Mercurial > hg > js-dsp-test

comparison fft/fftw/fftw-3.3.4/rdft/scalar/r2cf/r2cf_14.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 14 -name r2cf_14 -include r2cf.h */
29
30 /*
31 * This function contains 62 FP additions, 36 FP multiplications,
32 * (or, 32 additions, 6 multiplications, 30 fused multiply/add),
33 * 45 stack variables, 6 constants, and 28 memory accesses
34 */
35 #include "r2cf.h"
36
37 static void r2cf_14(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP900968867, +0.900968867902419126236102319507445051165919162);
40 DK(KP692021471, +0.692021471630095869627814897002069140197260599);
41 DK(KP801937735, +0.801937735804838252472204639014890102331838324);
42 DK(KP974927912, +0.974927912181823607018131682993931217232785801);
43 DK(KP356895867, +0.356895867892209443894399510021300583399127187);
44 DK(KP554958132, +0.554958132087371191422194871006410481067288862);
45 {
46 INT i;
47 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
48 E TN, T3, TG, TQ, Tx, To, TH, Td, TD, TO, Tw, Ta, TL, Ty, TT;
49 E TI, Tg, Tr, Te, Tf, TP, TJ;
50 {
51 E Tl, TE, Tk, Tm;
52 {
53 E T1, T2, Ti, Tj;
54 T1 = R0[0];
55 T2 = R1[WS(rs, 3)];
56 Ti = R0[WS(rs, 3)];
57 Tj = R1[WS(rs, 6)];
58 Tl = R0[WS(rs, 4)];
59 TN = T1 + T2;
60 T3 = T1 - T2;
61 TE = Ti + Tj;
62 Tk = Ti - Tj;
63 Tm = R1[0];
64 }
65 {
66 E T7, TC, T6, T8;
67 {
68 E T4, T5, TF, Tn;
69 T4 = R0[WS(rs, 1)];
70 T5 = R1[WS(rs, 4)];
71 T7 = R0[WS(rs, 6)];
72 TF = Tl + Tm;
73 Tn = Tl - Tm;
74 TC = T4 + T5;
75 T6 = T4 - T5;
76 TG = TE - TF;
77 TQ = TE + TF;
78 Tx = Tn - Tk;
79 To = Tk + Tn;
80 T8 = R1[WS(rs, 2)];
81 }
82 {
83 E Tb, Tc, TB, T9;
84 Tb = R0[WS(rs, 2)];
85 Tc = R1[WS(rs, 5)];
86 Te = R0[WS(rs, 5)];
87 TB = T7 + T8;
88 T9 = T7 - T8;
89 TH = Tb + Tc;
90 Td = Tb - Tc;
91 TD = TB - TC;
92 TO = TC + TB;
93 Tw = T6 - T9;
94 Ta = T6 + T9;
95 Tf = R1[WS(rs, 1)];
96 }
97 }
98 }
99 TL = FNMS(KP554958132, TG, TD);
100 Ty = FNMS(KP554958132, Tx, Tw);
101 TT = FNMS(KP356895867, TO, TQ);
102 TI = Te + Tf;
103 Tg = Te - Tf;
104 Tr = FNMS(KP356895867, Ta, To);
105 TP = TH + TI;
106 TJ = TH - TI;
107 {
108 E Th, Tv, TK, TM;
109 Th = Td + Tg;
110 Tv = Tg - Td;
111 TK = FMA(KP554958132, TJ, TG);
112 TM = FMA(KP554958132, TD, TJ);
113 Ci[WS(csi, 6)] = KP974927912 * (FNMS(KP801937735, TL, TJ));
114 {
115 E TR, TV, TU, Tz;
116 TR = FNMS(KP356895867, TQ, TP);
117 TV = FNMS(KP356895867, TP, TO);
118 TU = FNMS(KP692021471, TT, TP);
119 Cr[0] = TN + TO + TP + TQ;
120 Tz = FMA(KP554958132, Tv, Tx);
121 Ci[WS(csi, 1)] = KP974927912 * (FNMS(KP801937735, Ty, Tv));
122 {
123 E TA, Ts, Tt, Tp;
124 TA = FMA(KP554958132, Tw, Tv);
125 Ts = FNMS(KP692021471, Tr, Th);
126 Tt = FNMS(KP356895867, Th, Ta);
127 Tp = FNMS(KP356895867, To, Th);
128 Cr[WS(csr, 7)] = T3 + Ta + Th + To;
129 Ci[WS(csi, 2)] = KP974927912 * (FMA(KP801937735, TK, TD));
130 Ci[WS(csi, 4)] = KP974927912 * (FNMS(KP801937735, TM, TG));
131 {
132 E TS, TW, Tu, Tq;
133 TS = FNMS(KP692021471, TR, TO);
134 TW = FNMS(KP692021471, TV, TQ);
135 Cr[WS(csr, 2)] = FNMS(KP900968867, TU, TN);
136 Ci[WS(csi, 5)] = KP974927912 * (FMA(KP801937735, Tz, Tw));
137 Ci[WS(csi, 3)] = KP974927912 * (FNMS(KP801937735, TA, Tx));
138 Cr[WS(csr, 5)] = FNMS(KP900968867, Ts, T3);
139 Tu = FNMS(KP692021471, Tt, To);
140 Tq = FNMS(KP692021471, Tp, Ta);
141 Cr[WS(csr, 4)] = FNMS(KP900968867, TS, TN);
142 Cr[WS(csr, 6)] = FNMS(KP900968867, TW, TN);
143 Cr[WS(csr, 1)] = FNMS(KP900968867, Tu, T3);
144 Cr[WS(csr, 3)] = FNMS(KP900968867, Tq, T3);
145 }
146 }
147 }
148 }
149 }
150 }
