Mercurial > hg > sv-dependency-builds
comparison src/fftw-3.3.8/rdft/scalar/r2cf/hc2cfdft_10.c @ 167:bd3cc4d1df30
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 08:07:11 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_hc2cdft.native -fma -compact -variables 4 -pipeline-latency 4 -n 10 -dit -name hc2cfdft_10 -include rdft/scalar/hc2cf.h */ | |
29 | |
30 /* | |
31 * This function contains 122 FP additions, 92 FP multiplications, | |
32 * (or, 68 additions, 38 multiplications, 54 fused multiply/add), | |
33 * 81 stack variables, 5 constants, and 40 memory accesses | |
34 */ | |
35 #include "rdft/scalar/hc2cf.h" | |
36 | |
37 static void hc2cfdft_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms) | |
38 { | |
39 DK(KP951056516, +0.951056516295153572116439333379382143405698634); | |
40 DK(KP559016994, +0.559016994374947424102293417182819058860154590); | |
41 DK(KP500000000, +0.500000000000000000000000000000000000000000000); | |
42 DK(KP250000000, +0.250000000000000000000000000000000000000000000); | |
43 DK(KP618033988, +0.618033988749894848204586834365638117720309180); | |
44 { | |
45 INT m; | |
46 for (m = mb, W = W + ((mb - 1) * 18); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 18, MAKE_VOLATILE_STRIDE(40, rs)) { | |
47 E T3, T1u, Td, T1w, T1S, T2f, T14, T1p, T1j, T1q, T1N, T2e, TQ, T2i, T1n; | |
48 E T1H, Tz, T2h, T1m, T1C; | |
49 { | |
50 E T1, T2, T1h, Tc, TW, T1c, T1d, T1b, T1f, T1g, T1Q, T7, TV, T1J, TS; | |
51 E TU, Ts, Tx, T19, T18, T1O, T15, T17, Tt, T1A, Ti, Tn, TE, TD, T1F; | |
52 E TA, TC, Tj, T1y, TJ, TO, T12, T11, T1L, TY, T10, TK, T1D; | |
53 { | |
54 E Ta, Tb, T1e, T5, T6, TT; | |
55 T1 = Ip[0]; | |
56 T2 = Im[0]; | |
57 T1h = T1 + T2; | |
58 Ta = Rp[WS(rs, 2)]; | |
59 Tb = Rm[WS(rs, 2)]; | |
60 Tc = Ta - Tb; | |
61 TW = Ta + Tb; | |
62 T1c = Rm[0]; | |
63 T1d = Rp[0]; | |
64 T1e = T1c - T1d; | |
65 T1b = W[0]; | |
66 T1f = T1b * T1e; | |
67 T1g = W[1]; | |
68 T1Q = T1g * T1e; | |
69 T5 = Ip[WS(rs, 2)]; | |
70 T6 = Im[WS(rs, 2)]; | |
71 TT = T5 - T6; | |
72 T7 = T5 + T6; | |
73 TV = W[7]; | |
74 T1J = TV * TT; | |
75 TS = W[6]; | |
76 TU = TS * TT; | |
77 { | |
78 E Tq, Tr, T16, Tv, Tw, Tp; | |
79 Tq = Rm[WS(rs, 3)]; | |
80 Tr = Rp[WS(rs, 3)]; | |
81 Ts = Tq - Tr; | |
82 Tv = Ip[WS(rs, 3)]; | |
83 Tw = Im[WS(rs, 3)]; | |
84 Tx = Tv + Tw; | |
85 T16 = Tv - Tw; | |
86 T19 = Tr + Tq; | |
87 T18 = W[11]; | |
88 T1O = T18 * T16; | |
89 T15 = W[10]; | |
90 T17 = T15 * T16; | |
91 Tp = W[12]; | |
92 Tt = Tp * Ts; | |
93 T1A = Tp * Tx; | |
94 } | |
95 { | |
96 E Tg, Th, TB, Tl, Tm, Tf; | |
97 Tg = Ip[WS(rs, 1)]; | |
98 Th = Im[WS(rs, 1)]; | |
99 Ti = Tg - Th; | |
