Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/scalar/codelets/n1_10.c @ 10:37bf6b4a2645

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW3
author Chris Cannam
date Wed, 20 Mar 2013 15:35:50 +0000
parents
children
comparison
equal deleted inserted replaced
9:c0fb53affa76 10:37bf6b4a2645
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 10 -name n1_10 -include n.h */
29
30 /*
31 * This function contains 84 FP additions, 36 FP multiplications,
32 * (or, 48 additions, 0 multiplications, 36 fused multiply/add),
33 * 59 stack variables, 4 constants, and 40 memory accesses
34 */
35 #include "n.h"
36
37 static void n1_10(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(KP250000000, +0.250000000000000000000000000000000000000000000);
42 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
43 {
44 INT i;
45 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(40, is), MAKE_VOLATILE_STRIDE(40, os)) {
46 E T1g, T1a, T18, T1m, T1k, T1f, T19, T11, T1h, T1l;
47 {
48 E Tj, T3, T1b, TN, T1j, TU, T1i, TV, Tq, T10, Ti, Ts, Tw, T15, Tx;
49 E T13, TG, Ty, TB, TC;
50 {
51 E T1, T2, TL, TM;
52 T1 = ri[0];
53 T2 = ri[WS(is, 5)];
54 TL = ii[0];
55 TM = ii[WS(is, 5)];
56 {
57 E T7, Tk, T6, To, Tg, T8, Tb, Tc;
58 {
59 E T4, T5, Te, Tf;
60 T4 = ri[WS(is, 2)];
61 Tj = T1 + T2;
62 T3 = T1 - T2;
63 T1b = TL + TM;
64 TN = TL - TM;
65 T5 = ri[WS(is, 7)];
66 Te = ri[WS(is, 6)];
67 Tf = ri[WS(is, 1)];
68 T7 = ri[WS(is, 8)];
69 Tk = T4 + T5;
70 T6 = T4 - T5;
71 To = Te + Tf;
72 Tg = Te - Tf;
73 T8 = ri[WS(is, 3)];
74 Tb = ri[WS(is, 4)];
75 Tc = ri[WS(is, 9)];
76 }
77 {
78 E TE, TF, Tu, Tv;
79 {
80 E Ta, Th, Tl, T9;
81 Tu = ii[WS(is, 2)];
82 Tl = T7 + T8;
83 T9 = T7 - T8;
84 {
85 E Tn, Td, Tm, Tp;
86 Tn = Tb + Tc;
87 Td = Tb - Tc;
88 Tm = Tk + Tl;
89 T1j = Tk - Tl;
90 Ta = T6 + T9;
91 TU = T6 - T9;
92 Tp = Tn + To;
93 T1i = Tn - To;
94 Th = Td + Tg;
95 TV = Td - Tg;
96 Tq = Tm + Tp;
97 T10 = Tm - Tp;
98 Tv = ii[WS(is, 7)];
99 }
100 Ti = Ta + Th;
101 Ts = Ta - Th;
102 }
103 TE = ii[WS(is, 6)];
104 TF = ii[WS(is, 1)];
105 Tw = Tu - Tv;
106 T15 = Tu + Tv;
107 Tx = ii[WS(is, 8)];
108 T13 = TE + TF;
109 TG = TE - TF;
110 Ty = ii[WS(is, 3)];
111 TB = ii[WS(is, 4)];
112 TC = ii[WS(is, 9)];
113 }
114 }
115 }
116 {
117 E T17, TA, T14, TH, T1e, TQ, TS;
118 {
119 E TO, TP, T16, Tz;
120 ro[WS(os, 5)] = T3 + Ti;
121 T16 = Tx + Ty;
122 Tz = Tx - Ty;
123 {
124 E T12, TD, T1c, T1d;
125 T12 = TB + TC;
126 TD = TB - TC;
127 T1c = T15 + T16;
128 T17 = T15 - T16;
129 TO = Tw + Tz;
130 TA = Tw - Tz;
131 T1d = T12 + T13;
132 T14 = T12 - T13;
133 TP = TD + TG;
134 TH = TD - TG;
135 T1e = T1c + T1d;
136 T1g = T1c - T1d;
137 }
138 ro[0] = Tj + Tq;
139 TQ = TO + TP;
140 TS = TO - TP;
141 }
142 {
143 E TK, TI, TY, TW, TR, TJ, Tt, Tr, TZ, TX, TT;
144 TK = FNMS(KP618033988, TA, TH);
145 TI = FMA(KP618033988, TH, TA);
146 io[0] = T1b + T1e;
147 io[WS(os, 5)] = TN + TQ;
148 Tr = FNMS(KP250000000, Ti, T3);
