Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/rdft/scalar/r2cb/r2cb_32.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 08:07:29 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_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 32 -name r2cb_32 -include rdft/scalar/r2cb.h */
29
30 /*
31 * This function contains 156 FP additions, 84 FP multiplications,
32 * (or, 72 additions, 0 multiplications, 84 fused multiply/add),
33 * 54 stack variables, 9 constants, and 64 memory accesses
34 */
35 #include "rdft/scalar/r2cb.h"
36
37 static void r2cb_32(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP1_662939224, +1.662939224605090474157576755235811513477121624);
40 DK(KP668178637, +0.668178637919298919997757686523080761552472251);
41 DK(KP1_961570560, +1.961570560806460898252364472268478073947867462);
42 DK(KP198912367, +0.198912367379658006911597622644676228597850501);
43 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
44 DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
45 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
46 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
47 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
48 {
49 INT i;
50 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(128, rs), MAKE_VOLATILE_STRIDE(128, csr), MAKE_VOLATILE_STRIDE(128, csi)) {
51 E T5, T1R, Tz, T1t, T8, T1S, TE, T1u, Tg, T1X, T2m, TK, TP, T1x, T1U;
52 E T1w, To, T28, T2p, TW, T1d, T1D, T20, T1A, Tv, T23, T2q, T25, T1g, T1B;
53 E T17, T1E;
54 {
55 E T4, Ty, T3, Tx, T1, T2;
56 T4 = Cr[WS(csr, 8)];
57 Ty = Ci[WS(csi, 8)];
58 T1 = Cr[0];
59 T2 = Cr[WS(csr, 16)];
60 T3 = T1 + T2;
61 Tx = T1 - T2;
62 T5 = FMA(KP2_000000000, T4, T3);
63 T1R = FNMS(KP2_000000000, T4, T3);
64 Tz = FNMS(KP2_000000000, Ty, Tx);
65 T1t = FMA(KP2_000000000, Ty, Tx);
66 }
67 {
68 E T6, T7, TA, TB, TC, TD;
69 T6 = Cr[WS(csr, 4)];
70 T7 = Cr[WS(csr, 12)];
71 TA = T6 - T7;
72 TB = Ci[WS(csi, 4)];
73 TC = Ci[WS(csi, 12)];
74 TD = TB + TC;
75 T8 = T6 + T7;
76 T1S = TB - TC;
77 TE = TA - TD;
78 T1u = TA + TD;
79 }
80 {
81 E Tc, TG, TO, T1V, Tf, TL, TJ, T1W;
82 {
83 E Ta, Tb, TM, TN;
84 Ta = Cr[WS(csr, 2)];
85 Tb = Cr[WS(csr, 14)];
86 Tc = Ta + Tb;
87 TG = Ta - Tb;
88 TM = Ci[WS(csi, 2)];
89 TN = Ci[WS(csi, 14)];
90 TO = TM + TN;
91 T1V = TM - TN;
92 }
93 {
94 E Td, Te, TH, TI;
95 Td = Cr[WS(csr, 10)];
