Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/rdft/scalar/r2cf/r2cf_16.c @ 10:37bf6b4a2645

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW3
author Chris Cannam
date Wed, 20 Mar 2013 15:35:50 +0000
parents
children
comparison
equal deleted inserted replaced
9:c0fb53affa76 10:37bf6b4a2645
1 /*
2 * Copyright (c) 2003, 2007-11 Matteo Frigo
3 * Copyright (c) 2003, 2007-11 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 Sun Nov 25 07:39:47 EST 2012 */
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 16 -name r2cf_16 -include 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 * 38 stack variables, 3 constants, and 32 memory accesses
34 */
35 #include "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 TQ, TP;
46 {
47 E TB, TN, Tf, T7, Te, Tv, TO, TE, Tq, TJ, Tp, TI, TT, Ty, Tm;
48 E Tr, TK, Ts;
49 {
50 E TC, Ta, Td, TD;
51 {
52 E T1, T2, T4, T5;
53 T1 = R0[0];
54 T2 = R0[WS(rs, 4)];
55 T4 = R0[WS(rs, 2)];
56 T5 = R0[WS(rs, 6)];
57 {
58 E T8, T3, T6, T9, Tb, Tc;
59 T8 = R0[WS(rs, 1)];
60 TB = T1 - T2;
61 T3 = T1 + T2;
62 TN = T4 - T5;
63 T6 = T4 + T5;
64 T9 = R0[WS(rs, 5)];
65 Tb = R0[WS(rs, 7)];
66 Tc = R0[WS(rs, 3)];
67 Tf = T3 - T6;
68 T7 = T3 + T6;
69 TC = T8 - T9;
70 Ta = T8 + T9;
71 Td = Tb + Tc;
72 TD = Tb - Tc;
73 }
74 }
75 {
76 E TG, Ti, Tj, Tk, Tg, Th;
77 Tg = R1[0];
78 Th = R1[WS(rs, 4)];
79 Te = Ta + Td;
80 Tv = Td - Ta;
81 TO = TD - TC;
82 TE = TC + TD;
83 TG = Tg - Th;
84 Ti = Tg + Th;
85 Tj = R1[WS(rs, 2)];
86 Tk = R1[WS(rs, 6)];
87 {
88 E Tn, To, TH, Tl;
89 Tn = R1[WS(rs, 7)];
90 To = R1[WS(rs, 3)];
91 Tq = R1[WS(rs, 1)];
92 TH = Tj - Tk;
93 Tl = Tj + Tk;
94 TJ = Tn - To;
95 Tp = Tn + To;
96 TI = FNMS(KP414213562, TH, TG);
97 TT = FMA(KP414213562, TG, TH);
98 Ty = Ti + Tl;
99 Tm = Ti - Tl;
100 Tr = R1[WS(rs, 5)];
101 }
102 }
103 }
104 Cr[WS(csr, 4)] = T7 - Te;
105 TK = Tr - Tq;
106 Ts = Tq + Tr;
107 {
108 E Tx, TV, TF, TS, Tz, Tt, TM, TL;
109 Tx = T7 + Te;
110 TV = FNMS(KP707106781, TE, TB);
111 TF = FMA(KP707106781, TE, TB);
112 TL = FNMS(KP414213562, TK, TJ);
113 TS = FMA(KP414213562, TJ, TK);
114 Tz = Tp + Ts;
115 Tt = Tp - Ts;
116 TM = TI + TL;
117 TQ = TL - TI;
118 {
119 E TR, TU, TW, TA, Tw, Tu;
120 TP = FMA(KP707106781, TO, TN);
121 TR = FNMS(KP707106781, TO, TN);
122 TA = Ty + Tz;
123 Ci[WS(csi, 4)] = Tz - Ty;
124 Tw = Tt - Tm;
125 Tu = Tm + Tt;
126 Cr[WS(csr, 1)] = FMA(KP923879532, TM, TF);
127 Cr[WS(csr, 7)] = FNMS(KP923879532, TM, TF);
128 Cr[0] = Tx + TA;
129 Cr[WS(csr, 8)] = Tx - TA;
130 Ci[WS(csi, 6)] = FMS(KP707106781, Tw, Tv);
131 Ci[WS(csi, 2)] = FMA(KP707106781, Tw, Tv);
132 Cr[WS(csr, 2)] = FMA(KP707106781, Tu, Tf);
133 Cr[WS(csr, 6)] = FNMS(KP707106781, Tu, Tf);
134 TU = TS - TT;
135 TW = TT + TS;
136 Ci[WS(csi, 7)] = FMA(KP923879532, TU, TR);
137 Ci[WS(csi, 1)] = FMS(KP923879532, TU, TR);
138 Cr[WS(csr, 3)] = FMA(KP923879532, TW, TV);
139 Cr[WS(csr, 5)] = FNMS(KP923879532, TW, TV);
140 }
141 }
142 }
143 Ci[WS(csi, 5)] = FMS(KP923879532, TQ, TP);
144 Ci[WS(csi, 3)] = FMA(KP923879532, TQ, TP);
145 }
146 }
147 }
148
149 static const kr2c_desc desc = { 16, "r2cf_16", {38, 0, 20, 0}, &GENUS };
150
151 void X(codelet_r2cf_16) (planner *p) {
