Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/rdft/scalar/r2cf/r2cf_16.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 08:06:26 EDT 2018 */
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_r2cf.native -fma -compact -variables 4 -pipeline-latency 4 -n 16 -name r2cf_16 -include rdft/scalar/r2cf.h */
29
30 /*
31 * This function contains 58 FP additions, 20 FP multiplications,
32 * (or, 38 additions, 0 multiplications, 20 fused multiply/add),
33 * 34 stack variables, 3 constants, and 32 memory accesses
34 */
35 #include "rdft/scalar/r2cf.h"
36
37 static void r2cf_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP923879532, +0.923879532511286756128183189396788286822416626);
40 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
41 DK(KP414213562, +0.414213562373095048801688724209698078569671875);
42 {
43 INT i;
44 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) {
45 E T3, T6, T7, TN, TB, Ta, Td, Te, TO, TE, Tm, TT, Ty, TI, Tt;
46 E TS, Tz, TL, TC, TD, TR, TU;
47 {
48 E T1, T2, T4, T5;
49 T1 = R0[0];
50 T2 = R0[WS(rs, 4)];
51 T3 = T1 + T2;
52 T4 = R0[WS(rs, 2)];
53 T5 = R0[WS(rs, 6)];
54 T6 = T4 + T5;
55 T7 = T3 + T6;
56 TN = T4 - T5;
57 TB = T1 - T2;
58 }
59 {
60 E T8, T9, Tb, Tc;
61 T8 = R0[WS(rs, 1)];
62 T9 = R0[WS(rs, 5)];
63 Ta = T8 + T9;
64 TC = T8 - T9;
65 Tb = R0[WS(rs, 7)];
66 Tc = R0[WS(rs, 3)];
67 Td = Tb + Tc;
68 TD = Tb - Tc;
69 }
70 Te = Ta + Td;
71 TO = TD - TC;
72 TE = TC + TD;
73 {
74 E Ti, TG, Tl, TH;
75 {
76 E Tg, Th, Tj, Tk;
77 Tg = R1[0];
78 Th = R1[WS(rs, 4)];
79 Ti = Tg + Th;
80 TG = Tg - Th;
81 Tj = R1[WS(rs, 2)];
82 Tk = R1[WS(rs, 6)];
83 Tl = Tj + Tk;
84 TH = Tj - Tk;
85 }
86 Tm = Ti - Tl;
87 TT = FMA(KP414213562, TG, TH);
88 Ty = Ti + Tl;
89 TI = FNMS(KP414213562, TH, TG);
90 }
91 {
92 E Tp, TJ, Ts, TK;
93 {
94 E Tn, To, Tq, Tr;
95 Tn = R1[WS(rs, 7)];
96 To = R1[WS(rs, 3)];
97 Tp = Tn + To;
98 TJ = Tn - To;
99 Tq = R1[WS(rs, 1)];
100 Tr = R1[WS(rs, 5)];
101 Ts = Tq + Tr;
102 TK = Tr - Tq;
103 }
104 Tt = Tp - Ts;
105 TS = FMA(KP414213562, TJ, TK);
106 Tz = Tp + Ts;
107 TL = FNMS(KP414213562, TK, TJ);
108 }
109 Cr[WS(csr, 4)] = T7 - Te;
110 Ci[WS(csi, 4)] = Tz - Ty;
111 {
112 E Tf, Tu, Tv, Tw;
113 Tf = T3 - T6;
114 Tu = Tm + Tt;
115 Cr[WS(csr, 6)] = FNMS(KP707106781, Tu, Tf);
116 Cr[WS(csr, 2)] = FMA(KP707106781, Tu, Tf);
117 Tv = Td - Ta;
118 Tw = Tt - Tm;
119 Ci[WS(csi, 2)] = FMA(KP707106781, Tw, Tv);
120 Ci[WS(csi, 6)] = FMS(KP707106781, Tw, Tv);
121 }
122 {
123 E Tx, TA, TF, TM;
124 Tx = T7 + Te;
125 TA = Ty + Tz;
126 Cr[WS(csr, 8)] = Tx - TA;
127 Cr[0] = Tx + TA;
128 TF = FMA(KP707106781, TE, TB);
129 TM = TI + TL;
130 Cr[WS(csr, 7)] = FNMS(KP923879532, TM, TF);
131 Cr[WS(csr, 1)] = FMA(KP923879532, TM, TF);
132 }
133 TR = FNMS(KP707106781, TO, TN);
134 TU = TS - TT;
135 Ci[WS(csi, 1)] = FMS(KP923879532, TU, TR);
136 Ci[WS(csi, 7)] = FMA(KP923879532, TU, TR);
137 {
138 E TV, TW, TP, TQ;
139 TV = FNMS(KP707106781, TE, TB);
140 TW = TT + TS;
141 Cr[WS(csr, 5)] = FNMS(KP923879532, TW, TV);
142 Cr[WS(csr, 3)] = FMA(KP923879532, TW, TV);
143 TP = FMA(KP707106781, TO, TN);
144 TQ = TL - TI;
145 Ci[WS(csi, 3)] = FMA(KP923879532, TQ, TP);
146 Ci[WS(csi, 5)] = FMS(KP923879532, TQ, TP);
147 }
148 }
149 }
150 }
151
152 static const kr2c_desc desc = { 16, "r2cf_16", {38, 0, 20, 0}, &GENUS };
