Mercurial > hg > js-dsp-test

comparison fft/fftw/fftw-3.3.4/rdft/scalar/r2cb/r2cbIII_10.c @ 19:26056e866c29

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW to comparison table
author Chris Cannam
date Tue, 06 Oct 2015 13:08:39 +0100
parents
children
comparison
equal deleted inserted replaced
18:8db794ca3e0b 19:26056e866c29
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 Tue Mar 4 13:50:33 EST 2014 */
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 10 -name r2cbIII_10 -dft-III -include r2cbIII.h */
29
30 /*
31 * This function contains 32 FP additions, 28 FP multiplications,
32 * (or, 14 additions, 10 multiplications, 18 fused multiply/add),
33 * 38 stack variables, 5 constants, and 20 memory accesses
34 */
35 #include "r2cbIII.h"
36
37 static void r2cbIII_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
41 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
42 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
43 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
44 {
45 INT i;
46 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) {
47 E Tq, Ti, Tk, Tu, Tw, Tp, Tb, Tj, Tr, Tv;
48 {
49 E T1, To, Ts, Tt, T8, Ta, Te, Tl, Tm, Th, Tn, T9;
50 T1 = Cr[WS(csr, 2)];
51 To = Ci[WS(csi, 2)];
52 {
53 E T2, T3, T5, T6;
54 T2 = Cr[WS(csr, 4)];
55 T3 = Cr[0];
56 T5 = Cr[WS(csr, 3)];
57 T6 = Cr[WS(csr, 1)];
58 {
59 E Tc, T4, T7, Td, Tf, Tg;
60 Tc = Ci[WS(csi, 3)];
61 Ts = T2 - T3;
62 T4 = T2 + T3;
63 Tt = T5 - T6;
64 T7 = T5 + T6;
65 Td = Ci[WS(csi, 1)];
66 Tf = Ci[WS(csi, 4)];
67 Tg = Ci[0];
68 T8 = T4 + T7;
69 Ta = T7 - T4;
70 Te = Tc - Td;
71 Tl = Tc + Td;
72 Tm = Tf + Tg;
73 Th = Tf - Tg;
74 }
75 }
76 R0[0] = KP2_000000000 * (T1 + T8);
77 Tn = Tl - Tm;
78 Tq = Tl + Tm;
79 Ti = FMA(KP618033988, Th, Te);
80 Tk = FNMS(KP618033988, Te, Th);
81 R1[WS(rs, 2)] = KP2_000000000 * (Tn - To);
82 T9 = FMS(KP250000000, T8, T1);
83 Tu = FMA(KP618033988, Tt, Ts);
84 Tw = FNMS(KP618033988, Ts, Tt);
85 Tp = FMA(KP250000000, Tn, To);
86 Tb = FNMS(KP559016994, Ta, T9);
87 Tj = FMA(KP559016994, Ta, T9);
88 }
89 Tr = FMA(KP559016994, Tq, Tp);
90 Tv = FNMS(KP559016994, Tq, Tp);
91 R0[WS(rs, 2)] = -(KP2_000000000 * (FNMS(KP951056516, Tk, Tj)));
92 R0[WS(rs, 3)] = KP2_000000000 * (FMA(KP951056516, Tk, Tj));
93 R0[WS(rs, 4)] = -(KP2_000000000 * (FNMS(KP951056516, Ti, Tb)));
94 R0[WS(rs, 1)] = KP2_000000000 * (FMA(KP951056516, Ti, Tb));
95 R1[WS(rs, 1)] = KP2_000000000 * (FMA(KP951056516, Tw, Tv));
96 R1[WS(rs, 3)] = KP2_000000000 * (FNMS(KP951056516, Tw, Tv));
97 R1[WS(rs, 4)] = -(KP2_000000000 * (FNMS(KP951056516, Tu, Tr)));
98 R1[0] = -(KP2_000000000 * (FMA(KP951056516, Tu, Tr)));
99 }
100 }
