Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/rdft/scalar/r2cf/r2cfII_20.c @ 167:bd3cc4d1df30

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 <cannam@all-day-breakfast.com>
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
comparison
equal deleted inserted replaced
166:cbd6d7e562c7 167:bd3cc4d1df30
1 /*
2 * Copyright (c) 2003, 2007-14 Matteo Frigo
3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18 *
19 */
20
21 /* This file was automatically generated --- DO NOT EDIT */
22 /* Generated on Thu May 24 08:06:48 EDT 2018 */
23
24 #include "rdft/codelet-rdft.h"
25
26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
27
28 /* Generated by: ../../../genfft/gen_r2cf.native -fma -compact -variables 4 -pipeline-latency 4 -n 20 -name r2cfII_20 -dft-II -include rdft/scalar/r2cfII.h */
29
30 /*
31 * This function contains 102 FP additions, 63 FP multiplications,
32 * (or, 39 additions, 0 multiplications, 63 fused multiply/add),
33 * 53 stack variables, 10 constants, and 40 memory accesses
34 */
35 #include "rdft/scalar/r2cfII.h"
36
37 static void r2cfII_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
41 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
42 DK(KP690983005, +0.690983005625052575897706582817180941139845410);
43 DK(KP447213595, +0.447213595499957939281834733746255247088123672);
44 DK(KP552786404, +0.552786404500042060718165266253744752911876328);
45 DK(KP809016994, +0.809016994374947424102293417182819058860154590);
46 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
47 DK(KP381966011, +0.381966011250105151795413165634361882279690820);
48 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
49 {
50 INT i;
51 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(80, rs), MAKE_VOLATILE_STRIDE(80, csr), MAKE_VOLATILE_STRIDE(80, csi)) {
52 E Ti, T1d, T1f, T1e, Tg, T1p, TS, T1g, T1, T6, T7, T1r, T1k, T8, To;
53 E Tp, Tv, TX, Tr, TV, Tx, TF, TC, TD, T12, TG, TK, T10, Tc, Tf;
54 Ti = R1[WS(rs, 2)];
55 T1d = R0[WS(rs, 5)];
56 {
57 E Ta, Tb, Td, Te;
58 Ta = R0[WS(rs, 9)];
59 Tb = R0[WS(rs, 1)];
60 Tc = Ta - Tb;
61 T1f = Ta + Tb;
62 Td = R0[WS(rs, 3)];
63 Te = R0[WS(rs, 7)];
64 Tf = Td - Te;
65 T1e = Td + Te;
66 }
67 Tg = FNMS(KP618033988, Tf, Tc);
68 T1p = FMA(KP381966011, T1e, T1f);
69 TS = FMA(KP618033988, Tc, Tf);
70 T1g = FMA(KP381966011, T1f, T1e);
71 {
72 E T2, T5, T3, T4, T1i, T1j;
73 T1 = R0[0];
74 T2 = R0[WS(rs, 4)];
75 T5 = R0[WS(rs, 6)];
76 T3 = R0[WS(rs, 8)];
77 T4 = R0[WS(rs, 2)];
78 T1i = T2 + T5;
79 T1j = T3 + T4;
80 T6 = T2 + T3 - T4 - T5;
81 T7 = FNMS(KP250000000, T6, T1);
82 T1r = FNMS(KP618033988, T1i, T1j);
83 T1k = FMA(KP618033988, T1j, T1i);
84 T8 = (T3 + T5 - T2) - T4;
85 }
86 {
87 E Tn, Tu, Tt, Tq, TU;
88 {
89 E Tj, Tk, Tl, Tm;
90 Tj = R1[WS(rs, 8)];
91 To = R1[WS(rs, 6)];
92 Tk = R1[0];
93 Tl = R1[WS(rs, 4)];
94 Tm = Tk + Tl;
95 Tn = Tj - Tm;
96 Tu = Tk - Tl;
97 Tp = Tj + Tm;