96 Te = Cr[WS(csr, 6)];
97 Tf = Td + Te;
98 TL = Td - Te;
99 TH = Ci[WS(csi, 10)];
100 TI = Ci[WS(csi, 6)];
101 TJ = TH + TI;
102 T1W = TH - TI;
103 }
104 Tg = Tc + Tf;
105 T1X = T1V - T1W;
106 T2m = T1W + T1V;
107 TK = TG - TJ;
108 TP = TL + TO;
109 T1x = TG + TJ;
110 T1U = Tc - Tf;
111 T1w = TO - TL;
112 }
113 {
114 E Tk, TS, T1c, T26, Tn, T19, TV, T27;
115 {
116 E Ti, Tj, T1a, T1b;
117 Ti = Cr[WS(csr, 1)];
118 Tj = Cr[WS(csr, 15)];
119 Tk = Ti + Tj;
120 TS = Ti - Tj;
121 T1a = Ci[WS(csi, 1)];
122 T1b = Ci[WS(csi, 15)];
123 T1c = T1a + T1b;
124 T26 = T1a - T1b;
125 }
126 {
127 E Tl, Tm, TT, TU;
128 Tl = Cr[WS(csr, 9)];
129 Tm = Cr[WS(csr, 7)];
130 Tn = Tl + Tm;
131 T19 = Tl - Tm;
132 TT = Ci[WS(csi, 9)];
133 TU = Ci[WS(csi, 7)];
134 TV = TT + TU;
135 T27 = TT - TU;
136 }
137 To = Tk + Tn;
138 T28 = T26 - T27;
139 T2p = T27 + T26;
140 TW = TS - TV;
141 T1d = T19 + T1c;
142 T1D = T1c - T19;
143 T20 = Tk - Tn;
144 T1A = TS + TV;
145 }
146 {
147 E Tr, TX, T10, T22, Tu, T12, T15, T21;
148 {
149 E Tp, Tq, TY, TZ;
150 Tp = Cr[WS(csr, 5)];
151 Tq = Cr[WS(csr, 11)];
152 Tr = Tp + Tq;
153 TX = Tp - Tq;
154 TY = Ci[WS(csi, 5)];
155 TZ = Ci[WS(csi, 11)];
156 T10 = TY + TZ;
157 T22 = TY - TZ;
158 }
159 {
160 E Ts, Tt, T13, T14;
161 Ts = Cr[WS(csr, 3)];
162 Tt = Cr[WS(csr, 13)];
163 Tu = Ts + Tt;
164 T12 = Ts - Tt;
165 T13 = Ci[WS(csi, 3)];
166 T14 = Ci[WS(csi, 13)];
167 T15 = T13 + T14;
168 T21 = T14 - T13;
169 }
170 Tv = Tr + Tu;
171 T23 = T21 - T22;
172 T2q = T22 + T21;
173 T25 = Tr - Tu;
174 {
175 E T1e, T1f, T11, T16;
176 T1e = TX + T10;
177 T1f = T12 + T15;
178 T1g = T1e - T1f;
179 T1B = T1e + T1f;
180 T11 = TX - T10;
181 T16 = T12 - T15;
182 T17 = T11 + T16;
183 T1E = T16 - T11;
184 }
185 }
186 {
187 E Tw, T2w, Th, T2v, T9;
188 Tw = To + Tv;
189 T2w = T2q + T2p;
190 T9 = FMA(KP2_000000000, T8, T5);
191 Th = FMA(KP2_000000000, Tg, T9);
192 T2v = FNMS(KP2_000000000, Tg, T9);
193 R0[WS(rs, 8)] = FNMS(KP2_000000000, Tw, Th);
194 R0[WS(rs, 12)] = FMA(KP2_000000000, T2w, T2v);
195 R0[0] = FMA(KP2_000000000, Tw, Th);
196 R0[WS(rs, 4)] = FNMS(KP2_000000000, T2w, T2v);
197 }
198 {
199 E T2n, T2t, T2s, T2u, T2l, T2o, T2r;
200 T2l = FNMS(KP2_000000000, T8, T5);
201 T2n = FNMS(KP2_000000000, T2m, T2l);
202 T2t = FMA(KP2_000000000, T2m, T2l);
203 T2o = To - Tv;
204 T2r = T2p - T2q;
205 T2s = T2o - T2r;
