Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/dft/simd/common/n1fv_10.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 08:04:51 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_notw_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 10 -name n1fv_10 -include dft/simd/n1f.h */
29
30 /*
31 * This function contains 42 FP additions, 22 FP multiplications,
32 * (or, 24 additions, 4 multiplications, 18 fused multiply/add),
33 * 33 stack variables, 4 constants, and 20 memory accesses
34 */
35 #include "dft/simd/n1f.h"
36
37 static void n1fv_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
40 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
41 DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
42 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
43 {
44 INT i;
45 const R *xi;
46 R *xo;
47 xi = ri;
48 xo = ro;
49 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(20, is), MAKE_VOLATILE_STRIDE(20, os)) {
50 V T3, Tr, Tm, Tn, TD, TC, Tu, Tx, Ty, Ta, Th, Ti, T1, T2;
51 T1 = LD(&(xi[0]), ivs, &(xi[0]));
52 T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
53 T3 = VSUB(T1, T2);
54 Tr = VADD(T1, T2);
55 {
56 V T6, Ts, Tg, Tw, T9, Tt, Td, Tv;
57 {
58 V T4, T5, Te, Tf;
59 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
60 T5 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
61 T6 = VSUB(T4, T5);
62 Ts = VADD(T4, T5);
63 Te = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
64 Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
65 Tg = VSUB(Te, Tf);
66 Tw = VADD(Te, Tf);
67 }
68 {
69 V T7, T8, Tb, Tc;
70 T7 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
71 T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
72 T9 = VSUB(T7, T8);
73 Tt = VADD(T7, T8);
74 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
75 Tc = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
76 Td = VSUB(Tb, Tc);
77 Tv = VADD(Tb, Tc);
78 }
79 Tm = VSUB(T6, T9);
80 Tn = VSUB(Td, Tg);
81 TD = VSUB(Ts, Tt);
82 TC = VSUB(Tv, Tw);
83 Tu = VADD(Ts, Tt);
84 Tx = VADD(Tv, Tw);
85 Ty = VADD(Tu, Tx);
86 Ta = VADD(T6, T9);
87 Th = VADD(Td, Tg);
88 Ti = VADD(Ta, Th);
89 }
90 ST(&(xo[WS(os, 5)]), VADD(T3, Ti), ovs, &(xo[WS(os, 1)]));
91 ST(&(xo[0]), VADD(Tr, Ty), ovs, &(xo[0]));
92 {
93 V To, Tq, Tl, Tp, Tj, Tk;
94 To = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Tn, Tm));
95 Tq = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tm, Tn));
96 Tj = VFNMS(LDK(KP250000000), Ti, T3);
97 Tk = VSUB(Ta, Th);
98 Tl = VFMA(LDK(KP559016994), Tk, Tj);
99 Tp = VFNMS(LDK(KP559016994), Tk, Tj);
100 ST(&(xo[WS(os, 1)]), VFNMSI(To, Tl), ovs, &(xo[WS(os, 1)]));
101 ST(&(xo[WS(os, 7)]), VFMAI(Tq, Tp), ovs, &(xo[WS(os, 1)]));
102 ST(&(xo[WS(os, 9)]), VFMAI(To, Tl), ovs, &(xo[WS(os, 1)]));
103 ST(&(xo[WS(os, 3)]), VFNMSI(Tq, Tp), ovs, &(xo[WS(os, 1)]));
104 }
105 {
106 V TE, TG, TB, TF, Tz, TA;
107 TE = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TD, TC));
108 TG = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TC, TD));
109 Tz = VFNMS(LDK(KP250000000), Ty, Tr);
110 TA = VSUB(Tu, Tx);
111 TB = VFNMS(LDK(KP559016994), TA, Tz);
112 TF = VFMA(LDK(KP559016994), TA, Tz);
113 ST(&(xo[WS(os, 2)]), VFMAI(TE, TB), ovs, &(xo[0]));
114 ST(&(xo[WS(os, 6)]), VFNMSI(TG, TF), ovs, &(xo[0]));
115 ST(&(xo[WS(os, 8)]), VFNMSI(TE, TB), ovs, &(xo[0]));
116 ST(&(xo[WS(os, 4)]), VFMAI(TG, TF), ovs, &(xo[0]));
117 }
118 }
119 }
120 VLEAVE();
121 }
122
123 static const kdft_desc desc = { 10, XSIMD_STRING("n1fv_10"), {24, 4, 18, 0}, &GENUS, 0, 0, 0, 0 };
124
125 void XSIMD(codelet_n1fv_10) (planner *p) {
