Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/rdft/scalar/r2cb/r2cb_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 08:07:30 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_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 20 -name r2cb_20 -include rdft/scalar/r2cb.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 86 FP additions, 44 FP multiplications,
Chris@82 32 * (or, 42 additions, 0 multiplications, 44 fused multiply/add),
Chris@82 33 * 50 stack variables, 5 constants, and 40 memory accesses
Chris@82 34 */
Chris@82 35 #include "rdft/scalar/r2cb.h"
Chris@82 36
Chris@82 37 static void r2cb_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 38 {
Chris@82 39 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
Chris@82 40 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
Chris@82 41 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 42 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82 43 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 44 {
Chris@82 45 INT i;
Chris@82 46 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(80, rs), MAKE_VOLATILE_STRIDE(80, csr), MAKE_VOLATILE_STRIDE(80, csi)) {
Chris@82 47 E T5, TD, Tl, Tr, TO, T1l, T1d, T10, T1k, TT, T11, T1a, Tc, Tj, Tk;
Chris@82 48 E Tw, TB, TC, Tm, Tn, To, TE, TF, TG;
Chris@82 49 {
Chris@82 50 E T4, Tq, T3, Tp, T1, T2;
Chris@82 51 T4 = Cr[WS(csr, 5)];
Chris@82 52 Tq = Ci[WS(csi, 5)];
Chris@82 53 T1 = Cr[0];
Chris@82 54 T2 = Cr[WS(csr, 10)];
Chris@82 55 T3 = T1 + T2;
Chris@82 56 Tp = T1 - T2;
Chris@82 57 T5 = FNMS(KP2_000000000, T4, T3);
Chris@82 58 TD = FNMS(KP2_000000000, Tq, Tp);
Chris@82 59 Tl = FMA(KP2_000000000, T4, T3);
Chris@82 60 Tr = FMA(KP2_000000000, Tq, Tp);
Chris@82 61 }
Chris@82 62 {
Chris@82 63 E T8, Ts, TR, T19, Tb, T18, Tv, TS, Tf, Tx, TM, T1c, Ti, T1b, TA;
Chris@82 64 E TN;
Chris@82 65 {
Chris@82 66 E T6, T7, TP, TQ;
Chris@82 67 T6 = Cr[WS(csr, 4)];
Chris@82 68 T7 = Cr[WS(csr, 6)];
Chris@82 69 T8 = T6 + T7;
Chris@82 70 Ts = T6 - T7;
Chris@82 71 TP = Ci[WS(csi, 4)];
Chris@82 72 TQ = Ci[WS(csi, 6)];
Chris@82 73 TR = TP - TQ;
Chris@82 74 T19 = TP + TQ;
Chris@82 75 }
Chris@82 76 {
Chris@82 77 E T9, Ta, Tt, Tu;
Chris@82 78 T9 = Cr[WS(csr, 9)];
Chris@82 79 Ta = Cr[WS(csr, 1)];
Chris@82 80 Tb = T9 + Ta;
Chris@82 81 T18 = T9 - Ta;
Chris@82 82 Tt = Ci[WS(csi, 9)];
Chris@82 83 Tu = Ci[WS(csi, 1)];
Chris@82 84 Tv = Tt + Tu;
Chris@82 85 TS = Tt - Tu;
Chris@82 86 }
Chris@82 87 {
Chris@82 88 E Td, Te, TK, TL;
Chris@82 89 Td = Cr[WS(csr, 8)];
