Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/rdft/scalar/r2cf/r2cf_15.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 15 -name r2cf_15 -include rdft/scalar/r2cf.h */
29
30 /*
31 * This function contains 64 FP additions, 35 FP multiplications,
32 * (or, 36 additions, 7 multiplications, 28 fused multiply/add),
33 * 45 stack variables, 8 constants, and 30 memory accesses
34 */
35 #include "rdft/scalar/r2cf.h"
36
37 static void r2cf_15(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP910592997, +0.910592997310029334643087372129977886038870291);
40 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
41 DK(KP823639103, +0.823639103546331925877420039278190003029660514);
42 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
43 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
44 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
45 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
46 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
47 {
48 INT i;
49 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(60, rs), MAKE_VOLATILE_STRIDE(60, csr), MAKE_VOLATILE_STRIDE(60, csi)) {
50 E Ti, TR, TF, TM, TN, T7, Te, Tf, TV, TW, TX, Ts, Tv, TH, Tl;
51 E To, TG, TS, TT, TU;
52 {
53 E TD, Tg, Th, TE;
54 TD = R0[0];
55 Tg = R0[WS(rs, 5)];
56 Th = R1[WS(rs, 2)];
57 TE = Th + Tg;
58 Ti = Tg - Th;
59 TR = TD + TE;
60 TF = FNMS(KP500000000, TE, TD);
61 }
62 {
63 E Tj, Tq, Tt, Tm, T3, Tk, Ta, Tr, Td, Tu, T6, Tn;
64 Tj = R1[WS(rs, 1)];
65 Tq = R0[WS(rs, 3)];
66 Tt = R1[WS(rs, 4)];
67 Tm = R0[WS(rs, 6)];
68 {
69 E T1, T2, T8, T9;
70 T1 = R0[WS(rs, 4)];
71 T2 = R1[WS(rs, 6)];
72 T3 = T1 - T2;
73 Tk = T1 + T2;
74 T8 = R1[WS(rs, 5)];
75 T9 = R1[0];
76 Ta = T8 - T9;
77 Tr = T8 + T9;
78 }
79 {
80 E Tb, Tc, T4, T5;
81 Tb = R0[WS(rs, 7)];
82 Tc = R0[WS(rs, 2)];
83 Td = Tb - Tc;
84 Tu = Tb + Tc;
85 T4 = R0[WS(rs, 1)];
86 T5 = R1[WS(rs, 3)];
87 T6 = T4 - T5;
88 Tn = T4 + T5;
89 }
90 TM = T6 - T3;
91 TN = Td - Ta;
92 T7 = T3 + T6;
93 Te = Ta + Td;
94 Tf = T7 + Te;
95 TV = Tq + Tr;
96 TW = Tt + Tu;
97 TX = TV + TW;
98 Ts = FNMS(KP500000000, Tr, Tq);
99 Tv = FNMS(KP500000000, Tu, Tt);
100 TH = Ts + Tv;
101 Tl = FNMS(KP500000000, Tk, Tj);
102 To = FNMS(KP500000000, Tn, Tm);
103 TG = Tl + To;
104 TS = Tj + Tk;
105 TT = Tm + Tn;
106 TU = TS + TT;
107 }
108 Ci[WS(csi, 5)] = KP866025403 * (Tf - Ti);
109 {
110 E TK, TQ, TO, TI, TJ, TP, TL;
111 TK = TG - TH;
112 TQ = FNMS(KP618033988, TM, TN);
113 TO = FMA(KP618033988, TN, TM);
114 TI = TG + TH;
115 TJ = FNMS(KP250000000, TI, TF);
116 Cr[WS(csr, 5)] = TF + TI;
117 TP = FNMS(KP559016994, TK, TJ);
118 Cr[WS(csr, 2)] = FMA(KP823639103, TQ, TP);