100 Tl = Rp[WS(rs, 1)]; | |
101 Tm = Rm[WS(rs, 1)]; | |
102 Tn = Tl + Tm; | |
103 TB = Tm - Tl; | |
104 TE = Tg + Th; | |
105 TD = W[5]; | |
106 T1F = TD * TB; | |
107 TA = W[4]; | |
108 TC = TA * TB; | |
109 Tf = W[2]; | |
110 Tj = Tf * Ti; | |
111 T1y = Tf * Tn; | |
112 } | |
113 { | |
114 E TH, TI, TZ, TM, TN, TG; | |
115 TH = Ip[WS(rs, 4)]; | |
116 TI = Im[WS(rs, 4)]; | |
117 TJ = TH - TI; | |
118 TM = Rp[WS(rs, 4)]; | |
119 TN = Rm[WS(rs, 4)]; | |
120 TO = TM + TN; | |
121 TZ = TN - TM; | |
122 T12 = TH + TI; | |
123 T11 = W[17]; | |
124 T1L = T11 * TZ; | |
125 TY = W[16]; | |
126 T10 = TY * TZ; | |
127 TG = W[14]; | |
128 TK = TG * TJ; | |
129 T1D = TG * TO; | |
130 } | |
131 } | |
132 { | |
133 E T1P, T1R, T1K, T1M; | |
134 T3 = T1 - T2; | |
135 T1u = T1d + T1c; | |
136 { | |
137 E T4, T8, T9, T1v; | |
138 T4 = W[9]; | |
139 T8 = T4 * T7; | |
140 T9 = W[8]; | |
141 T1v = T9 * T7; | |
142 Td = FMA(T9, Tc, T8); | |
143 T1w = FNMS(T4, Tc, T1v); | |
144 } | |
145 T1P = FMA(T15, T19, T1O); | |
146 T1R = FMA(T1b, T1h, T1Q); | |
147 T1S = T1P - T1R; | |
148 T2f = T1P + T1R; | |
149 { | |
150 E TX, T13, T1a, T1i; | |
151 TX = FNMS(TV, TW, TU); | |
152 T13 = FNMS(T11, T12, T10); | |
153 T14 = TX + T13; | |
154 T1p = T13 - TX; | |
155 T1a = FNMS(T18, T19, T17); | |
156 T1i = FNMS(T1g, T1h, T1f); | |
157 T1j = T1a + T1i; | |
158 T1q = T1i - T1a; | |
159 } | |
160 T1K = FMA(TS, TW, T1J); | |
161 T1M = FMA(TY, T12, T1L); | |
162 T1N = T1K - T1M; | |
163 T2e = T1K + T1M; | |
164 { | |
165 E TF, T1G, TP, T1E, TL; | |
166 TF = FNMS(TD, TE, TC); | |
167 T1G = FMA(TA, TE, T1F); | |
168 TL = W[15]; | |
169 TP = FNMS(TL, TO, TK); | |
170 T1E = FMA(TL, TJ, T1D); | |
171 TQ = TF + TP; | |
172 T2i = T1G + T1E; | |
173 T1n = TF - TP; | |
174 T1H = T1E - T1G; | |
175 } | |
176 { | |
177 E To, T1z, Ty, T1B, Tk, Tu; | |
178 Tk = W[3]; | |
179 To = FNMS(Tk, Tn, Tj); | |
180 T1z = FMA(Tk, Ti, T1y); | |
181 Tu = W[13]; | |
182 Ty = FNMS(Tu, Tx, Tt); | |
183 T1B = FMA(Tu, Ts, T1A); | |
184 Tz = To + Ty; | |
185 T2h = T1z + T1B; | |
186 T1m = Ty - To; | |
187 T1C = T1z - T1B; | |
188 } | |
189 } | |
190 } | |
191 { | |
192 E T2k, T2m, Te, T1l, T2b, T2c, T2l, T2d; | |
193 { | |
194 E T2g, T2j, TR, T1k; | |
195 T2g = T2e - T2f; | |
196 T2j = T2h - T2i; | |
197 T2k = FNMS(KP618033988, T2j, T2g); | |
198 T2m = FMA(KP618033988, T2g, T2j); | |
199 Te = T3 - Td; | |
200 TR = Tz + TQ; | |
201 T1k = T14 + T1j; | |
202 T1l = TR + T1k; | |
203 T2b = FNMS(KP250000000, T1l, Te); | |
204 T2c = TR - T1k; | |
205 } | |
206 Ip[0] = KP500000000 * (Te + T1l); | |
207 T2l = FMA(KP559016994, T2c, T2b); | |
208 Ip[WS(rs, 4)] = KP500000000 * (FMA(KP951056516, T2m, T2l)); | |
