Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/rdft/scalar/r2cb/r2cb_12.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 07:41:07 EST 2012 */
23
24 #include "codelet-rdft.h"
25
26 #ifdef HAVE_FMA
27
28 /* Generated by: ../../../genfft/gen_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 12 -name r2cb_12 -include r2cb.h */
29
30 /*
31 * This function contains 38 FP additions, 16 FP multiplications,
32 * (or, 22 additions, 0 multiplications, 16 fused multiply/add),
33 * 31 stack variables, 2 constants, and 24 memory accesses
34 */
35 #include "r2cb.h"
36
37 static void r2cb_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
40 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
41 {
42 INT i;
43 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) {
44 E Ts, Tr;
45 {
46 E Tz, Te, Tn, Tk, Tc, Tw, Ty, Th, T4, T3, Td, T5;
47 {
48 E T8, Tu, Tl, Tm, Tb, T9, Ta, T1, T2, Tv;
49 T8 = Cr[WS(csr, 3)];
50 T9 = Cr[WS(csr, 5)];
51 Ta = Cr[WS(csr, 1)];
52 Tu = Ci[WS(csi, 3)];
53 Tl = Ci[WS(csi, 5)];
54 Tm = Ci[WS(csi, 1)];
55 Tb = T9 + Ta;
56 Tz = T9 - Ta;
57 Te = Ci[WS(csi, 4)];
58 Tn = Tl - Tm;
59 Tv = Tl + Tm;
60 Tk = FNMS(KP2_000000000, T8, Tb);
61 Tc = T8 + Tb;
62 T1 = Cr[0];
63 T2 = Cr[WS(csr, 4)];
64 Tw = Tu - Tv;
65 Ty = FMA(KP2_000000000, Tu, Tv);
66 Th = Ci[WS(csi, 2)];
67 T4 = Cr[WS(csr, 6)];
68 T3 = FMA(KP2_000000000, T2, T1);
69 Td = T1 - T2;
70 T5 = Cr[WS(csr, 2)];
71 }
72 {
73 E To, Tp, Tf, Tg, T6, TA, TC;
74 To = FMA(KP1_732050807, Tn, Tk);
75 Ts = FNMS(KP1_732050807, Tn, Tk);
76 Tp = FNMS(KP1_732050807, Te, Td);
77 Tf = FMA(KP1_732050807, Te, Td);
78 Tg = T4 - T5;
79 T6 = FMA(KP2_000000000, T5, T4);
80 TA = FMA(KP1_732050807, Tz, Ty);
81 TC = FNMS(KP1_732050807, Tz, Ty);
82 {
83 E Tt, T7, Ti, Tq, Tj, TB, Tx;
84 Tt = T3 - T6;
85 T7 = T3 + T6;
86 Ti = FNMS(KP1_732050807, Th, Tg);
87 Tq = FMA(KP1_732050807, Th, Tg);
88 R0[0] = FMA(KP2_000000000, Tc, T7);
89 R0[WS(rs, 3)] = FNMS(KP2_000000000, Tc, T7);
90 Tj = Tf + Ti;
91 TB = Tf - Ti;
92 Tr = Tp + Tq;
93 Tx = Tp - Tq;
94 R1[WS(rs, 5)] = TB + TC;
95 R1[WS(rs, 2)] = TB - TC;
96 R0[WS(rs, 4)] = Tj - To;
97 R0[WS(rs, 1)] = Tj + To;
98 R1[WS(rs, 3)] = Tx + TA;
99 R1[0] = Tx - TA;
100 R1[WS(rs, 4)] = FNMS(KP2_000000000, Tw, Tt);
101 R1[WS(rs, 1)] = FMA(KP2_000000000, Tw, Tt);
102 }
103 }
104 }
105 R0[WS(rs, 2)] = Tr - Ts;
106 R0[WS(rs, 5)] = Tr + Ts;
107 }
108 }
109 }
110
111 static const kr2c_desc desc = { 12, "r2cb_12", {22, 0, 16, 0}, &GENUS };
112
113 void X(codelet_r2cb_12) (planner *p) {
