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 Dec 10 07:06:03 EST 2020 */
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_r2cf.native -fma -compact -variables 4 -pipeline-latency 4 -n 10 -name r2cfII_10 -dft-II -include rdft/scalar/r2cfII.h */
29
30 /*
31 * This function contains 32 FP additions, 18 FP multiplications,
32 * (or, 14 additions, 0 multiplications, 18 fused multiply/add),
33 * 21 stack variables, 4 constants, and 20 memory accesses
34 */
35 #include "rdft/scalar/r2cfII.h"
36
r2cfII_10(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)37 static void r2cfII_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
40 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
41 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
42 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
43 {
44 INT i;
45 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) {
46 E T1, To, T8, Tt, Ta, Ts, Te, Tq, Th, Tn;
47 T1 = R0[0];
48 To = R1[WS(rs, 2)];
49 {
50 E T2, T3, T4, T5, T6, T7;
51 T2 = R0[WS(rs, 2)];
52 T3 = R0[WS(rs, 3)];
53 T4 = T2 - T3;
54 T5 = R0[WS(rs, 4)];
55 T6 = R0[WS(rs, 1)];
56 T7 = T5 - T6;
57 T8 = T4 + T7;
58 Tt = T5 + T6;
59 Ta = T4 - T7;
60 Ts = T2 + T3;
61 }
62 {
63 E Tc, Td, Tm, Tf, Tg, Tl;
64 Tc = R1[0];
65 Td = R1[WS(rs, 4)];
66 Tm = Tc + Td;
67 Tf = R1[WS(rs, 1)];
68 Tg = R1[WS(rs, 3)];
69 Tl = Tf + Tg;
70 Te = Tc - Td;
71 Tq = Tm + Tl;
72 Th = Tf - Tg;
73 Tn = Tl - Tm;
74 }
75 Cr[WS(csr, 2)] = T1 + T8;
76 Ci[WS(csi, 2)] = Tn - To;
77 {
78 E Ti, Tk, Tb, Tj, T9;
79 Ti = FMA(KP618033988, Th, Te);
80 Tk = FNMS(KP618033988, Te, Th);
81 T9 = FNMS(KP250000000, T8, T1);
82 Tb = FMA(KP559016994, Ta, T9);
83 Tj = FNMS(KP559016994, Ta, T9);
84 Cr[WS(csr, 4)] = FNMS(KP951056516, Ti, Tb);
85 Cr[WS(csr, 3)] = FMA(KP951056516, Tk, Tj);
86 Cr[0] = FMA(KP951056516, Ti, Tb);
87 Cr[WS(csr, 1)] = FNMS(KP951056516, Tk, Tj);
88 }
89 {
90 E Tu, Tw, Tr, Tv, Tp;
91 Tu = FMA(KP618033988, Tt, Ts);
92 Tw = FNMS(KP618033988, Ts, Tt);
93 Tp = FMA(KP250000000, Tn, To);
94 Tr = FMA(KP559016994, Tq, Tp);
95 Tv = FNMS(KP559016994, Tq, Tp);
96 Ci[0] = -(FMA(KP951056516, Tu, Tr));
97 Ci[WS(csi, 3)] = FMA(KP951056516, Tw, Tv);
98 Ci[WS(csi, 4)] = FMS(KP951056516, Tu, Tr);
99 Ci[WS(csi, 1)] = FNMS(KP951056516, Tw, Tv);
100 }
101 }
102 }
103 }
104
105 static const kr2c_desc desc = { 10, "r2cfII_10", { 14, 0, 18, 0 }, &GENUS };
106
X(codelet_r2cfII_10)107 void X(codelet_r2cfII_10) (planner *p) { X(kr2c_register) (p, r2cfII_10, &desc);
108 }
109
110 #else
111
112 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 10 -name r2cfII_10 -dft-II -include rdft/scalar/r2cfII.h */
113
114 /*
115 * This function contains 32 FP additions, 12 FP multiplications,
116 * (or, 26 additions, 6 multiplications, 6 fused multiply/add),
117 * 21 stack variables, 4 constants, and 20 memory accesses
118 */
119 #include "rdft/scalar/r2cfII.h"
120
r2cfII_10(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)121 static void r2cfII_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
122 {
123 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
124 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
125 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
126 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
127 {
128 INT i;
129 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) {
130 E T1, To, T8, Tq, T9, Tp, Te, Ts, Th, Tn;
131 T1 = R0[0];
132 To = R1[WS(rs, 2)];
133 {
134 E T2, T3, T4, T5, T6, T7;
135 T2 = R0[WS(rs, 2)];
136 T3 = R0[WS(rs, 3)];
137 T4 = T2 - T3;
138 T5 = R0[WS(rs, 4)];
139 T6 = R0[WS(rs, 1)];
140 T7 = T5 - T6;
141 T8 = T4 + T7;
142 Tq = T5 + T6;
143 T9 = KP559016994 * (T4 - T7);
144 Tp = T2 + T3;
145 }
146 {
147 E Tc, Td, Tm, Tf, Tg, Tl;
148 Tc = R1[0];
149 Td = R1[WS(rs, 4)];
150 Tm = Tc + Td;
151 Tf = R1[WS(rs, 1)];
152 Tg = R1[WS(rs, 3)];
153 Tl = Tf + Tg;
154 Te = Tc - Td;
155 Ts = KP559016994 * (Tm + Tl);
156 Th = Tf - Tg;
157 Tn = Tl - Tm;
158 }
159 Cr[WS(csr, 2)] = T1 + T8;
160 Ci[WS(csi, 2)] = Tn - To;
161 {
162 E Ti, Tk, Tb, Tj, Ta;
163 Ti = FMA(KP951056516, Te, KP587785252 * Th);
164 Tk = FNMS(KP587785252, Te, KP951056516 * Th);
165 Ta = FNMS(KP250000000, T8, T1);
166 Tb = T9 + Ta;
167 Tj = Ta - T9;
168 Cr[WS(csr, 4)] = Tb - Ti;
169 Cr[WS(csr, 3)] = Tj + Tk;
170 Cr[0] = Tb + Ti;
171 Cr[WS(csr, 1)] = Tj - Tk;
172 }
173 {
174 E Tr, Tw, Tu, Tv, Tt;
175 Tr = FMA(KP951056516, Tp, KP587785252 * Tq);
176 Tw = FNMS(KP587785252, Tp, KP951056516 * Tq);
177 Tt = FMA(KP250000000, Tn, To);
178 Tu = Ts + Tt;
179 Tv = Tt - Ts;
180 Ci[0] = -(Tr + Tu);
181 Ci[WS(csi, 3)] = Tw + Tv;
182 Ci[WS(csi, 4)] = Tr - Tu;
183 Ci[WS(csi, 1)] = Tv - Tw;
184 }
185 }
186 }
187 }
188
189 static const kr2c_desc desc = { 10, "r2cfII_10", { 26, 6, 6, 0 }, &GENUS };
190
X(codelet_r2cfII_10)191 void X(codelet_r2cfII_10) (planner *p) { X(kr2c_register) (p, r2cfII_10, &desc);
192 }
193
194 #endif
195