209 Im[WS(rs, 3)] = -(KP500000000 * (FNMS(KP951056516, T2m, T2l))); | |
210 T2d = FNMS(KP559016994, T2c, T2b); | |
211 Ip[WS(rs, 2)] = KP500000000 * (FMA(KP951056516, T2k, T2d)); | |
212 Im[WS(rs, 1)] = -(KP500000000 * (FNMS(KP951056516, T2k, T2d))); | |
213 } | |
214 { | |
215 E T2w, T2y, T2n, T2q, T2r, T2s, T2x, T2t; | |
216 { | |
217 E T2u, T2v, T2o, T2p; | |
218 T2u = T14 - T1j; | |
219 T2v = Tz - TQ; | |
220 T2w = FNMS(KP618033988, T2v, T2u); | |
221 T2y = FMA(KP618033988, T2u, T2v); | |
222 T2n = T1u + T1w; | |
223 T2o = T2h + T2i; | |
224 T2p = T2e + T2f; | |
225 T2q = T2o + T2p; | |
226 T2r = FNMS(KP250000000, T2q, T2n); | |
227 T2s = T2o - T2p; | |
228 } | |
229 Rp[0] = KP500000000 * (T2n + T2q); | |
230 T2x = FMA(KP559016994, T2s, T2r); | |
231 Rp[WS(rs, 4)] = KP500000000 * (FNMS(KP951056516, T2y, T2x)); | |
232 Rm[WS(rs, 3)] = KP500000000 * (FMA(KP951056516, T2y, T2x)); | |
233 T2t = FNMS(KP559016994, T2s, T2r); | |
234 Rp[WS(rs, 2)] = KP500000000 * (FNMS(KP951056516, T2w, T2t)); | |
235 Rm[WS(rs, 1)] = KP500000000 * (FMA(KP951056516, T2w, T2t)); | |
236 } | |
237 { | |
238 E T28, T2a, T1t, T1s, T23, T24, T29, T25; | |
239 { | |
240 E T26, T27, T1o, T1r; | |
241 T26 = T1H - T1C; | |
242 T27 = T1S - T1N; | |
243 T28 = FMA(KP618033988, T27, T26); | |
244 T2a = FNMS(KP618033988, T26, T27); | |
245 T1t = Td + T3; | |
246 T1o = T1m + T1n; | |
247 T1r = T1p + T1q; | |
248 T1s = T1o + T1r; | |
249 T23 = FMA(KP250000000, T1s, T1t); | |
250 T24 = T1r - T1o; | |
251 } | |
252 Im[WS(rs, 4)] = KP500000000 * (T1s - T1t); | |
253 T29 = FNMS(KP559016994, T24, T23); | |
254 Ip[WS(rs, 3)] = KP500000000 * (FMA(KP951056516, T2a, T29)); | |
255 Im[WS(rs, 2)] = -(KP500000000 * (FNMS(KP951056516, T2a, T29))); | |
256 T25 = FMA(KP559016994, T24, T23); | |
257 Ip[WS(rs, 1)] = KP500000000 * (FMA(KP951056516, T28, T25)); | |
258 Im[0] = -(KP500000000 * (FNMS(KP951056516, T28, T25))); | |
259 } | |
260 { | |
261 E T20, T22, T1x, T1U, T1V, T1W, T21, T1X; | |
262 { | |
263 E T1Y, T1Z, T1I, T1T; | |
264 T1Y = T1n - T1m; | |
265 T1Z = T1q - T1p; | |
266 T20 = FMA(KP618033988, T1Z, T1Y); | |
267 T22 = FNMS(KP618033988, T1Y, T1Z); | |
268 T1x = T1u - T1w; | |
269 T1I = T1C + T1H; | |
270 T1T = T1N + T1S; | |
271 T1U = T1I + T1T; | |
272 T1V = FNMS(KP250000000, T1U, T1x); | |
273 T1W = T1I - T1T; | |
274 } | |
275 Rm[WS(rs, 4)] = KP500000000 * (T1x + T1U); | |
276 T21 = FNMS(KP559016994, T1W, T1V); | |
277 Rp[WS(rs, 3)] = KP500000000 * (FMA(KP951056516, T22, T21)); | |
278 Rm[WS(rs, 2)] = KP500000000 * (FNMS(KP951056516, T22, T21)); | |
279 T1X = FMA(KP559016994, T1W, T1V); | |
