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 14:55:04 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 4 -name t2fv_4 -include t2f.h */
31
* This function contains 11 FP additions, 8 FP multiplications,
32
* (or, 9 additions, 6 multiplications, 2 fused multiply/add),
33
* 13 stack variables, and 8 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 *t2fv_4(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
49
for (i = m; i > 0; i = i - VL, x = x + (VL * dist), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(ios)) {
50
V T1, T7, T2, T5, T8, T3, T6;
51
T1 = LD(&(x[0]), dist, &(x[0]));
52
T7 = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)]));
53
T2 = LD(&(x[WS(ios, 2)]), dist, &(x[0]));
54
T5 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)]));
55
T8 = BYTWJ(&(W[TWVL * 4]), T7);
56
T3 = BYTWJ(&(W[TWVL * 2]), T2);
57
T6 = BYTWJ(&(W[0]), T5);
64
ST(&(x[0]), VADD(Ta, Tb), dist, &(x[0]));
65
ST(&(x[WS(ios, 2)]), VSUB(Ta, Tb), dist, &(x[0]));
66
ST(&(x[WS(ios, 3)]), VFMAI(T9, T4), dist, &(x[WS(ios, 1)]));
67
ST(&(x[WS(ios, 1)]), VFNMSI(T9, T4), dist, &(x[WS(ios, 1)]));
73
static const tw_instr twinstr[] = {
80
static const ct_desc desc = { 4, "t2fv_4", twinstr, &GENUS, {9, 6, 2, 0}, 0, 0, 0 };
82
void X(codelet_t2fv_4) (planner *p) {
83
X(kdft_dit_register) (p, t2fv_4, &desc);
87
/* Generated by: ../../../genfft/gen_twiddle_c -simd -compact -variables 4 -pipeline-latency 8 -n 4 -name t2fv_4 -include t2f.h */
90
* This function contains 11 FP additions, 6 FP multiplications,
91
* (or, 11 additions, 6 multiplications, 0 fused multiply/add),
92
* 13 stack variables, and 8 memory accesses
96
* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
97
* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
98
* $Id: gen_twiddle_c.ml,v 1.14 2006-02-12 23:34:12 athena Exp $
103
static const R *t2fv_4(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
108
for (i = m; i > 0; i = i - VL, x = x + (VL * dist), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(ios)) {
109
V T1, T8, T3, T6, T7, T2, T5;
110
T1 = LD(&(x[0]), dist, &(x[0]));
111
T7 = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)]));
112
T8 = BYTWJ(&(W[TWVL * 4]), T7);
113
T2 = LD(&(x[WS(ios, 2)]), dist, &(x[0]));
114
T3 = BYTWJ(&(W[TWVL * 2]), T2);
115
T5 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)]));
116
T6 = BYTWJ(&(W[0]), T5);
120
T9 = VBYI(VSUB(T6, T8));
121
ST(&(x[WS(ios, 1)]), VSUB(T4, T9), dist, &(x[WS(ios, 1)]));
122
ST(&(x[WS(ios, 3)]), VADD(T4, T9), dist, &(x[WS(ios, 1)]));
125
ST(&(x[WS(ios, 2)]), VSUB(Ta, Tb), dist, &(x[0]));
126
ST(&(x[0]), VADD(Ta, Tb), dist, &(x[0]));
132
static const tw_instr twinstr[] = {
139
static const ct_desc desc = { 4, "t2fv_4", twinstr, &GENUS, {11, 6, 0, 0}, 0, 0, 0 };
141
void X(codelet_t2fv_4) (planner *p) {
142
X(kdft_dit_register) (p, t2fv_4, &desc);
144
#endif /* HAVE_FMA */