206 T2u = T2o + T2r;
207 R0[WS(rs, 10)] = FNMS(KP1_414213562, T2s, T2n);
208 R0[WS(rs, 14)] = FMA(KP1_414213562, T2u, T2t);
209 R0[WS(rs, 2)] = FMA(KP1_414213562, T2s, T2n);
210 R0[WS(rs, 6)] = FNMS(KP1_414213562, T2u, T2t);
211 }
212 {
213 E TR, T1j, T1i, T1k;
214 {
215 E TF, TQ, T18, T1h;
216 TF = FMA(KP1_414213562, TE, Tz);
217 TQ = FNMS(KP414213562, TP, TK);
218 TR = FMA(KP1_847759065, TQ, TF);
219 T1j = FNMS(KP1_847759065, TQ, TF);
220 T18 = FMA(KP707106781, T17, TW);
221 T1h = FMA(KP707106781, T1g, T1d);
222 T1i = FNMS(KP198912367, T1h, T18);
223 T1k = FMA(KP198912367, T18, T1h);
224 }
225 R1[WS(rs, 8)] = FNMS(KP1_961570560, T1i, TR);
226 R1[WS(rs, 12)] = FMA(KP1_961570560, T1k, T1j);
227 R1[0] = FMA(KP1_961570560, T1i, TR);
228 R1[WS(rs, 4)] = FNMS(KP1_961570560, T1k, T1j);
229 }
230 {
231 E T2f, T2j, T2i, T2k;
232 {
233 E T2d, T2e, T2g, T2h;
234 T2d = FMA(KP2_000000000, T1S, T1R);
235 T2e = T1U + T1X;
236 T2f = FNMS(KP1_414213562, T2e, T2d);
237 T2j = FMA(KP1_414213562, T2e, T2d);
238 T2g = T28 - T25;
239 T2h = T20 - T23;
240 T2i = FNMS(KP414213562, T2h, T2g);
241 T2k = FMA(KP414213562, T2g, T2h);
242 }
243 R0[WS(rs, 3)] = FNMS(KP1_847759065, T2i, T2f);
244 R0[WS(rs, 15)] = FMA(KP1_847759065, T2k, T2j);
245 R0[WS(rs, 11)] = FMA(KP1_847759065, T2i, T2f);
246 R0[WS(rs, 7)] = FNMS(KP1_847759065, T2k, T2j);
247 }
248 {
249 E T1n, T1r, T1q, T1s;
250 {
251 E T1l, T1m, T1o, T1p;
252 T1l = FNMS(KP1_414213562, TE, Tz);
253 T1m = FMA(KP414213562, TK, TP);
254 T1n = FNMS(KP1_847759065, T1m, T1l);
255 T1r = FMA(KP1_847759065, T1m, T1l);
256 T1o = FNMS(KP707106781, T1g, T1d);
257 T1p = FNMS(KP707106781, T17, TW);
258 T1q = FNMS(KP668178637, T1p, T1o);
259 T1s = FMA(KP668178637, T1o, T1p);
260 }
261 R1[WS(rs, 2)] = FNMS(KP1_662939224, T1q, T1n);
262 R1[WS(rs, 14)] = FMA(KP1_662939224, T1s, T1r);
263 R1[WS(rs, 10)] = FMA(KP1_662939224, T1q, T1n);
264 R1[WS(rs, 6)] = FNMS(KP1_662939224, T1s, T1r);
265 }
266 {
267 E T1L, T1P, T1O, T1Q;
268 {
269 E T1J, T1K, T1M, T1N;
270 T1J = FMA(KP1_414213562, T1u, T1t);
271 T1K = FMA(KP414213562, T1w, T1x);
272 T1L = FNMS(KP1_847759065, T1K, T1J);
273 T1P = FMA(KP1_847759065, T1K, T1J);
274 T1M = FMA(KP707106781, T1E, T1D);
275 T1N = FMA(KP707106781, T1B, T1A);
276 T1O = FNMS(KP198912367, T1N, T1M);
277 T1Q = FMA(KP198912367, T1M, T1N);
278 }
