Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/scalar/codelets/n1_6.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 07:35:42 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 6 -name n1_6 -include n.h */
29
30 /*
31 * This function contains 36 FP additions, 12 FP multiplications,
32 * (or, 24 additions, 0 multiplications, 12 fused multiply/add),
33 * 30 stack variables, 2 constants, and 24 memory accesses
34 */
35 #include "n.h"
36
37 static void n1_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
40 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
41 {
42 INT i;
43 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) {
44 E TA, Tz;
45 {
46 E Tb, T3, Tx, Tp, Tj, Te, Ts, Ta, Tu, Ti, Tk;
47 {
48 E T1, T2, Tn, To;
49 T1 = ri[0];
50 T2 = ri[WS(is, 3)];
51 Tn = ii[0];
52 To = ii[WS(is, 3)];
53 {
54 E T4, T5, T7, T8;
55 T4 = ri[WS(is, 2)];
56 Tb = T1 + T2;
57 T3 = T1 - T2;
58 Tx = Tn + To;
59 Tp = Tn - To;
60 T5 = ri[WS(is, 5)];
61 T7 = ri[WS(is, 4)];
62 T8 = ri[WS(is, 1)];
63 {
64 E Tg, Tc, T6, Td, T9, Th;
65 Tg = ii[WS(is, 2)];
66 Tc = T4 + T5;
67 T6 = T4 - T5;
68 Td = T7 + T8;
69 T9 = T7 - T8;
70 Th = ii[WS(is, 5)];
71 Tj = ii[WS(is, 4)];
72 Te = Tc + Td;
73 TA = Td - Tc;
74 Ts = T9 - T6;
75 Ta = T6 + T9;
76 Tu = Tg + Th;
77 Ti = Tg - Th;
78 Tk = ii[WS(is, 1)];
79 }
80 }
81 }
82 ro[WS(os, 3)] = T3 + Ta;
83 ro[0] = Tb + Te;
84 {
85 E Tf, Tv, Tl, Ty, Tr;
86 Tf = FNMS(KP500000000, Ta, T3);
87 Tv = Tj + Tk;
88 Tl = Tj - Tk;
89 {
90 E Tt, Tw, Tq, Tm;
91 Tt = FNMS(KP500000000, Te, Tb);
92 Ty = Tu + Tv;
93 Tw = Tu - Tv;
94 Tq = Ti + Tl;
95 Tm = Ti - Tl;
96 io[0] = Tx + Ty;
97 ro[WS(os, 1)] = FMA(KP866025403, Tm, Tf);
98 ro[WS(os, 5)] = FNMS(KP866025403, Tm, Tf);
99 Tr = FNMS(KP500000000, Tq, Tp);
100 io[WS(os, 3)] = Tp + Tq;
101 ro[WS(os, 2)] = FNMS(KP866025403, Tw, Tt);
102 ro[WS(os, 4)] = FMA(KP866025403, Tw, Tt);
103 }
104 io[WS(os, 5)] = FNMS(KP866025403, Ts, Tr);
105 io[WS(os, 1)] = FMA(KP866025403, Ts, Tr);
106 Tz = FNMS(KP500000000, Ty, Tx);
107 }
108 }
109 io[WS(os, 4)] = FMA(KP866025403, TA, Tz);
110 io[WS(os, 2)] = FNMS(KP866025403, TA, Tz);
111 }
112 }
113 }
114
115 static const kdft_desc desc = { 6, "n1_6", {24, 0, 12, 0}, &GENUS, 0, 0, 0, 0 };
116
117 void X(codelet_n1_6) (planner *p) {
118 X(kdft_register) (p, n1_6, &desc);
119 }
120
121 #else /* HAVE_FMA */
122
123 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 6 -name n1_6 -include n.h */
124
125 /*
126 * This function contains 36 FP additions, 8 FP multiplications,
127 * (or, 32 additions, 4 multiplications, 4 fused multiply/add),
128 * 23 stack variables, 2 constants, and 24 memory accesses
129 */
130 #include "n.h"
131
132 static void n1_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
133 {
134 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
135 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
136 {
137 INT i;
138 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) {
139 E T3, Tb, Tq, Tx, T6, Tc, T9, Td, Ta, Te, Ti, Tu, Tl, Tv, Tr;
140 E Ty;
141 {
142 E T1, T2, To, Tp;
143 T1 = ri[0];
144 T2 = ri[WS(is, 3)];
145 T3 = T1 - T2;
146 Tb = T1 + T2;
147 To = ii[0];
148 Tp = ii[WS(is, 3)];
149 Tq = To - Tp;
150 Tx = To + Tp;
151 }
152 {
153 E T4, T5, T7, T8;
154 T4 = ri[WS(is, 2)];
155 T5 = ri[WS(is, 5)];
156 T6 = T4 - T5;
157 Tc = T4 + T5;
158 T7 = ri[WS(is, 4)];
159 T8 = ri[WS(is, 1)];
160 T9 = T7 - T8;
161 Td = T7 + T8;
162 }
163 Ta = T6 + T9;
164 Te = Tc + Td;
165 {
166 E Tg, Th, Tj, Tk;
167 Tg = ii[WS(is, 2)];
168 Th = ii[WS(is, 5)];
169 Ti = Tg - Th;
170 Tu = Tg + Th;
171 Tj = ii[WS(is, 4)];
172 Tk = ii[WS(is, 1)];
173 Tl = Tj - Tk;
174 Tv = Tj + Tk;
175 }
176 Tr = Ti + Tl;
177 Ty = Tu + Tv;
178 ro[WS(os, 3)] = T3 + Ta;
179 io[WS(os, 3)] = Tq + Tr;
180 ro[0] = Tb + Te;
181 io[0] = Tx + Ty;
182 {
183 E Tf, Tm, Tn, Ts;
184 Tf = FNMS(KP500000000, Ta, T3);
185 Tm = KP866025403 * (Ti - Tl);
186 ro[WS(os, 5)] = Tf - Tm;
187 ro[WS(os, 1)] = Tf + Tm;
188 Tn = KP866025403 * (T9 - T6);
189 Ts = FNMS(KP500000000, Tr, Tq);
190 io[WS(os, 1)] = Tn + Ts;
191 io[WS(os, 5)] = Ts - Tn;
192 }
193 {
194 E Tt, Tw, Tz, TA;
195 Tt = FNMS(KP500000000, Te, Tb);
196 Tw = KP866025403 * (Tu - Tv);
197 ro[WS(os, 2)] = Tt - Tw;
198 ro[WS(os, 4)] = Tt + Tw;
199 Tz = FNMS(KP500000000, Ty, Tx);
200 TA = KP866025403 * (Td - Tc);
201 io[WS(os, 2)] = Tz - TA;
202 io[WS(os, 4)] = TA + Tz;
203 }
204 }
205 }
206 }
207
208 static const kdft_desc desc = { 6, "n1_6", {32, 4, 4, 0}, &GENUS, 0, 0, 0, 0 };
209
210 void X(codelet_n1_6) (planner *p) {
211 X(kdft_register) (p, n1_6, &desc);
212 }
213
214 #endif /* HAVE_FMA */