Mercurial > hg > js-dsp-test

comparison fft/fftw/fftw-3.3.4/rdft/scalar/r2cb/r2cb_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 13:50:24 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 r2cb_10 -include r2cb.h */
29
30 /*
31 * This function contains 34 FP additions, 20 FP multiplications,
32 * (or, 14 additions, 0 multiplications, 20 fused multiply/add),
33 * 30 stack variables, 5 constants, and 20 memory accesses
34 */
35 #include "r2cb.h"
36
37 static void r2cb_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
40 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
41 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
42 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
43 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
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 Tb, T3, Tc, T6, Tq, To, Ty, Tw, Td, T9;
48 {
49 E Tu, Tn, T7, Tv, Tk, T8;
50 {
51 E T1, T2, Tl, Tm;
52 T1 = Cr[0];
53 T2 = Cr[WS(csr, 5)];
54 Tl = Ci[WS(csi, 2)];
55 Tm = Ci[WS(csi, 3)];
56 {
57 E Ti, Tj, T4, T5;
58 Ti = Ci[WS(csi, 4)];
59 Tb = T1 + T2;
60 T3 = T1 - T2;
61 Tu = Tl + Tm;
62 Tn = Tl - Tm;
63 Tj = Ci[WS(csi, 1)];
64 T4 = Cr[WS(csr, 2)];
65 T5 = Cr[WS(csr, 3)];
66 T7 = Cr[WS(csr, 4)];
67 Tv = Ti + Tj;
68 Tk = Ti - Tj;
69 Tc = T4 + T5;
70 T6 = T4 - T5;
71 T8 = Cr[WS(csr, 1)];
72 }
73 }
74 Tq = FMA(KP618033988, Tk, Tn);
75 To = FNMS(KP618033988, Tn, Tk);
76 Ty = FNMS(KP618033988, Tu, Tv);
77 Tw = FMA(KP618033988, Tv, Tu);
78 Td = T7 + T8;
79 T9 = T7 - T8;
80 }
81 {
82 E Te, Tg, Ta, Ts, Tf, Tr;
83 Te = Tc + Td;
84 Tg = Tc - Td;
85 Ta = T6 + T9;
86 Ts = T6 - T9;
87 Tf = FNMS(KP500000000, Te, Tb);
88 R0[0] = FMA(KP2_000000000, Te, Tb);
89 Tr = FNMS(KP500000000, Ta, T3);
90 R1[WS(rs, 2)] = FMA(KP2_000000000, Ta, T3);
91 {
92 E Th, Tp, Tt, Tx;
93 Th = FNMS(KP1_118033988, Tg, Tf);
94 Tp = FMA(KP1_118033988, Tg, Tf);
95 Tt = FMA(KP1_118033988, Ts, Tr);
96 Tx = FNMS(KP1_118033988, Ts, Tr);
97 R0[WS(rs, 3)] = FNMS(KP1_902113032, Tq, Tp);
98 R0[WS(rs, 2)] = FMA(KP1_902113032, Tq, Tp);
99 R0[WS(rs, 1)] = FMA(KP1_902113032, To, Th);
100 R0[WS(rs, 4)] = FNMS(KP1_902113032, To, Th);
101 R1[WS(rs, 1)] = FNMS(KP1_902113032, Ty, Tx);
102 R1[WS(rs, 3)] = FMA(KP1_902113032, Ty, Tx);
103 R1[WS(rs, 4)] = FMA(KP1_902113032, Tw, Tt);
104 R1[0] = FNMS(KP1_902113032, Tw, Tt);
105 }
106 }
107 }
108 }
109 }
110
111 static const kr2c_desc desc = { 10, "r2cb_10", {14, 0, 20, 0}, &GENUS };
112
113 void X(codelet_r2cb_10) (planner *p) {
114 X(kr2c_register) (p, r2cb_10, &desc);
115 }
116
117 #else /* HAVE_FMA */
118
119 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cb_10 -include r2cb.h */
120
121 /*
122 * This function contains 34 FP additions, 14 FP multiplications,
123 * (or, 26 additions, 6 multiplications, 8 fused multiply/add),
124 * 26 stack variables, 5 constants, and 20 memory accesses
125 */
126 #include "r2cb.h"
127
128 static void r2cb_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
129 {
130 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
131 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
132 DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
133 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
134 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
135 {
136 INT i;
137 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)) {
138 E T3, Tb, Tn, Tv, Tk, Tu, Ta, Ts, Te, Tg, Ti, Tj;
139 {
140 E T1, T2, Tl, Tm;
141 T1 = Cr[0];
142 T2 = Cr[WS(csr, 5)];
143 T3 = T1 - T2;
144 Tb = T1 + T2;
145 Tl = Ci[WS(csi, 4)];
146 Tm = Ci[WS(csi, 1)];
147 Tn = Tl - Tm;
148 Tv = Tl + Tm;
149 }
150 Ti = Ci[WS(csi, 2)];
151 Tj = Ci[WS(csi, 3)];
152 Tk = Ti - Tj;
153 Tu = Ti + Tj;
154 {
155 E T6, Tc, T9, Td;
156 {
157 E T4, T5, T7, T8;
158 T4 = Cr[WS(csr, 2)];
159 T5 = Cr[WS(csr, 3)];
160 T6 = T4 - T5;
161 Tc = T4 + T5;
162 T7 = Cr[WS(csr, 4)];
163 T8 = Cr[WS(csr, 1)];
164 T9 = T7 - T8;
165 Td = T7 + T8;
166 }
167 Ta = T6 + T9;
168 Ts = KP1_118033988 * (T6 - T9);
169 Te = Tc + Td;
170 Tg = KP1_118033988 * (Tc - Td);
171 }
172 R1[WS(rs, 2)] = FMA(KP2_000000000, Ta, T3);
173 R0[0] = FMA(KP2_000000000, Te, Tb);
174 {
175 E To, Tq, Th, Tp, Tf;
176 To = FNMS(KP1_902113032, Tn, KP1_175570504 * Tk);
177 Tq = FMA(KP1_902113032, Tk, KP1_175570504 * Tn);
178 Tf = FNMS(KP500000000, Te, Tb);
179 Th = Tf - Tg;
180 Tp = Tg + Tf;
181 R0[WS(rs, 1)] = Th - To;
182 R0[WS(rs, 2)] = Tp + Tq;
183 R0[WS(rs, 4)] = Th + To;
184 R0[WS(rs, 3)] = Tp - Tq;
185 }
186 {
187 E Tw, Ty, Tt, Tx, Tr;
188 Tw = FNMS(KP1_902113032, Tv, KP1_175570504 * Tu);
189 Ty = FMA(KP1_902113032, Tu, KP1_175570504 * Tv);
190 Tr = FNMS(KP500000000, Ta, T3);
191 Tt = Tr - Ts;
192 Tx = Ts + Tr;
193 R1[WS(rs, 3)] = Tt - Tw;
194 R1[WS(rs, 4)] = Tx + Ty;
195 R1[WS(rs, 1)] = Tt + Tw;
196 R1[0] = Tx - Ty;
197 }
198 }
199 }
200 }
201
202 static const kr2c_desc desc = { 10, "r2cb_10", {26, 6, 8, 0}, &GENUS };
203
204 void X(codelet_r2cb_10) (planner *p) {
205 X(kr2c_register) (p, r2cb_10, &desc);
206 }
207
208 #endif /* HAVE_FMA */