279 R1[WS(rs, 3)] = FNMS(KP1_961570560, T1O, T1L);
280 R1[WS(rs, 15)] = FMA(KP1_961570560, T1Q, T1P);
281 R1[WS(rs, 11)] = FMA(KP1_961570560, T1O, T1L);
282 R1[WS(rs, 7)] = FNMS(KP1_961570560, T1Q, T1P);
283 }
284 {
285 E T1Z, T2b, T2a, T2c;
286 {
287 E T1T, T1Y, T24, T29;
288 T1T = FNMS(KP2_000000000, T1S, T1R);
289 T1Y = T1U - T1X;
290 T1Z = FMA(KP1_414213562, T1Y, T1T);
291 T2b = FNMS(KP1_414213562, T1Y, T1T);
292 T24 = T20 + T23;
293 T29 = T25 + T28;
294 T2a = FNMS(KP414213562, T29, T24);
295 T2c = FMA(KP414213562, T24, T29);
296 }
297 R0[WS(rs, 9)] = FNMS(KP1_847759065, T2a, T1Z);
298 R0[WS(rs, 13)] = FMA(KP1_847759065, T2c, T2b);
299 R0[WS(rs, 1)] = FMA(KP1_847759065, T2a, T1Z);
300 R0[WS(rs, 5)] = FNMS(KP1_847759065, T2c, T2b);
301 }
302 {
303 E T1z, T1H, T1G, T1I;
304 {
305 E T1v, T1y, T1C, T1F;
306 T1v = FNMS(KP1_414213562, T1u, T1t);
307 T1y = FNMS(KP414213562, T1x, T1w);
308 T1z = FNMS(KP1_847759065, T1y, T1v);
309 T1H = FMA(KP1_847759065, T1y, T1v);
310 T1C = FNMS(KP707106781, T1B, T1A);
311 T1F = FNMS(KP707106781, T1E, T1D);
312 T1G = FNMS(KP668178637, T1F, T1C);
313 T1I = FMA(KP668178637, T1C, T1F);
314 }
315 R1[WS(rs, 9)] = FNMS(KP1_662939224, T1G, T1z);
316 R1[WS(rs, 13)] = FMA(KP1_662939224, T1I, T1H);
317 R1[WS(rs, 1)] = FMA(KP1_662939224, T1G, T1z);
318 R1[WS(rs, 5)] = FNMS(KP1_662939224, T1I, T1H);
319 }
320 }
321 }
322 }
323
324 static const kr2c_desc desc = { 32, "r2cb_32", {72, 0, 84, 0}, &GENUS };
325
326 void X(codelet_r2cb_32) (planner *p) {
327 X(kr2c_register) (p, r2cb_32, &desc);
328 }
329
330 #else
331
332 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 32 -name r2cb_32 -include rdft/scalar/r2cb.h */
333
334 /*
335 * This function contains 156 FP additions, 50 FP multiplications,
336 * (or, 140 additions, 34 multiplications, 16 fused multiply/add),
337 * 54 stack variables, 9 constants, and 64 memory accesses
338 */
339 #include "rdft/scalar/r2cb.h"
340
341 static void r2cb_32(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
342 {
343 DK(KP1_662939224, +1.662939224605090474157576755235811513477121624);
344 DK(KP1_111140466, +1.111140466039204449485661627897065748749874382);
345 DK(KP1_961570560, +1.961570560806460898252364472268478073947867462);
346 DK(KP390180644, +0.390180644032256535696569736954044481855383236);
347 DK(KP765366864, +0.765366864730179543456919968060797733522689125);