152 X(kr2c_register) (p, r2cf_16, &desc);
153 }
154
155 #else /* HAVE_FMA */
156
157 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 16 -name r2cf_16 -include r2cf.h */
158
159 /*
160 * This function contains 58 FP additions, 12 FP multiplications,
161 * (or, 54 additions, 8 multiplications, 4 fused multiply/add),
162 * 34 stack variables, 3 constants, and 32 memory accesses
163 */
164 #include "r2cf.h"
165
166 static void r2cf_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
167 {
168 DK(KP923879532, +0.923879532511286756128183189396788286822416626);
169 DK(KP382683432, +0.382683432365089771728459984030398866761344562);
170 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
171 {
172 INT i;
173 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)) {
174 E T3, T6, T7, Tz, Ti, Ta, Td, Te, TA, Th, Tq, TV, TF, TP, Tx;
175 E TU, TE, TM, Tg, Tf, TJ, TQ;
176 {
177 E T1, T2, T4, T5;
178 T1 = R0[0];
179 T2 = R0[WS(rs, 4)];
180 T3 = T1 + T2;
181 T4 = R0[WS(rs, 2)];
182 T5 = R0[WS(rs, 6)];
183 T6 = T4 + T5;
184 T7 = T3 + T6;
185 Tz = T1 - T2;
186 Ti = T4 - T5;
187 }
188 {
189 E T8, T9, Tb, Tc;
190 T8 = R0[WS(rs, 1)];
191 T9 = R0[WS(rs, 5)];
192 Ta = T8 + T9;
193 Tg = T8 - T9;
194 Tb = R0[WS(rs, 7)];
195 Tc = R0[WS(rs, 3)];
196 Td = Tb + Tc;
197 Tf = Tb - Tc;
198 }
199 Te = Ta + Td;
200 TA = KP707106781 * (Tg + Tf);
201 Th = KP707106781 * (Tf - Tg);
202 {
203 E Tm, TN, Tp, TO;
204 {
205 E Tk, Tl, Tn, To;
206 Tk = R1[WS(rs, 7)];
207 Tl = R1[WS(rs, 3)];
208 Tm = Tk - Tl;
209 TN = Tk + Tl;
210 Tn = R1[WS(rs, 1)];
211 To = R1[WS(rs, 5)];
212 Tp = Tn - To;
213 TO = Tn + To;
214 }
215 Tq = FNMS(KP923879532, Tp, KP382683432 * Tm);
216 TV = TN + TO;
217 TF = FMA(KP923879532, Tm, KP382683432 * Tp);
218 TP = TN - TO;
219 }
220 {
221 E Tt, TK, Tw, TL;
222 {
223 E Tr, Ts, Tu, Tv;
224 Tr = R1[0];
225 Ts = R1[WS(rs, 4)];
226 Tt = Tr - Ts;
227 TK = Tr + Ts;
228 Tu = R1[WS(rs, 2)];
229 Tv = R1[WS(rs, 6)];
230 Tw = Tu - Tv;
231 TL = Tu + Tv;
232 }
233 Tx = FMA(KP382683432, Tt, KP923879532 * Tw);
234 TU = TK + TL;
235 TE = FNMS(KP382683432, Tw, KP923879532 * Tt);
236 TM = TK - TL;
237 }
238 Cr[WS(csr, 4)] = T7 - Te;
239 Ci[WS(csi, 4)] = TV - TU;
240 {
241 E Tj, Ty, TD, TG;
242 Tj = Th - Ti;
243 Ty = Tq - Tx;
244 Ci[WS(csi, 1)] = Tj + Ty;
245 Ci[WS(csi, 7)] = Ty - Tj;
246 TD = Tz + TA;
247 TG = TE + TF;
248 Cr[WS(csr, 7)] = TD - TG;
249 Cr[WS(csr, 1)] = TD + TG;
250 }
251 {
252 E TB, TC, TH, TI;
253 TB = Tz - TA;
254 TC = Tx + Tq;
255 Cr[WS(csr, 5)] = TB - TC;
256 Cr[WS(csr, 3)] = TB + TC;
257 TH = Ti + Th;
258 TI = TF - TE;
259 Ci[WS(csi, 3)] = TH + TI;
260 Ci[WS(csi, 5)] = TI - TH;
261 }
262 TJ = T3 - T6;
263 TQ = KP707106781 * (TM + TP);
264 Cr[WS(csr, 6)] = TJ - TQ;
265 Cr[WS(csr, 2)] = TJ + TQ;
266 {
267 E TR, TS, TT, TW;
268 TR = Td - Ta;
269 TS = KP707106781 * (TP - TM);
270 Ci[WS(csi, 2)] = TR + TS;
271 Ci[WS(csi, 6)] = TS - TR;
272 TT = T7 + Te;
273 TW = TU + TV;
274 Cr[WS(csr, 8)] = TT - TW;
275 Cr[0] = TT + TW;
276 }
277 }
278 }
279 }
280
281 static const kr2c_desc desc = { 16, "r2cf_16", {54, 8, 4, 0}, &GENUS };
282
283 void X(codelet_r2cf_16) (planner *p) {
284 X(kr2c_register) (p, r2cf_16, &desc);
285 }
286
287 #endif /* HAVE_FMA */