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 07:06:37 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_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cbIII_10 -dft-III -include rdft/scalar/r2cbIII.h */
29
30 /*
31 * This function contains 32 FP additions, 28 FP multiplications,
32 * (or, 14 additions, 10 multiplications, 18 fused multiply/add),
33 * 22 stack variables, 5 constants, and 20 memory accesses
34 */
35 #include "rdft/scalar/r2cbIII.h"
36
r2cbIII_10(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)37 static void r2cbIII_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 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
44 {
45 INT i;
46 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) {
47 E T1, To, T8, Tt, Ta, Ts, Te, Tq, Th, Tn;
48 T1 = Cr[WS(csr, 2)];
49 To = Ci[WS(csi, 2)];
50 {
51 E T2, T3, T4, T5, T6, T7;
52 T2 = Cr[WS(csr, 4)];
53 T3 = Cr[0];
54 T4 = T2 + T3;
55 T5 = Cr[WS(csr, 3)];
56 T6 = Cr[WS(csr, 1)];
57 T7 = T5 + T6;
58 T8 = T4 + T7;
59 Tt = T5 - T6;
60 Ta = T7 - T4;
61 Ts = T2 - T3;
62 }
63 {
64 E Tc, Td, Tl, Tf, Tg, Tm;
65 Tc = Ci[WS(csi, 3)];
66 Td = Ci[WS(csi, 1)];
67 Tl = Tc + Td;
68 Tf = Ci[WS(csi, 4)];
69 Tg = Ci[0];
70 Tm = Tf + Tg;
71 Te = Tc - Td;
72 Tq = Tl + Tm;
73 Th = Tf - Tg;
74 Tn = Tl - Tm;
75 }
76 R0[0] = KP2_000000000 * (T1 + T8);
77 R1[WS(rs, 2)] = KP2_000000000 * (Tn - To);
78 {
79 E Ti, Tk, Tb, Tj, T9;
80 Ti = FMA(KP618033988, Th, Te);
81 Tk = FNMS(KP618033988, Te, Th);
82 T9 = FMS(KP250000000, T8, T1);
83 Tb = FNMS(KP559016994, Ta, T9);
84 Tj = FMA(KP559016994, Ta, T9);
85 R0[WS(rs, 1)] = KP2_000000000 * (FMA(KP951056516, Ti, Tb));
86 R0[WS(rs, 3)] = KP2_000000000 * (FMA(KP951056516, Tk, Tj));
87 R0[WS(rs, 4)] = -(KP2_000000000 * (FNMS(KP951056516, Ti, Tb)));
88 R0[WS(rs, 2)] = -(KP2_000000000 * (FNMS(KP951056516, Tk, Tj)));
89 }
90 {
91 E Tu, Tw, Tr, Tv, Tp;
92 Tu = FMA(KP618033988, Tt, Ts);
93 Tw = FNMS(KP618033988, Ts, Tt);
94 Tp = FMA(KP250000000, Tn, To);
95 Tr = FMA(KP559016994, Tq, Tp);
96 Tv = FNMS(KP559016994, Tq, Tp);
97 R1[0] = -(KP2_000000000 * (FMA(KP951056516, Tu, Tr)));
98 R1[WS(rs, 3)] = KP2_000000000 * (FNMS(KP951056516, Tw, Tv));
99 R1[WS(rs, 4)] = -(KP2_000000000 * (FNMS(KP951056516, Tu, Tr)));
100 R1[WS(rs, 1)] = KP2_000000000 * (FMA(KP951056516, Tw, Tv));
101 }
102 }
103 }
104 }
105
106 static const kr2c_desc desc = { 10, "r2cbIII_10", { 14, 10, 18, 0 }, &GENUS };
107
X(codelet_r2cbIII_10)108 void X(codelet_r2cbIII_10) (planner *p) { X(kr2c_register) (p, r2cbIII_10, &desc);
109 }
110
111 #else
112
113 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 10 -name r2cbIII_10 -dft-III -include rdft/scalar/r2cbIII.h */
114
115 /*
116 * This function contains 32 FP additions, 16 FP multiplications,
117 * (or, 26 additions, 10 multiplications, 6 fused multiply/add),
118 * 22 stack variables, 5 constants, and 20 memory accesses
119 */
120 #include "rdft/scalar/r2cbIII.h"
121
r2cbIII_10(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)122 static void r2cbIII_10(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
123 {
124 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
125 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
126 DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
127 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
128 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
129 {
130 INT i;
131 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(40, rs), MAKE_VOLATILE_STRIDE(40, csr), MAKE_VOLATILE_STRIDE(40, csi)) {
132 E T1, To, T8, Tq, Ta, Tp, Te, Ts, Th, Tn;
133 T1 = Cr[WS(csr, 2)];
134 To = Ci[WS(csi, 2)];
135 {
136 E T2, T3, T4, T5, T6, T7;
137 T2 = Cr[WS(csr, 4)];
138 T3 = Cr[0];
139 T4 = T2 + T3;
140 T5 = Cr[WS(csr, 3)];
141 T6 = Cr[WS(csr, 1)];
142 T7 = T5 + T6;
143 T8 = T4 + T7;
144 Tq = T5 - T6;
145 Ta = KP1_118033988 * (T7 - T4);
146 Tp = T2 - T3;
147 }
148 {
149 E Tc, Td, Tm, Tf, Tg, Tl;
150 Tc = Ci[WS(csi, 4)];
151 Td = Ci[0];
152 Tm = Tc + Td;
153 Tf = Ci[WS(csi, 1)];
154 Tg = Ci[WS(csi, 3)];
155 Tl = Tg + Tf;
156 Te = Tc - Td;
157 Ts = KP1_118033988 * (Tl + Tm);
158 Th = Tf - Tg;
159 Tn = Tl - Tm;
160 }
161 R0[0] = KP2_000000000 * (T1 + T8);
162 R1[WS(rs, 2)] = KP2_000000000 * (Tn - To);
163 {
164 E Ti, Tj, Tb, Tk, T9;
165 Ti = FNMS(KP1_902113032, Th, KP1_175570504 * Te);
166 Tj = FMA(KP1_175570504, Th, KP1_902113032 * Te);
167 T9 = FNMS(KP2_000000000, T1, KP500000000 * T8);
168 Tb = T9 - Ta;
169 Tk = T9 + Ta;
170 R0[WS(rs, 1)] = Tb + Ti;
171 R0[WS(rs, 3)] = Tk + Tj;
172 R0[WS(rs, 4)] = Ti - Tb;
173 R0[WS(rs, 2)] = Tj - Tk;
174 }
175 {
176 E Tr, Tv, Tu, Tw, Tt;
177 Tr = FMA(KP1_902113032, Tp, KP1_175570504 * Tq);
178 Tv = FNMS(KP1_175570504, Tp, KP1_902113032 * Tq);
179 Tt = FMA(KP500000000, Tn, KP2_000000000 * To);
180 Tu = Ts + Tt;
181 Tw = Tt - Ts;
182 R1[0] = -(Tr + Tu);
183 R1[WS(rs, 3)] = Tw - Tv;
184 R1[WS(rs, 4)] = Tr - Tu;
185 R1[WS(rs, 1)] = Tv + Tw;
186 }
187 }
188 }
189 }
190
191 static const kr2c_desc desc = { 10, "r2cbIII_10", { 26, 10, 6, 0 }, &GENUS };
192
X(codelet_r2cbIII_10)193 void X(codelet_r2cbIII_10) (planner *p) { X(kr2c_register) (p, r2cbIII_10, &desc);
194 }
195
196 #endif
197