348 DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
349 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
350 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
351 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
352 {
353 INT i;
354 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(128, rs), MAKE_VOLATILE_STRIDE(128, csr), MAKE_VOLATILE_STRIDE(128, csi)) {
355 E T9, T2c, TB, T1y, T6, T2b, Ty, T1v, Th, T2e, T2f, TD, TK, T1C, T1F;
356 E T1h, Tp, T2i, T2m, TN, T13, T1K, T1Y, T1k, Tw, TU, T1l, TW, T1V, T2j;
357 E T1R, T2l;
358 {
359 E T7, T8, T1w, Tz, TA, T1x;
360 T7 = Cr[WS(csr, 4)];
361 T8 = Cr[WS(csr, 12)];
362 T1w = T7 - T8;
363 Tz = Ci[WS(csi, 4)];
364 TA = Ci[WS(csi, 12)];
365 T1x = Tz + TA;
366 T9 = KP2_000000000 * (T7 + T8);
367 T2c = KP1_414213562 * (T1w + T1x);
368 TB = KP2_000000000 * (Tz - TA);
369 T1y = KP1_414213562 * (T1w - T1x);
370 }
371 {
372 E T5, T1u, T3, T1s;
373 {
374 E T4, T1t, T1, T2;
375 T4 = Cr[WS(csr, 8)];
376 T5 = KP2_000000000 * T4;
377 T1t = Ci[WS(csi, 8)];
378 T1u = KP2_000000000 * T1t;
379 T1 = Cr[0];
380 T2 = Cr[WS(csr, 16)];
381 T3 = T1 + T2;
382 T1s = T1 - T2;
383 }
384 T6 = T3 + T5;
385 T2b = T1s + T1u;
386 Ty = T3 - T5;
387 T1v = T1s - T1u;
388 }
389 {
390 E Td, T1A, TG, T1E, Tg, T1D, TJ, T1B;
391 {
392 E Tb, Tc, TE, TF;
393 Tb = Cr[WS(csr, 2)];
394 Tc = Cr[WS(csr, 14)];
395 Td = Tb + Tc;
396 T1A = Tb - Tc;
397 TE = Ci[WS(csi, 2)];
398 TF = Ci[WS(csi, 14)];
399 TG = TE - TF;
400 T1E = TE + TF;
401 }
402 {
403 E Te, Tf, TH, TI;
404 Te = Cr[WS(csr, 10)];
405 Tf = Cr[WS(csr, 6)];
406 Tg = Te + Tf;
407 T1D = Te - Tf;
408 TH = Ci[WS(csi, 10)];
409 TI = Ci[WS(csi, 6)];
410 TJ = TH - TI;
411 T1B = TH + TI;
412 }
413 Th = KP2_000000000 * (Td + Tg);
414 T2e = T1A + T1B;
415 T2f = T1E - T1D;
416 TD = Td - Tg;
417 TK = TG - TJ;
418 T1C = T1A - T1B;
419 T1F = T1D + T1E;
420 T1h = KP2_000000000 * (TJ + TG);
421 }
422 {
423 E Tl, T1I, TZ, T1X, To, T1W, T12, T1J;
424 {
425 E Tj, Tk, TX, TY;
426 Tj = Cr[WS(csr, 1)];
427 Tk = Cr[WS(csr, 15)];
428 Tl = Tj + Tk;
429 T1I = Tj - Tk;
430 TX = Ci[WS(csi, 1)];
431 TY = Ci[WS(csi, 15)];
432 TZ = TX - TY;
433 T1X = TX + TY;
434 }
435 {
436 E Tm, Tn, T10, T11;
437 Tm = Cr[WS(csr, 9)];
438 Tn = Cr[WS(csr, 7)];
439 To = Tm + Tn;
440 T1W = Tm - Tn;
441 T10 = Ci[WS(csi, 9)];
