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 07:04:43 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_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 4 -name n1bv_4 -include dft/simd/n1b.h */
29
30 /*
31 * This function contains 8 FP additions, 2 FP multiplications,
32 * (or, 6 additions, 0 multiplications, 2 fused multiply/add),
33 * 11 stack variables, 0 constants, and 8 memory accesses
34 */
35 #include "dft/simd/n1b.h"
36
n1bv_4(const R * ri,const R * ii,R * ro,R * io,stride is,stride os,INT v,INT ivs,INT ovs)37 static void n1bv_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 {
40 INT i;
41 const R *xi;
42 R *xo;
43 xi = ii;
44 xo = io;
45 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(8, is), MAKE_VOLATILE_STRIDE(8, os)) {
46 V T3, T7, T6, T8;
47 {
48 V T1, T2, T4, T5;
49 T1 = LD(&(xi[0]), ivs, &(xi[0]));
50 T2 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
51 T3 = VSUB(T1, T2);
52 T7 = VADD(T1, T2);
53 T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
54 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
55 T6 = VSUB(T4, T5);
56 T8 = VADD(T4, T5);
57 }
58 ST(&(xo[WS(os, 3)]), VFNMSI(T6, T3), ovs, &(xo[WS(os, 1)]));
59 ST(&(xo[0]), VADD(T7, T8), ovs, &(xo[0]));
60 ST(&(xo[WS(os, 1)]), VFMAI(T6, T3), ovs, &(xo[WS(os, 1)]));
61 ST(&(xo[WS(os, 2)]), VSUB(T7, T8), ovs, &(xo[0]));
62 }
63 }
64 VLEAVE();
65 }
66
67 static const kdft_desc desc = { 4, XSIMD_STRING("n1bv_4"), { 6, 0, 2, 0 }, &GENUS, 0, 0, 0, 0 };
68
XSIMD(codelet_n1bv_4)69 void XSIMD(codelet_n1bv_4) (planner *p) { X(kdft_register) (p, n1bv_4, &desc);
70 }
71
72 #else
73
74 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 4 -name n1bv_4 -include dft/simd/n1b.h */
75
76 /*
77 * This function contains 8 FP additions, 0 FP multiplications,
78 * (or, 8 additions, 0 multiplications, 0 fused multiply/add),
79 * 11 stack variables, 0 constants, and 8 memory accesses
80 */
81 #include "dft/simd/n1b.h"
82
n1bv_4(const R * ri,const R * ii,R * ro,R * io,stride is,stride os,INT v,INT ivs,INT ovs)83 static void n1bv_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
84 {
85 {
86 INT i;
87 const R *xi;
88 R *xo;
89 xi = ii;
90 xo = io;
91 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(8, is), MAKE_VOLATILE_STRIDE(8, os)) {
92 V T3, T7, T6, T8;
93 {
94 V T1, T2, T4, T5;
95 T1 = LD(&(xi[0]), ivs, &(xi[0]));
96 T2 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
97 T3 = VSUB(T1, T2);
98 T7 = VADD(T1, T2);
99 T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
100 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
101 T6 = VBYI(VSUB(T4, T5));
102 T8 = VADD(T4, T5);
103 }
104 ST(&(xo[WS(os, 3)]), VSUB(T3, T6), ovs, &(xo[WS(os, 1)]));
105 ST(&(xo[0]), VADD(T7, T8), ovs, &(xo[0]));
106 ST(&(xo[WS(os, 1)]), VADD(T3, T6), ovs, &(xo[WS(os, 1)]));
107 ST(&(xo[WS(os, 2)]), VSUB(T7, T8), ovs, &(xo[0]));
108 }
109 }
110 VLEAVE();
111 }
112
113 static const kdft_desc desc = { 4, XSIMD_STRING("n1bv_4"), { 8, 0, 0, 0 }, &GENUS, 0, 0, 0, 0 };
114
XSIMD(codelet_n1bv_4)115 void XSIMD(codelet_n1bv_4) (planner *p) { X(kdft_register) (p, n1bv_4, &desc);
116 }
117
118 #endif
119