Chris@82 90 Te = Cr[WS(csr, 2)];
Chris@82 91 Tf = Td + Te;
Chris@82 92 Tx = Td - Te;
Chris@82 93 TK = Ci[WS(csi, 8)];
Chris@82 94 TL = Ci[WS(csi, 2)];
Chris@82 95 TM = TK - TL;
Chris@82 96 T1c = TK + TL;
Chris@82 97 }
Chris@82 98 {
Chris@82 99 E Tg, Th, Ty, Tz;
Chris@82 100 Tg = Cr[WS(csr, 7)];
Chris@82 101 Th = Cr[WS(csr, 3)];
Chris@82 102 Ti = Tg + Th;
Chris@82 103 T1b = Tg - Th;
Chris@82 104 Ty = Ci[WS(csi, 7)];
Chris@82 105 Tz = Ci[WS(csi, 3)];
Chris@82 106 TA = Ty + Tz;
Chris@82 107 TN = Tz - Ty;
Chris@82 108 }
Chris@82 109 TO = TM - TN;
Chris@82 110 T1l = T19 - T18;
Chris@82 111 T1d = T1b + T1c;
Chris@82 112 T10 = TS + TR;
Chris@82 113 T1k = T1c - T1b;
Chris@82 114 TT = TR - TS;
Chris@82 115 T11 = TN + TM;
Chris@82 116 T1a = T18 + T19;
Chris@82 117 Tc = T8 - Tb;
Chris@82 118 Tj = Tf - Ti;
Chris@82 119 Tk = Tc + Tj;
Chris@82 120 Tw = Ts + Tv;
Chris@82 121 TB = Tx - TA;
Chris@82 122 TC = Tw + TB;
Chris@82 123 Tm = T8 + Tb;
Chris@82 124 Tn = Tf + Ti;
Chris@82 125 To = Tm + Tn;
Chris@82 126 TE = Ts - Tv;
Chris@82 127 TF = Tx + TA;
Chris@82 128 TG = TE + TF;
Chris@82 129 }
Chris@82 130 R0[WS(rs, 5)] = FMA(KP2_000000000, Tk, T5);
Chris@82 131 R1[WS(rs, 7)] = FMA(KP2_000000000, TC, Tr);
Chris@82 132 R1[WS(rs, 2)] = FMA(KP2_000000000, TG, TD);
Chris@82 133 R0[0] = FMA(KP2_000000000, To, Tl);
Chris@82 134 {
Chris@82 135 E TU, TW, TJ, TV, TH, TI;
Chris@82 136 TU = FNMS(KP618033988, TT, TO);
Chris@82 137 TW = FMA(KP618033988, TO, TT);
Chris@82 138 TH = FNMS(KP500000000, Tk, T5);
Chris@82 139 TI = Tc - Tj;
Chris@82 140 TJ = FNMS(KP1_118033988, TI, TH);
Chris@82 141 TV = FMA(KP1_118033988, TI, TH);
Chris@82 142 R0[WS(rs, 9)] = FNMS(KP1_902113032, TU, TJ);
Chris@82 143 R0[WS(rs, 7)] = FMA(KP1_902113032, TW, TV);
Chris@82 144 R0[WS(rs, 1)] = FMA(KP1_902113032, TU, TJ);
Chris@82 145 R0[WS(rs, 3)] = FNMS(KP1_902113032, TW, TV);
Chris@82 146 }
Chris@82 147 {
Chris@82 148 E T1e, T1g, T17, T1f, T15, T16;
Chris@82 149 T1e = FMA(KP618033988, T1d, T1a);
Chris@82 150 T1g = FNMS(KP618033988, T1a, T1d);
Chris@82 151 T15 = FNMS(KP500000000, TG, TD);
Chris@82 152 T16 = TE - TF;
Chris@82 153 T17 = FMA(KP1_118033988, T16, T15);
Chris@82 154 T1f = FNMS(KP1_118033988, T16, T15);
Chris@82 155 R1[0] = FNMS(KP1_902113032, T1e, T17);
Chris@82 156 R1[WS(rs, 8)] = FMA(KP1_902113032, T1g, T1f);
Chris@82 157 R1[WS(rs, 4)] = FMA(KP1_902113032, T1e, T17);
Chris@82 158 R1[WS(rs, 6)] = FNMS(KP1_902113032, T1g, T1f);
Chris@82 159 }
Chris@82 160 {
Chris@82 161 E T1m, T1o, T1j, T1n, T1h, T1i;