119 Cr[WS(csr, 7)] = FNMS(KP823639103, TQ, TP);
120 TL = FMA(KP559016994, TK, TJ);
121 Cr[WS(csr, 1)] = FMA(KP823639103, TO, TL);
122 Cr[WS(csr, 4)] = FNMS(KP823639103, TO, TL);
123 }
124 {
125 E T11, T12, T10, TY, TZ;
126 T11 = TW - TV;
127 T12 = TS - TT;
128 Ci[WS(csi, 3)] = KP951056516 * (FMA(KP618033988, T12, T11));
129 Ci[WS(csi, 6)] = -(KP951056516 * (FNMS(KP618033988, T11, T12)));
130 T10 = TU - TX;
131 TY = TU + TX;
132 TZ = FNMS(KP250000000, TY, TR);
133 Cr[WS(csr, 3)] = FNMS(KP559016994, T10, TZ);
134 Cr[0] = TR + TY;
135 Cr[WS(csr, 6)] = FMA(KP559016994, T10, TZ);
136 {
137 E Tx, TB, TA, TC;
138 {
139 E Tp, Tw, Ty, Tz;
140 Tp = Tl - To;
141 Tw = Ts - Tv;
142 Tx = FMA(KP618033988, Tw, Tp);
143 TB = FNMS(KP618033988, Tp, Tw);
144 Ty = FMA(KP250000000, Tf, Ti);
145 Tz = Te - T7;
146 TA = FMA(KP559016994, Tz, Ty);
147 TC = FNMS(KP559016994, Tz, Ty);
148 }
149 Ci[WS(csi, 1)] = -(KP951056516 * (FNMS(KP910592997, TA, Tx)));
150 Ci[WS(csi, 7)] = KP951056516 * (FMA(KP910592997, TC, TB));
151 Ci[WS(csi, 4)] = KP951056516 * (FMA(KP910592997, TA, Tx));
152 Ci[WS(csi, 2)] = KP951056516 * (FNMS(KP910592997, TC, TB));
153 }
154 }
155 }
156 }
157 }
158
159 static const kr2c_desc desc = { 15, "r2cf_15", {36, 7, 28, 0}, &GENUS };
160
161 void X(codelet_r2cf_15) (planner *p) {
162 X(kr2c_register) (p, r2cf_15, &desc);
163 }
164
165 #else
166
167 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 15 -name r2cf_15 -include rdft/scalar/r2cf.h */
168
169 /*
170 * This function contains 64 FP additions, 25 FP multiplications,
171 * (or, 50 additions, 11 multiplications, 14 fused multiply/add),
172 * 47 stack variables, 10 constants, and 30 memory accesses
173 */
174 #include "rdft/scalar/r2cf.h"
175
176 static void r2cf_15(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
177 {
178 DK(KP484122918, +0.484122918275927110647408174972799951354115213);
179 DK(KP216506350, +0.216506350946109661690930792688234045867850657);
180 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
181 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
182 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
183 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
184 DK(KP509036960, +0.509036960455127183450980863393907648510733164);
185 DK(KP823639103, +0.823639103546331925877420039278190003029660514);
186 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
187 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
188 {
189 INT i;
190 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(60, rs), MAKE_VOLATILE_STRIDE(60, csr), MAKE_VOLATILE_STRIDE(60, csi)) {
191 E Ti, TR, TL, TD, TE, T7, Te, Tf, TV, TW, TX, Tv, Ty, TH, To;
