Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/rdft/scalar/r2r/e01_8.c @ 167:bd3cc4d1df30

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW 3.3.8 source, and a Linux build
author Chris Cannam <cannam@all-day-breakfast.com>
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
comparison
equal deleted inserted replaced
166:cbd6d7e562c7 167:bd3cc4d1df30
1 /*
2 * Copyright (c) 2003, 2007-14 Matteo Frigo
3 * Copyright (c) 2003, 2007-14 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 Thu May 24 08:08:10 EDT 2018 */
23
24 #include "rdft/codelet-rdft.h"
25
26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
27
28 /* Generated by: ../../../genfft/gen_r2r.native -fma -compact -variables 4 -pipeline-latency 4 -redft01 -n 8 -name e01_8 -include rdft/scalar/r2r.h */
29
30 /*
31 * This function contains 26 FP additions, 24 FP multiplications,
32 * (or, 2 additions, 0 multiplications, 24 fused multiply/add),
33 * 27 stack variables, 8 constants, and 16 memory accesses
34 */
35 #include "rdft/scalar/r2r.h"
36
37 static void e01_8(const R *I, R *O, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DK(KP1_662939224, +1.662939224605090474157576755235811513477121624);
40 DK(KP668178637, +0.668178637919298919997757686523080761552472251);
41 DK(KP1_961570560, +1.961570560806460898252364472268478073947867462);
42 DK(KP198912367, +0.198912367379658006911597622644676228597850501);
43 DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
44 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
45 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
46 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
47 {
48 INT i;
49 for (i = v; i > 0; i = i - 1, I = I + ivs, O = O + ovs, MAKE_VOLATILE_STRIDE(16, is), MAKE_VOLATILE_STRIDE(16, os)) {
50 E T3, Tj, T6, Tk, Tc, Tn, Tf, Tm;
51 {
52 E T1, T2, T4, T5;
53 T1 = I[0];
54 T2 = I[WS(is, 4)];
55 T3 = FMA(KP1_414213562, T2, T1);
56 Tj = FNMS(KP1_414213562, T2, T1);
57 T4 = I[WS(is, 2)];
58 T5 = I[WS(is, 6)];
59 T6 = FMA(KP414213562, T5, T4);
60 Tk = FMS(KP414213562, T4, T5);
61 {
62 E T8, Td, Tb, Te, T9, Ta;
63 T8 = I[WS(is, 1)];
64 Td = I[WS(is, 7)];
65 T9 = I[WS(is, 5)];
66 Ta = I[WS(is, 3)];
67 Tb = T9 + Ta;
68 Te = Ta - T9;
69 Tc = FMA(KP707106781, Tb, T8);
70 Tn = FNMS(KP707106781, Te, Td);
71 Tf = FMA(KP707106781, Te, Td);
72 Tm = FNMS(KP707106781, Tb, T8);
73 }
74 }
75 {
76 E T7, Tg, Tp, Tq;
77 T7 = FMA(KP1_847759065, T6, T3);
78 Tg = FMA(KP198912367, Tf, Tc);
79 O[WS(os, 7)] = FNMS(KP1_961570560, Tg, T7);
80 O[0] = FMA(KP1_961570560, Tg, T7);
81 Tp = FNMS(KP1_847759065, Tk, Tj);
82 Tq = FMA(KP668178637, Tm, Tn);
83 O[WS(os, 5)] = FNMS(KP1_662939224, Tq, Tp);
84 O[WS(os, 2)] = FMA(KP1_662939224, Tq, Tp);
85 }
86 {
87 E Th, Ti, Tl, To;
88 Th = FNMS(KP1_847759065, T6, T3);
89 Ti = FNMS(KP198912367, Tc, Tf);
90 O[WS(os, 3)] = FNMS(KP1_961570560, Ti, Th);
91 O[WS(os, 4)] = FMA(KP1_961570560, Ti, Th);