149 TY = FNMS(KP618033988, TU, TV);
150 TW = FMA(KP618033988, TV, TU);
151 TR = FNMS(KP250000000, TQ, TN);
152 TJ = FNMS(KP559016994, Ts, Tr);
153 Tt = FMA(KP559016994, Ts, Tr);
154 T1a = FMA(KP618033988, T14, T17);
155 T18 = FNMS(KP618033988, T17, T14);
156 ro[WS(os, 7)] = FNMS(KP951056516, TK, TJ);
157 ro[WS(os, 3)] = FMA(KP951056516, TK, TJ);
158 ro[WS(os, 1)] = FMA(KP951056516, TI, Tt);
159 ro[WS(os, 9)] = FNMS(KP951056516, TI, Tt);
160 TX = FNMS(KP559016994, TS, TR);
161 TT = FMA(KP559016994, TS, TR);
162 TZ = FNMS(KP250000000, Tq, Tj);
163 io[WS(os, 3)] = FNMS(KP951056516, TY, TX);
164 io[WS(os, 7)] = FMA(KP951056516, TY, TX);
165 io[WS(os, 9)] = FMA(KP951056516, TW, TT);
166 io[WS(os, 1)] = FNMS(KP951056516, TW, TT);
167 T1m = FMA(KP618033988, T1i, T1j);
168 T1k = FNMS(KP618033988, T1j, T1i);
169 T1f = FNMS(KP250000000, T1e, T1b);
170 T19 = FMA(KP559016994, T10, TZ);
171 T11 = FNMS(KP559016994, T10, TZ);
172 }
173 }
174 }
175 ro[WS(os, 4)] = FNMS(KP951056516, T1a, T19);
176 ro[WS(os, 6)] = FMA(KP951056516, T1a, T19);
177 ro[WS(os, 8)] = FMA(KP951056516, T18, T11);
178 ro[WS(os, 2)] = FNMS(KP951056516, T18, T11);
179 T1h = FNMS(KP559016994, T1g, T1f);
180 T1l = FMA(KP559016994, T1g, T1f);
181 io[WS(os, 4)] = FMA(KP951056516, T1m, T1l);
182 io[WS(os, 6)] = FNMS(KP951056516, T1m, T1l);
183 io[WS(os, 8)] = FNMS(KP951056516, T1k, T1h);
184 io[WS(os, 2)] = FMA(KP951056516, T1k, T1h);
185 }
186 }
187 }
188
189 static const kdft_desc desc = { 10, "n1_10", {48, 0, 36, 0}, &GENUS, 0, 0, 0, 0 };
190
191 void X(codelet_n1_10) (planner *p) {
192 X(kdft_register) (p, n1_10, &desc);
193 }
194
195 #else /* HAVE_FMA */
196
197 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 10 -name n1_10 -include n.h */
198
199 /*
200 * This function contains 84 FP additions, 24 FP multiplications,
201 * (or, 72 additions, 12 multiplications, 12 fused multiply/add),
202 * 41 stack variables, 4 constants, and 40 memory accesses
203 */
204 #include "n.h"
205
206 static void n1_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
207 {
208 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
209 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
210 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
211 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
212 {
213 INT i;
214 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(40, is), MAKE_VOLATILE_STRIDE(40, os)) {
215 E T3, Tj, TQ, T1e, TU, TV, T1c, T1b, Tm, Tp, Tq, Ta, Th, Ti, TA;
216 E TH, T17, T14, T1f, T1g, T1h, TL, TM, TR;
217 {
218 E T1, T2, TO, TP;
219 T1 = ri[0];
220 T2 = ri[WS(is, 5)];
221 T3 = T1 - T2;
222 Tj = T1 + T2;
223 TO = ii[0];
224 TP = ii[WS(is, 5)];
225 TQ = TO - TP;
226 T1e = TO + TP;
227 }
228 {
229 E T6, Tk, Tg, To, T9, Tl, Td, Tn;
230 {
231 E T4, T5, Te, Tf;
232 T4 = ri[WS(is, 2)];
233 T5 = ri[WS(is, 7)];
234 T6 = T4 - T5;
235 Tk = T4 + T5;
236 Te = ri[WS(is, 6)];
237 Tf = ri[WS(is, 1)];
238 Tg = Te - Tf;