192 E Tr, TG, TS, TT, TU;
193 {
194 E TJ, Tg, Th, TK;
195 TJ = R0[0];
196 Tg = R0[WS(rs, 5)];
197 Th = R1[WS(rs, 2)];
198 TK = Th + Tg;
199 Ti = Tg - Th;
200 TR = TJ + TK;
201 TL = FNMS(KP500000000, TK, TJ);
202 }
203 {
204 E Tm, Tt, Tw, Tp, T3, Tx, Ta, Tn, Td, Tq, T6, Tu;
205 Tm = R1[WS(rs, 1)];
206 Tt = R0[WS(rs, 3)];
207 Tw = R1[WS(rs, 4)];
208 Tp = R0[WS(rs, 6)];
209 {
210 E T1, T2, T8, T9;
211 T1 = R0[WS(rs, 7)];
212 T2 = R0[WS(rs, 2)];
213 T3 = T1 - T2;
214 Tx = T1 + T2;
215 T8 = R1[WS(rs, 6)];
216 T9 = R0[WS(rs, 4)];
217 Ta = T8 - T9;
218 Tn = T9 + T8;
219 }
220 {
221 E Tb, Tc, T4, T5;
222 Tb = R1[WS(rs, 3)];
223 Tc = R0[WS(rs, 1)];
224 Td = Tb - Tc;
225 Tq = Tc + Tb;
226 T4 = R1[0];
227 T5 = R1[WS(rs, 5)];
228 T6 = T4 - T5;
229 Tu = T5 + T4;
230 }
231 TD = Ta - Td;
232 TE = T6 + T3;
233 T7 = T3 - T6;
234 Te = Ta + Td;
235 Tf = T7 - Te;
236 TV = Tt + Tu;
237 TW = Tw + Tx;
238 TX = TV + TW;
239 Tv = FNMS(KP500000000, Tu, Tt);
240 Ty = FNMS(KP500000000, Tx, Tw);
241 TH = Tv + Ty;
242 To = FNMS(KP500000000, Tn, Tm);
243 Tr = FNMS(KP500000000, Tq, Tp);
244 TG = To + Tr;
245 TS = Tm + Tn;
246 TT = Tp + Tq;
247 TU = TS + TT;
248 }
249 Ci[WS(csi, 5)] = KP866025403 * (Tf - Ti);
250 {
251 E TF, TP, TI, TM, TN, TQ, TO;
252 TF = FMA(KP823639103, TD, KP509036960 * TE);
253 TP = FNMS(KP509036960, TD, KP823639103 * TE);
254 TI = KP559016994 * (TG - TH);
255 TM = TG + TH;
256 TN = FNMS(KP250000000, TM, TL);
257 Cr[WS(csr, 5)] = TL + TM;
258 TQ = TN - TI;
259 Cr[WS(csr, 2)] = TP + TQ;
260 Cr[WS(csr, 7)] = TQ - TP;
261 TO = TI + TN;
262 Cr[WS(csr, 1)] = TF + TO;
263 Cr[WS(csr, 4)] = TO - TF;
264 }
265 {
266 E T11, T12, T10, TY, TZ;
267 T11 = TS - TT;
268 T12 = TW - TV;
269 Ci[WS(csi, 3)] = FMA(KP587785252, T11, KP951056516 * T12);
270 Ci[WS(csi, 6)] = FNMS(KP951056516, T11, KP587785252 * T12);
271 T10 = KP559016994 * (TU - TX);
272 TY = TU + TX;
273 TZ = FNMS(KP250000000, TY, TR);
274 Cr[WS(csr, 3)] = TZ - T10;
275 Cr[0] = TR + TY;
276 Cr[WS(csr, 6)] = T10 + TZ;
277 {
278 E Tl, TB, TA, TC;
279 {
280 E Tj, Tk, Ts, Tz;
281 Tj = FMA(KP866025403, Ti, KP216506350 * Tf);
282 Tk = KP484122918 * (Te + T7);
283 Tl = Tj + Tk;
284 TB = Tk - Tj;
285 Ts = To - Tr;
286 Tz = Tv - Ty;
287 TA = FMA(KP951056516, Ts, KP587785252 * Tz);
288 TC = FNMS(KP587785252, Ts, KP951056516 * Tz);
289 }
290 Ci[WS(csi, 1)] = Tl - TA;
291 Ci[WS(csi, 7)] = TC - TB;
292 Ci[WS(csi, 4)] = Tl + TA;
293 Ci[WS(csi, 2)] = TB + TC;
294 }
295 }
296 }
297 }
298 }
299
300 static const kr2c_desc desc = { 15, "r2cf_15", {50, 11, 14, 0}, &GENUS };
301
302 void X(codelet_r2cf_15) (planner *p) {
303 X(kr2c_register) (p, r2cf_15, &desc);
304 }
305
306 #endif