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 07:04:08 EST 2020 */
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 3 -name n1_3 -include dft/scalar/n.h */
29 
30 /*
31  * This function contains 12 FP additions, 6 FP multiplications,
32  * (or, 6 additions, 0 multiplications, 6 fused multiply/add),
33  * 15 stack variables, 2 constants, and 12 memory accesses
34  */
35 #include "dft/scalar/n.h"
36 
n1_3(const R * ri,const R * ii,R * ro,R * io,stride is,stride os,INT v,INT ivs,INT ovs)37 static void n1_3(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(12, is), MAKE_VOLATILE_STRIDE(12, os)) {
44 	       E T1, T9, T4, Tc, T8, Ta, T5, Tb;
45 	       T1 = ri[0];
46 	       T9 = ii[0];
47 	       {
48 		    E T2, T3, T6, T7;
49 		    T2 = ri[WS(is, 1)];
50 		    T3 = ri[WS(is, 2)];
51 		    T4 = T2 + T3;
52 		    Tc = T3 - T2;
53 		    T6 = ii[WS(is, 1)];
54 		    T7 = ii[WS(is, 2)];
55 		    T8 = T6 - T7;
56 		    Ta = T6 + T7;
57 	       }
58 	       ro[0] = T1 + T4;
59 	       io[0] = T9 + Ta;
60 	       T5 = FNMS(KP500000000, T4, T1);
61 	       ro[WS(os, 2)] = FNMS(KP866025403, T8, T5);
62 	       ro[WS(os, 1)] = FMA(KP866025403, T8, T5);
63 	       Tb = FNMS(KP500000000, Ta, T9);
64 	       io[WS(os, 1)] = FMA(KP866025403, Tc, Tb);
65 	       io[WS(os, 2)] = FNMS(KP866025403, Tc, Tb);
66 	  }
67      }
68 }
69 
70 static const kdft_desc desc = { 3, "n1_3", { 6, 0, 6, 0 }, &GENUS, 0, 0, 0, 0 };
71 
X(codelet_n1_3)72 void X(codelet_n1_3) (planner *p) { X(kdft_register) (p, n1_3, &desc);
73 }
74 
75 #else
76 
77 /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 3 -name n1_3 -include dft/scalar/n.h */
78 
79 /*
80  * This function contains 12 FP additions, 4 FP multiplications,
81  * (or, 10 additions, 2 multiplications, 2 fused multiply/add),
82  * 15 stack variables, 2 constants, and 12 memory accesses
83  */
84 #include "dft/scalar/n.h"
85 
n1_3(const R * ri,const R * ii,R * ro,R * io,stride is,stride os,INT v,INT ivs,INT ovs)86 static void n1_3(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
87 {
88      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
89      DK(KP866025403, +0.866025403784438646763723170752936183471402627);
90      {
91 	  INT i;
92 	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(12, is), MAKE_VOLATILE_STRIDE(12, os)) {
93 	       E T1, Ta, T4, T9, T8, Tb, T5, Tc;
94 	       T1 = ri[0];
95 	       Ta = ii[0];
96 	       {
97 		    E T2, T3, T6, T7;
98 		    T2 = ri[WS(is, 1)];
99 		    T3 = ri[WS(is, 2)];
100 		    T4 = T2 + T3;
101 		    T9 = KP866025403 * (T3 - T2);
102 		    T6 = ii[WS(is, 1)];
103 		    T7 = ii[WS(is, 2)];
104 		    T8 = KP866025403 * (T6 - T7);
105 		    Tb = T6 + T7;
106 	       }
107 	       ro[0] = T1 + T4;
108 	       io[0] = Ta + Tb;
109 	       T5 = FNMS(KP500000000, T4, T1);
110 	       ro[WS(os, 2)] = T5 - T8;
111 	       ro[WS(os, 1)] = T5 + T8;
112 	       Tc = FNMS(KP500000000, Tb, Ta);
113 	       io[WS(os, 1)] = T9 + Tc;
114 	       io[WS(os, 2)] = Tc - T9;
115 	  }
116      }
117 }
118 
119 static const kdft_desc desc = { 3, "n1_3", { 10, 2, 2, 0 }, &GENUS, 0, 0, 0, 0 };
120 
X(codelet_n1_3)121 void X(codelet_n1_3) (planner *p) { X(kdft_register) (p, n1_3, &desc);
122 }
123 
124 #endif
125