Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/scalar/codelets/n1_15.c @ 95:89f5e221ed7b

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW3
author Chris Cannam <cannam@all-day-breakfast.com>
date Wed, 20 Mar 2013 15:35:50 +0000
parents
children
comparison
equal deleted inserted replaced
94:d278df1123f9 95:89f5e221ed7b
1 /*
2 * Copyright (c) 2003, 2007-11 Matteo Frigo
3 * Copyright (c) 2003, 2007-11 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 Sun Nov 25 07:35:43 EST 2012 */
23
24 #include "codelet-dft.h"
25
26 #ifdef HAVE_FMA
27
28 /* Generated by: ../../../genfft/gen_notw.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 15 -name n1_15 -include n.h */
29
30 /*
31 * This function contains 156 FP additions, 84 FP multiplications,
32 * (or, 72 additions, 0 multiplications, 84 fused multiply/add),
33 * 75 stack variables, 6 constants, and 60 memory accesses
34 */
35 #include "n.h"
36
37 static void n1_15(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
41 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
42 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
43 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
44 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
45 {
46 INT i;
47 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(60, is), MAKE_VOLATILE_STRIDE(60, os)) {
48 E T1r, T1g, T14, T13;
49 {
50 E T5, T2l, Tx, TV, T1z, T1X, T2s, Tr, T24, TT, T2e, T2n, T1Z, T1Q, T1B;
51 E T11, T1H, TW, T2t, Tg, TX, T25, TI, T2h, T2m, T1Y, T1T, T1A;
52 {
53 E T1, T1v, T2, T3, Tu, Tv, TZ, T10;
54 T1 = ri[0];
55 T1v = ii[0];
56 T2 = ri[WS(is, 5)];
57 T3 = ri[WS(is, 10)];
58 Tu = ii[WS(is, 5)];
59 Tv = ii[WS(is, 10)];
60 {
61 E T1k, Tm, TM, TJ, Tl, T2c, T1j, T1m, TP, T1p, Tp, TQ;
62 {
63 E Th, T1h, TK, TL, Tk, Tn, To, T1i;
64 {
65 E Ti, Tj, T1y, T4;
66 Th = ri[WS(is, 6)];
67 T1y = T3 - T2;
68 T4 = T2 + T3;
69 {
70 E T1w, Tw, Tt, T1x;
71 T1w = Tu + Tv;
72 Tw = Tu - Tv;
73 Ti = ri[WS(is, 11)];
74 T5 = T1 + T4;
75 Tt = FNMS(KP500000000, T4, T1);
76 T2l = T1v + T1w;
77 T1x = FNMS(KP500000000, T1w, T1v);
78 Tx = FNMS(KP866025403, Tw, Tt);
79 TV = FMA(KP866025403, Tw, Tt);
80 T1z = FMA(KP866025403, T1y, T1x);
81 T1X = FNMS(KP866025403, T1y, T1x);
82 Tj = ri[WS(is, 1)];
83 }
84 T1h = ii[WS(is, 6)];
85 TK = ii[WS(is, 11)];
86 TL = ii[WS(is, 1)];
87 Tk = Ti + Tj;
88 T1k = Tj - Ti;
89 }
90 Tm = ri[WS(is, 9)];
91 TM = TK - TL;
92 T1i = TK + TL;
93 TJ = FNMS(KP500000000, Tk, Th);
94 Tl = Th + Tk;
95 Tn = ri[WS(is, 14)];
96 To = ri[WS(is, 4)];
97 T2c = T1h + T1i;
98 T1j = FNMS(KP500000000, T1i, T1h);
