Mercurial > hg > js-dsp-test

comparison fft/fftw/fftw-3.3.4/rdft/scalar/r2cf/r2cf_15.c @ 19:26056e866c29

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW to comparison table
author Chris Cannam
date Tue, 06 Oct 2015 13:08:39 +0100
parents
children
comparison
equal deleted inserted replaced
18:8db794ca3e0b 19:26056e866c29
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 Tue Mar 4 13:49:07 EST 2014 */
23
24 #include "codelet-rdft.h"
25
26 #ifdef HAVE_FMA
27
28 /* Generated by: ../../../genfft/gen_r2cf.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 15 -name r2cf_15 -include 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 * 50 stack variables, 8 constants, and 30 memory accesses
34 */
35 #include "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(KP823639103, +0.823639103546331925877420039278190003029660514);
41 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
42 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
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 Tw, Tz, Tp, Ty;
51 {
52 E Ti, TF, TR, TN, TX, T11, TM, TS, Tl, TH, Tf, To, TT, TD, Tg;
53 E Th;
54 TD = R0[0];
55 Tg = R0[WS(rs, 5)];
56 Th = R1[WS(rs, 2)];
57 {
58 E Tj, Tq, Tt, Tm, T3, Tk, T4, Ta, Tr, Td, Tu, T5, TE;
59 Tj = R1[WS(rs, 1)];
60 Tq = R0[WS(rs, 3)];
61 Tt = R1[WS(rs, 4)];
62 TE = Th + Tg;
63 Ti = Tg - Th;
64 Tm = R0[WS(rs, 6)];
65 {
66 E T8, T9, T1, T2, Tb, Tc;
67 T1 = R0[WS(rs, 4)];
68 T2 = R1[WS(rs, 6)];
69 TF = FNMS(KP500000000, TE, TD);
70 TR = TD + TE;
71 T8 = R1[WS(rs, 5)];
72 T3 = T1 - T2;
73 Tk = T1 + T2;
74 T9 = R1[0];
75 Tb = R0[WS(rs, 7)];
76 Tc = R0[WS(rs, 2)];
77 T4 = R0[WS(rs, 1)];
78 Ta = T8 - T9;
79 Tr = T8 + T9;
80 Td = Tb - Tc;
81 Tu = Tb + Tc;
82 T5 = R1[WS(rs, 3)];
83 }
84 {
85 E Ts, Tv, Te, Tn, T7, T6, TV, TW;
86 TV = Tq + Tr;
87 Ts = FNMS(KP500000000, Tr, Tq);
88 Tv = FNMS(KP500000000, Tu, Tt);
89 TW = Tt + Tu;
90 Te = Ta + Td;
91 TN = Td - Ta;
92 Tn = T4 + T5;
93 T6 = T4 - T5;
94 TX = TV + TW;
95 T11 = TW - TV;
96 TM = T6 - T3;
97 T7 = T3 + T6;
98 TS = Tj + Tk;
99 Tl = FNMS(KP500000000, Tk, Tj);
100 TH = Ts + Tv;
101 Tw = Ts - Tv;
102 Tz = Te - T7;
103 Tf = T7 + Te;
104 To = FNMS(KP500000000, Tn, Tm);
105 TT = Tm + Tn;
106 }
107 }
108 {
109 E TO, TQ, TU, T12, TK, TI, TG;
110 Ci[WS(csi, 5)] = KP866025403 * (Tf - Ti);
111 TG = Tl + To;
112 Tp = Tl - To;
113 TO = FMA(KP618033988, TN, TM);
114 TQ = FNMS(KP618033988, TM, TN);
115 TU = TS + TT;
116 T12 = TS - TT;
117 TK = TG - TH;
118 TI = TG + TH;
119 {
120 E T10, TY, TL, TP, TJ, TZ;
121 T10 = TU - TX;
122 TY = TU + TX;
123 Cr[WS(csr, 5)] = TF + TI;
124 TJ = FNMS(KP250000000, TI, TF);
