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 Sun Jul 2 15:50:11 EDT 2006 */
24
#include "codelet-rdft.h"
28
/* Generated by: ../../../genfft/gen_hc2hc -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 4 -dit -name hf_4 -include hf.h */
31
* This function contains 22 FP additions, 12 FP multiplications,
32
* (or, 16 additions, 6 multiplications, 6 fused multiply/add),
33
* 31 stack variables, and 16 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_hc2hc.ml,v 1.16 2006-02-12 23:34:12 athena Exp $
44
static const R *hf_4(R *rio, R *iio, const R *W, stride ios, INT m, INT dist)
47
for (i = m - 2; i > 0; i = i - 2, rio = rio + dist, iio = iio - dist, W = W + 6, MAKE_VOLATILE_STRIDE(ios)) {
48
E Tq, Tk, Ty, Tw, T8, Tm, To, Te;
50
E T1, Tv, Ta, Td, Tu, T7, T9, Tc, Tn, Tb;
58
Tv = iio[-WS(ios, 3)];
60
E T3, T6, T5, Tt, T4, T2, Tp, Th;
64
T6 = iio[-WS(ios, 1)];
66
Tq = FNMS(Ti, Tg, Tp);
72
Td = iio[-WS(ios, 2)];
73
Tu = FNMS(T5, T3, Tt);
85
To = FNMS(Tc, Ta, Tn);
94
rio[WS(ios, 1)] = Tm + Tr;
95
iio[-WS(ios, 3)] = Tm - Tr;
97
rio[WS(ios, 2)] = Ts - Tw;
98
iio[-WS(ios, 1)] = Tx + Ty;
99
rio[WS(ios, 3)] = Tx - Ty;
101
iio[-WS(ios, 2)] = T8 - Tl;
107
static const tw_instr twinstr[] = {
112
static const hc2hc_desc desc = { 4, "hf_4", twinstr, &GENUS, {16, 6, 6, 0}, 0, 0, 0 };
114
void X(codelet_hf_4) (planner *p) {
115
X(khc2hc_register) (p, hf_4, &desc);
119
/* Generated by: ../../../genfft/gen_hc2hc -compact -variables 4 -pipeline-latency 4 -n 4 -dit -name hf_4 -include hf.h */
122
* This function contains 22 FP additions, 12 FP multiplications,
123
* (or, 16 additions, 6 multiplications, 6 fused multiply/add),
124
* 13 stack variables, and 16 memory accesses
128
* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
129
* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
130
* $Id: gen_hc2hc.ml,v 1.16 2006-02-12 23:34:12 athena Exp $
135
static const R *hf_4(R *rio, R *iio, const R *W, stride ios, INT m, INT dist)
138
for (i = m - 2; i > 0; i = i - 2, rio = rio + dist, iio = iio - dist, W = W + 6, MAKE_VOLATILE_STRIDE(ios)) {
139
E T1, Th, Tl, Tp, T6, To, Tc, Tk;
143
Te = rio[WS(ios, 3)];
147
Th = FMA(Td, Te, Tf * Tg);
148
Tl = FNMS(Tf, Te, Td * Tg);
150
Tp = iio[-WS(ios, 3)];
153
T3 = rio[WS(ios, 2)];
154
T5 = iio[-WS(ios, 1)];
157
T6 = FMA(T2, T3, T4 * T5);
158
To = FNMS(T4, T3, T2 * T5);
162
T9 = rio[WS(ios, 1)];
163
Tb = iio[-WS(ios, 2)];
166
Tc = FMA(T8, T9, Ta * Tb);
167
Tk = FNMS(Ta, T9, T8 * Tb);
173
iio[-WS(ios, 2)] = T7 - Ti;
177
rio[WS(ios, 2)] = Tn - Tq;
184
iio[-WS(ios, 3)] = Tj - Tm;
185
rio[WS(ios, 1)] = Tj + Tm;
188
rio[WS(ios, 3)] = Tr - Ts;
189
iio[-WS(ios, 1)] = Tr + Ts;
195
static const tw_instr twinstr[] = {
200
static const hc2hc_desc desc = { 4, "hf_4", twinstr, &GENUS, {16, 6, 6, 0}, 0, 0, 0 };
202
void X(codelet_hf_4) (planner *p) {
203
X(khc2hc_register) (p, hf_4, &desc);
205
#endif /* HAVE_FMA */