92 Tl = FMA(KP1_847759065, Tk, Tj);
93 To = FNMS(KP668178637, Tn, Tm);
94 O[WS(os, 6)] = FNMS(KP1_662939224, To, Tl);
95 O[WS(os, 1)] = FMA(KP1_662939224, To, Tl);
96 }
97 }
98 }
99 }
100
101 static const kr2r_desc desc = { 8, "e01_8", {2, 0, 24, 0}, &GENUS, REDFT01 };
102
103 void X(codelet_e01_8) (planner *p) {
104 X(kr2r_register) (p, e01_8, &desc);
105 }
106
107 #else
108
109 /* Generated by: ../../../genfft/gen_r2r.native -compact -variables 4 -pipeline-latency 4 -redft01 -n 8 -name e01_8 -include rdft/scalar/r2r.h */
110
111 /*
112 * This function contains 26 FP additions, 15 FP multiplications,
113 * (or, 20 additions, 9 multiplications, 6 fused multiply/add),
114 * 28 stack variables, 8 constants, and 16 memory accesses
115 */
116 #include "rdft/scalar/r2r.h"
117
118 static void e01_8(const R *I, R *O, stride is, stride os, INT v, INT ivs, INT ovs)
119 {
120 DK(KP1_662939224, +1.662939224605090474157576755235811513477121624);
121 DK(KP1_111140466, +1.111140466039204449485661627897065748749874382);
122 DK(KP390180644, +0.390180644032256535696569736954044481855383236);
123 DK(KP1_961570560, +1.961570560806460898252364472268478073947867462);
124 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
125 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
126 DK(KP765366864, +0.765366864730179543456919968060797733522689125);
127 DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
128 {
129 INT i;
130 for (i = v; i > 0; i = i - 1, I = I + ivs, O = O + ovs, MAKE_VOLATILE_STRIDE(16, is), MAKE_VOLATILE_STRIDE(16, os)) {
131 E T7, Tl, T4, Tk, Td, To, Tg, Tn;
132 {
133 E T5, T6, T1, T3, T2;
134 T5 = I[WS(is, 2)];
135 T6 = I[WS(is, 6)];
136 T7 = FMA(KP1_847759065, T5, KP765366864 * T6);
137 Tl = FNMS(KP1_847759065, T6, KP765366864 * T5);
138 T1 = I[0];
139 T2 = I[WS(is, 4)];
140 T3 = KP1_414213562 * T2;
141 T4 = T1 + T3;
142 Tk = T1 - T3;
143 {
144 E T9, Tf, Tc, Te, Ta, Tb;
145 T9 = I[WS(is, 1)];
146 Tf = I[WS(is, 7)];
147 Ta = I[WS(is, 5)];
148 Tb = I[WS(is, 3)];
149 Tc = KP707106781 * (Ta + Tb);
150 Te = KP707106781 * (Ta - Tb);
151 Td = T9 + Tc;
152 To = Te + Tf;
153 Tg = Te - Tf;
154 Tn = T9 - Tc;
155 }
156 }
157 {
158 E T8, Th, Tq, Tr;
159 T8 = T4 + T7;
160 Th = FNMS(KP390180644, Tg, KP1_961570560 * Td);
161 O[WS(os, 7)] = T8 - Th;
162 O[0] = T8 + Th;
163 Tq = Tk - Tl;
164 Tr = FMA(KP1_111140466, Tn, KP1_662939224 * To);
165 O[WS(os, 5)] = Tq - Tr;
166 O[WS(os, 2)] = Tq + Tr;
167 }
168 {
169 E Ti, Tj, Tm, Tp;
170 Ti = T4 - T7;
171 Tj = FMA(KP390180644, Td, KP1_961570560 * Tg);
172 O[WS(os, 4)] = Ti - Tj;
173 O[WS(os, 3)] = Ti + Tj;
174 Tm = Tk + Tl;
175 Tp = FNMS(KP1_111140466, To, KP1_662939224 * Tn);
176 O[WS(os, 6)] = Tm - Tp;
177 O[WS(os, 1)] = Tm + Tp;
178 }
179 }
180 }
181 }
182
183 static const kr2r_desc desc = { 8, "e01_8", {20, 9, 6, 0}, &GENUS, REDFT01 };
184
185 void X(codelet_e01_8) (planner *p) {
186 X(kr2r_register) (p, e01_8, &desc);
187 }
188
189 #endif