280 Rp[WS(rs, 1)] = KP500000000 * (FMA(KP951056516, T20, T1X)); | |
281 Rm[0] = KP500000000 * (FNMS(KP951056516, T20, T1X)); | |
282 } | |
283 } | |
284 } | |
285 } | |
286 | |
287 static const tw_instr twinstr[] = { | |
288 {TW_FULL, 1, 10}, | |
289 {TW_NEXT, 1, 0} | |
290 }; | |
291 | |
292 static const hc2c_desc desc = { 10, "hc2cfdft_10", twinstr, &GENUS, {68, 38, 54, 0} }; | |
293 | |
294 void X(codelet_hc2cfdft_10) (planner *p) { | |
295 X(khc2c_register) (p, hc2cfdft_10, &desc, HC2C_VIA_DFT); | |
296 } | |
297 #else | |
298 | |
299 /* Generated by: ../../../genfft/gen_hc2cdft.native -compact -variables 4 -pipeline-latency 4 -n 10 -dit -name hc2cfdft_10 -include rdft/scalar/hc2cf.h */ | |
300 | |
301 /* | |
302 * This function contains 122 FP additions, 68 FP multiplications, | |
303 * (or, 92 additions, 38 multiplications, 30 fused multiply/add), | |
304 * 62 stack variables, 5 constants, and 40 memory accesses | |
305 */ | |
306 #include "rdft/scalar/hc2cf.h" | |
307 | |
308 static void hc2cfdft_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms) | |
309 { | |
310 DK(KP293892626, +0.293892626146236564584352977319536384298826219); | |
311 DK(KP475528258, +0.475528258147576786058219666689691071702849317); | |
312 DK(KP125000000, +0.125000000000000000000000000000000000000000000); | |
313 DK(KP500000000, +0.500000000000000000000000000000000000000000000); | |
314 DK(KP279508497, +0.279508497187473712051146708591409529430077295); | |
315 { | |
316 INT m; | |
317 for (m = mb, W = W + ((mb - 1) * 18); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 18, MAKE_VOLATILE_STRIDE(40, rs)) { | |
318 E Tw, TL, TM, T1W, T1X, T27, T1Z, T20, T26, TX, T1a, T1b, T1d, T1e, T1f; | |
319 E T1q, T1t, T1u, T1x, T1A, T1B, T1g, T1h, T1i, Td, T25, T1k, T1F; | |
320 { | |
321 E T3, T1D, T19, T1z, T7, Tb, TR, T1v, Tm, T1o, TK, T1s, Tv, T1p, T12; | |
322 E T1y, TF, T1r, TW, T1w; | |
323 { | |
324 E T1, T2, T18, T14, T15, T16, T13, T17; | |
325 T1 = Ip[0]; | |
326 T2 = Im[0]; | |
327 T18 = T1 + T2; | |
328 T14 = Rm[0]; | |
329 T15 = Rp[0]; | |
330 T16 = T14 - T15; | |
331 T3 = T1 - T2; | |
332 T1D = T15 + T14; | |
333 T13 = W[0]; | |
334 T17 = W[1]; | |
335 T19 = FNMS(T17, T18, T13 * T16); | |
336 T1z = FMA(T17, T16, T13 * T18); | |
337 } | |
338 { | |
339 E T5, T6, TO, T9, Ta, TQ, TN, TP; | |
340 T5 = Ip[WS(rs, 2)]; | |
341 T6 = Im[WS(rs, 2)]; | |
342 TO = T5 - T6; | |
343 T9 = Rp[WS(rs, 2)]; | |
344 Ta = Rm[WS(rs, 2)]; | |
345 TQ = T9 + Ta; | |
346 T7 = T5 + T6; | |
347 Tb = T9 - Ta; | |
348 TN = W[6]; | |
349 TP = W[7]; | |
350 TR = FNMS(TP, TQ, TN * TO); | |
351 T1v = FMA(TP, TO, TN * TQ); | |
352 } | |
353 { | |
