Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/scalar/codelets/n1_7.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 07:35:42 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 7 -name n1_7 -include n.h */
29
30 /*
31 * This function contains 60 FP additions, 42 FP multiplications,
32 * (or, 18 additions, 0 multiplications, 42 fused multiply/add),
33 * 51 stack variables, 6 constants, and 28 memory accesses
34 */
35 #include "n.h"
36
37 static void n1_7(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DK(KP974927912, +0.974927912181823607018131682993931217232785801);
40 DK(KP900968867, +0.900968867902419126236102319507445051165919162);
41 DK(KP801937735, +0.801937735804838252472204639014890102331838324);
42 DK(KP692021471, +0.692021471630095869627814897002069140197260599);
43 DK(KP356895867, +0.356895867892209443894399510021300583399127187);
44 DK(KP554958132, +0.554958132087371191422194871006410481067288862);
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(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
48 E Tz, TP, Ty, TK, TN, TE, Tw, TF;
49 {
50 E T1, TI, T4, TG, Ta, TT, Tp, TH, T7, Tk, TJ, TO, Tu, Tb, TB;
51 E Tg, Tl, Th, Ti;
52 T1 = ri[0];
53 Tz = ii[0];
54 {
55 E T5, T6, Te, Tf;
56 {
57 E T2, T3, T8, T9;
58 T2 = ri[WS(is, 1)];
59 T3 = ri[WS(is, 6)];
60 T8 = ri[WS(is, 3)];
61 T9 = ri[WS(is, 4)];
62 T5 = ri[WS(is, 2)];
63 TI = T3 - T2;
64 T4 = T2 + T3;
65 TG = T9 - T8;
66 Ta = T8 + T9;
67 T6 = ri[WS(is, 5)];
68 }
69 Te = ii[WS(is, 2)];
70 TT = FMA(KP554958132, TG, TI);
71 Tp = FNMS(KP356895867, T4, Ta);
72 TH = T6 - T5;
73 T7 = T5 + T6;
74 Tf = ii[WS(is, 5)];
75 Tk = ii[WS(is, 3)];
76 TJ = FNMS(KP554958132, TI, TH);
77 TO = FMA(KP554958132, TH, TG);
78 Tu = FNMS(KP356895867, Ta, T7);
79 Tb = FNMS(KP356895867, T7, T4);
80 TB = Te + Tf;
81 Tg = Te - Tf;
82 Tl = ii[WS(is, 4)];
83 Th = ii[WS(is, 1)];
84 Ti = ii[WS(is, 6)];
85 }
86 {
87 E Tm, TA, Tj, TD, Ts, TL, Tx, TU, To, TR, Td, TM, Tv;
88 {
89 E TC, TQ, Tn, Tc;
90 ro[0] = T1 + T4 + T7 + Ta;
91 TC = Tk + Tl;
92 Tm = Tk - Tl;
93 TA = Th + Ti;
94 Tj = Th - Ti;
95 TD = FNMS(KP356895867, TC, TB);
96 Ts = FMA(KP554958132, Tg, Tm);
97 TL = FNMS(KP356895867, TA, TC);
98 TQ = FNMS(KP356895867, TB, TA);
99 Tx = FNMS(KP554958132, Tj, Tg);
100 Tn = FMA(KP554958132, Tm, Tj);
101 io[0] = Tz + TA + TB + TC;
102 Tc = FNMS(KP692021471, Tb, Ta);
103 TU = FMA(KP801937735, TT, TH);
104 To = FMA(KP801937735, Tn, Tg);
105 TR = FNMS(KP692021471, TQ, TC);
106 Td = FNMS(KP900968867, Tc, T1);
107 }
108 {
109 E Tt, Tr, TS, Tq;
110 Tt = FNMS(KP801937735, Ts, Tj);
111 Tq = FNMS(KP692021471, Tp, T7);
112 TS = FNMS(KP900968867, TR, Tz);
113 ro[WS(os, 1)] = FMA(KP974927912, To, Td);
114 ro[WS(os, 6)] = FNMS(KP974927912, To, Td);
115 Tr = FNMS(KP900968867, Tq, T1);
116 io[WS(os, 6)] = FNMS(KP974927912, TU, TS);
117 io[WS(os, 1)] = FMA(KP974927912, TU, TS);
118 TP = FNMS(KP801937735, TO, TI);
119 ro[WS(os, 2)] = FMA(KP974927912, Tt, Tr);
120 ro[WS(os, 5)] = FNMS(KP974927912, Tt, Tr);
121 TM = FNMS(KP692021471, TL, TB);
122 }
123 Ty = FNMS(KP801937735, Tx, Tm);
124 Tv = FNMS(KP692021471, Tu, T4);
125 TK = FNMS(KP801937735, TJ, TG);
126 TN = FNMS(KP900968867, TM, Tz);
127 TE = FNMS(KP692021471, TD, TA);
128 Tw = FNMS(KP900968867, Tv, T1);
129 }
130 }
131 io[WS(os, 5)] = FNMS(KP974927912, TP, TN);
132 io[WS(os, 2)] = FMA(KP974927912, TP, TN);
133 TF = FNMS(KP900968867, TE, Tz);
134 ro[WS(os, 3)] = FMA(KP974927912, Ty, Tw);
135 ro[WS(os, 4)] = FNMS(KP974927912, Ty, Tw);
136 io[WS(os, 4)] = FNMS(KP974927912, TK, TF);
