW);
Chris@82 116 Tw = W[9];
Chris@82 117 Ty = FMA(Tw, Tx, Tv);
Chris@82 118 TZ = FNMS(Tw, Tu, TY);
Chris@82 119 TV = Ts - Ty;
Chris@82 120 T10 = TX - TZ;
Chris@82 121 }
Chris@82 122 {
Chris@82 123 E TU, T1a, T1t, T1v, T19, T1w, T1d, T1u;
Chris@82 124 {
Chris@82 125 E TO, TT, T1r, T1s;
Chris@82 126 TO = T1 - T7;
Chris@82 127 TT = TQ - TS;
Chris@82 128 TU = TO + TT;
Chris@82 129 T1a = TO - TT;
Chris@82 130 T1r = T1m - T1l;
Chris@82 131 T1s = Te - Tk;
Chris@82 132 T1t = T1r - T1s;
Chris@82 133 T1v = T1s + T1r;
Chris@82 134 }
Chris@82 135 {
Chris@82 136 E T11, T18, T1b, T1c;
Chris@82 137 T11 = TV + T10;
Chris@82 138 T18 = T12 - T17;
Chris@82 139 T19 = T11 + T18;
Chris@82 140 T1w = T18 - T11;
Chris@82 141 T1b = T10 - TV;
Chris@82 142 T1c = T12 + T17;
Chris@82 143 T1d = T1b - T1c;
Chris@82 144 T1u = T1b + T1c;
Chris@82 145 }
Chris@82 146 ri[WS(rs, 5)] = FNMS(KP707106781, T19, TU);
Chris@82 147 ii[WS(rs, 5)] = FNMS(KP707106781, T1u, T1t);
Chris@82 148 ri[WS(rs, 1)] = FMA(KP707106781, T19, TU);
Chris@82 149 ii[WS(rs, 1)] = FMA(KP707106781, T1u, T1t);
Chris@82 150 ri[WS(rs, 7)] = FNMS(KP707106781, T1d, T1a);
Chris@82 151 ii[WS(rs, 7)] = FNMS(KP707106781, T1w, T1v);
Chris@82 152 ri[WS(rs, 3)] = FMA(KP707106781, T1d, T1a);
Chris@82 153 ii[WS(rs, 3)] = FMA(KP707106781, T1w, T1v);
Chris@82 154 }
Chris@82 155 {
Chris@82 156 E Tm, T1e, T1o, T1q, TN, T1p, T1h, T1i;
Chris@82 157 {
Chris@82 158 E T8, Tl, T1j, T1n;
Chris@82 159 T8 = T1 + T7;
Chris@82 160 Tl = Te + Tk;
Chris@82 161 Tm = T8 + Tl;
Chris@82 162 T1e = T8 - Tl;
Chris@82 163 T1j = TQ + TS;
Chris@82 164 T1n = T1l + T1m;
Chris@82 165 T1o = T1j + T1n;
Chris@82 166 T1q = T1n - T1j;
Chris@82 167 }
Chris@82 168 {
Chris@82 169 E Tz, TM, T1f, T1g;
Chris@82 170 Tz = Ts + Ty;
Chris@82 171 TM = TF + TL;
Chris@82 172 TN = Tz + TM;
Chris@82 173 T1p = TM - Tz;
Chris@82 174 T1f = TX + TZ;
Chris@82 175 T1g = T14 + T16;
Chris@82 176 T1h = T1f - T1g;
Chris@82 177 T1i = T1f + T1g;
Chris@82 178 }
Chris@82 179 ri[WS(rs, 4)] = Tm - TN;
Chris@82 180 ii[WS(rs, 4)] = T1o - T1i;
Chris@82 181 ri[0] = Tm + TN;
Chris@82 182 ii[0] = T1i + T1o;
Chris@82 183 ri[WS(rs, 6)] = T1e - T1h;
Chris@82 184 ii[WS(rs, 6)] = T1q - T1p;
Chris@82 185 ri[WS(rs, 2)] = T1e + T1h;
Chris@82 186 ii[WS(rs, 2)] = T1p + T1q;
Chris@82 187 }
Chris@82 188 }
Chris@82 189 }
Chris@82 190 }
Chris@82 191
Chris@82 192 static const tw_instr twinstr[] = {
Chris@82 193 {TW_FULL, 0, 8},
Chris@82 194 {TW_NEXT, 1, 0}
Chris@82 195 };
Chris@82 196
Chris@82 197 static const ct_desc desc = { 8, "t1_8", twinstr, &GENUS, {44, 14, 22, 0}, 0, 0, 0 };
Chris@82 198
Chris@82 199 void X(codelet_t1_8) (planner *p) {
Chris@82 200 X(kdft_dit_register) (p, t1_8, &desc);
Chris@82 201 }
Chris@82 202 #else