151 }
152
153 static const kr2c_desc desc = { 14, "r2cf_14", {32, 6, 30, 0}, &GENUS };
154
155 void X(codelet_r2cf_14) (planner *p) {
156 X(kr2c_register) (p, r2cf_14, &desc);
157 }
158
159 #else /* HAVE_FMA */
160
161 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 14 -name r2cf_14 -include r2cf.h */
162
163 /*
164 * This function contains 62 FP additions, 36 FP multiplications,
165 * (or, 38 additions, 12 multiplications, 24 fused multiply/add),
166 * 29 stack variables, 6 constants, and 28 memory accesses
167 */
168 #include "r2cf.h"
169
170 static void r2cf_14(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
171 {
172 DK(KP900968867, +0.900968867902419126236102319507445051165919162);
173 DK(KP222520933, +0.222520933956314404288902564496794759466355569);
174 DK(KP623489801, +0.623489801858733530525004884004239810632274731);
175 DK(KP433883739, +0.433883739117558120475768332848358754609990728);
176 DK(KP974927912, +0.974927912181823607018131682993931217232785801);
177 DK(KP781831482, +0.781831482468029808708444526674057750232334519);
178 {
179 INT i;
180 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
181 E T3, TB, T6, Tv, Tn, Ts, Tk, Tt, Td, Ty, T9, Tw, Tg, Tz, T1;
182 E T2;
183 T1 = R0[0];
184 T2 = R1[WS(rs, 3)];
185 T3 = T1 - T2;
186 TB = T1 + T2;
187 {
188 E T4, T5, Tl, Tm;
189 T4 = R0[WS(rs, 2)];
190 T5 = R1[WS(rs, 5)];
191 T6 = T4 - T5;
192 Tv = T4 + T5;
193 Tl = R0[WS(rs, 6)];
194 Tm = R1[WS(rs, 2)];
195 Tn = Tl - Tm;
196 Ts = Tl + Tm;
197 }
198 {
199 E Ti, Tj, Tb, Tc;
200 Ti = R0[WS(rs, 1)];
201 Tj = R1[WS(rs, 4)];
202 Tk = Ti - Tj;
203 Tt = Ti + Tj;
204 Tb = R0[WS(rs, 3)];
205 Tc = R1[WS(rs, 6)];
206 Td = Tb - Tc;
207 Ty = Tb + Tc;
208 }
209 {
210 E T7, T8, Te, Tf;
211 T7 = R0[WS(rs, 5)];
212 T8 = R1[WS(rs, 1)];
213 T9 = T7 - T8;
214 Tw = T7 + T8;
215 Te = R0[WS(rs, 4)];
216 Tf = R1[0];
217 Tg = Te - Tf;
218 Tz = Te + Tf;
219 }
220 {
221 E Tp, Tr, Tq, Ta, To, Th;
222 Tp = Tn - Tk;
223 Tr = Tg - Td;
224 Tq = T9 - T6;
225 Ci[WS(csi, 1)] = FMA(KP781831482, Tp, KP974927912 * Tq) + (KP433883739 * Tr);
226 Ci[WS(csi, 5)] = FMA(KP433883739, Tq, KP781831482 * Tr) - (KP974927912 * Tp);
227 Ci[WS(csi, 3)] = FMA(KP433883739, Tp, KP974927912 * Tr) - (KP781831482 * Tq);
228 Ta = T6 + T9;
229 To = Tk + Tn;
230 Th = Td + Tg;
231 Cr[WS(csr, 3)] = FMA(KP623489801, Ta, T3) + FNMA(KP222520933, Th, KP900968867 * To);
232 Cr[WS(csr, 7)] = T3 + To + Ta + Th;
233 Cr[WS(csr, 1)] = FMA(KP623489801, To, T3) + FNMA(KP900968867, Th, KP222520933 * Ta);
234 Cr[WS(csr, 5)] = FMA(KP623489801, Th, T3) + FNMA(KP900968867, Ta, KP222520933 * To);
235 }
236 {
237 E Tu, TA, Tx, TC, TE, TD;
238 Tu = Ts - Tt;
239 TA = Ty - Tz;
240 Tx = Tv - Tw;
241 Ci[WS(csi, 2)] = FMA(KP974927912, Tu, KP433883739 * Tx) + (KP781831482 * TA);
242 Ci[WS(csi, 6)] = FMA(KP974927912, Tx, KP433883739 * TA) - (KP781831482 * Tu);
243 Ci[WS(csi, 4)] = FNMS(KP781831482, Tx, KP974927912 * TA) - (KP433883739 * Tu);
244 TC = Tt + Ts;
245 TE = Tv + Tw;
246 TD = Ty + Tz;
247 Cr[WS(csr, 6)] = FMA(KP623489801, TC, TB) + FNMA(KP900968867, TD, KP222520933 * TE);
248 Cr[WS(csr, 2)] = FMA(KP623489801, TD, TB) + FNMA(KP900968867, TE, KP222520933 * TC);
249 Cr[WS(csr, 4)] = FMA(KP623489801, TE, TB) + FNMA(KP222520933, TD, KP900968867 * TC);
250 Cr[0] = TB + TC + TE + TD;
251 }
252 }
253 }
254 }
255
256 static const kr2c_desc desc = { 14, "r2cf_14", {38, 12, 24, 0}, &GENUS };
257
258 void X(codelet_r2cf_14) (planner *p) {
259 X(kr2c_register) (p, r2cf_14, &desc);
260 }
261
262 #endif /* HAVE_FMA */