101 }
102
103 static const kr2c_desc desc = { 10, "r2cbIII_10", {14, 10, 18, 0}, &GENUS };
104
105 void X(codelet_r2cbIII_10) (planner *p) {
106 X(kr2c_register) (p, r2cbIII_10, &desc);
107 }
108
109 #else /* HAVE_FMA */
110
111 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cbIII_10 -dft-III -include r2cbIII.h */
112
113 /*
114 * This function contains 32 FP additions, 16 FP multiplications,
115 * (or, 26 additions, 10 multiplications, 6 fused multiply/add),
116 * 22 stack variables, 5 constants, and 20 memory accesses
117 */
118 #include "r2cbIII.h"
119
120 static void r2cbIII_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
121 {
122 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
123 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
124 DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
125 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
126 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
127 {
128 INT i;
129 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) {
130 E T1, To, T8, Tq, Ta, Tp, Te, Ts, Th, Tn;
131 T1 = Cr[WS(csr, 2)];
132 To = Ci[WS(csi, 2)];
133 {
134 E T2, T3, T4, T5, T6, T7;
135 T2 = Cr[WS(csr, 4)];
136 T3 = Cr[0];
137 T4 = T2 + T3;
138 T5 = Cr[WS(csr, 3)];
139 T6 = Cr[WS(csr, 1)];
140 T7 = T5 + T6;
141 T8 = T4 + T7;
142 Tq = T5 - T6;
143 Ta = KP1_118033988 * (T7 - T4);
144 Tp = T2 - T3;
145 }
146 {
147 E Tc, Td, Tm, Tf, Tg, Tl;
148 Tc = Ci[WS(csi, 4)];
149 Td = Ci[0];
150 Tm = Tc + Td;
151 Tf = Ci[WS(csi, 1)];
152 Tg = Ci[WS(csi, 3)];
153 Tl = Tg + Tf;
154 Te = Tc - Td;
155 Ts = KP1_118033988 * (Tl + Tm);
156 Th = Tf - Tg;
157 Tn = Tl - Tm;
158 }
159 R0[0] = KP2_000000000 * (T1 + T8);
160 R1[WS(rs, 2)] = KP2_000000000 * (Tn - To);
161 {
162 E Ti, Tj, Tb, Tk, T9;
163 Ti = FNMS(KP1_902113032, Th, KP1_175570504 * Te);
164 Tj = FMA(KP1_175570504, Th, KP1_902113032 * Te);
165 T9 = FNMS(KP2_000000000, T1, KP500000000 * T8);
166 Tb = T9 - Ta;
167 Tk = T9 + Ta;
168 R0[WS(rs, 1)] = Tb + Ti;
169 R0[WS(rs, 3)] = Tk + Tj;
170 R0[WS(rs, 4)] = Ti - Tb;
171 R0[WS(rs, 2)] = Tj - Tk;
172 }
173 {
174 E Tr, Tv, Tu, Tw, Tt;
175 Tr = FMA(KP1_902113032, Tp, KP1_175570504 * Tq);
176 Tv = FNMS(KP1_175570504, Tp, KP1_902113032 * Tq);
177 Tt = FMA(KP500000000, Tn, KP2_000000000 * To);
178 Tu = Ts + Tt;
179 Tw = Tt - Ts;
180 R1[0] = -(Tr + Tu);
181 R1[WS(rs, 3)] = Tw - Tv;
182 R1[WS(rs, 4)] = Tr - Tu;
183 R1[WS(rs, 1)] = Tv + Tw;
184 }
185 }
186 }
187 }
188
189 static const kr2c_desc desc = { 10, "r2cbIII_10", {26, 10, 6, 0}, &GENUS };
190
191 void X(codelet_r2cbIII_10) (planner *p) {
192 X(kr2c_register) (p, r2cbIII_10, &desc);
193 }
194
195 #endif /* HAVE_FMA */