98 Tt = To + Tj;
99 }
100 Tv = FNMS(KP618033988, Tu, Tt);
101 TX = FMA(KP618033988, Tt, Tu);
102 Tq = FMA(KP809016994, Tp, To);
103 Tr = FNMS(KP552786404, Tq, Tn);
104 TU = FMA(KP447213595, Tp, Tn);
105 TV = FNMS(KP690983005, TU, To);
106 }
107 {
108 E TJ, TE, TI, TZ;
109 Tx = R1[WS(rs, 7)];
110 {
111 E Ty, Tz, TA, TB;
112 Ty = R1[WS(rs, 1)];
113 TF = R1[WS(rs, 3)];
114 Tz = R1[WS(rs, 5)];
115 TA = R1[WS(rs, 9)];
116 TB = Tz + TA;
117 TC = Ty + TB;
118 TJ = Tz - TA;
119 TE = Ty - TB;
120 TI = TF + Ty;
121 }
122 TD = FMA(KP250000000, TC, Tx);
123 T12 = FNMS(KP618033988, TI, TJ);
124 TG = FNMS(KP552786404, TF, TE);
125 TK = FMA(KP618033988, TJ, TI);
126 TZ = FMA(KP447213595, TC, TE);
127 T10 = FNMS(KP690983005, TZ, TF);
128 }
129 {
130 E T19, T1w, T1c, T1x, T1a, T1b;
131 T19 = T1 + T6;
132 T1w = T1f + T1d - T1e;
133 T1a = Ti + To - Tp;
134 T1b = TC - TF - Tx;
135 T1c = T1a + T1b;
136 T1x = T1a - T1b;
137 Cr[WS(csr, 2)] = FNMS(KP707106781, T1c, T19);
138 Ci[WS(csi, 2)] = FMS(KP707106781, T1x, T1w);
139 Cr[WS(csr, 7)] = FMA(KP707106781, T1c, T19);
140 Ci[WS(csi, 7)] = FMA(KP707106781, T1x, T1w);
141 }
142 {
143 E TT, T15, T1s, T1u, TY, T17, T13, T16;
144 {
145 E TR, T1q, TW, T11;
146 TR = FMA(KP559016994, T8, T7);
147 TT = FMA(KP951056516, TS, TR);
148 T15 = FNMS(KP951056516, TS, TR);
149 T1q = FNMS(KP809016994, T1p, T1d);
150 T1s = FNMS(KP951056516, T1r, T1q);
151 T1u = FMA(KP951056516, T1r, T1q);
152 TW = FNMS(KP809016994, TV, Ti);
153 TY = FMA(KP951056516, TX, TW);
154 T17 = FNMS(KP951056516, TX, TW);
155 T11 = FNMS(KP809016994, T10, Tx);
156 T13 = FNMS(KP951056516, T12, T11);
157 T16 = FMA(KP951056516, T12, T11);
158 }
159 {
160 E T14, T1v, T18, T1t;
161 T14 = TY - T13;
162 Cr[WS(csr, 6)] = FNMS(KP707106781, T14, TT);
163 Cr[WS(csr, 3)] = FMA(KP707106781, T14, TT);
164 T1v = T17 + T16;
165 Ci[WS(csi, 6)] = FMS(KP707106781, T1v, T1u);
166 Ci[WS(csi, 3)] = FMA(KP707106781, T1v, T1u);
167 T18 = T16 - T17;
168 Cr[WS(csr, 8)] = FNMS(KP707106781, T18, T15);
169 Cr[WS(csr, 1)] = FMA(KP707106781, T18, T15);
170 T1t = TY + T13;
171 Ci[WS(csi, 8)] = -(FMA(KP707106781, T1t, T1s));
172 Ci[WS(csi, 1)] = FNMS(KP707106781, T1t, T1s);
173 }
174 }
175 {
176 E Th, TN, T1l, T1n, Tw, TO, TL, TP;
177 {
178 E T9, T1h, Ts, TH;
179 T9 = FNMS(KP559016994, T8, T7);
180 Th = FNMS(KP951056516, Tg, T9);
181 TN = FMA(KP951056516, Tg, T9);
182 T1h = FMA(KP809016994, T1g, T1d);
183 T1l = FMA(KP951056516, T1k, T1h);
184 T1n = FNMS(KP951056516, T1k, T1h);
185 Ts = FNMS(KP559016994, Tr, Ti);
186 Tw = FNMS(KP951056516, Tv, Ts);
187 TO = FMA(KP951056516, Tv, Ts);
188 TH = FNMS(KP559016994, TG, TD);
189 TL = FNMS(KP951056516, TK, TH);
190 TP = FMA(KP951056516, TK, TH);
191 }
192 {
193 E TM, T1m, TQ, T1o;
194 TM = Tw - TL;
195 Cr[WS(csr, 9)] = FNMS(KP707106781, TM, Th);
196 Cr[0] = FMA(KP707106781, TM, Th);
197 T1m = TO + TP;
198 Ci[0] = -(FMA(KP707106781, T1m, T1l));
