|
Chris@42
|
1 /*
|
|
Chris@42
|
2 * Copyright (c) 2003, 2007-14 Matteo Frigo
|
|
Chris@42
|
3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
|
|
Chris@42
|
4 *
|
|
Chris@42
|
5 * This program is free software; you can redistribute it and/or modify
|
|
Chris@42
|
6 * it under the terms of the GNU General Public License as published by
|
|
Chris@42
|
7 * the Free Software Foundation; either version 2 of the License, or
|
|
Chris@42
|
8 * (at your option) any later version.
|
|
Chris@42
|
9 *
|
|
Chris@42
|
10 * This program is distributed in the hope that it will be useful,
|
|
Chris@42
|
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
Chris@42
|
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
Chris@42
|
13 * GNU General Public License for more details.
|
|
Chris@42
|
14 *
|
|
Chris@42
|
15 * You should have received a copy of the GNU General Public License
|
|
Chris@42
|
16 * along with this program; if not, write to the Free Software
|
|
Chris@42
|
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
Chris@42
|
18 *
|
|
Chris@42
|
19 */
|
|
Chris@42
|
20
|
|
Chris@42
|
21
|
|
Chris@42
|
22 #include "verify.h"
|
|
Chris@42
|
23
|
|
Chris@42
|
24 static void recur(int rnk, const bench_iodim *dims0, const bench_iodim *dims1,
|
|
Chris@42
|
25 dotens2_closure *k,
|
|
Chris@42
|
26 int indx0, int ondx0, int indx1, int ondx1)
|
|
Chris@42
|
27 {
|
|
Chris@42
|
28 if (rnk == 0)
|
|
Chris@42
|
29 k->apply(k, indx0, ondx0, indx1, ondx1);
|
|
Chris@42
|
30 else {
|
|
Chris@42
|
31 int i, n = dims0[0].n;
|
|
Chris@42
|
32 int is0 = dims0[0].is;
|
|
Chris@42
|
33 int os0 = dims0[0].os;
|
|
Chris@42
|
34 int is1 = dims1[0].is;
|
|
Chris@42
|
35 int os1 = dims1[0].os;
|
|
Chris@42
|
36
|
|
Chris@42
|
37 BENCH_ASSERT(n == dims1[0].n);
|
|
Chris@42
|
38
|
|
Chris@42
|
39 for (i = 0; i < n; ++i) {
|
|
Chris@42
|
40 recur(rnk - 1, dims0 + 1, dims1 + 1, k,
|
|
Chris@42
|
41 indx0, ondx0, indx1, ondx1);
|
|
Chris@42
|
42 indx0 += is0; ondx0 += os0;
|
|
Chris@42
|
43 indx1 += is1; ondx1 += os1;
|
|
Chris@42
|
44 }
|
|
Chris@42
|
45 }
|
|
Chris@42
|
46 }
|
|
Chris@42
|
47
|
|
Chris@42
|
48 void bench_dotens2(const bench_tensor *sz0, const bench_tensor *sz1, dotens2_closure *k)
|
|
Chris@42
|
49 {
|
|
Chris@42
|
50 BENCH_ASSERT(sz0->rnk == sz1->rnk);
|
|
Chris@42
|
51 if (sz0->rnk == BENCH_RNK_MINFTY)
|
|
Chris@42
|
52 return;
|
|
Chris@42
|
53 recur(sz0->rnk, sz0->dims, sz1->dims, k, 0, 0, 0, 0);
|
|
Chris@42
|
54 }
|
