Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/rdft/scalar/r2cf/hc2cfdft_6.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 08:07:10 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 6 -dit -name hc2cfdft_6 -include rdft/scalar/hc2cf.h */
29
30 /*
31 * This function contains 58 FP additions, 44 FP multiplications,
32 * (or, 36 additions, 22 multiplications, 22 fused multiply/add),
33 * 27 stack variables, 2 constants, and 24 memory accesses
34 */
35 #include "rdft/scalar/hc2cf.h"
36
37 static void hc2cfdft_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
38 {
39 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
40 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
41 {
42 INT m;
43 for (m = mb, W = W + ((mb - 1) * 10); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 10, MAKE_VOLATILE_STRIDE(24, rs)) {
44 E T3, TQ, TJ, T12, Tu, TX, TB, T10, Td, TS, Tk, TV;
45 {
46 E T1, T2, TI, TD, TE, TF;
47 T1 = Ip[0];
48 T2 = Im[0];
49 TI = T1 + T2;
50 TD = Rm[0];
51 TE = Rp[0];
52 TF = TD - TE;
53 T3 = T1 - T2;
54 TQ = TE + TD;
55 {
56 E TC, TG, TH, T11;
57 TC = W[0];
58 TG = TC * TF;
59 TH = W[1];
60 T11 = TH * TF;
61 TJ = FNMS(TH, TI, TG);
62 T12 = FMA(TC, TI, T11);
63 }
64 }
65 {
66 E To, TA, Tt, Tx;
67 {
68 E Tm, Tn, Tr, Ts;
69 Tm = Rm[WS(rs, 2)];
70 Tn = Rp[WS(rs, 2)];
71 To = Tm - Tn;
72 TA = Tn + Tm;
73 Tr = Ip[WS(rs, 2)];
74 Ts = Im[WS(rs, 2)];
75 Tt = Tr + Ts;
76 Tx = Tr - Ts;
77 }
78 {
79 E Tp, TW, Tl, Tq;
80 Tl = W[8];
81 Tp = Tl * To;
82 TW = Tl * Tt;
83 Tq = W[9];
84 Tu = FNMS(Tq, Tt, Tp);
85 TX = FMA(Tq, To, TW);
86 }
87 {
88 E Tw, Ty, Tz, TZ;
89 Tw = W[6];
90 Ty = Tw * Tx;
91 Tz = W[7];
92 TZ = Tz * Tx;
93 TB = FNMS(Tz, TA, Ty);
94 T10 = FMA(Tw, TA, TZ);
95 }
96 }
97 {
98 E T7, Tg, Tc, Tj;
99 {
100 E T5, T6, Ta, Tb;
101 T5 = Ip[WS(rs, 1)];
102 T6 = Im[WS(rs, 1)];
103 T7 = T5 + T6;
104 Tg = T5 - T6;
105 Ta = Rp[WS(rs, 1)];
106 Tb = Rm[WS(rs, 1)];
107 Tc = Ta - Tb;
108 Tj = Ta + Tb;
109 }
110 {
111 E T4, T8, T9, TR;
112 T4 = W[5];
113 T8 = T4 * T7;
114 T9 = W[4];
115 TR = T9 * T7;
116 Td = FMA(T9, Tc, T8);
117 TS = FNMS(T4, Tc, TR);
118 }
119 {
120 E Tf, Th, Ti, TU;
121 Tf = W[2];
122 Th = Tf * Tg;
123 Ti = W[3];
124 TU = Ti * Tg;
125 Tk = FNMS(Ti, Tj, Th);
126 TV = FMA(Tf, Tj, TU);
127 }
128 }
129 {
130 E Te, T1d, TL, T1g, T1c, T1e, T19, T1f;
131 Te = T3 - Td;
132 T1d = TQ + TS;
133 {
134 E Tv, TK, T1a, T1b;
135 Tv = Tk + Tu;
136 TK = TB + TJ;
137 TL = Tv + TK;
138 T1g = Tv - TK;
139 T1a = TV + TX;
140 T1b = T10 + T12;
141 T1c = T1a - T1b;
142 T1e = T1a + T1b;