239 To = Te + Tf;
240 }
241 {
242 E T7, T8, Tb, Tc;
243 T7 = ri[WS(is, 8)];
244 T8 = ri[WS(is, 3)];
245 T9 = T7 - T8;
246 Tl = T7 + T8;
247 Tb = ri[WS(is, 4)];
248 Tc = ri[WS(is, 9)];
249 Td = Tb - Tc;
250 Tn = Tb + Tc;
251 }
252 TU = T6 - T9;
253 TV = Td - Tg;
254 T1c = Tk - Tl;
255 T1b = Tn - To;
256 Tm = Tk + Tl;
257 Tp = Tn + To;
258 Tq = Tm + Tp;
259 Ta = T6 + T9;
260 Th = Td + Tg;
261 Ti = Ta + Th;
262 }
263 {
264 E Tw, T15, TG, T13, Tz, T16, TD, T12;
265 {
266 E Tu, Tv, TE, TF;
267 Tu = ii[WS(is, 2)];
268 Tv = ii[WS(is, 7)];
269 Tw = Tu - Tv;
270 T15 = Tu + Tv;
271 TE = ii[WS(is, 6)];
272 TF = ii[WS(is, 1)];
273 TG = TE - TF;
274 T13 = TE + TF;
275 }
276 {
277 E Tx, Ty, TB, TC;
278 Tx = ii[WS(is, 8)];
279 Ty = ii[WS(is, 3)];
280 Tz = Tx - Ty;
281 T16 = Tx + Ty;
282 TB = ii[WS(is, 4)];
283 TC = ii[WS(is, 9)];
284 TD = TB - TC;
285 T12 = TB + TC;
286 }
287 TA = Tw - Tz;
288 TH = TD - TG;
289 T17 = T15 - T16;
290 T14 = T12 - T13;
291 T1f = T15 + T16;
292 T1g = T12 + T13;
293 T1h = T1f + T1g;
294 TL = Tw + Tz;
295 TM = TD + TG;
296 TR = TL + TM;
297 }
298 ro[WS(os, 5)] = T3 + Ti;
299 io[WS(os, 5)] = TQ + TR;
300 ro[0] = Tj + Tq;
301 io[0] = T1e + T1h;
302 {
303 E TI, TK, Tt, TJ, Tr, Ts;
304 TI = FMA(KP951056516, TA, KP587785252 * TH);
305 TK = FNMS(KP587785252, TA, KP951056516 * TH);
306 Tr = KP559016994 * (Ta - Th);
307 Ts = FNMS(KP250000000, Ti, T3);
308 Tt = Tr + Ts;
309 TJ = Ts - Tr;
310 ro[WS(os, 9)] = Tt - TI;
311 ro[WS(os, 3)] = TJ + TK;
312 ro[WS(os, 1)] = Tt + TI;
313 ro[WS(os, 7)] = TJ - TK;
314 }
315 {
316 E TW, TY, TT, TX, TN, TS;
317 TW = FMA(KP951056516, TU, KP587785252 * TV);
318 TY = FNMS(KP587785252, TU, KP951056516 * TV);
319 TN = KP559016994 * (TL - TM);
320 TS = FNMS(KP250000000, TR, TQ);
321 TT = TN + TS;
322 TX = TS - TN;
323 io[WS(os, 1)] = TT - TW;
324 io[WS(os, 7)] = TY + TX;
325 io[WS(os, 9)] = TW + TT;
326 io[WS(os, 3)] = TX - TY;
327 }
328 {
329 E T18, T1a, T11, T19, TZ, T10;
330 T18 = FNMS(KP587785252, T17, KP951056516 * T14);
331 T1a = FMA(KP951056516, T17, KP587785252 * T14);
332 TZ = FNMS(KP250000000, Tq, Tj);
333 T10 = KP559016994 * (Tm - Tp);
334 T11 = TZ - T10;
335 T19 = T10 + TZ;
336 ro[WS(os, 2)] = T11 - T18;
337 ro[WS(os, 6)] = T19 + T1a;
338 ro[WS(os, 8)] = T11 + T18;
339 ro[WS(os, 4)] = T19 - T1a;
340 }
341 {
342 E T1d, T1l, T1k, T1m, T1i, T1j;
343 T1d = FNMS(KP587785252, T1c, KP951056516 * T1b);
344 T1l = FMA(KP951056516, T1c, KP587785252 * T1b);
345 T1i = FNMS(KP250000000, T1h, T1e);
346 T1j = KP559016994 * (T1f - T1g);
347 T1k = T1i - T1j;
348 T1m = T1j + T1i;
349 io[WS(os, 2)] = T1d + T1k;
350 io[WS(os, 6)] = T1m - T1l;
351 io[WS(os, 8)] = T1k - T1d;
352 io[WS(os, 4)] = T1l + T1m;
353 }
354 }
355 }
356 }
357
358 static const kdft_desc desc = { 10, "n1_10", {72, 12, 12, 0}, &GENUS, 0, 0, 0, 0 };
359
360 void X(codelet_n1_10) (planner *p) {
361 X(kdft_register) (p, n1_10, &desc);
362 }
363
364 #endif /* HAVE_FMA */