Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/simd/common/t2fv_5.c @ 10:37bf6b4a2645

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW3
author Chris Cannam
date Wed, 20 Mar 2013 15:35:50 +0000
parents
children
comparison
equal deleted inserted replaced
9:c0fb53affa76 10:37bf6b4a2645
1 /*
2 * Copyright (c) 2003, 2007-11 Matteo Frigo
3 * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology
4 *
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.
9 *
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.
14 *
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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18 *
19 */
20
21 /* This file was automatically generated --- DO NOT EDIT */
22 /* Generated on Sun Nov 25 07:38:41 EST 2012 */
23
24 #include "codelet-dft.h"
25
26 #ifdef HAVE_FMA
27
28 /* Generated by: ../../../genfft/gen_twiddle_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 5 -name t2fv_5 -include t2f.h */
29
30 /*
31 * This function contains 20 FP additions, 19 FP multiplications,
32 * (or, 11 additions, 10 multiplications, 9 fused multiply/add),
33 * 26 stack variables, 4 constants, and 10 memory accesses
34 */
35 #include "t2f.h"
36
37 static void t2fv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
38 {
39 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
40 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
41 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
42 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
43 {
44 INT m;
45 R *x;
46 x = ri;
47 for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(5, rs)) {
48 V T1, T2, T9, T4, T7;
49 T1 = LD(&(x[0]), ms, &(x[0]));
50 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
51 T9 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
52 T4 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
53 T7 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
54 {
55 V T3, Ta, T5, T8;
56 T3 = BYTWJ(&(W[0]), T2);
57 Ta = BYTWJ(&(W[TWVL * 4]), T9);
58 T5 = BYTWJ(&(W[TWVL * 6]), T4);
59 T8 = BYTWJ(&(W[TWVL * 2]), T7);
60 {
61 V T6, Tg, Tb, Th;
62 T6 = VADD(T3, T5);
63 Tg = VSUB(T3, T5);
64 Tb = VADD(T8, Ta);
65 Th = VSUB(T8, Ta);
66 {
67 V Te, Tc, Tk, Ti, Td, Tj, Tf;
68 Te = VSUB(T6, Tb);
69 Tc = VADD(T6, Tb);
70 Tk = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tg, Th));
71 Ti = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Th, Tg));
72 Td = VFNMS(LDK(KP250000000), Tc, T1);
73 ST(&(x[0]), VADD(T1, Tc), ms, &(x[0]));
74 Tj = VFNMS(LDK(KP559016994), Te, Td);
75 Tf = VFMA(LDK(KP559016994), Te, Td);
76 ST(&(x[WS(rs, 2)]), VFMAI(Tk, Tj), ms, &(x[0]));
77 ST(&(x[WS(rs, 3)]), VFNMSI(Tk, Tj), ms, &(x[WS(rs, 1)]));
78 ST(&(x[WS(rs, 4)]), VFMAI(Ti, Tf), ms, &(x[0]));
79 ST(&(x[WS(rs, 1)]), VFNMSI(Ti, Tf), ms, &(x[WS(rs, 1)]));
80 }
81 }
82 }
83 }
84 }
85 VLEAVE();
86 }
87
88 static const tw_instr twinstr[] = {
89 VTW(0, 1),
90 VTW(0, 2),
91 VTW(0, 3),
92 VTW(0, 4),
93 {TW_NEXT, VL, 0}
94 };
95
96 static const ct_desc desc = { 5, XSIMD_STRING("t2fv_5"), twinstr, &GENUS, {11, 10, 9, 0}, 0, 0, 0 };
97
98 void XSIMD(codelet_t2fv_5) (planner *p) {
99 X(kdft_dit_register) (p, t2fv_5, &desc);
100 }
101 #else /* HAVE_FMA */
102
103 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 5 -name t2fv_5 -include t2f.h */
104
105 /*
106 * This function contains 20 FP additions, 14 FP multiplications,
107 * (or, 17 additions, 11 multiplications, 3 fused multiply/add),
108 * 20 stack variables, 4 constants, and 10 memory accesses
109 */
110 #include "t2f.h"
111
112 static void t2fv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
113 {
114 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
115 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
116 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
117 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
118 {
119 INT m;
120 R *x;
121 x = ri;
122 for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(5, rs)) {
123 V Tc, Tg, Th, T5, Ta, Td;
124 Tc = LD(&(x[0]), ms, &(x[0]));
125 {
126 V T2, T9, T4, T7;
127 {
128 V T1, T8, T3, T6;
129 T1 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
130 T2 = BYTWJ(&(W[0]), T1);
131 T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
132 T9 = BYTWJ(&(W[TWVL * 4]), T8);
133 T3 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
134 T4 = BYTWJ(&(W[TWVL * 6]), T3);
135 T6 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
136 T7 = BYTWJ(&(W[TWVL * 2]), T6);
137 }
138 Tg = VSUB(T2, T4);
139 Th = VSUB(T7, T9);
140 T5 = VADD(T2, T4);
141 Ta = VADD(T7, T9);
142 Td = VADD(T5, Ta);
143 }
144 ST(&(x[0]), VADD(Tc, Td), ms, &(x[0]));
145 {
146 V Ti, Tj, Tf, Tk, Tb, Te;
147 Ti = VBYI(VFMA(LDK(KP951056516), Tg, VMUL(LDK(KP587785252), Th)));
148 Tj = VBYI(VFNMS(LDK(KP587785252), Tg, VMUL(LDK(KP951056516), Th)));
149 Tb = VMUL(LDK(KP559016994), VSUB(T5, Ta));
150 Te = VFNMS(LDK(KP250000000), Td, Tc);
151 Tf = VADD(Tb, Te);
152 Tk = VSUB(Te, Tb);
153 ST(&(x[WS(rs, 1)]), VSUB(Tf, Ti), ms, &(x[WS(rs, 1)]));
154 ST(&(x[WS(rs, 3)]), VSUB(Tk, Tj), ms, &(x[WS(rs, 1)]));
155 ST(&(x[WS(rs, 4)]), VADD(Ti, Tf), ms, &(x[0]));
156 ST(&(x[WS(rs, 2)]), VADD(Tj, Tk), ms, &(x[0]));
157 }
158 }
159 }
160 VLEAVE();
161 }
162
163 static const tw_instr twinstr[] = {
164 VTW(0, 1),
165 VTW(0, 2),
166 VTW(0, 3),
167 VTW(0, 4),
168 {TW_NEXT, VL, 0}
169 };
170
171 static const ct_desc desc = { 5, XSIMD_STRING("t2fv_5"), twinstr, &GENUS, {17, 11, 3, 0}, 0, 0, 0 };
172
173 void XSIMD(codelet_t2fv_5) (planner *p) {
174 X(kdft_dit_register) (p, t2fv_5, &desc);
175 }
176 #endif /* HAVE_FMA */