114 X(kr2c_register) (p, r2cb_12, &desc);
115 }
116
117 #else /* HAVE_FMA */
118
119 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 12 -name r2cb_12 -include r2cb.h */
120
121 /*
122 * This function contains 38 FP additions, 10 FP multiplications,
123 * (or, 34 additions, 6 multiplications, 4 fused multiply/add),
124 * 25 stack variables, 2 constants, and 24 memory accesses
125 */
126 #include "r2cb.h"
127
128 static void r2cb_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
129 {
130 DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
131 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
132 {
133 INT i;
134 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) {
135 E T8, Tb, Tm, TA, Tw, Tx, Tp, TB, T3, Tr, Tg, T6, Ts, Tk;
136 {
137 E T9, Ta, Tn, To;
138 T8 = Cr[WS(csr, 3)];
139 T9 = Cr[WS(csr, 5)];
140 Ta = Cr[WS(csr, 1)];
141 Tb = T9 + Ta;
142 Tm = FMS(KP2_000000000, T8, Tb);
143 TA = KP1_732050807 * (T9 - Ta);
144 Tw = Ci[WS(csi, 3)];
145 Tn = Ci[WS(csi, 5)];
146 To = Ci[WS(csi, 1)];
147 Tx = Tn + To;
148 Tp = KP1_732050807 * (Tn - To);
149 TB = FMA(KP2_000000000, Tw, Tx);
150 }
151 {
152 E Tf, T1, T2, Td, Te;
153 Te = Ci[WS(csi, 4)];
154 Tf = KP1_732050807 * Te;
155 T1 = Cr[0];
156 T2 = Cr[WS(csr, 4)];
157 Td = T1 - T2;
158 T3 = FMA(KP2_000000000, T2, T1);
159 Tr = Td - Tf;
160 Tg = Td + Tf;
161 }
162 {
163 E Tj, T4, T5, Th, Ti;
164 Ti = Ci[WS(csi, 2)];
165 Tj = KP1_732050807 * Ti;
166 T4 = Cr[WS(csr, 6)];
167 T5 = Cr[WS(csr, 2)];
168 Th = T4 - T5;
169 T6 = FMA(KP2_000000000, T5, T4);
170 Ts = Th + Tj;
171 Tk = Th - Tj;
172 }
173 {
174 E T7, Tc, Tz, TC;
175 T7 = T3 + T6;
176 Tc = KP2_000000000 * (T8 + Tb);
177 R0[WS(rs, 3)] = T7 - Tc;
178 R0[0] = T7 + Tc;
179 {
180 E Tl, Tq, TD, TE;
181 Tl = Tg + Tk;
182 Tq = Tm - Tp;
183 R0[WS(rs, 1)] = Tl - Tq;
184 R0[WS(rs, 4)] = Tl + Tq;
185 TD = Tg - Tk;
186 TE = TB - TA;
187 R1[WS(rs, 2)] = TD - TE;
188 R1[WS(rs, 5)] = TD + TE;
189 }
190 Tz = Tr - Ts;
191 TC = TA + TB;
192 R1[0] = Tz - TC;
193 R1[WS(rs, 3)] = Tz + TC;
194 {
195 E Tv, Ty, Tt, Tu;
196 Tv = T3 - T6;
197 Ty = KP2_000000000 * (Tw - Tx);
198 R1[WS(rs, 4)] = Tv - Ty;
199 R1[WS(rs, 1)] = Tv + Ty;
200 Tt = Tr + Ts;
201 Tu = Tm + Tp;
202 R0[WS(rs, 5)] = Tt - Tu;
203 R0[WS(rs, 2)] = Tt + Tu;
204 }
205 }
206 }
207 }
208 }
209
210 static const kr2c_desc desc = { 12, "r2cb_12", {34, 6, 4, 0}, &GENUS };
211
212 void X(codelet_r2cb_12) (planner *p) {
213 X(kr2c_register) (p, r2cb_12, &desc);
214 }
215
216 #endif /* HAVE_FMA */