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 14:45:52 EDT 2006 */
24
#include "codelet-dft.h"
28
/* Generated by: ../../../genfft/gen_twiddle_c -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 5 -name t1fv_5 -include t1f.h */
31
* This function contains 20 FP additions, 19 FP multiplications,
32
* (or, 11 additions, 10 multiplications, 9 fused multiply/add),
33
* 26 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_twiddle_c.ml,v 1.14 2006-02-12 23:34:12 athena Exp $
44
static const R *t1fv_5(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
46
DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
47
DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
48
DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
49
DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
53
for (i = m; i > 0; i = i - VL, x = x + (VL * dist), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(ios)) {
55
T1 = LD(&(x[0]), dist, &(x[0]));
56
T2 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)]));
57
T9 = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)]));
58
T4 = LD(&(x[WS(ios, 4)]), dist, &(x[0]));
59
T7 = LD(&(x[WS(ios, 2)]), dist, &(x[0]));
62
T3 = BYTWJ(&(W[0]), T2);
63
Ta = BYTWJ(&(W[TWVL * 4]), T9);
64
T5 = BYTWJ(&(W[TWVL * 6]), T4);
65
T8 = BYTWJ(&(W[TWVL * 2]), T7);
73
V Te, Tc, Tk, Ti, Td, Tj, Tf;
76
Tk = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tg, Th));
77
Ti = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Th, Tg));
78
Td = VFNMS(LDK(KP250000000), Tc, T1);
79
ST(&(x[0]), VADD(T1, Tc), dist, &(x[0]));
80
Tj = VFNMS(LDK(KP559016994), Te, Td);
81
Tf = VFMA(LDK(KP559016994), Te, Td);
82
ST(&(x[WS(ios, 2)]), VFMAI(Tk, Tj), dist, &(x[0]));
83
ST(&(x[WS(ios, 3)]), VFNMSI(Tk, Tj), dist, &(x[WS(ios, 1)]));
84
ST(&(x[WS(ios, 4)]), VFMAI(Ti, Tf), dist, &(x[0]));
85
ST(&(x[WS(ios, 1)]), VFNMSI(Ti, Tf), dist, &(x[WS(ios, 1)]));
93
static const tw_instr twinstr[] = {
101
static const ct_desc desc = { 5, "t1fv_5", twinstr, &GENUS, {11, 10, 9, 0}, 0, 0, 0 };
103
void X(codelet_t1fv_5) (planner *p) {
104
X(kdft_dit_register) (p, t1fv_5, &desc);
108
/* Generated by: ../../../genfft/gen_twiddle_c -simd -compact -variables 4 -pipeline-latency 8 -n 5 -name t1fv_5 -include t1f.h */
111
* This function contains 20 FP additions, 14 FP multiplications,
112
* (or, 17 additions, 11 multiplications, 3 fused multiply/add),
113
* 20 stack variables, and 10 memory accesses
117
* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
118
* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
119
* $Id: gen_twiddle_c.ml,v 1.14 2006-02-12 23:34:12 athena Exp $
124
static const R *t1fv_5(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
126
DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
127
DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
128
DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
129
DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
133
for (i = m; i > 0; i = i - VL, x = x + (VL * dist), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(ios)) {
134
V Tc, Tg, Th, T5, Ta, Td;
135
Tc = LD(&(x[0]), dist, &(x[0]));
140
T1 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)]));
141
T2 = BYTWJ(&(W[0]), T1);
142
T8 = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)]));
143
T9 = BYTWJ(&(W[TWVL * 4]), T8);
144
T3 = LD(&(x[WS(ios, 4)]), dist, &(x[0]));
145
T4 = BYTWJ(&(W[TWVL * 6]), T3);
146
T6 = LD(&(x[WS(ios, 2)]), dist, &(x[0]));
147
T7 = BYTWJ(&(W[TWVL * 2]), T6);
155
ST(&(x[0]), VADD(Tc, Td), dist, &(x[0]));
157
V Ti, Tj, Tf, Tk, Tb, Te;
158
Ti = VBYI(VFMA(LDK(KP951056516), Tg, VMUL(LDK(KP587785252), Th)));
159
Tj = VBYI(VFNMS(LDK(KP587785252), Tg, VMUL(LDK(KP951056516), Th)));
160
Tb = VMUL(LDK(KP559016994), VSUB(T5, Ta));
161
Te = VFNMS(LDK(KP250000000), Td, Tc);
164
ST(&(x[WS(ios, 1)]), VSUB(Tf, Ti), dist, &(x[WS(ios, 1)]));
165
ST(&(x[WS(ios, 3)]), VSUB(Tk, Tj), dist, &(x[WS(ios, 1)]));
166
ST(&(x[WS(ios, 4)]), VADD(Ti, Tf), dist, &(x[0]));
167
ST(&(x[WS(ios, 2)]), VADD(Tj, Tk), dist, &(x[0]));
173
static const tw_instr twinstr[] = {
181
static const ct_desc desc = { 5, "t1fv_5", twinstr, &GENUS, {17, 11, 3, 0}, 0, 0, 0 };
183
void X(codelet_t1fv_5) (planner *p) {
184
X(kdft_dit_register) (p, t1fv_5, &desc);
186
#endif /* HAVE_FMA */