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 07:06:25 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 8 -name r2cb_8 -include rdft/scalar/r2cb.h */
29
30 /*
31 * This function contains 20 FP additions, 12 FP multiplications,
32 * (or, 8 additions, 0 multiplications, 12 fused multiply/add),
33 * 19 stack variables, 2 constants, and 16 memory accesses
34 */
35 #include "rdft/scalar/r2cb.h"
36
r2cb_8(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)37 static void r2cb_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
40 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
41 {
42 INT i;
43 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) {
44 E T4, Ta, T3, T9, T8, Tc, Tf, Tk, T1, T2, T5, Tj;
45 T4 = Cr[WS(csr, 2)];
46 Ta = Ci[WS(csi, 2)];
47 T1 = Cr[0];
48 T2 = Cr[WS(csr, 4)];
49 T3 = T1 + T2;
50 T9 = T1 - T2;
51 {
52 E T6, T7, Td, Te;
53 T6 = Cr[WS(csr, 1)];
54 T7 = Cr[WS(csr, 3)];
55 T8 = T6 + T7;
56 Tc = T6 - T7;
57 Td = Ci[WS(csi, 1)];
58 Te = Ci[WS(csi, 3)];
59 Tf = Td + Te;
60 Tk = Td - Te;
61 }
62 T5 = FMA(KP2_000000000, T4, T3);
63 R0[WS(rs, 2)] = FNMS(KP2_000000000, T8, T5);
64 R0[0] = FMA(KP2_000000000, T8, T5);
65 Tj = FNMS(KP2_000000000, T4, T3);
66 R0[WS(rs, 1)] = FNMS(KP2_000000000, Tk, Tj);
67 R0[WS(rs, 3)] = FMA(KP2_000000000, Tk, Tj);
68 {
69 E Tb, Tg, Th, Ti;
70 Tb = FNMS(KP2_000000000, Ta, T9);
71 Tg = Tc - Tf;
72 R1[WS(rs, 2)] = FNMS(KP1_414213562, Tg, Tb);
73 R1[0] = FMA(KP1_414213562, Tg, Tb);
74 Th = FMA(KP2_000000000, Ta, T9);
75 Ti = Tc + Tf;
76 R1[WS(rs, 1)] = FNMS(KP1_414213562, Ti, Th);
77 R1[WS(rs, 3)] = FMA(KP1_414213562, Ti, Th);
78 }
79 }
80 }
81 }
82
83 static const kr2c_desc desc = { 8, "r2cb_8", { 8, 0, 12, 0 }, &GENUS };
84
X(codelet_r2cb_8)85 void X(codelet_r2cb_8) (planner *p) { X(kr2c_register) (p, r2cb_8, &desc);
86 }
87
88 #else
89
90 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 8 -name r2cb_8 -include rdft/scalar/r2cb.h */
91
92 /*
93 * This function contains 20 FP additions, 6 FP multiplications,
94 * (or, 20 additions, 6 multiplications, 0 fused multiply/add),
95 * 21 stack variables, 2 constants, and 16 memory accesses
96 */
97 #include "rdft/scalar/r2cb.h"
98
r2cb_8(R * R0,R * R1,R * Cr,R * Ci,stride rs,stride csr,stride csi,INT v,INT ivs,INT ovs)99 static void r2cb_8(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
100 {
101 DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
102 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
103 {
104 INT i;
105 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(32, rs), MAKE_VOLATILE_STRIDE(32, csr), MAKE_VOLATILE_STRIDE(32, csi)) {
106 E T5, Tg, T3, Te, T9, Ti, Td, Tj, T6, Ta;
107 {
108 E T4, Tf, T1, T2;
109 T4 = Cr[WS(csr, 2)];
110 T5 = KP2_000000000 * T4;
111 Tf = Ci[WS(csi, 2)];
112 Tg = KP2_000000000 * Tf;
113 T1 = Cr[0];
114 T2 = Cr[WS(csr, 4)];
115 T3 = T1 + T2;
116 Te = T1 - T2;
117 {
118 E T7, T8, Tb, Tc;
119 T7 = Cr[WS(csr, 1)];
120 T8 = Cr[WS(csr, 3)];
121 T9 = KP2_000000000 * (T7 + T8);
122 Ti = T7 - T8;
123 Tb = Ci[WS(csi, 1)];
124 Tc = Ci[WS(csi, 3)];
125 Td = KP2_000000000 * (Tb - Tc);
126 Tj = Tb + Tc;
127 }
128 }
129 T6 = T3 + T5;
130 R0[WS(rs, 2)] = T6 - T9;
131 R0[0] = T6 + T9;
132 Ta = T3 - T5;
133 R0[WS(rs, 1)] = Ta - Td;
134 R0[WS(rs, 3)] = Ta + Td;
135 {
136 E Th, Tk, Tl, Tm;
137 Th = Te - Tg;
138 Tk = KP1_414213562 * (Ti - Tj);
139 R1[WS(rs, 2)] = Th - Tk;
140 R1[0] = Th + Tk;
141 Tl = Te + Tg;
142 Tm = KP1_414213562 * (Ti + Tj);
143 R1[WS(rs, 1)] = Tl - Tm;
144 R1[WS(rs, 3)] = Tl + Tm;
145 }
146 }
147 }
148 }
149
150 static const kr2c_desc desc = { 8, "r2cb_8", { 20, 6, 0, 0 }, &GENUS };
151
X(codelet_r2cb_8)152 void X(codelet_r2cb_8) (planner *p) { X(kr2c_register) (p, r2cb_8, &desc);
153 }
154
155 #endif
156