Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/simd/common/t3fv_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 07:38:54 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 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 5 -name t3fv_5 -include t3f.h */
29
30 /*
31 * This function contains 22 FP additions, 23 FP multiplications,
32 * (or, 13 additions, 14 multiplications, 9 fused multiply/add),
33 * 30 stack variables, 4 constants, and 10 memory accesses
34 */
35 #include "t3f.h"
36
37 static void t3fv_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) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(5, rs)) {
48 V T2, T5, T1, T3, Td, T7, Tb;
49 T2 = LDW(&(W[0]));
50 T5 = LDW(&(W[TWVL * 2]));
51 T1 = LD(&(x[0]), ms, &(x[0]));
52 T3 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
53 Td = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
54 T7 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
55 Tb = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
56 {
57 V Ta, T6, T4, Te, Tc, T8;
58 Ta = VZMULJ(T2, T5);
59 T6 = VZMUL(T2, T5);
60 T4 = VZMULJ(T2, T3);
61 Te = VZMULJ(T5, Td);
62 Tc = VZMULJ(Ta, Tb);
63 T8 = VZMULJ(T6, T7);
64 {
65 V Tf, Tl, T9, Tk;
66 Tf = VADD(Tc, Te);
67 Tl = VSUB(Tc, Te);
68 T9 = VADD(T4, T8);
69 Tk = VSUB(T4, T8);
70 {
71 V Ti, Tg, To, Tm, Th, Tn, Tj;
72 Ti = VSUB(T9, Tf);
73 Tg = VADD(T9, Tf);
74 To = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tk, Tl));
75 Tm = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Tl, Tk));
76 Th = VFNMS(LDK(KP250000000), Tg, T1);
77 ST(&(x[0]), VADD(T1, Tg), ms, &(x[0]));
78 Tn = VFNMS(LDK(KP559016994), Ti, Th);
79 Tj = VFMA(LDK(KP559016994), Ti, Th);
80 ST(&(x[WS(rs, 2)]), VFMAI(To, Tn), ms, &(x[0]));
81 ST(&(x[WS(rs, 3)]), VFNMSI(To, Tn), ms, &(x[WS(rs, 1)]));
82 ST(&(x[WS(rs, 4)]), VFMAI(Tm, Tj), ms, &(x[0]));
83 ST(&(x[WS(rs, 1)]), VFNMSI(Tm, Tj), ms, &(x[WS(rs, 1)]));
84 }
85 }
86 }
87 }
88 }
89 VLEAVE();
90 }
91
92 static const tw_instr twinstr[] = {
93 VTW(0, 1),
94 VTW(0, 3),
95 {TW_NEXT, VL, 0}
96 };
97
98 static const ct_desc desc = { 5, XSIMD_STRING("t3fv_5"), twinstr, &GENUS, {13, 14, 9, 0}, 0, 0, 0 };
99
100 void XSIMD(codelet_t3fv_5) (planner *p) {
101 X(kdft_dit_register) (p, t3fv_5, &desc);
102 }
103 #else /* HAVE_FMA */
104
105 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 5 -name t3fv_5 -include t3f.h */
106
107 /*
108 * This function contains 22 FP additions, 18 FP multiplications,
109 * (or, 19 additions, 15 multiplications, 3 fused multiply/add),
110 * 24 stack variables, 4 constants, and 10 memory accesses
111 */
112 #include "t3f.h"
113
114 static void t3fv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
115 {
116 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
117 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
118 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
119 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
120 {
121 INT m;
122 R *x;
123 x = ri;
124 for (m = mb, W = W + (mb * ((TWVL / VL) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(5, rs)) {
125 V T1, T4, T5, T9;
126 T1 = LDW(&(W[0]));
127 T4 = LDW(&(W[TWVL * 2]));
128 T5 = VZMUL(T1, T4);
129 T9 = VZMULJ(T1, T4);
130 {
131 V Tg, Tk, Tl, T8, Te, Th;
132 Tg = LD(&(x[0]), ms, &(x[0]));
133 {
134 V T3, Td, T7, Tb;
135 {
136 V T2, Tc, T6, Ta;
137 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
138 T3 = VZMULJ(T1, T2);
139 Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
140 Td = VZMULJ(T4, Tc);
141 T6 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
142 T7 = VZMULJ(T5, T6);
143 Ta = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
144 Tb = VZMULJ(T9, Ta);
145 }
146 Tk = VSUB(T3, T7);
147 Tl = VSUB(Tb, Td);
148 T8 = VADD(T3, T7);
149 Te = VADD(Tb, Td);
150 Th = VADD(T8, Te);
151 }
152 ST(&(x[0]), VADD(Tg, Th), ms, &(x[0]));
153 {
154 V Tm, Tn, Tj, To, Tf, Ti;
155 Tm = VBYI(VFMA(LDK(KP951056516), Tk, VMUL(LDK(KP587785252), Tl)));
156 Tn = VBYI(VFNMS(LDK(KP587785252), Tk, VMUL(LDK(KP951056516), Tl)));
157 Tf = VMUL(LDK(KP559016994), VSUB(T8, Te));
158 Ti = VFNMS(LDK(KP250000000), Th, Tg);
159 Tj = VADD(Tf, Ti);
160 To = VSUB(Ti, Tf);
161 ST(&(x[WS(rs, 1)]), VSUB(Tj, Tm), ms, &(x[WS(rs, 1)]));
162 ST(&(x[WS(rs, 3)]), VSUB(To, Tn), ms, &(x[WS(rs, 1)]));
163 ST(&(x[WS(rs, 4)]), VADD(Tm, Tj), ms, &(x[0]));
164 ST(&(x[WS(rs, 2)]), VADD(Tn, To), ms, &(x[0]));
165 }
166 }
167 }
168 }
169 VLEAVE();
170 }
171
172 static const tw_instr twinstr[] = {
173 VTW(0, 1),
174 VTW(0, 3),
175 {TW_NEXT, VL, 0}
176 };
177
178 static const ct_desc desc = { 5, XSIMD_STRING("t3fv_5"), twinstr, &GENUS, {19, 15, 3, 0}, 0, 0, 0 };
179
180 void XSIMD(codelet_t3fv_5) (planner *p) {
181 X(kdft_dit_register) (p, t3fv_5, &desc);
182 }
183 #endif /* HAVE_FMA */