125 Ci[WS(csi, 6)] = -(KP951056516 * (FNMS(KP618033988, T11, T12)));
126 Ci[WS(csi, 3)] = KP951056516 * (FMA(KP618033988, T12, T11));
127 TL = FMA(KP559016994, TK, TJ);
128 TP = FNMS(KP559016994, TK, TJ);
129 Cr[0] = TR + TY;
130 TZ = FNMS(KP250000000, TY, TR);
131 Cr[WS(csr, 4)] = FNMS(KP823639103, TO, TL);
132 Cr[WS(csr, 1)] = FMA(KP823639103, TO, TL);
133 Cr[WS(csr, 7)] = FNMS(KP823639103, TQ, TP);
134 Cr[WS(csr, 2)] = FMA(KP823639103, TQ, TP);
135 Cr[WS(csr, 6)] = FMA(KP559016994, T10, TZ);
136 Cr[WS(csr, 3)] = FNMS(KP559016994, T10, TZ);
137 Ty = FMA(KP250000000, Tf, Ti);
138 }
139 }
140 }
141 {
142 E TB, Tx, TC, TA;
143 TB = FNMS(KP618033988, Tp, Tw);
144 Tx = FMA(KP618033988, Tw, Tp);
145 TC = FNMS(KP559016994, Tz, Ty);
146 TA = FMA(KP559016994, Tz, Ty);
147 Ci[WS(csi, 2)] = KP951056516 * (FNMS(KP910592997, TC, TB));
148 Ci[WS(csi, 7)] = KP951056516 * (FMA(KP910592997, TC, TB));
149 Ci[WS(csi, 4)] = KP951056516 * (FMA(KP910592997, TA, Tx));
150 Ci[WS(csi, 1)] = -(KP951056516 * (FNMS(KP910592997, TA, Tx)));
151 }
152 }
153 }
154 }
155
156 static const kr2c_desc desc = { 15, "r2cf_15", {36, 7, 28, 0}, &GENUS };
157
158 void X(codelet_r2cf_15) (planner *p) {
159 X(kr2c_register) (p, r2cf_15, &desc);
160 }
161
162 #else /* HAVE_FMA */
163
164 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 15 -name r2cf_15 -include r2cf.h */
165
166 /*
167 * This function contains 64 FP additions, 25 FP multiplications,
168 * (or, 50 additions, 11 multiplications, 14 fused multiply/add),
169 * 47 stack variables, 10 constants, and 30 memory accesses
170 */
171 #include "r2cf.h"
172
173 static void r2cf_15(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
174 {
175 DK(KP484122918, +0.484122918275927110647408174972799951354115213);
176 DK(KP216506350, +0.216506350946109661690930792688234045867850657);
177 DK(KP951056516, +0.951056516295153572116439333379382143405698634);
178 DK(KP587785252, +0.587785252292473129168705954639072768597652438);
179 DK(KP250000000, +0.250000000000000000000000000000000000000000000);
180 DK(KP559016994, +0.559016994374947424102293417182819058860154590);
181 DK(KP509036960, +0.509036960455127183450980863393907648510733164);
182 DK(KP823639103, +0.823639103546331925877420039278190003029660514);
183 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
184 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
185 {
186 INT i;
187 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)) {
188 E Ti, TR, TL, TD, TE, T7, Te, Tf, TV, TW, TX, Tv, Ty, TH, To;
189 E Tr, TG, TS, TT, TU;
190 {
191 E TJ, Tg, Th, TK;
192 TJ = R0[0];