354 E Th, TJ, Tl, TH; | |
355 { | |
356 E Tf, Tg, Tj, Tk; | |
357 Tf = Ip[WS(rs, 1)]; | |
358 Tg = Im[WS(rs, 1)]; | |
359 Th = Tf - Tg; | |
360 TJ = Tf + Tg; | |
361 Tj = Rp[WS(rs, 1)]; | |
362 Tk = Rm[WS(rs, 1)]; | |
363 Tl = Tj + Tk; | |
364 TH = Tj - Tk; | |
365 } | |
366 { | |
367 E Te, Ti, TG, TI; | |
368 Te = W[2]; | |
369 Ti = W[3]; | |
370 Tm = FNMS(Ti, Tl, Te * Th); | |
371 T1o = FMA(Te, Tl, Ti * Th); | |
372 TG = W[4]; | |
373 TI = W[5]; | |
374 TK = FMA(TG, TH, TI * TJ); | |
375 T1s = FNMS(TI, TH, TG * TJ); | |
376 } | |
377 } | |
378 { | |
379 E Tq, TZ, Tu, T11; | |
380 { | |
381 E To, Tp, Ts, Tt; | |
382 To = Ip[WS(rs, 3)]; | |
383 Tp = Im[WS(rs, 3)]; | |
384 Tq = To + Tp; | |
385 TZ = To - Tp; | |
386 Ts = Rp[WS(rs, 3)]; | |
387 Tt = Rm[WS(rs, 3)]; | |
388 Tu = Ts - Tt; | |
389 T11 = Ts + Tt; | |
390 } | |
391 { | |
392 E Tn, Tr, TY, T10; | |
393 Tn = W[13]; | |
394 Tr = W[12]; | |
395 Tv = FMA(Tn, Tq, Tr * Tu); | |
396 T1p = FNMS(Tn, Tu, Tr * Tq); | |
397 TY = W[10]; | |
398 T10 = W[11]; | |
399 T12 = FNMS(T10, T11, TY * TZ); | |
400 T1y = FMA(T10, TZ, TY * T11); | |
401 } | |
402 } | |
403 { | |
404 E TA, TV, TE, TT; | |
405 { | |
406 E Ty, Tz, TC, TD; | |
407 Ty = Ip[WS(rs, 4)]; | |
408 Tz = Im[WS(rs, 4)]; | |
409 TA = Ty - Tz; | |
410 TV = Ty + Tz; | |
411 TC = Rp[WS(rs, 4)]; | |
412 TD = Rm[WS(rs, 4)]; | |
413 TE = TC + TD; | |
414 TT = TC - TD; | |
415 } | |
416 { | |
417 E Tx, TB, TS, TU; | |
418 Tx = W[14]; | |
419 TB = W[15]; | |
420 TF = FNMS(TB, TE, Tx * TA); | |
421 T1r = FMA(Tx, TE, TB * TA); | |
422 TS = W[16]; | |
423 TU = W[17]; | |
424 TW = FMA(TS, TT, TU * TV); | |
425 T1w = FNMS(TU, TT, TS * TV); | |
426 } | |
427 } | |
428 Tw = Tm - Tv; | |
429 TL = TF - TK; | |
430 TM = Tw + TL; | |
431 T1W = T1v + T1w; | |
432 T1X = T1y + T1z; | |
433 T27 = T1W + T1X; | |
434 T1Z = T1o + T1p; | |
435 T20 = T1s + T1r; | |
436 T26 = T1Z + T20; | |
437 TX = TR - TW; | |
438 T1a = T12 + T19; | |
439 T1b = TX + T1a; | |
440 T1d = T19 - T12; | |
441 T1e = TR + TW; | |
442 T1f = T1d - T1e; | |
443 T1q = T1o - T1p; | |
444 T1t = T1r - T1s; | |
445 T1u = T1q + T1t; | |
446 T1x = T1v - T1w; | |
447 T1A = T1y - T1z; | |
448 T1B = T1x + T1A; | |
449 T1g = Tm + Tv; | |
450 T1h = TK + TF; | |
451 T1i = T1g + T1h; | |
452 { | |
453 E Tc, T1E, T4, T8; | |
454 T4 = W[9]; | |
455 T8 = W[8]; | |
456 Tc = FMA(T4, T7, T8 * Tb); | |
457 T1E = FNMS(T4, Tb, T8 * T7); | |
458 Td = T3 - Tc; | |
459 T25 = T1D + T1E; | |
460 T1k = Tc + T3; | |
461 T1F = T1D - T1E; | |
462 } | |
463 } | |
464 { | |
465 E T1U, T1c, T1T, T22, T24, T1Y, T21, T23, T1V; | |
466 T1U = KP279508497 * (TM - T1b); | |
467 T1c = TM + T1b; | |