99 T1m = ii[WS(is, 9)];
100 TP = ii[WS(is, 14)];
101 T1p = To - Tn;
102 Tp = Tn + To;
103 TQ = ii[WS(is, 4)];
104 }
105 {
106 E TN, TS, T1o, T2d;
107 {
108 E TO, T1n, TR, Tq;
109 TN = FNMS(KP866025403, TM, TJ);
110 TZ = FMA(KP866025403, TM, TJ);
111 TO = FNMS(KP500000000, Tp, Tm);
112 Tq = Tm + Tp;
113 T1n = TP + TQ;
114 TR = TP - TQ;
115 T2s = Tl - Tq;
116 Tr = Tl + Tq;
117 T10 = FMA(KP866025403, TR, TO);
118 TS = FNMS(KP866025403, TR, TO);
119 T1o = FNMS(KP500000000, T1n, T1m);
120 T2d = T1m + T1n;
121 }
122 {
123 E T1O, T1l, T1P, T1q;
124 T1O = FNMS(KP866025403, T1k, T1j);
125 T1l = FMA(KP866025403, T1k, T1j);
126 T24 = TN - TS;
127 TT = TN + TS;
128 T1P = FNMS(KP866025403, T1p, T1o);
129 T1q = FMA(KP866025403, T1p, T1o);
130 T2e = T2c - T2d;
131 T2n = T2c + T2d;
132 T1Z = T1O + T1P;
133 T1Q = T1O - T1P;
134 T1r = T1l - T1q;
135 T1B = T1l + T1q;
136 }
137 }
138 }
139 {
140 E T19, Tb, TB, Ty, Ta, T2f, T18, T1b, TE, T1e, Te, TF;
141 {
142 E T6, T16, Tz, TA, T9, T7, T8, Tc, Td, T17;
143 T6 = ri[WS(is, 3)];
144 T7 = ri[WS(is, 8)];
145 T11 = TZ + T10;
146 T1H = TZ - T10;
147 T8 = ri[WS(is, 13)];
148 T16 = ii[WS(is, 3)];
149 Tz = ii[WS(is, 8)];
150 TA = ii[WS(is, 13)];
151 T9 = T7 + T8;
152 T19 = T8 - T7;
153 Tb = ri[WS(is, 12)];
154 TB = Tz - TA;
155 T17 = Tz + TA;
156 Ty = FNMS(KP500000000, T9, T6);
157 Ta = T6 + T9;
158 Tc = ri[WS(is, 2)];
159 Td = ri[WS(is, 7)];
160 T2f = T16 + T17;
161 T18 = FNMS(KP500000000, T17, T16);
162 T1b = ii[WS(is, 12)];
163 TE = ii[WS(is, 2)];
164 T1e = Td - Tc;
165 Te = Tc + Td;
166 TF = ii[WS(is, 7)];
167 }
168 {
169 E TC, TH, T1d, T2g;
170 {
171 E TD, T1c, TG, Tf;
172 TC = FNMS(KP866025403, TB, Ty);
173 TW = FMA(KP866025403, TB, Ty);
174 TD = FNMS(KP500000000, Te, Tb);
175 Tf = Tb + Te;
176 T1c = TE + TF;
177 TG = TE - TF;
178 T2t = Ta - Tf;
179 Tg = Ta + Tf;
180 TX = FMA(KP866025403, TG, TD);
181 TH = FNMS(KP866025403, TG, TD);
182 T1d = FNMS(KP500000000, T1c, T1b);
183 T2g = T1b + T1c;
184 }
185 {
186 E T1R, T1a, T1S, T1f;
187 T1R = FNMS(KP866025403, T19, T18);
188 T1a = FMA(KP866025403, T19, T18);
189 T25 = TC - TH;
190 TI = TC + TH;
191 T1S = FNMS(KP866025403, T1e, T1d);
192 T1f = FMA(KP866025403, T1e, T1d);
193 T2h = T2f - T2g;
194 T2m = T2f + T2g;
195 T1Y = T1R + T1S;
196 T1T = T1R - T1S;
197 T1g = T1a - T1f;
198 T1A = T1a + T1f;
199 }
200 }
201 }
202 }
203 {
204 E TY, T1G, T1M, T1L, T2a, T29, Ts, T22, T21, T20;
205 T2a = Tg - Tr;
206 Ts = Tg + Tr;
207 TY = TW + TX;
208 T1G = TW - TX;
209 T29 = FNMS(KP250000000, Ts, T5);
210 ro[0] = T5 + Ts;
211 {
212 E T2q, T2p, T2o, TU;
213 T2o = T2m + T2n;