193 Tg = R0[WS(rs, 5)];
194 Th = R1[WS(rs, 2)];
195 TK = Th + Tg;
196 Ti = Tg - Th;
197 TR = TJ + TK;
198 TL = FNMS(KP500000000, TK, TJ);
199 }
200 {
201 E Tm, Tt, Tw, Tp, T3, Tx, Ta, Tn, Td, Tq, T6, Tu;
202 Tm = R1[WS(rs, 1)];
203 Tt = R0[WS(rs, 3)];
204 Tw = R1[WS(rs, 4)];
205 Tp = R0[WS(rs, 6)];
206 {
207 E T1, T2, T8, T9;
208 T1 = R0[WS(rs, 7)];
209 T2 = R0[WS(rs, 2)];
210 T3 = T1 - T2;
211 Tx = T1 + T2;
212 T8 = R1[WS(rs, 6)];
213 T9 = R0[WS(rs, 4)];
214 Ta = T8 - T9;
215 Tn = T9 + T8;
216 }
217 {
218 E Tb, Tc, T4, T5;
219 Tb = R1[WS(rs, 3)];
220 Tc = R0[WS(rs, 1)];
221 Td = Tb - Tc;
222 Tq = Tc + Tb;
223 T4 = R1[0];
224 T5 = R1[WS(rs, 5)];
225 T6 = T4 - T5;
226 Tu = T5 + T4;
227 }
228 TD = Ta - Td;
229 TE = T6 + T3;
230 T7 = T3 - T6;
231 Te = Ta + Td;
232 Tf = T7 - Te;
233 TV = Tt + Tu;
234 TW = Tw + Tx;
235 TX = TV + TW;
236 Tv = FNMS(KP500000000, Tu, Tt);
237 Ty = FNMS(KP500000000, Tx, Tw);
238 TH = Tv + Ty;
239 To = FNMS(KP500000000, Tn, Tm);
240 Tr = FNMS(KP500000000, Tq, Tp);
241 TG = To + Tr;
242 TS = Tm + Tn;
243 TT = Tp + Tq;
244 TU = TS + TT;
245 }
246 Ci[WS(csi, 5)] = KP866025403 * (Tf - Ti);
247 {
248 E TF, TP, TI, TM, TN, TQ, TO;
249 TF = FMA(KP823639103, TD, KP509036960 * TE);
250 TP = FNMS(KP509036960, TD, KP823639103 * TE);
251 TI = KP559016994 * (TG - TH);
252 TM = TG + TH;
253 TN = FNMS(KP250000000, TM, TL);
254 Cr[WS(csr, 5)] = TL + TM;
255 TQ = TN - TI;
256 Cr[WS(csr, 2)] = TP + TQ;
257 Cr[WS(csr, 7)] = TQ - TP;
258 TO = TI + TN;
259 Cr[WS(csr, 1)] = TF + TO;
260 Cr[WS(csr, 4)] = TO - TF;
261 }
262 {
263 E T11, T12, T10, TY, TZ;
264 T11 = TS - TT;
265 T12 = TW - TV;
266 Ci[WS(csi, 3)] = FMA(KP587785252, T11, KP951056516 * T12);
267 Ci[WS(csi, 6)] = FNMS(KP951056516, T11, KP587785252 * T12);
268 T10 = KP559016994 * (TU - TX);
269 TY = TU + TX;
270 TZ = FNMS(KP250000000, TY, TR);
271 Cr[WS(csr, 3)] = TZ - T10;
272 Cr[0] = TR + TY;
273 Cr[WS(csr, 6)] = T10 + TZ;
274 {
275 E Tl, TB, TA, TC;
276 {
277 E Tj, Tk, Ts, Tz;
278 Tj = FMA(KP866025403, Ti, KP216506350 * Tf);
279 Tk = KP484122918 * (Te + T7);
280 Tl = Tj + Tk;
281 TB = Tk - Tj;
282 Ts = To - Tr;
283 Tz = Tv - Ty;
284 TA = FMA(KP951056516, Ts, KP587785252 * Tz);
285 TC = FNMS(KP587785252, Ts, KP951056516 * Tz);
286 }
287 Ci[WS(csi, 1)] = Tl - TA;
288 Ci[WS(csi, 7)] = TC - TB;
289 Ci[WS(csi, 4)] = Tl + TA;
290 Ci[WS(csi, 2)] = TB + TC;
291 }
292 }
293 }
294 }
295 }
296
297 static const kr2c_desc desc = { 15, "r2cf_15", {50, 11, 14, 0}, &GENUS };
298
299 void X(codelet_r2cf_15) (planner *p) {
300 X(kr2c_register) (p, r2cf_15, &desc);
301 }
302
303 #endif /* HAVE_FMA */