468 T1T = FNMS(KP125000000, T1c, KP500000000 * Td); | |
469 T1Y = T1W - T1X; | |
470 T21 = T1Z - T20; | |
471 T22 = FNMS(KP293892626, T21, KP475528258 * T1Y); | |
472 T24 = FMA(KP475528258, T21, KP293892626 * T1Y); | |
473 Ip[0] = KP500000000 * (Td + T1c); | |
474 T23 = T1U + T1T; | |
475 Ip[WS(rs, 4)] = T23 + T24; | |
476 Im[WS(rs, 3)] = T24 - T23; | |
477 T1V = T1T - T1U; | |
478 Ip[WS(rs, 2)] = T1V + T22; | |
479 Im[WS(rs, 1)] = T22 - T1V; | |
480 } | |
481 { | |
482 E T2a, T28, T29, T2e, T2g, T2c, T2d, T2f, T2b; | |
483 T2a = KP279508497 * (T26 - T27); | |
484 T28 = T26 + T27; | |
485 T29 = FNMS(KP125000000, T28, KP500000000 * T25); | |
486 T2c = TX - T1a; | |
487 T2d = Tw - TL; | |
488 T2e = FNMS(KP293892626, T2d, KP475528258 * T2c); | |
489 T2g = FMA(KP475528258, T2d, KP293892626 * T2c); | |
490 Rp[0] = KP500000000 * (T25 + T28); | |
491 T2f = T2a + T29; | |
492 Rp[WS(rs, 4)] = T2f - T2g; | |
493 Rm[WS(rs, 3)] = T2g + T2f; | |
494 T2b = T29 - T2a; | |
495 Rp[WS(rs, 2)] = T2b - T2e; | |
496 Rm[WS(rs, 1)] = T2e + T2b; | |
497 } | |
498 { | |
499 E T1M, T1j, T1L, T1Q, T1S, T1O, T1P, T1R, T1N; | |
500 T1M = KP279508497 * (T1i + T1f); | |
501 T1j = T1f - T1i; | |
502 T1L = FMA(KP500000000, T1k, KP125000000 * T1j); | |
503 T1O = T1A - T1x; | |
504 T1P = T1q - T1t; | |
505 T1Q = FNMS(KP475528258, T1P, KP293892626 * T1O); | |
506 T1S = FMA(KP293892626, T1P, KP475528258 * T1O); | |
507 Im[WS(rs, 4)] = KP500000000 * (T1j - T1k); | |
508 T1R = T1L - T1M; | |
509 Ip[WS(rs, 3)] = T1R + T1S; | |
510 Im[WS(rs, 2)] = T1S - T1R; | |
511 T1N = T1L + T1M; | |
512 Ip[WS(rs, 1)] = T1N + T1Q; | |
513 Im[0] = T1Q - T1N; | |
514 } | |
515 { | |
516 E T1C, T1G, T1H, T1n, T1J, T1l, T1m, T1K, T1I; | |
517 T1C = KP279508497 * (T1u - T1B); | |
518 T1G = T1u + T1B; | |
519 T1H = FNMS(KP125000000, T1G, KP500000000 * T1F); | |
520 T1l = T1g - T1h; | |
521 T1m = T1e + T1d; | |
522 T1n = FMA(KP475528258, T1l, KP293892626 * T1m); | |
523 T1J = FNMS(KP293892626, T1l, KP475528258 * T1m); | |
524 Rm[WS(rs, 4)] = KP500000000 * (T1F + T1G); | |
525 T1K = T1H - T1C; | |
526 Rp[WS(rs, 3)] = T1J + T1K; | |
527 Rm[WS(rs, 2)] = T1K - T1J; | |
528 T1I = T1C + T1H; | |
529 Rp[WS(rs, 1)] = T1n + T1I; | |
530 Rm[0] = T1I - T1n; | |
531 } | |
532 } | |
533 } | |
534 } | |
535 | |
536 static const tw_instr twinstr[] = { | |
537 {TW_FULL, 1, 10}, | |
538 {TW_NEXT, 1, 0} | |
539 }; | |
540 | |
541 static const hc2c_desc desc = { 10, "hc2cfdft_10", twinstr, &GENUS, {92, 38, 30, 0} }; | |
542 | |
543 void X(codelet_hc2cfdft_10) (planner *p) { | |
544 X(khc2c_register) (p, hc2cfdft_10, &desc, HC2C_VIA_DFT); | |
545 } | |
546 #endif |