Chris@82 203
Chris@82 204 /* Generated by: ../../../genfft/gen_twiddle.native -compact -variables 4 -pipeline-latency 4 -n 8 -name t1_8 -include dft/scalar/t.h */
Chris@82 205
Chris@82 206 /*
Chris@82 207 * This function contains 66 FP additions, 32 FP multiplications,
Chris@82 208 * (or, 52 additions, 18 multiplications, 14 fused multiply/add),
Chris@82 209 * 28 stack variables, 1 constants, and 32 memory accesses
Chris@82 210 */
Chris@82 211 #include "dft/scalar/t.h"
Chris@82 212
Chris@82 213 static void t1_8(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82 214 {
Chris@82 215 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82 216 {
Chris@82 217 INT m;
Chris@82 218 for (m = mb, W = W + (mb * 14); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 14, MAKE_VOLATILE_STRIDE(16, rs)) {
Chris@82 219 E T7, T1e, TH, T19, TF, T13, TR, TU, Ti, T1f, TK, T16, Tu, T12, TM;
Chris@82 220 E TP;
Chris@82 221 {
Chris@82 222 E T1, T18, T6, T17;
Chris@82 223 T1 = ri[0];
Chris@82 224 T18 = ii[0];
Chris@82 225 {
Chris@82 226 E T3, T5, T2, T4;
Chris@82 227 T3 = ri[WS(rs, 4)];
Chris@82 228 T5 = ii[WS(rs, 4)];
Chris@82 229 T2 = W[6];
Chris@82 230 T4 = W[7];
Chris@82 231 T6 = FMA(T2, T3, T4 * T5);
Chris@82 232 T17 = FNMS(T4, T3, T2 * T5);
Chris@82 233 }
Chris@82 234 T7 = T1 + T6;
Chris@82 235 T1e = T18 - T17;
Chris@82 236 TH = T1 - T6;
Chris@82 237 T19 = T17 + T18;
Chris@82 238 }
Chris@82 239 {
Chris@82 240 E Tz, TS, TE, TT;
Chris@82 241 {
Chris@82 242 E Tw, Ty, Tv, Tx;
Chris@82 243 Tw = ri[WS(rs, 7)];
Chris@82 244 Ty = ii[WS(rs, 7)];
Chris@82 245 Tv = W[12];
Chris@82 246 Tx = W[13];
Chris@82 247 Tz = FMA(Tv, Tw, Tx * Ty);
Chris@82 248 TS = FNMS(Tx, Tw, Tv * Ty);
Chris@82 249 }
Chris@82 250 {
Chris@82 251 E TB, TD, TA, TC;
Chris@82 252 TB = ri[WS(rs, 3)];
Chris@82 253 TD = ii[WS(rs, 3)];
Chris@82 254 TA = W[4];
Chris@82 255 TC = W[5];
Chris@82 256 TE = FMA(TA, TB, TC * TD);
Chris@82 257 TT = FNMS(TC, TB, TA * TD);
Chris@82 258 }
Chris@82 259 TF = Tz + TE;
Chris@82 260 T13 = TS + TT;
Chris@82 261 TR = Tz - TE;
Chris@82 262 TU = TS - TT;
Chris@82 263 }
Chris@82 264 {
Chris@82 265 E Tc, TI, Th, TJ;
Chris@82 266 {
Chris@82 267 E T9, Tb, T8, Ta;
Chris@82 268 T9 = ri[WS(rs, 2)];
Chris@82 269 Tb = ii[WS(rs, 2)];
Chris@82 270 T8 = W[2];
Chris@82 271 Ta = W[3];
Chris@82 272 Tc = FMA(T8, T9, Ta * Tb);
Chris@82 273 TI = FNMS(Ta, T9, T8 * Tb);
Chris@82 274 }
Chris@82 275 {
Chris@82 276 E Te, Tg, Td, Tf;
Chris@82 277 Te = ri[WS(rs, 6)];
Chris@82 278 Tg = ii[WS(rs, 6)];
Chris@82 279 Td = W[10];
Chris@82 280 Tf = W[11];
Chris@82 281 Th = FMA(Td, Te, Tf * Tg);
Chris@82 282 TJ = FNMS(Tf, Te, Td * Tg);
Chris@82 283 }
Chris@82 284 Ti = Tc + Th;
Chris@82 285 T1f = Tc - Th;