199 Ci[WS(csi, 9)] = FNMS(KP707106781, T1m, T1l);
200 TQ = TO - TP;
201 Cr[WS(csr, 5)] = FNMS(KP707106781, TQ, TN);
202 Cr[WS(csr, 4)] = FMA(KP707106781, TQ, TN);
203 T1o = Tw + TL;
204 Ci[WS(csi, 4)] = -(FMA(KP707106781, T1o, T1n));
205 Ci[WS(csi, 5)] = FNMS(KP707106781, T1o, T1n);
206 }
207 }
208 }
209 }
210 }
211
212 static const kr2c_desc desc = { 20, "r2cfII_20", {39, 0, 63, 0}, &GENUS };
213
214 void X(codelet_r2cfII_20) (planner *p) {
215 X(kr2c_register) (p, r2cfII_20, &desc);
216 }
217
218 #else
219
220 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 20 -name r2cfII_20 -dft-II -include rdft/scalar/r2cfII.h */
221
222 /*
223 * This function contains 102 FP additions, 34 FP multiplications,
224 * (or, 86 additions, 18 multiplications, 16 fused multiply/add),
225 * 60 stack variables, 13 constants, and 40 memory accesses
226 */
227 #include "rdft/scalar/r2cfII.h"
228
229 static void r2cfII_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
230 {
231 DK(KP572061402, +0.572061402817684297600072783580302076536153377);
232 DK(KP218508012, +0.218508012224410535399650602527877556893735408);
233 DK(KP309016994, +0.309016994374947424102293417182819058860154590);
234 DK(KP809016994, +0.809016994374947424102293417182819058860154590);
235 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
236 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
237 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
238 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
239 DK(KP176776695, +0.176776695296636881100211090526212259821208984);
240 DK(KP395284707, +0.395284707521047416499861693054089816714944392);
241 DK(KP672498511, +0.672498511963957326960058968885748755876783111);
242 DK(KP415626937, +0.415626937777453428589967464113135184222253485);
243 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
244 {
245 INT i;
246 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(80, rs), MAKE_VOLATILE_STRIDE(80, csr), MAKE_VOLATILE_STRIDE(80, csi)) {
247 E T8, TD, Tm, TN, T9, TC, TY, TE, Te, TF, Tl, TK, T12, TL, Tk;
248 E TM, T1, T6, Tq, T1l, T1c, Tp, T1f, T1e, T1d, Ty, TW, T1g, T1m, Tx;
249 E Tu;
250 T8 = R1[WS(rs, 2)];
251 TD = KP707106781 * T8;
252 Tm = R1[WS(rs, 7)];
253 TN = KP707106781 * Tm;
254 {
255 E Ta, TA, Td, TB, Tb, Tc;
256 T9 = R1[WS(rs, 6)];
257 Ta = R1[WS(rs, 8)];
258 TA = T9 + Ta;
259 Tb = R1[0];
260 Tc = R1[WS(rs, 4)];
261 Td = Tb + Tc;
262 TB = Tb - Tc;
263 TC = FMA(KP415626937, TA, KP672498511 * TB);
264 TY = FNMS(KP415626937, TB, KP672498511 * TA);
265 TE = KP395284707 * (Ta - Td);
266 Te = Ta + Td;
267 TF = KP176776695 * Te;
268 }
269 {
270 E Tg, TJ, Tj, TI, Th, Ti;
271 Tg = R1[WS(rs, 1)];
272 Tl = R1[WS(rs, 3)];
273 TJ = Tg + Tl;
274 Th = R1[WS(rs, 5)];
275 Ti = R1[WS(rs, 9)];
276 Tj = Th + Ti;
277 TI = Th - Ti;
278 TK = FNMS(KP415626937, TJ, KP672498511 * TI);
279 T12 = FMA(KP415626937, TI, KP672498511 * TJ);