Chris@82 162 T1m = FNMS(KP618033988, T1l, T1k);
Chris@82 163 T1o = FMA(KP618033988, T1k, T1l);
Chris@82 164 T1h = FNMS(KP500000000, TC, Tr);
Chris@82 165 T1i = Tw - TB;
Chris@82 166 T1j = FNMS(KP1_118033988, T1i, T1h);
Chris@82 167 T1n = FMA(KP1_118033988, T1i, T1h);
Chris@82 168 R1[WS(rs, 1)] = FNMS(KP1_902113032, T1m, T1j);
Chris@82 169 R1[WS(rs, 9)] = FMA(KP1_902113032, T1o, T1n);
Chris@82 170 R1[WS(rs, 3)] = FMA(KP1_902113032, T1m, T1j);
Chris@82 171 R1[WS(rs, 5)] = FNMS(KP1_902113032, T1o, T1n);
Chris@82 172 }
Chris@82 173 {
Chris@82 174 E T12, T14, TZ, T13, TX, TY;
Chris@82 175 T12 = FMA(KP618033988, T11, T10);
Chris@82 176 T14 = FNMS(KP618033988, T10, T11);
Chris@82 177 TX = FNMS(KP500000000, To, Tl);
Chris@82 178 TY = Tm - Tn;
Chris@82 179 TZ = FMA(KP1_118033988, TY, TX);
Chris@82 180 T13 = FNMS(KP1_118033988, TY, TX);
Chris@82 181 R0[WS(rs, 8)] = FNMS(KP1_902113032, T12, TZ);
Chris@82 182 R0[WS(rs, 6)] = FMA(KP1_902113032, T14, T13);
Chris@82 183 R0[WS(rs, 2)] = FMA(KP1_902113032, T12, TZ);
Chris@82 184 R0[WS(rs, 4)] = FNMS(KP1_902113032, T14, T13);
Chris@82 185 }
Chris@82 186 }
Chris@82 187 }
Chris@82 188 }
Chris@82 189
Chris@82 190 static const kr2c_desc desc = { 20, "r2cb_20", {42, 0, 44, 0}, &GENUS };
Chris@82 191
Chris@82 192 void X(codelet_r2cb_20) (planner *p) {
Chris@82 193 X(kr2c_register) (p, r2cb_20, &desc);
Chris@82 194 }
Chris@82 195
Chris@82 196 #else
Chris@82 197
Chris@82 198 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 20 -name r2cb_20 -include rdft/scalar/r2cb.h */
Chris@82 199
Chris@82 200 /*
Chris@82 201 * This function contains 86 FP additions, 30 FP multiplications,
Chris@82 202 * (or, 70 additions, 14 multiplications, 16 fused multiply/add),
Chris@82 203 * 50 stack variables, 5 constants, and 40 memory accesses
Chris@82 204 */
Chris@82 205 #include "rdft/scalar/r2cb.h"
Chris@82 206
Chris@82 207 static void r2cb_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@82 208 {
Chris@82 209 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
Chris@82 210 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 211 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
Chris@82 212 DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
Chris@82 213 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@82 214 {
Chris@82 215 INT i;
Chris@82 216 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(80, rs), MAKE_VOLATILE_STRIDE(80, csr), MAKE_VOLATILE_STRIDE(80, csi)) {
Chris@82 217 E T6, TF, Tm, Tt, TQ, T1n, T1f, T12, T1m, TV, T13, T1c, Td, Tk, Tl;