153
154 void X(codelet_r2cf_16) (planner *p) {
155 X(kr2c_register) (p, r2cf_16, &desc);
156 }
157
158 #else
159
160 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 16 -name r2cf_16 -include rdft/scalar/r2cf.h */
161
162 /*
163 * This function contains 58 FP additions, 12 FP multiplications,
164 * (or, 54 additions, 8 multiplications, 4 fused multiply/add),
165 * 34 stack variables, 3 constants, and 32 memory accesses
166 */
167 #include "rdft/scalar/r2cf.h"
168
169 static void r2cf_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
170 {
171 DK(KP923879532, +0.923879532511286756128183189396788286822416626);
172 DK(KP382683432, +0.382683432365089771728459984030398866761344562);
173 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
174 {
175 INT i;
176 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) {
177 E T3, T6, T7, Tz, Ti, Ta, Td, Te, TA, Th, Tq, TV, TF, TP, Tx;
178 E TU, TE, TM, Tg, Tf, TJ, TQ;
179 {
180 E T1, T2, T4, T5;
181 T1 = R0[0];
182 T2 = R0[WS(rs, 4)];
183 T3 = T1 + T2;
184 T4 = R0[WS(rs, 2)];
185 T5 = R0[WS(rs, 6)];
186 T6 = T4 + T5;
187 T7 = T3 + T6;
188 Tz = T1 - T2;
189 Ti = T4 - T5;
190 }
191 {
192 E T8, T9, Tb, Tc;
193 T8 = R0[WS(rs, 1)];
194 T9 = R0[WS(rs, 5)];
195 Ta = T8 + T9;
196 Tg = T8 - T9;
197 Tb = R0[WS(rs, 7)];
198 Tc = R0[WS(rs, 3)];
199 Td = Tb + Tc;
200 Tf = Tb - Tc;
201 }
202 Te = Ta + Td;
203 TA = KP707106781 * (Tg + Tf);
204 Th = KP707106781 * (Tf - Tg);
205 {
206 E Tm, TN, Tp, TO;
207 {
208 E Tk, Tl, Tn, To;
209 Tk = R1[WS(rs, 7)];
210 Tl = R1[WS(rs, 3)];
211 Tm = Tk - Tl;
212 TN = Tk + Tl;
213 Tn = R1[WS(rs, 1)];
214 To = R1[WS(rs, 5)];
215 Tp = Tn - To;
216 TO = Tn + To;
217 }
218 Tq = FNMS(KP923879532, Tp, KP382683432 * Tm);
219 TV = TN + TO;
220 TF = FMA(KP923879532, Tm, KP382683432 * Tp);
221 TP = TN - TO;
222 }
223 {
224 E Tt, TK, Tw, TL;
225 {
226 E Tr, Ts, Tu, Tv;
227 Tr = R1[0];
228 Ts = R1[WS(rs, 4)];
229 Tt = Tr - Ts;
230 TK = Tr + Ts;
231 Tu = R1[WS(rs, 2)];
232 Tv = R1[WS(rs, 6)];
233 Tw = Tu - Tv;
234 TL = Tu + Tv;
235 }
236 Tx = FMA(KP382683432, Tt, KP923879532 * Tw);
237 TU = TK + TL;
238 TE = FNMS(KP382683432, Tw, KP923879532 * Tt);
239 TM = TK - TL;
240 }
241 Cr[WS(csr, 4)] = T7 - Te;
242 Ci[WS(csi, 4)] = TV - TU;
243 {
244 E Tj, Ty, TD, TG;
245 Tj = Th - Ti;
246 Ty = Tq - Tx;
247 Ci[WS(csi, 1)] = Tj + Ty;
248 Ci[WS(csi, 7)] = Ty - Tj;
249 TD = Tz + TA;
250 TG = TE + TF;
251 Cr[WS(csr, 7)] = TD - TG;
252 Cr[WS(csr, 1)] = TD + TG;
253 }
254 {
255 E TB, TC, TH, TI;
256 TB = Tz - TA;
257 TC = Tx + Tq;
258 Cr[WS(csr, 5)] = TB - TC;
259 Cr[WS(csr, 3)] = TB + TC;
260 TH = Ti + Th;
261 TI = TF - TE;
262 Ci[WS(csi, 3)] = TH + TI;
263 Ci[WS(csi, 5)] = TI - TH;
264 }
265 TJ = T3 - T6;
266 TQ = KP707106781 * (TM + TP);
267 Cr[WS(csr, 6)] = TJ - TQ;
268 Cr[WS(csr, 2)] = TJ + TQ;
269 {
270 E TR, TS, TT, TW;
271 TR = Td - Ta;
272 TS = KP707106781 * (TP - TM);
273 Ci[WS(csi, 2)] = TR + TS;
274 Ci[WS(csi, 6)] = TS - TR;
275 TT = T7 + Te;
276 TW = TU + TV;
277 Cr[WS(csr, 8)] = TT - TW;
278 Cr[0] = TT + TW;
279 }
280 }
281 }
282 }
283
284 static const kr2c_desc desc = { 16, "r2cf_16", {54, 8, 4, 0}, &GENUS };
285
286 void X(codelet_r2cf_16) (planner *p) {
287 X(kr2c_register) (p, r2cf_16, &desc);
288 }
289
290 #endif