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 14:27:09 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 6 -name n1bv_6 -include n1b.h */
31
* This function contains 18 FP additions, 8 FP multiplications,
32
* (or, 12 additions, 2 multiplications, 6 fused multiply/add),
33
* 23 stack variables, and 12 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_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
46
DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
47
DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
53
for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
54
V T1, T2, T4, T5, T7, T8;
55
T1 = LD(&(xi[0]), ivs, &(xi[0]));
56
T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
57
T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
58
T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
59
T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
60
T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
62
V T3, Td, T6, Te, T9, Tf;
70
V Tg, Ti, Ta, Tc, Th, Tb;
72
Ti = VMUL(LDK(KP866025403), VSUB(Te, Tf));
74
Tc = VMUL(LDK(KP866025403), VSUB(T6, T9));
75
Th = VFNMS(LDK(KP500000000), Tg, Td);
76
ST(&(xo[0]), VADD(Td, Tg), ovs, &(xo[0]));
77
Tb = VFNMS(LDK(KP500000000), Ta, T3);
78
ST(&(xo[WS(os, 3)]), VADD(T3, Ta), ovs, &(xo[WS(os, 1)]));
79
ST(&(xo[WS(os, 4)]), VFMAI(Ti, Th), ovs, &(xo[0]));
80
ST(&(xo[WS(os, 2)]), VFNMSI(Ti, Th), ovs, &(xo[0]));
81
ST(&(xo[WS(os, 5)]), VFNMSI(Tc, Tb), ovs, &(xo[WS(os, 1)]));
82
ST(&(xo[WS(os, 1)]), VFMAI(Tc, Tb), ovs, &(xo[WS(os, 1)]));
88
static const kdft_desc desc = { 6, "n1bv_6", {12, 2, 6, 0}, &GENUS, 0, 0, 0, 0 };
89
void X(codelet_n1bv_6) (planner *p) {
90
X(kdft_register) (p, n1bv_6, &desc);
95
/* Generated by: ../../../genfft/gen_notw_c -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 6 -name n1bv_6 -include n1b.h */
98
* This function contains 18 FP additions, 4 FP multiplications,
99
* (or, 16 additions, 2 multiplications, 2 fused multiply/add),
100
* 19 stack variables, and 12 memory accesses
104
* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
105
* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
106
* $Id: gen_notw_c.ml,v 1.17 2006-02-12 23:34:12 athena Exp $
111
static void n1bv_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
113
DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
114
DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
120
for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
121
V Ta, Td, T3, Te, T6, Tf, Tb, Tg, T8, T9;
122
T8 = LD(&(xi[0]), ivs, &(xi[0]));
123
T9 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
128
T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
129
T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
132
T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
133
T5 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
139
ST(&(xo[WS(os, 3)]), VADD(Ta, Tb), ovs, &(xo[WS(os, 1)]));
140
ST(&(xo[0]), VADD(Td, Tg), ovs, &(xo[0]));
143
T7 = VBYI(VMUL(LDK(KP866025403), VSUB(T3, T6)));
144
Tc = VFNMS(LDK(KP500000000), Tb, Ta);
145
ST(&(xo[WS(os, 1)]), VADD(T7, Tc), ovs, &(xo[WS(os, 1)]));
146
ST(&(xo[WS(os, 5)]), VSUB(Tc, T7), ovs, &(xo[WS(os, 1)]));
147
Th = VFNMS(LDK(KP500000000), Tg, Td);
148
Ti = VBYI(VMUL(LDK(KP866025403), VSUB(Te, Tf)));
149
ST(&(xo[WS(os, 2)]), VSUB(Th, Ti), ovs, &(xo[0]));
150
ST(&(xo[WS(os, 4)]), VADD(Ti, Th), ovs, &(xo[0]));
155
static const kdft_desc desc = { 6, "n1bv_6", {16, 2, 2, 0}, &GENUS, 0, 0, 0, 0 };
156
void X(codelet_n1bv_6) (planner *p) {
157
X(kdft_register) (p, n1bv_6, &desc);
160
#endif /* HAVE_FMA */