Chris@82 218 E Ty, TD, TE, Tn, To, Tp, TG, TH, TI;
Chris@82 219 {
Chris@82 220 E T5, Ts, T3, Tq;
Chris@82 221 {
Chris@82 222 E T4, Tr, T1, T2;
Chris@82 223 T4 = Cr[WS(csr, 5)];
Chris@82 224 T5 = KP2_000000000 * T4;
Chris@82 225 Tr = Ci[WS(csi, 5)];
Chris@82 226 Ts = KP2_000000000 * Tr;
Chris@82 227 T1 = Cr[0];
Chris@82 228 T2 = Cr[WS(csr, 10)];
Chris@82 229 T3 = T1 + T2;
Chris@82 230 Tq = T1 - T2;
Chris@82 231 }
Chris@82 232 T6 = T3 - T5;
Chris@82 233 TF = Tq - Ts;
Chris@82 234 Tm = T3 + T5;
Chris@82 235 Tt = Tq + Ts;
Chris@82 236 }
Chris@82 237 {
Chris@82 238 E T9, Tu, TO, T1b, Tc, T1a, Tx, TP, Tg, Tz, TT, T1e, Tj, T1d, TC;
Chris@82 239 E TU;
Chris@82 240 {
Chris@82 241 E T7, T8, TM, TN;
Chris@82 242 T7 = Cr[WS(csr, 4)];
Chris@82 243 T8 = Cr[WS(csr, 6)];
Chris@82 244 T9 = T7 + T8;
Chris@82 245 Tu = T7 - T8;
Chris@82 246 TM = Ci[WS(csi, 4)];
Chris@82 247 TN = Ci[WS(csi, 6)];
Chris@82 248 TO = TM - TN;
Chris@82 249 T1b = TM + TN;
Chris@82 250 }
Chris@82 251 {
Chris@82 252 E Ta, Tb, Tv, Tw;
Chris@82 253 Ta = Cr[WS(csr, 9)];
Chris@82 254 Tb = Cr[WS(csr, 1)];
Chris@82 255 Tc = Ta + Tb;
Chris@82 256 T1a = Ta - Tb;
Chris@82 257 Tv = Ci[WS(csi, 9)];
Chris@82 258 Tw = Ci[WS(csi, 1)];
Chris@82 259 Tx = Tv + Tw;
Chris@82 260 TP = Tv - Tw;
Chris@82 261 }
Chris@82 262 {
Chris@82 263 E Te, Tf, TR, TS;
Chris@82 264 Te = Cr[WS(csr, 8)];
Chris@82 265 Tf = Cr[WS(csr, 2)];
Chris@82 266 Tg = Te + Tf;
Chris@82 267 Tz = Te - Tf;
Chris@82 268 TR = Ci[WS(csi, 8)];
Chris@82 269 TS = Ci[WS(csi, 2)];
Chris@82 270 TT = TR - TS;
Chris@82 271 T1e = TR + TS;
Chris@82 272 }
Chris@82 273 {
Chris@82 274 E Th, Ti, TA, TB;
Chris@82 275 Th = Cr[WS(csr, 7)];
Chris@82 276 Ti = Cr[WS(csr, 3)];
Chris@82 277 Tj = Th + Ti;
Chris@82 278 T1d = Th - Ti;
Chris@82 279 TA = Ci[WS(csi, 7)];
Chris@82 280 TB = Ci[WS(csi, 3)];
Chris@82 281 TC = TA + TB;
Chris@82 282 TU = TB - TA;
Chris@82 283 }
Chris@82 284 TQ = TO - TP;
Chris@82 285 T1n = T1e - T1d;
Chris@82 286 T1f = T1d + T1e;
Chris@82 287 T12 = TP + TO;
Chris@82 288 T1m = T1b - T1a;
Chris@82 289 TV = TT - TU;
Chris@82 290 T13 = TU + TT;
Chris@82 291 T1c = T1a + T1b;
Chris@82 292 Td = T9 - Tc;
Chris@82 293 Tk = Tg - Tj;
Chris@82 294 Tl = Td + Tk;
Chris@82 295 Ty = Tu + Tx;
Chris@82 296 TD = Tz - TC;
Chris@82 297 TE = Ty + TD;
Chris@82 298 Tn = T9 + Tc;
Chris@82 299 To = Tg + Tj;
Chris@82 300 Tp = Tn + To;
Chris@82 301 TG = Tu - Tx;
Chris@82 302 TH = Tz + TC;
Chris@82 303 TI = TG + TH;
Chris@82 304 }
Chris@82 305 R0[WS(rs, 5)] = FMA(KP2_000000000, Tl, T6);