126 X(kdft_register) (p, n1fv_10, &desc);
127 }
128
129 #else
130
131 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 10 -name n1fv_10 -include dft/simd/n1f.h */
132
133 /*
134 * This function contains 42 FP additions, 12 FP multiplications,
135 * (or, 36 additions, 6 multiplications, 6 fused multiply/add),
136 * 33 stack variables, 4 constants, and 20 memory accesses
137 */
138 #include "dft/simd/n1f.h"
139
140 static void n1fv_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
141 {
142 DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
143 DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
144 DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
145 DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
146 {
147 INT i;
148 const R *xi;
149 R *xo;
150 xi = ri;
151 xo = ro;
152 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(20, is), MAKE_VOLATILE_STRIDE(20, os)) {
153 V Ti, Ty, Tm, Tn, Tw, Tt, Tz, TA, TB, T7, Te, Tj, Tg, Th;
154 Tg = LD(&(xi[0]), ivs, &(xi[0]));
155 Th = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
156 Ti = VSUB(Tg, Th);
157 Ty = VADD(Tg, Th);
158 {
159 V T3, Tu, Td, Ts, T6, Tv, Ta, Tr;
160 {
161 V T1, T2, Tb, Tc;
162 T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
163 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
164 T3 = VSUB(T1, T2);
165 Tu = VADD(T1, T2);
166 Tb = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
167 Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
168 Td = VSUB(Tb, Tc);
169 Ts = VADD(Tb, Tc);
170 }
171 {
172 V T4, T5, T8, T9;
173 T4 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
174 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
175 T6 = VSUB(T4, T5);
176 Tv = VADD(T4, T5);
177 T8 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
178 T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
179 Ta = VSUB(T8, T9);
180 Tr = VADD(T8, T9);
181 }
182 Tm = VSUB(T3, T6);
183 Tn = VSUB(Ta, Td);
184 Tw = VSUB(Tu, Tv);
185 Tt = VSUB(Tr, Ts);
186 Tz = VADD(Tu, Tv);
187 TA = VADD(Tr, Ts);
188 TB = VADD(Tz, TA);
189 T7 = VADD(T3, T6);
190 Te = VADD(Ta, Td);
191 Tj = VADD(T7, Te);
192 }
193 ST(&(xo[WS(os, 5)]), VADD(Ti, Tj), ovs, &(xo[WS(os, 1)]));
194 ST(&(xo[0]), VADD(Ty, TB), ovs, &(xo[0]));
195 {
196 V To, Tq, Tl, Tp, Tf, Tk;
197 To = VBYI(VFMA(LDK(KP951056516), Tm, VMUL(LDK(KP587785252), Tn)));
198 Tq = VBYI(VFNMS(LDK(KP587785252), Tm, VMUL(LDK(KP951056516), Tn)));
199 Tf = VMUL(LDK(KP559016994), VSUB(T7, Te));
200 Tk = VFNMS(LDK(KP250000000), Tj, Ti);
201 Tl = VADD(Tf, Tk);
202 Tp = VSUB(Tk, Tf);
203 ST(&(xo[WS(os, 1)]), VSUB(Tl, To), ovs, &(xo[WS(os, 1)]));
204 ST(&(xo[WS(os, 7)]), VADD(Tq, Tp), ovs, &(xo[WS(os, 1)]));
205 ST(&(xo[WS(os, 9)]), VADD(To, Tl), ovs, &(xo[WS(os, 1)]));
206 ST(&(xo[WS(os, 3)]), VSUB(Tp, Tq), ovs, &(xo[WS(os, 1)]));
207 }
208 {
209 V Tx, TF, TE, TG, TC, TD;
210 Tx = VBYI(VFNMS(LDK(KP587785252), Tw, VMUL(LDK(KP951056516), Tt)));
211 TF = VBYI(VFMA(LDK(KP951056516), Tw, VMUL(LDK(KP587785252), Tt)));
212 TC = VFNMS(LDK(KP250000000), TB, Ty);
213 TD = VMUL(LDK(KP559016994), VSUB(Tz, TA));
214 TE = VSUB(TC, TD);
215 TG = VADD(TD, TC);
216 ST(&(xo[WS(os, 2)]), VADD(Tx, TE), ovs, &(xo[0]));
217 ST(&(xo[WS(os, 6)]), VSUB(TG, TF), ovs, &(xo[0]));
218 ST(&(xo[WS(os, 8)]), VSUB(TE, Tx), ovs, &(xo[0]));
219 ST(&(xo[WS(os, 4)]), VADD(TF, TG), ovs, &(xo[0]));
220 }
221 }
222 }
223 VLEAVE();
224 }
225
226 static const kdft_desc desc = { 10, XSIMD_STRING("n1fv_10"), {36, 6, 6, 0}, &GENUS, 0, 0, 0, 0 };
227
228 void XSIMD(codelet_n1fv_10) (planner *p) {
229 X(kdft_register) (p, n1fv_10, &desc);
230 }
231
232 #endif