442 T11 = Ci[WS(csi, 7)];
443 T12 = T10 - T11;
444 T1J = T10 + T11;
445 }
446 Tp = Tl + To;
447 T2i = T1I + T1J;
448 T2m = T1X - T1W;
449 TN = Tl - To;
450 T13 = TZ - T12;
451 T1K = T1I - T1J;
452 T1Y = T1W + T1X;
453 T1k = T12 + TZ;
454 }
455 {
456 E Ts, T1L, TT, T1M, Tv, T1O, TQ, T1P;
457 {
458 E Tq, Tr, TR, TS;
459 Tq = Cr[WS(csr, 5)];
460 Tr = Cr[WS(csr, 11)];
461 Ts = Tq + Tr;
462 T1L = Tq - Tr;
463 TR = Ci[WS(csi, 5)];
464 TS = Ci[WS(csi, 11)];
465 TT = TR - TS;
466 T1M = TR + TS;
467 }
468 {
469 E Tt, Tu, TO, TP;
470 Tt = Cr[WS(csr, 3)];
471 Tu = Cr[WS(csr, 13)];
472 Tv = Tt + Tu;
473 T1O = Tt - Tu;
474 TO = Ci[WS(csi, 13)];
475 TP = Ci[WS(csi, 3)];
476 TQ = TO - TP;
477 T1P = TP + TO;
478 }
479 Tw = Ts + Tv;
480 TU = TQ - TT;
481 T1l = TT + TQ;
482 TW = Ts - Tv;
483 {
484 E T1T, T1U, T1N, T1Q;
485 T1T = T1L + T1M;
486 T1U = T1O + T1P;
487 T1V = KP707106781 * (T1T - T1U);
488 T2j = KP707106781 * (T1T + T1U);
489 T1N = T1L - T1M;
490 T1Q = T1O - T1P;
491 T1R = KP707106781 * (T1N + T1Q);
492 T2l = KP707106781 * (T1N - T1Q);
493 }
494 }
495 {
496 E Tx, T1r, Ti, T1q, Ta;
497 Tx = KP2_000000000 * (Tp + Tw);
498 T1r = KP2_000000000 * (T1l + T1k);
499 Ta = T6 + T9;
500 Ti = Ta + Th;
501 T1q = Ta - Th;
502 R0[WS(rs, 8)] = Ti - Tx;
503 R0[WS(rs, 12)] = T1q + T1r;
504 R0[0] = Ti + Tx;
505 R0[WS(rs, 4)] = T1q - T1r;
506 }
507 {
508 E T1i, T1o, T1n, T1p, T1g, T1j, T1m;
509 T1g = T6 - T9;
510 T1i = T1g - T1h;
511 T1o = T1g + T1h;
512 T1j = Tp - Tw;
513 T1m = T1k - T1l;
514 T1n = KP1_414213562 * (T1j - T1m);
515 T1p = KP1_414213562 * (T1j + T1m);
516 R0[WS(rs, 10)] = T1i - T1n;
517 R0[WS(rs, 14)] = T1o + T1p;
518 R0[WS(rs, 2)] = T1i + T1n;
519 R0[WS(rs, 6)] = T1o - T1p;
520 }
521 {
522 E TM, T16, T15, T17;
523 {
524 E TC, TL, TV, T14;
525 TC = Ty - TB;
526 TL = KP1_414213562 * (TD - TK);
527 TM = TC + TL;
528 T16 = TC - TL;
529 TV = TN + TU;
530 T14 = TW + T13;
531 T15 = FNMS(KP765366864, T14, KP1_847759065 * TV);
532 T17 = FMA(KP765366864, TV, KP1_847759065 * T14);
533 }
534 R0[WS(rs, 9)] = TM - T15;
535 R0[WS(rs, 13)] = T16 + T17;
536 R0[WS(rs, 1)] = TM + T15;
537 R0[WS(rs, 5)] = T16 - T17;
538 }
539 {
540 E T2t, T2x, T2w, T2y;
541 {
542 E T2r, T2s, T2u, T2v;
543 T2r = T2b + T2c;
544 T2s = FMA(KP1_847759065, T2e, KP765366864 * T2f);
545 T2t = T2r - T2s;
546 T2x = T2r + T2s;
547 T2u = T2i + T2j;