214 T2q = T2m - T2n;
215 {
216 E T2k, T2i, T2b, T2j;
217 T2k = FMA(KP618033988, T2e, T2h);
218 T2i = FNMS(KP618033988, T2h, T2e);
219 T2b = FNMS(KP559016994, T2a, T29);
220 T2j = FMA(KP559016994, T2a, T29);
221 ro[WS(os, 3)] = FMA(KP951056516, T2i, T2b);
222 ro[WS(os, 12)] = FNMS(KP951056516, T2i, T2b);
223 ro[WS(os, 6)] = FMA(KP951056516, T2k, T2j);
224 ro[WS(os, 9)] = FNMS(KP951056516, T2k, T2j);
225 T2p = FNMS(KP250000000, T2o, T2l);
226 }
227 io[0] = T2l + T2o;
228 TU = TI + TT;
229 T1M = TI - TT;
230 {
231 E T2r, T2v, T2w, T2u;
232 T2r = FNMS(KP559016994, T2q, T2p);
233 T2v = FMA(KP559016994, T2q, T2p);
234 T2w = FMA(KP618033988, T2s, T2t);
235 T2u = FNMS(KP618033988, T2t, T2s);
236 io[WS(os, 9)] = FMA(KP951056516, T2w, T2v);
237 io[WS(os, 6)] = FNMS(KP951056516, T2w, T2v);
238 io[WS(os, 12)] = FMA(KP951056516, T2u, T2r);
239 io[WS(os, 3)] = FNMS(KP951056516, T2u, T2r);
240 T1L = FNMS(KP250000000, TU, Tx);
241 }
242 ro[WS(os, 5)] = Tx + TU;
243 }
244 T20 = T1Y + T1Z;
245 T22 = T1Y - T1Z;
246 {
247 E T1N, T1V, T1W, T1U;
248 T1N = FNMS(KP559016994, T1M, T1L);
249 T1V = FMA(KP559016994, T1M, T1L);
250 T1W = FMA(KP618033988, T1Q, T1T);
251 T1U = FNMS(KP618033988, T1T, T1Q);
252 ro[WS(os, 11)] = FMA(KP951056516, T1W, T1V);
253 ro[WS(os, 14)] = FNMS(KP951056516, T1W, T1V);
254 ro[WS(os, 8)] = FMA(KP951056516, T1U, T1N);
255 ro[WS(os, 2)] = FNMS(KP951056516, T1U, T1N);
256 T21 = FNMS(KP250000000, T20, T1X);
257 }
258 io[WS(os, 5)] = T1X + T20;
259 {
260 E T1E, T1D, T1C, T12;
261 T1C = T1A + T1B;
262 T1E = T1A - T1B;
263 {
264 E T23, T27, T28, T26;
265 T23 = FNMS(KP559016994, T22, T21);
266 T27 = FMA(KP559016994, T22, T21);
267 T28 = FMA(KP618033988, T24, T25);
268 T26 = FNMS(KP618033988, T25, T24);
269 io[WS(os, 14)] = FMA(KP951056516, T28, T27);
270 io[WS(os, 11)] = FNMS(KP951056516, T28, T27);
271 io[WS(os, 8)] = FNMS(KP951056516, T26, T23);
272 io[WS(os, 2)] = FMA(KP951056516, T26, T23);
273 T1D = FNMS(KP250000000, T1C, T1z);
274 }
275 io[WS(os, 10)] = T1z + T1C;
276 T12 = TY + T11;
277 T14 = TY - T11;
278 {
279 E T1F, T1J, T1K, T1I;
280 T1F = FMA(KP559016994, T1E, T1D);
281 T1J = FNMS(KP559016994, T1E, T1D);
282 T1K = FNMS(KP618033988, T1G, T1H);
283 T1I = FMA(KP618033988, T1H, T1G);
284 io[WS(os, 13)] = FNMS(KP951056516, T1K, T1J);
285 io[WS(os, 7)] = FMA(KP951056516, T1K, T1J);
286 io[WS(os, 4)] = FMA(KP951056516, T1I, T1F);
287 io[WS(os, 1)] = FNMS(KP951056516, T1I, T1F);
288 T13 = FNMS(KP250000000, T12, TV);
289 }
290 ro[WS(os, 10)] = TV + T12;
291 }
292 }
293 }
294 {
295 E T1t, T15, T1s, T1u;
296 T1t = FNMS(KP559016994, T14, T13);