143 }
144 Ip[0] = KP500000000 * (Te + TL);
145 Rp[0] = KP500000000 * (T1d + T1e);
146 T19 = FNMS(KP500000000, TL, Te);
147 Ip[WS(rs, 2)] = KP500000000 * (FMA(KP866025403, T1c, T19));
148 Im[WS(rs, 1)] = -(KP500000000 * (FNMS(KP866025403, T1c, T19)));
149 T1f = FNMS(KP500000000, T1e, T1d);
150 Rp[WS(rs, 2)] = KP500000000 * (FNMS(KP866025403, T1g, T1f));
151 Rm[WS(rs, 1)] = KP500000000 * (FMA(KP866025403, T1g, T1f));
152 }
153 {
154 E TP, TT, TO, T16, T14, T18, T15, T17;
155 TP = Td + T3;
156 TT = TQ - TS;
157 {
158 E TM, TN, TY, T13;
159 TM = Tu - Tk;
160 TN = TJ - TB;
161 TO = TM + TN;
162 T16 = TN - TM;
163 TY = TV - TX;
164 T13 = T10 - T12;
165 T14 = TY + T13;
166 T18 = T13 - TY;
167 }
168 Im[WS(rs, 2)] = KP500000000 * (TO - TP);
169 Rm[WS(rs, 2)] = KP500000000 * (TT + T14);
170 T15 = FNMS(KP500000000, T14, TT);
171 Rp[WS(rs, 1)] = KP500000000 * (FMA(KP866025403, T16, T15));
172 Rm[0] = KP500000000 * (FNMS(KP866025403, T16, T15));
173 T17 = FMA(KP500000000, TO, TP);
174 Ip[WS(rs, 1)] = KP500000000 * (FMA(KP866025403, T18, T17));
175 Im[0] = -(KP500000000 * (FNMS(KP866025403, T18, T17)));
176 }
177 }
178 }
179 }
180
181 static const tw_instr twinstr[] = {
182 {TW_FULL, 1, 6},
183 {TW_NEXT, 1, 0}
184 };
185
186 static const hc2c_desc desc = { 6, "hc2cfdft_6", twinstr, &GENUS, {36, 22, 22, 0} };
187
188 void X(codelet_hc2cfdft_6) (planner *p) {
189 X(khc2c_register) (p, hc2cfdft_6, &desc, HC2C_VIA_DFT);
190 }
191 #else
192
193 /* Generated by: ../../../genfft/gen_hc2cdft.native -compact -variables 4 -pipeline-latency 4 -n 6 -dit -name hc2cfdft_6 -include rdft/scalar/hc2cf.h */
194
195 /*
196 * This function contains 58 FP additions, 36 FP multiplications,
197 * (or, 44 additions, 22 multiplications, 14 fused multiply/add),
198 * 40 stack variables, 3 constants, and 24 memory accesses
199 */
200 #include "rdft/scalar/hc2cf.h"
201
202 static void hc2cfdft_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
203 {
204 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
205 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
206 DK(KP433012701, +0.433012701892219323381861585376468091735701313);
207 {
208 INT m;
209 for (m = mb, W = W + ((mb - 1) * 10); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 10, MAKE_VOLATILE_STRIDE(24, rs)) {
210 E T3, TM, Tc, TN, Ts, T10, TI, TR, TF, T11, TH, TU;
211 {
212 E T1, T2, TD, Tz, TA, TB, T7, Tf, Tb, Th, Tq, Tw, Tm, Tu, T4;
213 E T8;
214 {
215 E T5, T6, T9, Ta;
216 T1 = Ip[0];
217 T2 = Im[0];
218 TD = T1 + T2;
219 Tz = Rm[0];
220 TA = Rp[0];