548 T2v = T2m - T2l;
549 T2w = FNMS(KP1_961570560, T2v, KP390180644 * T2u);
550 T2y = FMA(KP1_961570560, T2u, KP390180644 * T2v);
551 }
552 R1[WS(rs, 11)] = T2t - T2w;
553 R1[WS(rs, 15)] = T2x + T2y;
554 R1[WS(rs, 3)] = T2t + T2w;
555 R1[WS(rs, 7)] = T2x - T2y;
556 }
557 {
558 E T1a, T1e, T1d, T1f;
559 {
560 E T18, T19, T1b, T1c;
561 T18 = Ty + TB;
562 T19 = KP1_414213562 * (TD + TK);
563 T1a = T18 - T19;
564 T1e = T18 + T19;
565 T1b = TN - TU;
566 T1c = T13 - TW;
567 T1d = FNMS(KP1_847759065, T1c, KP765366864 * T1b);
568 T1f = FMA(KP1_847759065, T1b, KP765366864 * T1c);
569 }
570 R0[WS(rs, 11)] = T1a - T1d;
571 R0[WS(rs, 15)] = T1e + T1f;
572 R0[WS(rs, 3)] = T1a + T1d;
573 R0[WS(rs, 7)] = T1e - T1f;
574 }
575 {
576 E T25, T29, T28, T2a;
577 {
578 E T23, T24, T26, T27;
579 T23 = T1v - T1y;
580 T24 = FMA(KP765366864, T1C, KP1_847759065 * T1F);
581 T25 = T23 - T24;
582 T29 = T23 + T24;
583 T26 = T1K - T1R;
584 T27 = T1Y - T1V;
585 T28 = FNMS(KP1_662939224, T27, KP1_111140466 * T26);
586 T2a = FMA(KP1_662939224, T26, KP1_111140466 * T27);
587 }
588 R1[WS(rs, 10)] = T25 - T28;
589 R1[WS(rs, 14)] = T29 + T2a;
590 R1[WS(rs, 2)] = T25 + T28;
591 R1[WS(rs, 6)] = T29 - T2a;
592 }
593 {
594 E T2h, T2p, T2o, T2q;
595 {
596 E T2d, T2g, T2k, T2n;
597 T2d = T2b - T2c;
598 T2g = FNMS(KP1_847759065, T2f, KP765366864 * T2e);
599 T2h = T2d + T2g;
600 T2p = T2d - T2g;
601 T2k = T2i - T2j;
602 T2n = T2l + T2m;
603 T2o = FNMS(KP1_111140466, T2n, KP1_662939224 * T2k);
604 T2q = FMA(KP1_111140466, T2k, KP1_662939224 * T2n);
605 }
606 R1[WS(rs, 9)] = T2h - T2o;
607 R1[WS(rs, 13)] = T2p + T2q;
608 R1[WS(rs, 1)] = T2h + T2o;
609 R1[WS(rs, 5)] = T2p - T2q;
610 }
611 {
612 E T1H, T21, T20, T22;
613 {
614 E T1z, T1G, T1S, T1Z;
615 T1z = T1v + T1y;
616 T1G = FNMS(KP765366864, T1F, KP1_847759065 * T1C);
617 T1H = T1z + T1G;
618 T21 = T1z - T1G;
619 T1S = T1K + T1R;
620 T1Z = T1V + T1Y;
621 T20 = FNMS(KP390180644, T1Z, KP1_961570560 * T1S);
622 T22 = FMA(KP390180644, T1S, KP1_961570560 * T1Z);
623 }
624 R1[WS(rs, 8)] = T1H - T20;
625 R1[WS(rs, 12)] = T21 + T22;
626 R1[0] = T1H + T20;
627 R1[WS(rs, 4)] = T21 - T22;
628 }
629 }
630 }
631 }
632
633 static const kr2c_desc desc = { 32, "r2cb_32", {140, 34, 16, 0}, &GENUS };
634
635 void X(codelet_r2cb_32) (planner *p) {
636 X(kr2c_register) (p, r2cb_32, &desc);
637 }
638
639 #endif