297 T15 = FMA(KP559016994, T14, T13);
298 T1s = FMA(KP618033988, T1r, T1g);
299 T1u = FNMS(KP618033988, T1g, T1r);
300 ro[WS(os, 13)] = FMA(KP951056516, T1u, T1t);
301 ro[WS(os, 7)] = FNMS(KP951056516, T1u, T1t);
302 ro[WS(os, 1)] = FMA(KP951056516, T1s, T15);
303 ro[WS(os, 4)] = FNMS(KP951056516, T1s, T15);
304 }
305 }
306 }
307 }
308
309 static const kdft_desc desc = { 15, "n1_15", {72, 0, 84, 0}, &GENUS, 0, 0, 0, 0 };
310
311 void X(codelet_n1_15) (planner *p) {
312 X(kdft_register) (p, n1_15, &desc);
313 }
314
315 #else /* HAVE_FMA */
316
317 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 15 -name n1_15 -include n.h */
318
319 /*
320 * This function contains 156 FP additions, 56 FP multiplications,
321 * (or, 128 additions, 28 multiplications, 28 fused multiply/add),
322 * 69 stack variables, 6 constants, and 60 memory accesses
323 */
324 #include "n.h"
325
326 static void n1_15(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
327 {
328 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
329 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
330 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
331 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
332 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
333 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
334 {
335 INT i;
336 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(60, is), MAKE_VOLATILE_STRIDE(60, os)) {
337 E T5, T2l, Tx, TV, T1C, T20, Tl, Tq, Tr, TN, TS, TT, T2c, T2d, T2n;
338 E T1O, T1P, T22, T1l, T1q, T1w, TZ, T10, T11, Ta, Tf, Tg, TC, TH, TI;
339 E T2f, T2g, T2m, T1R, T1S, T21, T1a, T1f, T1v, TW, TX, TY;
340 {
341 E T1, T1z, T4, T1y, Tw, T1A, Tt, T1B;
342 T1 = ri[0];
343 T1z = ii[0];
344 {
345 E T2, T3, Tu, Tv;
346 T2 = ri[WS(is, 5)];
347 T3 = ri[WS(is, 10)];
348 T4 = T2 + T3;
349 T1y = KP866025403 * (T3 - T2);
350 Tu = ii[WS(is, 5)];
351 Tv = ii[WS(is, 10)];
352 Tw = KP866025403 * (Tu - Tv);
353 T1A = Tu + Tv;
354 }
355 T5 = T1 + T4;
356 T2l = T1z + T1A;
357 Tt = FNMS(KP500000000, T4, T1);
358 Tx = Tt - Tw;
359 TV = Tt + Tw;
360 T1B = FNMS(KP500000000, T1A, T1z);
361 T1C = T1y + T1B;
362 T20 = T1B - T1y;
363 }
364 {
365 E Th, Tk, TJ, T1h, T1i, T1j, TM, T1k, Tm, Tp, TO, T1m, T1n, T1o, TR;
366 E T1p;
367 {
368 E Ti, Tj, TK, TL;
369 Th = ri[WS(is, 6)];
370 Ti = ri[WS(is, 11)];
371 Tj = ri[WS(is, 1)];
372 Tk = Ti + Tj;
373 TJ = FNMS(KP500000000, Tk, Th);
374 T1h = KP866025403 * (Tj - Ti);
375 T1i = ii[WS(is, 6)];
376 TK = ii[WS(is, 11)];
377 TL = ii[WS(is, 1)];
378 T1j = TK + TL;
379 TM = KP866025403 * (TK - TL);