280 TL = KP395284707 * (Tg - Tj);
281 Tk = Tg + Tj;
282 TM = KP176776695 * Tk;
283 }
284 {
285 E T2, T5, T3, T4, T1a, T1b;
286 T1 = R0[0];
287 T2 = R0[WS(rs, 6)];
288 T5 = R0[WS(rs, 8)];
289 T3 = R0[WS(rs, 2)];
290 T4 = R0[WS(rs, 4)];
291 T1a = T4 + T2;
292 T1b = T5 + T3;
293 T6 = T2 + T3 - (T4 + T5);
294 Tq = FMA(KP250000000, T6, T1);
295 T1l = FNMS(KP951056516, T1b, KP587785252 * T1a);
296 T1c = FMA(KP951056516, T1a, KP587785252 * T1b);
297 Tp = KP559016994 * (T5 + T2 - (T4 + T3));
298 }
299 T1f = R0[WS(rs, 5)];
300 {
301 E Tv, Tw, Ts, Tt;
302 Tv = R0[WS(rs, 9)];
303 Tw = R0[WS(rs, 1)];
304 Tx = Tv - Tw;
305 T1e = Tv + Tw;
306 Ts = R0[WS(rs, 3)];
307 Tt = R0[WS(rs, 7)];
308 Tu = Ts - Tt;
309 T1d = Ts + Tt;
310 }
311 Ty = FMA(KP951056516, Tu, KP587785252 * Tx);
312 TW = FNMS(KP951056516, Tx, KP587785252 * Tu);
313 T1g = FMA(KP809016994, T1d, KP309016994 * T1e) + T1f;
314 T1m = FNMS(KP809016994, T1e, T1f) - (KP309016994 * T1d);
315 {
316 E T7, T1r, To, T1q, Tf, Tn;
317 T7 = T1 - T6;
318 T1r = T1e + T1f - T1d;
319 Tf = T8 + (T9 - Te);
320 Tn = (Tk - Tl) - Tm;
321 To = KP707106781 * (Tf + Tn);
322 T1q = KP707106781 * (Tf - Tn);
323 Cr[WS(csr, 2)] = T7 - To;
324 Ci[WS(csi, 2)] = T1q - T1r;
325 Cr[WS(csr, 7)] = T7 + To;
326 Ci[WS(csi, 7)] = T1q + T1r;
327 }
328 {
329 E T1h, T1j, TX, T15, T10, T16, T13, T17, TV, TZ, T11;
330 T1h = T1c - T1g;
331 T1j = T1c + T1g;
332 TV = Tq - Tp;
333 TX = TV - TW;
334 T15 = TV + TW;
335 TZ = FMA(KP218508012, T9, TD) + TF - TE;
336 T10 = TY + TZ;
337 T16 = TZ - TY;
338 T11 = FNMS(KP218508012, Tl, TL) - (TM + TN);
339 T13 = T11 - T12;
340 T17 = T11 + T12;
341 {
342 E T14, T19, T18, T1i;
343 T14 = T10 + T13;
344 Cr[WS(csr, 5)] = TX - T14;
345 Cr[WS(csr, 4)] = TX + T14;
346 T19 = T17 - T16;
347 Ci[WS(csi, 5)] = T19 - T1h;
348 Ci[WS(csi, 4)] = T19 + T1h;
349 T18 = T16 + T17;
350 Cr[WS(csr, 9)] = T15 - T18;
351 Cr[0] = T15 + T18;
352 T1i = T13 - T10;
353 Ci[0] = T1i - T1j;
354 Ci[WS(csi, 9)] = T1i + T1j;
355 }
356 }
357 {
358 E T1n, T1p, Tz, TR, TH, TS, TP, TT, Tr, TG, TO;
359 T1n = T1l + T1m;
360 T1p = T1m - T1l;
361 Tr = Tp + Tq;
362 Tz = Tr + Ty;
363 TR = Tr - Ty;
364 TG = TD + TE + FNMS(KP572061402, T9, TF);
365 TH = TC + TG;
366 TS = TC - TG;
367 TO = TL + TM + FNMS(KP572061402, Tl, TN);
368 TP = TK - TO;
369 TT = TK + TO;
370 {
371 E TQ, T1o, TU, T1k;
372 TQ = TH + TP;
373 Cr[WS(csr, 6)] = Tz - TQ;
374 Cr[WS(csr, 3)] = Tz + TQ;
375 T1o = TT - TS;
376 Ci[WS(csi, 6)] = T1o - T1p;
377 Ci[WS(csi, 3)] = T1o + T1p;
378 TU = TS + TT;
379 Cr[WS(csr, 8)] = TR - TU;
380 Cr[WS(csr, 1)] = TR + TU;
381 T1k = TP - TH;
382 Ci[WS(csi, 8)] = T1k - T1n;
383 Ci[WS(csi, 1)] = T1k + T1n;
384 }
385 }
386 }
387 }
388 }
389
390 static const kr2c_desc desc = { 20, "r2cfII_20", {86, 18, 16, 0}, &GENUS };
391
392 void X(codelet_r2cfII_20) (planner *p) {
393 X(kr2c_register) (p, r2cfII_20, &desc);
394 }
395
396 #endif