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 Tue Mar 7 10:18:43 EST 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 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 4 -name t3fv_4 -include t3f.h */
31
* This function contains 12 FP additions, 10 FP multiplications,
32
* (or, 10 additions, 8 multiplications, 2 fused multiply/add),
33
* 16 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 *t3fv_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 * 4), MAKE_VOLATILE_STRIDE(ios)) {
50
V T2, T3, T1, Ta, T5, T8;
52
T3 = LDW(&(W[TWVL * 2]));
53
T1 = LD(&(x[0]), dist, &(x[0]));
54
Ta = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)]));
55
T5 = LD(&(x[WS(ios, 2)]), dist, &(x[0]));
56
T8 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)]));
69
ST(&(x[0]), VADD(Td, Te), dist, &(x[0]));
70
ST(&(x[WS(ios, 2)]), VSUB(Td, Te), dist, &(x[0]));
71
ST(&(x[WS(ios, 3)]), VFMAI(Tc, T7), dist, &(x[WS(ios, 1)]));
72
ST(&(x[WS(ios, 1)]), VFNMSI(Tc, T7), dist, &(x[WS(ios, 1)]));
79
static const tw_instr twinstr[] = {
85
static const ct_desc desc = { 4, "t3fv_4", twinstr, &GENUS, {10, 8, 2, 0}, 0, 0, 0 };
87
void X(codelet_t3fv_4) (planner *p) {
88
X(kdft_dit_register) (p, t3fv_4, &desc);
92
/* Generated by: ../../../genfft/gen_twiddle_c -simd -compact -variables 4 -pipeline-latency 8 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 4 -name t3fv_4 -include t3f.h */
95
* This function contains 12 FP additions, 8 FP multiplications,
96
* (or, 12 additions, 8 multiplications, 0 fused multiply/add),
97
* 16 stack variables, and 8 memory accesses
101
* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
102
* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
103
* $Id: gen_twiddle_c.ml,v 1.14 2006-02-12 23:34:12 athena Exp $
108
static const R *t3fv_4(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
113
for (i = m; i > 0; i = i - VL, x = x + (VL * dist), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(ios)) {
116
T3 = LDW(&(W[TWVL * 2]));
119
V T1, Tb, T6, T9, Ta, T5, T8;
120
T1 = LD(&(x[0]), dist, &(x[0]));
121
Ta = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)]));
123
T5 = LD(&(x[WS(ios, 2)]), dist, &(x[0]));
125
T8 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)]));
130
Tc = VBYI(VSUB(T9, Tb));
131
ST(&(x[WS(ios, 1)]), VSUB(T7, Tc), dist, &(x[WS(ios, 1)]));
132
ST(&(x[WS(ios, 3)]), VADD(T7, Tc), dist, &(x[WS(ios, 1)]));
135
ST(&(x[WS(ios, 2)]), VSUB(Td, Te), dist, &(x[0]));
136
ST(&(x[0]), VADD(Td, Te), dist, &(x[0]));
143
static const tw_instr twinstr[] = {
149
static const ct_desc desc = { 4, "t3fv_4", twinstr, &GENUS, {12, 8, 0, 0}, 0, 0, 0 };
151
void X(codelet_t3fv_4) (planner *p) {
152
X(kdft_dit_register) (p, t3fv_4, &desc);
154
#endif /* HAVE_FMA */