Chris@82 286 TK = TI - TJ;
Chris@82 287 T16 = TI + TJ;
Chris@82 288 }
Chris@82 289 {
Chris@82 290 E To, TN, Tt, TO;
Chris@82 291 {
Chris@82 292 E Tl, Tn, Tk, Tm;
Chris@82 293 Tl = ri[WS(rs, 1)];
Chris@82 294 Tn = ii[WS(rs, 1)];
Chris@82 295 Tk = W[0];
Chris@82 296 Tm = W[1];
Chris@82 297 To = FMA(Tk, Tl, Tm * Tn);
Chris@82 298 TN = FNMS(Tm, Tl, Tk * Tn);
Chris@82 299 }
Chris@82 300 {
Chris@82 301 E Tq, Ts, Tp, Tr;
Chris@82 302 Tq = ri[WS(rs, 5)];
Chris@82 303 Ts = ii[WS(rs, 5)];
Chris@82 304 Tp = W[8];
Chris@82 305 Tr = W[9];
Chris@82 306 Tt = FMA(Tp, Tq, Tr * Ts);
Chris@82 307 TO = FNMS(Tr, Tq, Tp * Ts);
Chris@82 308 }
Chris@82 309 Tu = To + Tt;
Chris@82 310 T12 = TN + TO;
Chris@82 311 TM = To - Tt;
Chris@82 312 TP = TN - TO;
Chris@82 313 }
Chris@82 314 {
Chris@82 315 E Tj, TG, T1b, T1c;
Chris@82 316 Tj = T7 + Ti;
Chris@82 317 TG = Tu + TF;
Chris@82 318 ri[WS(rs, 4)] = Tj - TG;
Chris@82 319 ri[0] = Tj + TG;
Chris@82 320 {
Chris@82 321 E T15, T1a, T11, T14;
Chris@82 322 T15 = T12 + T13;
Chris@82 323 T1a = T16 + T19;
Chris@82 324 ii[0] = T15 + T1a;
Chris@82 325 ii[WS(rs, 4)] = T1a - T15;
Chris@82 326 T11 = T7 - Ti;
Chris@82 327 T14 = T12 - T13;
Chris@82 328 ri[WS(rs, 6)] = T11 - T14;
Chris@82 329 ri[WS(rs, 2)] = T11 + T14;
Chris@82 330 }
Chris@82 331 T1b = TF - Tu;
Chris@82 332 T1c = T19 - T16;
Chris@82 333 ii[WS(rs, 2)] = T1b + T1c;
Chris@82 334 ii[WS(rs, 6)] = T1c - T1b;
Chris@82 335 {
Chris@82 336 E TX, T1g, T10, T1d, TY, TZ;
Chris@82 337 TX = TH - TK;
Chris@82 338 T1g = T1e - T1f;
Chris@82 339 TY = TP - TM;
Chris@82 340 TZ = TR + TU;
Chris@82 341 T10 = KP707106781 * (TY - TZ);
Chris@82 342 T1d = KP707106781 * (TY + TZ);
Chris@82 343 ri[WS(rs, 7)] = TX - T10;
Chris@82 344 ii[WS(rs, 5)] = T1g - T1d;
Chris@82 345 ri[WS(rs, 3)] = TX + T10;
Chris@82 346 ii[WS(rs, 1)] = T1d + T1g;
Chris@82 347 }
Chris@82 348 {
Chris@82 349 E TL, T1i, TW, T1h, TQ, TV;
Chris@82 350 TL = TH + TK;
Chris@82 351 T1i = T1f + T1e;
Chris@82 352 TQ = TM + TP;
Chris@82 353 TV = TR - TU;
Chris@82 354 TW = KP707106781 * (TQ + TV);
Chris@82 355 T1h = KP707106781 * (TV - TQ);
Chris@82 356 ri[WS(rs, 5)] = TL - TW;
Chris@82 357 ii[WS(rs, 7)] = T1i - T1h;
Chris@82 358 ri[WS(rs, 1)] = TL + TW;
Chris@82 359 ii[WS(rs, 3)] = T1h + T1i;
Chris@82 360 }
Chris@82 361 }
Chris@82 362 }
Chris@82 363 }
Chris@82 364 }
Chris@82 365
Chris@82 366 static const tw_instr twinstr[] = {
Chris@82 367 {TW_FULL, 0, 8},
Chris@82 368 {TW_NEXT, 1, 0}
Chris@82 369 };
Chris@82 370
Chris@82 371 static const ct_desc desc = { 8, "t1_8", twinstr, &GENUS, {52, 18, 14, 0}, 0, 0, 0 };
Chris@82 372
Chris@82 373 void X(codelet_t1_8) (planner *p) {
Chris@82 374 X(kdft_dit_register) (p, t1_8, &desc);
Chris@82 375 }
Chris@82 376 #endif