Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/dft/scalar/codelets/n1_6.c @ 82:d0c2a83c1364

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
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
comparison
equal deleted inserted replaced
81:7029a4916348 82:d0c2a83c1364
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 08:04:10 EDT 2018 */
23
24 #include "dft/codelet-dft.h"
25
26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
27
28 /* Generated by: ../../../genfft/gen_notw.native -fma -compact -variables 4 -pipeline-latency 4 -n 6 -name n1_6 -include dft/scalar/n.h */
29
30 /*
31 * This function contains 36 FP additions, 12 FP multiplications,
32 * (or, 24 additions, 0 multiplications, 12 fused multiply/add),
33 * 23 stack variables, 2 constants, and 24 memory accesses
34 */
35 #include "dft/scalar/n.h"
36
37 static void n1_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
40 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
41 {
42 INT i;
43 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) {
44 E T3, Tb, Tp, Tx, T6, Tc, T9, Td, Ta, Te, Ti, Tu, Tl, Tv, Tq;
45 E Ty;
46 {
47 E T1, T2, Tn, To;
48 T1 = ri[0];
49 T2 = ri[WS(is, 3)];
50 T3 = T1 - T2;
51 Tb = T1 + T2;
52 Tn = ii[0];
53 To = ii[WS(is, 3)];
54 Tp = Tn - To;
55 Tx = Tn + To;
56 }
57 {
58 E T4, T5, T7, T8;
59 T4 = ri[WS(is, 2)];
60 T5 = ri[WS(is, 5)];
61 T6 = T4 - T5;
62 Tc = T4 + T5;
63 T7 = ri[WS(is, 4)];
64 T8 = ri[WS(is, 1)];
65 T9 = T7 - T8;
66 Td = T7 + T8;
67 }
68 Ta = T6 + T9;
69 Te = Tc + Td;
70 {
71 E Tg, Th, Tj, Tk;
72 Tg = ii[WS(is, 2)];
73 Th = ii[WS(is, 5)];
74 Ti = Tg - Th;
75 Tu = Tg + Th;
76 Tj = ii[WS(is, 4)];
77 Tk = ii[WS(is, 1)];
78 Tl = Tj - Tk;
79 Tv = Tj + Tk;
80 }
81 Tq = Ti + Tl;
82 Ty = Tu + Tv;
83 ro[WS(os, 3)] = T3 + Ta;
84 io[WS(os, 3)] = Tp + Tq;
85 ro[0] = Tb + Te;
86 io[0] = Tx + Ty;
87 {
88 E Tf, Tm, Tr, Ts;
89 Tf = FNMS(KP500000000, Ta, T3);
90 Tm = Ti - Tl;
91 ro[WS(os, 5)] = FNMS(KP866025403, Tm, Tf);
92 ro[WS(os, 1)] = FMA(KP866025403, Tm, Tf);
93 Tr = FNMS(KP500000000, Tq, Tp);
94 Ts = T9 - T6;
95 io[WS(os, 1)] = FMA(KP866025403, Ts, Tr);
96 io[WS(os, 5)] = FNMS(KP866025403, Ts, Tr);
97 }
98 {
99 E Tt, Tw, Tz, TA;
100 Tt = FNMS(KP500000000, Te, Tb);
101 Tw = Tu - Tv;
102 ro[WS(os, 2)] = FNMS(KP866025403, Tw, Tt);
103 ro[WS(os, 4)] = FMA(KP866025403, Tw, Tt);
104 Tz = FNMS(KP500000000, Ty, Tx);
105 TA = Td - Tc;
106 io[WS(os, 2)] = FNMS(KP866025403, TA, Tz);
107 io[WS(os, 4)] = FMA(KP866025403, TA, Tz);
108 }
109 }
110 }
111 }
112
113 static const kdft_desc desc = { 6, "n1_6", {24, 0, 12, 0}, &GENUS, 0, 0, 0, 0 };
114
115 void X(codelet_n1_6) (planner *p) {
116 X(kdft_register) (p, n1_6, &desc);
117 }
118
119 #else
120
121 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 6 -name n1_6 -include dft/scalar/n.h */
122
123 /*
124 * This function contains 36 FP additions, 8 FP multiplications,
125 * (or, 32 additions, 4 multiplications, 4 fused multiply/add),
126 * 23 stack variables, 2 constants, and 24 memory accesses
127 */
128 #include "dft/scalar/n.h"
129
130 static void n1_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
131 {
132 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
133 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
134 {
135 INT i;
136 for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) {
137 E T3, Tb, Tq, Tx, T6, Tc, T9, Td, Ta, Te, Ti, Tu, Tl, Tv, Tr;
138 E Ty;
139 {
140 E T1, T2, To, Tp;
141 T1 = ri[0];
142 T2 = ri[WS(is, 3)];
143 T3 = T1 - T2;
144 Tb = T1 + T2;
145 To = ii[0];
146 Tp = ii[WS(is, 3)];
147 Tq = To - Tp;
148 Tx = To + Tp;
149 }
150 {
151 E T4, T5, T7, T8;
152 T4 = ri[WS(is, 2)];
153 T5 = ri[WS(is, 5)];
154 T6 = T4 - T5;
155 Tc = T4 + T5;
156 T7 = ri[WS(is, 4)];
157 T8 = ri[WS(is, 1)];
158 T9 = T7 - T8;
159 Td = T7 + T8;
160 }
161 Ta = T6 + T9;
162 Te = Tc + Td;
163 {
164 E Tg, Th, Tj, Tk;
165 Tg = ii[WS(is, 2)];
166 Th = ii[WS(is, 5)];
167 Ti = Tg - Th;
168 Tu = Tg + Th;
169 Tj = ii[WS(is, 4)];
170 Tk = ii[WS(is, 1)];
171 Tl = Tj - Tk;
172 Tv = Tj + Tk;
173 }
174 Tr = Ti + Tl;
175 Ty = Tu + Tv;
176 ro[WS(os, 3)] = T3 + Ta;
177 io[WS(os, 3)] = Tq + Tr;
178 ro[0] = Tb + Te;
179 io[0] = Tx + Ty;
180 {
181 E Tf, Tm, Tn, Ts;
182 Tf = FNMS(KP500000000, Ta, T3);
183 Tm = KP866025403 * (Ti - Tl);
184 ro[WS(os, 5)] = Tf - Tm;
185 ro[WS(os, 1)] = Tf + Tm;
186 Tn = KP866025403 * (T9 - T6);
187 Ts = FNMS(KP500000000, Tr, Tq);
188 io[WS(os, 1)] = Tn + Ts;
189 io[WS(os, 5)] = Ts - Tn;
190 }
191 {
192 E Tt, Tw, Tz, TA;
193 Tt = FNMS(KP500000000, Te, Tb);
194 Tw = KP866025403 * (Tu - Tv);
195 ro[WS(os, 2)] = Tt - Tw;
196 ro[WS(os, 4)] = Tt + Tw;
197 Tz = FNMS(KP500000000, Ty, Tx);
198 TA = KP866025403 * (Td - Tc);
199 io[WS(os, 2)] = Tz - TA;
200 io[WS(os, 4)] = TA + Tz;
201 }
202 }
203 }
204 }
205
206 static const kdft_desc desc = { 6, "n1_6", {32, 4, 4, 0}, &GENUS, 0, 0, 0, 0 };
207
208 void X(codelet_n1_6) (planner *p) {
209 X(kdft_register) (p, n1_6, &desc);
210 }
211
212 #endif