2
* Copyright (c) 2003, 2006 Matteo Frigo
3
* Copyright (c) 2003, 2006 Massachusetts Institute of Technology
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.
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.
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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
21
/* This file was automatically generated --- DO NOT EDIT */
22
/* Generated on Sat Jul 1 14:27:07 EDT 2006 */
24
#include "codelet-dft.h"
28
/* Generated by: ../../../genfft/gen_notw_c -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 5 -name n1bv_5 -include n1b.h */
31
* This function contains 16 FP additions, 11 FP multiplications,
32
* (or, 7 additions, 2 multiplications, 9 fused multiply/add),
33
* 23 stack variables, and 10 memory accesses
37
* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
38
* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
39
* $Id: gen_notw_c.ml,v 1.17 2006-02-12 23:34:12 athena Exp $
44
static void n1bv_5(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
46
DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
47
DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
48
DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
49
DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
55
for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
57
T1 = LD(&(xi[0]), ivs, &(xi[0]));
58
T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
59
T3 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
60
T5 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
61
T6 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
69
V Tg, Te, Ta, T8, T9, Tf, Tb;
70
Tg = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tc, Td));
71
Te = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Td, Tc));
74
T9 = VFNMS(LDK(KP250000000), T8, T1);
75
ST(&(xo[0]), VADD(T1, T8), ovs, &(xo[0]));
76
Tf = VFNMS(LDK(KP559016994), Ta, T9);
77
Tb = VFMA(LDK(KP559016994), Ta, T9);
78
ST(&(xo[WS(os, 2)]), VFNMSI(Tg, Tf), ovs, &(xo[0]));
79
ST(&(xo[WS(os, 3)]), VFMAI(Tg, Tf), ovs, &(xo[WS(os, 1)]));
80
ST(&(xo[WS(os, 4)]), VFNMSI(Te, Tb), ovs, &(xo[0]));
81
ST(&(xo[WS(os, 1)]), VFMAI(Te, Tb), ovs, &(xo[WS(os, 1)]));
87
static const kdft_desc desc = { 5, "n1bv_5", {7, 2, 9, 0}, &GENUS, 0, 0, 0, 0 };
88
void X(codelet_n1bv_5) (planner *p) {
89
X(kdft_register) (p, n1bv_5, &desc);
94
/* Generated by: ../../../genfft/gen_notw_c -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 5 -name n1bv_5 -include n1b.h */
97
* This function contains 16 FP additions, 6 FP multiplications,
98
* (or, 13 additions, 3 multiplications, 3 fused multiply/add),
99
* 18 stack variables, and 10 memory accesses
103
* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
104
* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
105
* $Id: gen_notw_c.ml,v 1.17 2006-02-12 23:34:12 athena Exp $
110
static void n1bv_5(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
112
DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
113
DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
114
DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
115
DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
121
for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
122
V Tb, T3, Tc, T6, Ta;
123
Tb = LD(&(xi[0]), ivs, &(xi[0]));
125
V T1, T2, T8, T4, T5, T9;
126
T1 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
127
T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
129
T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
130
T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
135
Ta = VMUL(LDK(KP559016994), VSUB(T8, T9));
137
ST(&(xo[0]), VADD(Tb, Tc), ovs, &(xo[0]));
139
V T7, Tf, Te, Tg, Td;
140
T7 = VBYI(VFMA(LDK(KP951056516), T3, VMUL(LDK(KP587785252), T6)));
141
Tf = VBYI(VFNMS(LDK(KP951056516), T6, VMUL(LDK(KP587785252), T3)));
142
Td = VFNMS(LDK(KP250000000), Tc, Tb);
145
ST(&(xo[WS(os, 1)]), VADD(T7, Te), ovs, &(xo[WS(os, 1)]));
146
ST(&(xo[WS(os, 3)]), VSUB(Tg, Tf), ovs, &(xo[WS(os, 1)]));
147
ST(&(xo[WS(os, 4)]), VSUB(Te, T7), ovs, &(xo[0]));
148
ST(&(xo[WS(os, 2)]), VADD(Tf, Tg), ovs, &(xo[0]));
153
static const kdft_desc desc = { 5, "n1bv_5", {13, 3, 3, 0}, &GENUS, 0, 0, 0, 0 };
154
void X(codelet_n1bv_5) (planner *p) {
155
X(kdft_register) (p, n1bv_5, &desc);
158
#endif /* HAVE_FMA */