380 T1k = FNMS(KP500000000, T1j, T1i);
381 }
382 {
383 E Tn, To, TP, TQ;
384 Tm = ri[WS(is, 9)];
385 Tn = ri[WS(is, 14)];
386 To = ri[WS(is, 4)];
387 Tp = Tn + To;
388 TO = FNMS(KP500000000, Tp, Tm);
389 T1m = KP866025403 * (To - Tn);
390 T1n = ii[WS(is, 9)];
391 TP = ii[WS(is, 14)];
392 TQ = ii[WS(is, 4)];
393 T1o = TP + TQ;
394 TR = KP866025403 * (TP - TQ);
395 T1p = FNMS(KP500000000, T1o, T1n);
396 }
397 Tl = Th + Tk;
398 Tq = Tm + Tp;
399 Tr = Tl + Tq;
400 TN = TJ - TM;
401 TS = TO - TR;
402 TT = TN + TS;
403 T2c = T1i + T1j;
404 T2d = T1n + T1o;
405 T2n = T2c + T2d;
406 T1O = T1k - T1h;
407 T1P = T1p - T1m;
408 T22 = T1O + T1P;
409 T1l = T1h + T1k;
410 T1q = T1m + T1p;
411 T1w = T1l + T1q;
412 TZ = TJ + TM;
413 T10 = TO + TR;
414 T11 = TZ + T10;
415 }
416 {
417 E T6, T9, Ty, T16, T17, T18, TB, T19, Tb, Te, TD, T1b, T1c, T1d, TG;
418 E T1e;
419 {
420 E T7, T8, Tz, TA;
421 T6 = ri[WS(is, 3)];
422 T7 = ri[WS(is, 8)];
423 T8 = ri[WS(is, 13)];
424 T9 = T7 + T8;
425 Ty = FNMS(KP500000000, T9, T6);
426 T16 = KP866025403 * (T8 - T7);
427 T17 = ii[WS(is, 3)];
428 Tz = ii[WS(is, 8)];
429 TA = ii[WS(is, 13)];
430 T18 = Tz + TA;
431 TB = KP866025403 * (Tz - TA);
432 T19 = FNMS(KP500000000, T18, T17);
433 }
434 {
435 E Tc, Td, TE, TF;
436 Tb = ri[WS(is, 12)];
437 Tc = ri[WS(is, 2)];
438 Td = ri[WS(is, 7)];
439 Te = Tc + Td;
440 TD = FNMS(KP500000000, Te, Tb);
441 T1b = KP866025403 * (Td - Tc);
442 T1c = ii[WS(is, 12)];
443 TE = ii[WS(is, 2)];
444 TF = ii[WS(is, 7)];
445 T1d = TE + TF;
446 TG = KP866025403 * (TE - TF);
447 T1e = FNMS(KP500000000, T1d, T1c);
448 }
449 Ta = T6 + T9;
450 Tf = Tb + Te;
451 Tg = Ta + Tf;
452 TC = Ty - TB;
453 TH = TD - TG;
454 TI = TC + TH;
455 T2f = T17 + T18;
456 T2g = T1c + T1d;
457 T2m = T2f + T2g;
458 T1R = T19 - T16;
459 T1S = T1e - T1b;
460 T21 = T1R + T1S;
461 T1a = T16 + T19;
462 T1f = T1b + T1e;
463 T1v = T1a + T1f;
464 TW = Ty + TB;
465 TX = TD + TG;
466 TY = TW + TX;
467 }
468 {
469 E T2a, Ts, T29, T2i, T2k, T2e, T2h, T2j, T2b;
470 T2a = KP559016994 * (Tg - Tr);
471 Ts = Tg + Tr;
472 T29 = FNMS(KP250000000, Ts, T5);
473 T2e = T2c - T2d;
474 T2h = T2f - T2g;
475 T2i = FNMS(KP587785252, T2h, KP951056516 * T2e);
476 T2k = FMA(KP951056516, T2h, KP587785252 * T2e);
477 ro[0] = T5 + Ts;
478 T2j = T2a + T29;
479 ro[WS(os, 9)] = T2j - T2k;
480 ro[WS(os, 6)] = T2j + T2k;
481 T2b = T29 - T2a;
482 ro[WS(os, 12)] = T2b - T2i;
483 ro[WS(os, 3)] = T2b + T2i;
484 }
485 {
486 E T2q, T2o, T2p, T2u, T2w, T2s, T2t, T2v, T2r;
487 T2q = KP559016994 * (T2m - T2n);
488 T2o = T2m + T2n;
489 T2p = FNMS(KP250000000, T2o, T2l);
