Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/rdft/scalar/r2cb/r2cbIII_8.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 07:41:34 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 8 -name r2cbIII_8 -dft-III -include r2cbIII.h */
29
30 /*
31 * This function contains 22 FP additions, 12 FP multiplications,
32 * (or, 18 additions, 8 multiplications, 4 fused multiply/add),
33 * 23 stack variables, 4 constants, and 16 memory accesses
34 */
35 #include "r2cbIII.h"
36
37 static void r2cbIII_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
40 DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
41 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
42 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
43 {
44 INT i;
45 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) {
46 E T4, T7, T3, Tl, Tf, T5, T8, T9, T6, Tc;
47 {
48 E T1, T2, Td, Te;
49 T1 = Cr[0];
50 T2 = Cr[WS(csr, 3)];
51 Td = Ci[0];
52 Te = Ci[WS(csi, 3)];
53 T4 = Cr[WS(csr, 2)];
54 T7 = T1 - T2;
55 T3 = T1 + T2;
56 Tl = Te - Td;
57 Tf = Td + Te;
58 T5 = Cr[WS(csr, 1)];
59 T8 = Ci[WS(csi, 2)];
60 T9 = Ci[WS(csi, 1)];
61 }
62 T6 = T4 + T5;
63 Tc = T4 - T5;
64 {
65 E Ta, Tk, Tg, Th;
66 Ta = T8 + T9;
67 Tk = T8 - T9;
68 Tg = Tc + Tf;
69 Th = Tc - Tf;
70 {
71 E Tj, Tm, Tb, Ti;
72 Tj = T3 - T6;
73 R0[0] = KP2_000000000 * (T3 + T6);
74 Tm = Tk + Tl;
75 R0[WS(rs, 2)] = KP2_000000000 * (Tl - Tk);
76 Tb = T7 - Ta;
77 Ti = T7 + Ta;
78 R0[WS(rs, 3)] = KP1_414213562 * (Tm - Tj);
79 R0[WS(rs, 1)] = KP1_414213562 * (Tj + Tm);
80 R1[WS(rs, 3)] = -(KP1_847759065 * (FNMS(KP414213562, Th, Ti)));
81 R1[WS(rs, 1)] = KP1_847759065 * (FMA(KP414213562, Ti, Th));
82 R1[WS(rs, 2)] = -(KP1_847759065 * (FMA(KP414213562, Tb, Tg)));
83 R1[0] = KP1_847759065 * (FNMS(KP414213562, Tg, Tb));
84 }
85 }
86 }
87 }
88 }
89
90 static const kr2c_desc desc = { 8, "r2cbIII_8", {18, 8, 4, 0}, &GENUS };
91
92 void X(codelet_r2cbIII_8) (planner *p) {
93 X(kr2c_register) (p, r2cbIII_8, &desc);
94 }
95
96 #else /* HAVE_FMA */
97
98 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 8 -name r2cbIII_8 -dft-III -include r2cbIII.h */
99
100 /*
101 * This function contains 22 FP additions, 12 FP multiplications,
102 * (or, 18 additions, 8 multiplications, 4 fused multiply/add),
103 * 19 stack variables, 4 constants, and 16 memory accesses
104 */
105 #include "r2cbIII.h"
106
107 static void r2cbIII_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
108 {
109 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
110 DK(KP765366864, +0.765366864730179543456919968060797733522689125);
111 DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
112 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
113 {
114 INT i;
115 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) {
116 E T3, T7, Tf, Tl, T6, Tc, Ta, Tk, Tb, Tg;
117 {
118 E T1, T2, Td, Te;
119 T1 = Cr[0];
120 T2 = Cr[WS(csr, 3)];
121 T3 = T1 + T2;
122 T7 = T1 - T2;
123 Td = Ci[0];
124 Te = Ci[WS(csi, 3)];
125 Tf = Td + Te;
126 Tl = Te - Td;
127 }
128 {
129 E T4, T5, T8, T9;
130 T4 = Cr[WS(csr, 2)];
131 T5 = Cr[WS(csr, 1)];
132 T6 = T4 + T5;
133 Tc = T4 - T5;
134 T8 = Ci[WS(csi, 2)];
135 T9 = Ci[WS(csi, 1)];
136 Ta = T8 + T9;
137 Tk = T8 - T9;
138 }
139 R0[0] = KP2_000000000 * (T3 + T6);
140 R0[WS(rs, 2)] = KP2_000000000 * (Tl - Tk);
141 Tb = T7 - Ta;
142 Tg = Tc + Tf;
143 R1[0] = FNMS(KP765366864, Tg, KP1_847759065 * Tb);
144 R1[WS(rs, 2)] = -(FMA(KP765366864, Tb, KP1_847759065 * Tg));
145 {
146 E Th, Ti, Tj, Tm;
147 Th = T7 + Ta;
148 Ti = Tc - Tf;
149 R1[WS(rs, 1)] = FMA(KP765366864, Th, KP1_847759065 * Ti);
150 R1[WS(rs, 3)] = FNMS(KP1_847759065, Th, KP765366864 * Ti);
151 Tj = T3 - T6;
152 Tm = Tk + Tl;
153 R0[WS(rs, 1)] = KP1_414213562 * (Tj + Tm);
154 R0[WS(rs, 3)] = KP1_414213562 * (Tm - Tj);
155 }
156 }
157 }
158 }
159
160 static const kr2c_desc desc = { 8, "r2cbIII_8", {18, 8, 4, 0}, &GENUS };
161
162 void X(codelet_r2cbIII_8) (planner *p) {
163 X(kr2c_register) (p, r2cbIII_8, &desc);
164 }
165
166 #endif /* HAVE_FMA */