Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.5/rdft/scalar/r2cf/r2cfII_12.c @ 42:2cd0e3b3e1fd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Current fftw source
author Chris Cannam
date Tue, 18 Oct 2016 13:40:26 +0100
parents
children
comparison
equal deleted inserted replaced
41:481f5f8c5634 42:2cd0e3b3e1fd
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 Sat Jul 30 16:47:26 EDT 2016 */
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 12 -name r2cfII_12 -dft-II -include r2cfII.h */
29
30 /*
31 * This function contains 45 FP additions, 24 FP multiplications,
32 * (or, 21 additions, 0 multiplications, 24 fused multiply/add),
33 * 37 stack variables, 3 constants, and 24 memory accesses
34 */
35 #include "r2cfII.h"
36
37 static void r2cfII_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
40 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
41 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
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(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) {
45 E TD, TB, Tp, T9, Tq, Tr, TE, To, Ts, TC;
46 {
47 E T8, T1, Tv, Tm, TF, Tz, Tl, Ta, Tb, Tt, TA, T4, Tc;
48 {
49 E Tx, Th, Ti, Tj, Ty, T6, T7, T2, T3, Tk;
50 Tx = R0[WS(rs, 3)];
51 T6 = R0[WS(rs, 5)];
52 T7 = R0[WS(rs, 1)];
53 Th = R1[WS(rs, 4)];
54 Ti = R1[WS(rs, 2)];
55 Tj = R1[0];
56 Ty = T6 + T7;
57 T8 = T6 - T7;
58 T1 = R0[0];
59 Tv = Ti - Tj - Th;
60 Tk = Ti - Tj;
61 Tm = Ti + Tj;
62 TF = Tx - Ty;
63 Tz = FMA(KP500000000, Ty, Tx);
64 T2 = R0[WS(rs, 2)];
65 T3 = R0[WS(rs, 4)];
66 Tl = FMA(KP500000000, Tk, Th);
67 Ta = R1[WS(rs, 1)];
68 Tb = R1[WS(rs, 3)];
69 Tt = T1 + T3 - T2;
70 TA = T3 + T2;
71 T4 = T2 - T3;
72 Tc = R1[WS(rs, 5)];
73 }
74 {
75 E Tn, Tg, T5, Tu;
76 TD = FNMS(KP866025403, TA, Tz);
77 TB = FMA(KP866025403, TA, Tz);
78 T5 = FMA(KP500000000, T4, T1);
79 Tu = Ta + Tc - Tb;
80 {
81 E Td, Tf, TG, Tw, Te;
82 Td = Tb - Tc;
83 Tf = Tc + Tb;
84 Tp = FMA(KP866025403, T8, T5);
85 T9 = FNMS(KP866025403, T8, T5);
86 TG = Tv - Tu;
87 Tw = Tu + Tv;
88 Te = FMA(KP500000000, Td, Ta);
89 Tq = FMA(KP866025403, Tm, Tl);
90 Tn = FNMS(KP866025403, Tm, Tl);
91 Ci[WS(csi, 1)] = FMA(KP707106781, TG, TF);
92 Ci[WS(csi, 4)] = FMS(KP707106781, TG, TF);
93 Cr[WS(csr, 4)] = FMA(KP707106781, Tw, Tt);
94 Cr[WS(csr, 1)] = FNMS(KP707106781, Tw, Tt);
95 Tg = FNMS(KP866025403, Tf, Te);
96 Tr = FMA(KP866025403, Tf, Te);
97 }
98 TE = Tg + Tn;
99 To = Tg - Tn;
100 }
101 }
102 Ci[WS(csi, 2)] = FMS(KP707106781, TE, TD);
103 Ci[WS(csi, 3)] = FMA(KP707106781, TE, TD);
104 Cr[0] = FMA(KP707106781, To, T9);
105 Cr[WS(csr, 5)] = FNMS(KP707106781, To, T9);
106 Ts = Tq - Tr;
107 TC = Tr + Tq;
108 Ci[0] = -(FMA(KP707106781, TC, TB));
109 Ci[WS(csi, 5)] = FNMS(KP707106781, TC, TB);
110 Cr[WS(csr, 2)] = FMA(KP707106781, Ts, Tp);
111 Cr[WS(csr, 3)] = FNMS(KP707106781, Ts, Tp);
112 }
113 }
114 }
115
