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 13:58:09 EDT 2006 */
24
#include "codelet-dft.h"
28
/* Generated by: ../../../genfft/gen_twiddle -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 3 -name t1_3 -include t.h */
31
* This function contains 16 FP additions, 14 FP multiplications,
32
* (or, 6 additions, 4 multiplications, 10 fused multiply/add),
33
* 21 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_twiddle.ml,v 1.24 2006-02-12 23:34:12 athena Exp $
44
static const R *t1_3(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
46
DK(KP866025403, +0.866025403784438646763723170752936183471402627);
47
DK(KP500000000, +0.500000000000000000000000000000000000000000000);
49
for (i = m; i > 0; i = i - 1, ri = ri + dist, ii = ii + dist, W = W + 4, MAKE_VOLATILE_STRIDE(ios)) {
50
E T1, Tm, T9, Tc, Tb, Th, T7, Ti, Ta, Tj, Td;
54
E T3, T6, T2, T5, Tg, T4, T8;
65
Th = FNMS(T5, T3, Tg);
70
Tj = FNMS(Tb, T9, Ti);
73
E Tk, Te, To, Tn, Tl, Tf;
79
Tn = FNMS(KP500000000, Tl, Tm);
81
Tf = FNMS(KP500000000, Te, T1);
82
ii[WS(ios, 1)] = FMA(KP866025403, To, Tn);
83
ii[WS(ios, 2)] = FNMS(KP866025403, To, Tn);
84
ri[WS(ios, 2)] = FNMS(KP866025403, Tk, Tf);
85
ri[WS(ios, 1)] = FMA(KP866025403, Tk, Tf);
91
static const tw_instr twinstr[] = {
96
static const ct_desc desc = { 3, "t1_3", twinstr, &GENUS, {6, 4, 10, 0}, 0, 0, 0 };
98
void X(codelet_t1_3) (planner *p) {
99
X(kdft_dit_register) (p, t1_3, &desc);
103
/* Generated by: ../../../genfft/gen_twiddle -compact -variables 4 -pipeline-latency 4 -n 3 -name t1_3 -include t.h */
106
* This function contains 16 FP additions, 12 FP multiplications,
107
* (or, 10 additions, 6 multiplications, 6 fused multiply/add),
108
* 15 stack variables, and 12 memory accesses
112
* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
113
* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
114
* $Id: gen_twiddle.ml,v 1.24 2006-02-12 23:34:12 athena Exp $
119
static const R *t1_3(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
121
DK(KP866025403, +0.866025403784438646763723170752936183471402627);
122
DK(KP500000000, +0.500000000000000000000000000000000000000000000);
124
for (i = m; i > 0; i = i - 1, ri = ri + dist, ii = ii + dist, W = W + 4, MAKE_VOLATILE_STRIDE(ios)) {
125
E T1, Ti, T6, Te, Tb, Tf, Tc, Th;
134
T6 = FMA(T2, T3, T4 * T5);
135
Te = FNMS(T4, T3, T2 * T5);
143
Tb = FMA(T7, T8, T9 * Ta);
144
Tf = FNMS(T9, T8, T7 * Ta);
152
Td = FNMS(KP500000000, Tc, T1);
153
Tg = KP866025403 * (Te - Tf);
154
ri[WS(ios, 2)] = Td - Tg;
155
ri[WS(ios, 1)] = Td + Tg;
156
Tj = KP866025403 * (Tb - T6);
157
Tk = FNMS(KP500000000, Th, Ti);
158
ii[WS(ios, 1)] = Tj + Tk;
159
ii[WS(ios, 2)] = Tk - Tj;
165
static const tw_instr twinstr[] = {
170
static const ct_desc desc = { 3, "t1_3", twinstr, &GENUS, {10, 6, 6, 0}, 0, 0, 0 };
172
void X(codelet_t1_3) (planner *p) {
173
X(kdft_dit_register) (p, t1_3, &desc);
175
#endif /* HAVE_FMA */