Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/dft/simd/common/t3fv_5.c @ 167:bd3cc4d1df30

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW 3.3.8 source, and a Linux build
author Chris Cannam <cannam@all-day-breakfast.com>
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
comparison
equal deleted inserted replaced
166:cbd6d7e562c7 167:bd3cc4d1df30
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 May 24 08:05:54 EDT 2018 */
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_twiddle_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 5 -name t3fv_5 -include dft/simd/t3f.h */
29
30 /*
31 * This function contains 22 FP additions, 23 FP multiplications,
32 * (or, 13 additions, 14 multiplications, 9 fused multiply/add),
33 * 24 stack variables, 4 constants, and 10 memory accesses
34 */
35 #include "dft/simd/t3f.h"
36
37 static void t3fv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
38 {
39 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
40 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
41 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
42 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
43 {
44 INT m;
45 R *x;
46 x = ri;
47 for (m = mb, W = W + (mb * ((TWVL / VL) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(5, rs)) {
48 V T2, T5, T6, Ta;
49 T2 = LDW(&(W[0]));
50 T5 = LDW(&(W[TWVL * 2]));
51 T6 = VZMUL(T2, T5);
52 Ta = VZMULJ(T2, T5);
53 {
54 V T1, Tk, Tl, T9, Tf, Tg;
55 T1 = LD(&(x[0]), ms, &(x[0]));
56 {
57 V T4, Te, T8, Tc;
58 {
59 V T3, Td, T7, Tb;
60 T3 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
61 T4 = VZMULJ(T2, T3);
62 Td = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
63 Te = VZMULJ(T5, Td);
64 T7 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
65 T8 = VZMULJ(T6, T7);
66 Tb = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
67 Tc = VZMULJ(Ta, Tb);
68 }
69 Tk = VSUB(T4, T8);
70 Tl = VSUB(Tc, Te);
71 T9 = VADD(T4, T8);
72 Tf = VADD(Tc, Te);
73 Tg = VADD(T9, Tf);
74 }
75 ST(&(x[0]), VADD(T1, Tg), ms, &(x[0]));
76 {
77 V Tm, To, Tj, Tn, Th, Ti;
78 Tm = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Tl, Tk));
79 To = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tk, Tl));
80 Th = VFNMS(LDK(KP250000000), Tg, T1);
81 Ti = VSUB(T9, Tf);
82 Tj = VFMA(LDK(KP559016994), Ti, Th);
83 Tn = VFNMS(LDK(KP559016994), Ti, Th);
84 ST(&(x[WS(rs, 1)]), VFNMSI(Tm, Tj), ms, &(x[WS(rs, 1)]));
85 ST(&(x[WS(rs, 3)]), VFNMSI(To, Tn), ms, &(x[WS(rs, 1)]));
86 ST(&(x[WS(rs, 4)]), VFMAI(Tm, Tj), ms, &(x[0]));
87 ST(&(x[WS(rs, 2)]), VFMAI(To, Tn), ms, &(x[0]));
88 }
89 }
90 }
91 }
92 VLEAVE();
93 }
94
95 static const tw_instr twinstr[] = {
96 VTW(0, 1),
97 VTW(0, 3),
98 {TW_NEXT, VL, 0}
99 };
100
101 static const ct_desc desc = { 5, XSIMD_STRING("t3fv_5"), twinstr, &GENUS, {13, 14, 9, 0}, 0, 0, 0 };
102
103 void XSIMD(codelet_t3fv_5) (planner *p) {
104 X(kdft_dit_register) (p, t3fv_5, &desc);
105 }
106 #else
107
108 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 5 -name t3fv_5 -include dft/simd/t3f.h */
109
110 /*
111 * This function contains 22 FP additions, 18 FP multiplications,
112 * (or, 19 additions, 15 multiplications, 3 fused multiply/add),
113 * 24 stack variables, 4 constants, and 10 memory accesses
114 */
115 #include "dft/simd/t3f.h"
116
117 static void t3fv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
118 {
119 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
120 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
121 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
122 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
123 {
124 INT m;
125 R *x;
126 x = ri;
127 for (m = mb, W = W + (mb * ((TWVL / VL) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(5, rs)) {
128 V T1, T4, T5, T9;
129 T1 = LDW(&(W[0]));
130 T4 = LDW(&(W[TWVL * 2]));
131 T5 = VZMUL(T1, T4);
132 T9 = VZMULJ(T1, T4);
133 {
134 V Tg, Tk, Tl, T8, Te, Th;
135 Tg = LD(&(x[0]), ms, &(x[0]));
136 {
137 V T3, Td, T7, Tb;
138 {
139 V T2, Tc, T6, Ta;
140 T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
141 T3 = VZMULJ(T1, T2);
142 Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
143 Td = VZMULJ(T4, Tc);
144 T6 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
145 T7 = VZMULJ(T5, T6);
146 Ta = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
147 Tb = VZMULJ(T9, Ta);
148 }
149 Tk = VSUB(T3, T7);
150 Tl = VSUB(Tb, Td);
151 T8 = VADD(T3, T7);
152 Te = VADD(Tb, Td);
153 Th = VADD(T8, Te);
154 }
155 ST(&(x[0]), VADD(Tg, Th), ms, &(x[0]));
156 {
157 V Tm, Tn, Tj, To, Tf, Ti;
158 Tm = VBYI(VFMA(LDK(KP951056516), Tk, VMUL(LDK(KP587785252), Tl)));
159 Tn = VBYI(VFNMS(LDK(KP587785252), Tk, VMUL(LDK(KP951056516), Tl)));
160 Tf = VMUL(LDK(KP559016994), VSUB(T8, Te));
161 Ti = VFNMS(LDK(KP250000000), Th, Tg);
162 Tj = VADD(Tf, Ti);
163 To = VSUB(Ti, Tf);
164 ST(&(x[WS(rs, 1)]), VSUB(Tj, Tm), ms, &(x[WS(rs, 1)]));
165 ST(&(x[WS(rs, 3)]), VSUB(To, Tn), ms, &(x[WS(rs, 1)]));
166 ST(&(x[WS(rs, 4)]), VADD(Tm, Tj), ms, &(x[0]));
167 ST(&(x[WS(rs, 2)]), VADD(Tn, To), ms, &(x[0]));
168 }
169 }
170 }
171 }
172 VLEAVE();
173 }
174
175 static const tw_instr twinstr[] = {
176 VTW(0, 1),
177 VTW(0, 3),
178 {TW_NEXT, VL, 0}
179 };
180
181 static const ct_desc desc = { 5, XSIMD_STRING("t3fv_5"), twinstr, &GENUS, {19, 15, 3, 0}, 0, 0, 0 };
182
183 void XSIMD(codelet_t3fv_5) (planner *p) {
184 X(kdft_dit_register) (p, t3fv_5, &desc);
185 }
186 #endif