137 io[WS(os, 3)] = FMA(KP974927912, TK, TF);
138 }
139 }
140 }
141
142 static const kdft_desc desc = { 7, "n1_7", {18, 0, 42, 0}, &GENUS, 0, 0, 0, 0 };
143
144 void X(codelet_n1_7) (planner *p) {
145 X(kdft_register) (p, n1_7, &desc);
146 }
147
148 #else /* HAVE_FMA */
149
150 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 7 -name n1_7 -include n.h */
151
152 /*
153 * This function contains 60 FP additions, 36 FP multiplications,
154 * (or, 36 additions, 12 multiplications, 24 fused multiply/add),
155 * 25 stack variables, 6 constants, and 28 memory accesses
156 */
157 #include "n.h"
158
159 static void n1_7(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
160 {
161 DK(KP222520933, +0.222520933956314404288902564496794759466355569);
162 DK(KP900968867, +0.900968867902419126236102319507445051165919162);
163 DK(KP623489801, +0.623489801858733530525004884004239810632274731);
164 DK(KP433883739, +0.433883739117558120475768332848358754609990728);
165 DK(KP781831482, +0.781831482468029808708444526674057750232334519);
166 DK(KP974927912, +0.974927912181823607018131682993931217232785801);
167 {
168 INT i;
169 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
170 E T1, Tu, T4, Tq, Te, Tx, T7, Ts, Tk, Tv, Ta, Tr, Th, Tw;
171 T1 = ri[0];
172 Tu = ii[0];
173 {
174 E T2, T3, Tc, Td;
175 T2 = ri[WS(is, 1)];
176 T3 = ri[WS(is, 6)];
177 T4 = T2 + T3;
178 Tq = T3 - T2;
179 Tc = ii[WS(is, 1)];
180 Td = ii[WS(is, 6)];
181 Te = Tc - Td;
182 Tx = Tc + Td;
183 }
184 {
185 E T5, T6, Ti, Tj;
186 T5 = ri[WS(is, 2)];
187 T6 = ri[WS(is, 5)];
188 T7 = T5 + T6;
189 Ts = T6 - T5;
190 Ti = ii[WS(is, 2)];
191 Tj = ii[WS(is, 5)];
192 Tk = Ti - Tj;
193 Tv = Ti + Tj;
194 }
195 {
196 E T8, T9, Tf, Tg;
197 T8 = ri[WS(is, 3)];
198 T9 = ri[WS(is, 4)];
199 Ta = T8 + T9;
200 Tr = T9 - T8;
201 Tf = ii[WS(is, 3)];
202 Tg = ii[WS(is, 4)];
203 Th = Tf - Tg;
204 Tw = Tf + Tg;
205 }
206 ro[0] = T1 + T4 + T7 + Ta;
207 io[0] = Tu + Tx + Tv + Tw;
208 {
209 E Tl, Tb, TB, TC;
210 Tl = FNMS(KP781831482, Th, KP974927912 * Te) - (KP433883739 * Tk);
211 Tb = FMA(KP623489801, Ta, T1) + FNMA(KP900968867, T7, KP222520933 * T4);
212 ro[WS(os, 5)] = Tb - Tl;
213 ro[WS(os, 2)] = Tb + Tl;
214 TB = FNMS(KP781831482, Tr, KP974927912 * Tq) - (KP433883739 * Ts);
215 TC = FMA(KP623489801, Tw, Tu) + FNMA(KP900968867, Tv, KP222520933 * Tx);
216 io[WS(os, 2)] = TB + TC;
217 io[WS(os, 5)] = TC - TB;
218 }
219 {
220 E Tn, Tm, Tz, TA;
221 Tn = FMA(KP781831482, Te, KP974927912 * Tk) + (KP433883739 * Th);
222 Tm = FMA(KP623489801, T4, T1) + FNMA(KP900968867, Ta, KP222520933 * T7);
223 ro[WS(os, 6)] = Tm - Tn;
224 ro[WS(os, 1)] = Tm + Tn;
225 Tz = FMA(KP781831482, Tq, KP974927912 * Ts) + (KP433883739 * Tr);
226 TA = FMA(KP623489801, Tx, Tu) + FNMA(KP900968867, Tw, KP222520933 * Tv);
227 io[WS(os, 1)] = Tz + TA;
228 io[WS(os, 6)] = TA - Tz;
229 }
230 {
231 E Tp, To, Tt, Ty;
232 Tp = FMA(KP433883739, Te, KP974927912 * Th) - (KP781831482 * Tk);
233 To = FMA(KP623489801, T7, T1) + FNMA(KP222520933, Ta, KP900968867 * T4);
234 ro[WS(os, 4)] = To - Tp;
235 ro[WS(os, 3)] = To + Tp;
236 Tt = FMA(KP433883739, Tq, KP974927912 * Tr) - (KP781831482 * Ts);
237 Ty = FMA(KP623489801, Tv, Tu) + FNMA(KP222520933, Tw, KP900968867 * Tx);
238 io[WS(os, 3)] = Tt + Ty;
239 io[WS(os, 4)] = Ty - Tt;
240 }
241 }
242 }
243 }
244
245 static const kdft_desc desc = { 7, "n1_7", {36, 12, 24, 0}, &GENUS, 0, 0, 0, 0 };
246
247 void X(codelet_n1_7) (planner *p) {
248 X(kdft_register) (p, n1_7, &desc);
249 }
250
251 #endif /* HAVE_FMA */