Chris@82 306 R1[WS(rs, 7)] = FMA(KP2_000000000, TE, Tt);
Chris@82 307 R1[WS(rs, 2)] = FMA(KP2_000000000, TI, TF);
Chris@82 308 R0[0] = FMA(KP2_000000000, Tp, Tm);
Chris@82 309 {
Chris@82 310 E TW, TY, TL, TX, TJ, TK;
Chris@82 311 TW = FNMS(KP1_902113032, TV, KP1_175570504 * TQ);
Chris@82 312 TY = FMA(KP1_902113032, TQ, KP1_175570504 * TV);
Chris@82 313 TJ = FNMS(KP500000000, Tl, T6);
Chris@82 314 TK = KP1_118033988 * (Td - Tk);
Chris@82 315 TL = TJ - TK;
Chris@82 316 TX = TK + TJ;
Chris@82 317 R0[WS(rs, 1)] = TL - TW;
Chris@82 318 R0[WS(rs, 7)] = TX + TY;
Chris@82 319 R0[WS(rs, 9)] = TL + TW;
Chris@82 320 R0[WS(rs, 3)] = TX - TY;
Chris@82 321 }
Chris@82 322 {
Chris@82 323 E T1g, T1i, T19, T1h, T17, T18;
Chris@82 324 T1g = FNMS(KP1_902113032, T1f, KP1_175570504 * T1c);
Chris@82 325 T1i = FMA(KP1_902113032, T1c, KP1_175570504 * T1f);
Chris@82 326 T17 = FNMS(KP500000000, TI, TF);
Chris@82 327 T18 = KP1_118033988 * (TG - TH);
Chris@82 328 T19 = T17 - T18;
Chris@82 329 T1h = T18 + T17;
Chris@82 330 R1[WS(rs, 8)] = T19 - T1g;
Chris@82 331 R1[WS(rs, 4)] = T1h + T1i;
Chris@82 332 R1[WS(rs, 6)] = T19 + T1g;
Chris@82 333 R1[0] = T1h - T1i;
Chris@82 334 }
Chris@82 335 {
Chris@82 336 E T1o, T1q, T1l, T1p, T1j, T1k;
Chris@82 337 T1o = FNMS(KP1_902113032, T1n, KP1_175570504 * T1m);
Chris@82 338 T1q = FMA(KP1_902113032, T1m, KP1_175570504 * T1n);
Chris@82 339 T1j = FNMS(KP500000000, TE, Tt);
Chris@82 340 T1k = KP1_118033988 * (Ty - TD);
Chris@82 341 T1l = T1j - T1k;
Chris@82 342 T1p = T1k + T1j;
Chris@82 343 R1[WS(rs, 3)] = T1l - T1o;
Chris@82 344 R1[WS(rs, 9)] = T1p + T1q;
Chris@82 345 R1[WS(rs, 1)] = T1l + T1o;
Chris@82 346 R1[WS(rs, 5)] = T1p - T1q;
Chris@82 347 }
Chris@82 348 {
Chris@82 349 E T14, T16, T11, T15, TZ, T10;
Chris@82 350 T14 = FNMS(KP1_902113032, T13, KP1_175570504 * T12);
Chris@82 351 T16 = FMA(KP1_902113032, T12, KP1_175570504 * T13);
Chris@82 352 TZ = FNMS(KP500000000, Tp, Tm);
Chris@82 353 T10 = KP1_118033988 * (Tn - To);
Chris@82 354 T11 = TZ - T10;
Chris@82 355 T15 = T10 + TZ;
Chris@82 356 R0[WS(rs, 6)] = T11 - T14;
Chris@82 357 R0[WS(rs, 2)] = T15 + T16;
Chris@82 358 R0[WS(rs, 4)] = T11 + T14;
Chris@82 359 R0[WS(rs, 8)] = T15 - T16;
Chris@82 360 }
Chris@82 361 }
Chris@82 362 }
Chris@82 363 }
Chris@82 364
Chris@82 365 static const kr2c_desc desc = { 20, "r2cb_20", {70, 14, 16, 0}, &GENUS };
Chris@82 366
Chris@82 367 void X(codelet_r2cb_20) (planner *p) {
Chris@82 368 X(kr2c_register) (p, r2cb_20, &desc);
Chris@82 369 }
Chris@82 370
Chris@82 371 #endif