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 07:05:49 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 8 -name r2cf_8 -include rdft/scalar/r2cf.h */
29
30 /*
31 * This function contains 20 FP additions, 4 FP multiplications,
32 * (or, 16 additions, 0 multiplications, 4 fused multiply/add),
33 * 14 stack variables, 1 constants, and 16 memory accesses
34 */
35 #include "rdft/scalar/r2cf.h"
36
r2cf_8(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)37 static void r2cf_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
40 {
41 INT i;
42 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) {
43 E T3, T7, Td, Tj, T6, Tf, Ta, Ti;
44 {
45 E T1, T2, Tb, Tc;
46 T1 = R0[0];
47 T2 = R0[WS(rs, 2)];
48 T3 = T1 + T2;
49 T7 = T1 - T2;
50 Tb = R1[WS(rs, 3)];
51 Tc = R1[WS(rs, 1)];
52 Td = Tb - Tc;
53 Tj = Tb + Tc;
54 }
55 {
56 E T4, T5, T8, T9;
57 T4 = R0[WS(rs, 1)];
58 T5 = R0[WS(rs, 3)];
59 T6 = T4 + T5;
60 Tf = T4 - T5;
61 T8 = R1[0];
62 T9 = R1[WS(rs, 2)];
63 Ta = T8 - T9;
64 Ti = T8 + T9;
65 }
66 Cr[WS(csr, 2)] = T3 - T6;
67 Ci[WS(csi, 2)] = Tj - Ti;
68 {
69 E Te, Tg, Th, Tk;
70 Te = Ta + Td;
71 Cr[WS(csr, 3)] = FNMS(KP707106781, Te, T7);
72 Cr[WS(csr, 1)] = FMA(KP707106781, Te, T7);
73 Tg = Td - Ta;
74 Ci[WS(csi, 1)] = FMS(KP707106781, Tg, Tf);
75 Ci[WS(csi, 3)] = FMA(KP707106781, Tg, Tf);
76 Th = T3 + T6;
77 Tk = Ti + Tj;
78 Cr[WS(csr, 4)] = Th - Tk;
79 Cr[0] = Th + Tk;
80 }
81 }
82 }
83 }
84
85 static const kr2c_desc desc = { 8, "r2cf_8", { 16, 0, 4, 0 }, &GENUS };
86
X(codelet_r2cf_8)87 void X(codelet_r2cf_8) (planner *p) { X(kr2c_register) (p, r2cf_8, &desc);
88 }
89
90 #else
91
92 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 8 -name r2cf_8 -include rdft/scalar/r2cf.h */
93
94 /*
95 * This function contains 20 FP additions, 2 FP multiplications,
96 * (or, 20 additions, 2 multiplications, 0 fused multiply/add),
97 * 14 stack variables, 1 constants, and 16 memory accesses
98 */
99 #include "rdft/scalar/r2cf.h"
100
r2cf_8(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)101 static void r2cf_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
102 {
103 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
104 {
105 INT i;
106 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) {
107 E T3, T7, Td, Tj, T6, Tg, Ta, Ti;
108 {
109 E T1, T2, Tb, Tc;
110 T1 = R0[0];
111 T2 = R0[WS(rs, 2)];
112 T3 = T1 + T2;
113 T7 = T1 - T2;
114 Tb = R1[WS(rs, 3)];
115 Tc = R1[WS(rs, 1)];
116 Td = Tb - Tc;
117 Tj = Tb + Tc;
118 }
119 {
120 E T4, T5, T8, T9;
121 T4 = R0[WS(rs, 1)];
122 T5 = R0[WS(rs, 3)];
123 T6 = T4 + T5;
124 Tg = T4 - T5;
125 T8 = R1[0];
126 T9 = R1[WS(rs, 2)];
127 Ta = T8 - T9;
128 Ti = T8 + T9;
129 }
130 Cr[WS(csr, 2)] = T3 - T6;
131 Ci[WS(csi, 2)] = Tj - Ti;
132 {
133 E Te, Tf, Th, Tk;
134 Te = KP707106781 * (Ta + Td);
135 Cr[WS(csr, 3)] = T7 - Te;
136 Cr[WS(csr, 1)] = T7 + Te;
137 Tf = KP707106781 * (Td - Ta);
138 Ci[WS(csi, 1)] = Tf - Tg;
139 Ci[WS(csi, 3)] = Tg + Tf;
140 Th = T3 + T6;
141 Tk = Ti + Tj;
142 Cr[WS(csr, 4)] = Th - Tk;
143 Cr[0] = Th + Tk;
144 }
145 }
146 }
147 }
148
149 static const kr2c_desc desc = { 8, "r2cf_8", { 20, 2, 0, 0 }, &GENUS };
150
X(codelet_r2cf_8)151 void X(codelet_r2cf_8) (planner *p) { X(kr2c_register) (p, r2cf_8, &desc);
152 }
153
154 #endif
155