221 TB = Tz - TA;
222 T5 = Ip[WS(rs, 1)];
223 T6 = Im[WS(rs, 1)];
224 T7 = T5 + T6;
225 Tf = T5 - T6;
226 T9 = Rp[WS(rs, 1)];
227 Ta = Rm[WS(rs, 1)];
228 Tb = T9 - Ta;
229 Th = T9 + Ta;
230 {
231 E To, Tp, Tk, Tl;
232 To = Rp[WS(rs, 2)];
233 Tp = Rm[WS(rs, 2)];
234 Tq = To - Tp;
235 Tw = To + Tp;
236 Tk = Ip[WS(rs, 2)];
237 Tl = Im[WS(rs, 2)];
238 Tm = Tk + Tl;
239 Tu = Tk - Tl;
240 }
241 }
242 T3 = T1 - T2;
243 TM = TA + Tz;
244 T4 = W[5];
245 T8 = W[4];
246 Tc = FMA(T4, T7, T8 * Tb);
247 TN = FNMS(T4, Tb, T8 * T7);
248 {
249 E Ti, TP, Tr, TQ;
250 {
251 E Te, Tg, Tj, Tn;
252 Te = W[2];
253 Tg = W[3];
254 Ti = FNMS(Tg, Th, Te * Tf);
255 TP = FMA(Tg, Tf, Te * Th);
256 Tj = W[9];
257 Tn = W[8];
258 Tr = FMA(Tj, Tm, Tn * Tq);
259 TQ = FNMS(Tj, Tq, Tn * Tm);
260 }
261 Ts = Ti - Tr;
262 T10 = TP + TQ;
263 TI = Ti + Tr;
264 TR = TP - TQ;
265 }
266 {
267 E Tx, TS, TE, TT;
268 {
269 E Tt, Tv, Ty, TC;
270 Tt = W[6];
271 Tv = W[7];
272 Tx = FNMS(Tv, Tw, Tt * Tu);
273 TS = FMA(Tv, Tu, Tt * Tw);
274 Ty = W[0];
275 TC = W[1];
276 TE = FNMS(TC, TD, Ty * TB);
277 TT = FMA(TC, TB, Ty * TD);
278 }
279 TF = Tx + TE;
280 T11 = TS + TT;
281 TH = TE - Tx;
282 TU = TS - TT;
283 }
284 }
285 {
286 E T12, Td, TG, TZ;
287 T12 = KP433012701 * (T10 - T11);
288 Td = T3 - Tc;
289 TG = Ts + TF;
290 TZ = FNMS(KP250000000, TG, KP500000000 * Td);
291 Ip[0] = KP500000000 * (Td + TG);
292 Im[WS(rs, 1)] = T12 - TZ;
293 Ip[WS(rs, 2)] = TZ + T12;
294 }
295 {
296 E T16, T13, T14, T15;
297 T16 = KP433012701 * (Ts - TF);
298 T13 = TM + TN;
299 T14 = T10 + T11;
300 T15 = FNMS(KP250000000, T14, KP500000000 * T13);
301 Rp[WS(rs, 2)] = T15 - T16;
302 Rp[0] = KP500000000 * (T13 + T14);
303 Rm[WS(rs, 1)] = T16 + T15;
304 }
305 {
306 E TY, TJ, TK, TX;
307 TY = KP433012701 * (TU - TR);
308 TJ = TH - TI;
309 TK = Tc + T3;
310 TX = FMA(KP500000000, TK, KP250000000 * TJ);
311 Im[WS(rs, 2)] = KP500000000 * (TJ - TK);
312 Im[0] = TY - TX;
313 Ip[WS(rs, 1)] = TX + TY;
314 }
315 {
316 E TL, TO, TV, TW;
317 TL = KP433012701 * (TI + TH);
318 TO = TM - TN;
319 TV = TR + TU;
320 TW = FNMS(KP250000000, TV, KP500000000 * TO);
321 Rp[WS(rs, 1)] = TL + TW;
322 Rm[WS(rs, 2)] = KP500000000 * (TO + TV);
323 Rm[0] = TW - TL;
324 }
325 }
326 }
327 }
328
329 static const tw_instr twinstr[] = {
330 {TW_FULL, 1, 6},
331 {TW_NEXT, 1, 0}
332 };
333
334 static const hc2c_desc desc = { 6, "hc2cfdft_6", twinstr, &GENUS, {44, 22, 14, 0} };
335
336 void X(codelet_hc2cfdft_6) (planner *p) {
337 X(khc2c_register) (p, hc2cfdft_6, &desc, HC2C_VIA_DFT);
338 }
339 #endif