116 static const kr2c_desc desc = { 12, "r2cfII_12", {21, 0, 24, 0}, &GENUS };
117
118 void X(codelet_r2cfII_12) (planner *p) {
119 X(kr2c_register) (p, r2cfII_12, &desc);
120 }
121
122 #else /* HAVE_FMA */
123
124 /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 12 -name r2cfII_12 -dft-II -include r2cfII.h */
125
126 /*
127 * This function contains 43 FP additions, 12 FP multiplications,
128 * (or, 39 additions, 8 multiplications, 4 fused multiply/add),
129 * 28 stack variables, 5 constants, and 24 memory accesses
130 */
131 #include "r2cfII.h"
132
133 static void r2cfII_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
134 {
135 DK(KP353553390, +0.353553390593273762200422181052424519642417969);
136 DK(KP707106781, +0.707106781186547524400844362104849039284835938);
137 DK(KP612372435, +0.612372435695794524549321018676472847991486870);
138 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
139 DK(KP866025403, +0.866025403784438646763723170752936183471402627);
140 {
141 INT i;
142 for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) {
143 E Tx, Tg, T4, Tz, Ty, Tj, TA, T9, Tm, Tl, Te, Tp, To, Tf, TE;
144 E TF;
145 {
146 E T1, T3, T2, Th, Ti;
147 T1 = R0[0];
148 T3 = R0[WS(rs, 2)];
149 T2 = R0[WS(rs, 4)];
150 Tx = KP866025403 * (T2 + T3);
151 Tg = FMA(KP500000000, T3 - T2, T1);
152 T4 = T1 + T2 - T3;
153 Tz = R0[WS(rs, 3)];
154 Th = R0[WS(rs, 5)];
155 Ti = R0[WS(rs, 1)];
156 Ty = Th + Ti;
157 Tj = KP866025403 * (Th - Ti);
158 TA = FMA(KP500000000, Ty, Tz);
159 }
160 {
161 E T5, T6, T7, T8;
162 T5 = R1[WS(rs, 1)];
163 T6 = R1[WS(rs, 5)];
164 T7 = R1[WS(rs, 3)];
165 T8 = T6 - T7;
166 T9 = T5 + T8;
167 Tm = KP612372435 * (T6 + T7);
168 Tl = FNMS(KP353553390, T8, KP707106781 * T5);
169 }
170 {
171 E Td, Ta, Tb, Tc;
172 Td = R1[WS(rs, 4)];
173 Ta = R1[WS(rs, 2)];
174 Tb = R1[0];
175 Tc = Ta - Tb;
176 Te = Tc - Td;
177 Tp = FMA(KP353553390, Tc, KP707106781 * Td);
178 To = KP612372435 * (Ta + Tb);
179 }
180 Tf = KP707106781 * (T9 + Te);
181 Cr[WS(csr, 1)] = T4 - Tf;
182 Cr[WS(csr, 4)] = T4 + Tf;
183 TE = KP707106781 * (Te - T9);
184 TF = Tz - Ty;
185 Ci[WS(csi, 4)] = TE - TF;
186 Ci[WS(csi, 1)] = TE + TF;
187 {
188 E Tk, TB, Tr, Tw, Tn, Tq;
189 Tk = Tg - Tj;
190 TB = Tx - TA;
191 Tn = Tl - Tm;
192 Tq = To - Tp;
193 Tr = Tn + Tq;
194 Tw = Tn - Tq;
195 Cr[WS(csr, 5)] = Tk - Tr;
196 Ci[WS(csi, 2)] = Tw + TB;
197 Cr[0] = Tk + Tr;
198 Ci[WS(csi, 3)] = Tw - TB;
199 }
200 {
201 E Ts, TD, Tv, TC, Tt, Tu;
202 Ts = Tg + Tj;
203 TD = Tx + TA;
204 Tt = To + Tp;
205 Tu = Tm + Tl;
206 Tv = Tt - Tu;
207 TC = Tu + Tt;
208 Cr[WS(csr, 3)] = Ts - Tv;
209 Ci[WS(csi, 5)] = TD - TC;
210 Cr[WS(csr, 2)] = Ts + Tv;
211 Ci[0] = -(TC + TD);
212 }
213 }
214 }
215 }
216
217 static const kr2c_desc desc = { 12, "r2cfII_12", {39, 8, 4, 0}, &GENUS };
218
219 void X(codelet_r2cfII_12) (planner *p) {
220 X(kr2c_register) (p, r2cfII_12, &desc);
221 }
222
223 #endif /* HAVE_FMA */