490 T2s = Tl - Tq;
491 T2t = Ta - Tf;
492 T2u = FNMS(KP587785252, T2t, KP951056516 * T2s);
493 T2w = FMA(KP951056516, T2t, KP587785252 * T2s);
494 io[0] = T2l + T2o;
495 T2v = T2q + T2p;
496 io[WS(os, 6)] = T2v - T2w;
497 io[WS(os, 9)] = T2w + T2v;
498 T2r = T2p - T2q;
499 io[WS(os, 3)] = T2r - T2u;
500 io[WS(os, 12)] = T2u + T2r;
501 }
502 {
503 E T1M, TU, T1L, T1U, T1W, T1Q, T1T, T1V, T1N;
504 T1M = KP559016994 * (TI - TT);
505 TU = TI + TT;
506 T1L = FNMS(KP250000000, TU, Tx);
507 T1Q = T1O - T1P;
508 T1T = T1R - T1S;
509 T1U = FNMS(KP587785252, T1T, KP951056516 * T1Q);
510 T1W = FMA(KP951056516, T1T, KP587785252 * T1Q);
511 ro[WS(os, 5)] = Tx + TU;
512 T1V = T1M + T1L;
513 ro[WS(os, 14)] = T1V - T1W;
514 ro[WS(os, 11)] = T1V + T1W;
515 T1N = T1L - T1M;
516 ro[WS(os, 2)] = T1N - T1U;
517 ro[WS(os, 8)] = T1N + T1U;
518 }
519 {
520 E T25, T23, T24, T1Z, T28, T1X, T1Y, T27, T26;
521 T25 = KP559016994 * (T21 - T22);
522 T23 = T21 + T22;
523 T24 = FNMS(KP250000000, T23, T20);
524 T1X = TN - TS;
525 T1Y = TC - TH;
526 T1Z = FNMS(KP587785252, T1Y, KP951056516 * T1X);
527 T28 = FMA(KP951056516, T1Y, KP587785252 * T1X);
528 io[WS(os, 5)] = T20 + T23;
529 T27 = T25 + T24;
530 io[WS(os, 11)] = T27 - T28;
531 io[WS(os, 14)] = T28 + T27;
532 T26 = T24 - T25;
533 io[WS(os, 2)] = T1Z + T26;
534 io[WS(os, 8)] = T26 - T1Z;
535 }
536 {
537 E T1x, T1D, T1E, T1I, T1J, T1G, T1H, T1K, T1F;
538 T1x = KP559016994 * (T1v - T1w);
539 T1D = T1v + T1w;
540 T1E = FNMS(KP250000000, T1D, T1C);
541 T1G = TW - TX;
542 T1H = TZ - T10;
543 T1I = FMA(KP951056516, T1G, KP587785252 * T1H);
544 T1J = FNMS(KP587785252, T1G, KP951056516 * T1H);
545 io[WS(os, 10)] = T1C + T1D;
546 T1K = T1E - T1x;
547 io[WS(os, 7)] = T1J + T1K;
548 io[WS(os, 13)] = T1K - T1J;
549 T1F = T1x + T1E;
550 io[WS(os, 1)] = T1F - T1I;
551 io[WS(os, 4)] = T1I + T1F;
552 }
553 {
554 E T13, T12, T14, T1s, T1u, T1g, T1r, T1t, T15;
555 T13 = KP559016994 * (TY - T11);
556 T12 = TY + T11;
557 T14 = FNMS(KP250000000, T12, TV);
558 T1g = T1a - T1f;
559 T1r = T1l - T1q;
560 T1s = FMA(KP951056516, T1g, KP587785252 * T1r);
561 T1u = FNMS(KP587785252, T1g, KP951056516 * T1r);
562 ro[WS(os, 10)] = TV + T12;
563 T1t = T14 - T13;
564 ro[WS(os, 7)] = T1t - T1u;
565 ro[WS(os, 13)] = T1t + T1u;
566 T15 = T13 + T14;
567 ro[WS(os, 4)] = T15 - T1s;
568 ro[WS(os, 1)] = T15 + T1s;
569 }
570 }
571 }
572 }
573
574 static const kdft_desc desc = { 15, "n1_15", {128, 28, 28, 0}, &GENUS, 0, 0, 0, 0 };
575
576 void X(codelet_n1_15) (planner *p) {
577 X(kdft_register) (p, n1_15, &desc);
578 }
579
580 #endif /* HAVE_FMA */