|
d@0
|
1 /*
|
|
d@0
|
2 * Copyright (c) 2003, 2007-8 Matteo Frigo
|
|
d@0
|
3 * Copyright (c) 2003, 2007-8 Massachusetts Institute of Technology
|
|
d@0
|
4 *
|
|
d@0
|
5 * This program is free software; you can redistribute it and/or modify
|
|
d@0
|
6 * it under the terms of the GNU General Public License as published by
|
|
d@0
|
7 * the Free Software Foundation; either version 2 of the License, or
|
|
d@0
|
8 * (at your option) any later version.
|
|
d@0
|
9 *
|
|
d@0
|
10 * This program is distributed in the hope that it will be useful,
|
|
d@0
|
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
d@0
|
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
d@0
|
13 * GNU General Public License for more details.
|
|
d@0
|
14 *
|
|
d@0
|
15 * You should have received a copy of the GNU General Public License
|
|
d@0
|
16 * along with this program; if not, write to the Free Software
|
|
d@0
|
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
d@0
|
18 *
|
|
d@0
|
19 */
|
|
d@0
|
20 /* Generated by: ../../genfft/gen_notw_c -standalone -fma -reorder-insns -simd -compact -variables 100000 -with-ostride 2 -include fftw-spu.h -store-multiple 2 -n 6 -name X(spu_n2fv_6) */
|
|
d@0
|
21
|
|
d@0
|
22 /*
|
|
d@0
|
23 * This function contains 18 FP additions, 8 FP multiplications,
|
|
d@0
|
24 * (or, 12 additions, 2 multiplications, 6 fused multiply/add),
|
|
d@0
|
25 * 29 stack variables, 2 constants, and 15 memory accesses
|
|
d@0
|
26 */
|
|
d@0
|
27 #include "fftw-spu.h"
|
|
d@0
|
28
|
|
d@0
|
29 void X(spu_n2fv_6) (const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) {
|
|
d@0
|
30 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
|
|
d@0
|
31 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
|
|
d@0
|
32 INT i;
|
|
d@0
|
33 const R *xi;
|
|
d@0
|
34 R *xo;
|
|
d@0
|
35 xi = ri;
|
|
d@0
|
36 xo = ro;
|
|
d@0
|
37 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
|
|
d@0
|
38 V Td, T3, Tc, Ta, Tg, Ti, T1, T2, Te, T6, Tf, T9, T4, T5, T7;
|
|
d@0
|
39 V T8, Tj, Tk, Tl, Tb, Tm, Th, Tn, To;
|
|
d@0
|
40 T1 = LD(&(xi[0]), ivs, &(xi[0]));
|
|
d@0
|
41 T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
|
|
d@0
|
42 Td = VADD(T1, T2);
|
|
d@0
|
43 T3 = VSUB(T1, T2);
|
|
d@0
|
44 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
|
|
d@0
|
45 T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
|
|
d@0
|
46 Te = VADD(T4, T5);
|
|
d@0
|
47 T6 = VSUB(T4, T5);
|
|
d@0
|
48 T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
|
|
d@0
|
49 T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
|
|
d@0
|
50 Tf = VADD(T7, T8);
|
|
d@0
|
51 T9 = VSUB(T7, T8);
|
|
d@0
|
52 Tc = VMUL(LDK(KP866025403), VSUB(T9, T6));
|
|
d@0
|
53 Ta = VADD(T6, T9);
|
|
d@0
|
54 Tg = VADD(Te, Tf);
|
|
d@0
|
55 Ti = VMUL(LDK(KP866025403), VSUB(Tf, Te));
|
|
d@0
|
56 Tj = VADD(T3, Ta);
|
|
d@0
|
57 STM2(&(xo[6]), Tj, ovs, &(xo[2]));
|
|
d@0
|
58 Tk = VADD(Td, Tg);
|
|
d@0
|
59 STM2(&(xo[0]), Tk, ovs, &(xo[0]));
|
|
d@0
|
60 Tb = VFNMS(LDK(KP500000000), Ta, T3);
|
|
d@0
|
61 Tl = VFNMSI(Tc, Tb);
|
|
d@0
|
62 STM2(&(xo[10]), Tl, ovs, &(xo[2]));
|
|
d@0
|
63 Tm = VFMAI(Tc, Tb);
|
|
d@0
|
64 STM2(&(xo[2]), Tm, ovs, &(xo[2]));
|
|
d@0
|
65 STN2(&(xo[0]), Tk, Tm, ovs);
|
|
d@0
|
66 Th = VFNMS(LDK(KP500000000), Tg, Td);
|
|
d@0
|
67 Tn = VFNMSI(Ti, Th);
|
|
d@0
|
68 STM2(&(xo[4]), Tn, ovs, &(xo[0]));
|
|
d@0
|
69 STN2(&(xo[4]), Tn, Tj, ovs);
|
|
d@0
|
70 To = VFMAI(Ti, Th);
|
|
d@0
|
71 STM2(&(xo[8]), To, ovs, &(xo[0]));
|
|
d@0
|
72 STN2(&(xo[8]), To, Tl, ovs);
|
|
d@0
|
73 }
|
|
d@0
|
74 }
|
