Mercurial > hg > sv-dependency-builds
comparison src/fftw-3.3.3/threads/openmp.c @ 10:37bf6b4a2645
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 /* openmp.c: thread spawning via OpenMP */ | |
22 | |
23 #include "threads.h" | |
24 | |
25 #if !defined(_OPENMP) | |
26 #error OpenMP enabled but not using an OpenMP compiler | |
27 #endif | |
28 | |
29 int X(ithreads_init)(void) | |
30 { | |
31 return 0; /* no error */ | |
32 } | |
33 | |
34 /* Distribute a loop from 0 to loopmax-1 over nthreads threads. | |
35 proc(d) is called to execute a block of iterations from d->min | |
36 to d->max-1. d->thr_num indicate the number of the thread | |
37 that is executing proc (from 0 to nthreads-1), and d->data is | |
38 the same as the data parameter passed to X(spawn_loop). | |
39 | |
40 This function returns only after all the threads have completed. */ | |
41 void X(spawn_loop)(int loopmax, int nthr, spawn_function proc, void *data) | |
42 { | |
43 int block_size; | |
44 spawn_data d; | |
45 int i; | |
46 | |
47 A(loopmax >= 0); | |
48 A(nthr > 0); | |
49 A(proc); | |
50 | |
51 if (!loopmax) return; | |
52 | |
53 /* Choose the block size and number of threads in order to (1) | |
54 minimize the critical path and (2) use the fewest threads that | |
55 achieve the same critical path (to minimize overhead). | |
56 e.g. if loopmax is 5 and nthr is 4, we should use only 3 | |
57 threads with block sizes of 2, 2, and 1. */ | |
58 block_size = (loopmax + nthr - 1) / nthr; | |
59 nthr = (loopmax + block_size - 1) / block_size; | |
60 | |
61 THREAD_ON; /* prevent debugging mode from failing under threads */ | |
62 #pragma omp parallel for private(d) | |
63 for (i = 0; i < nthr; ++i) { | |
64 d.max = (d.min = i * block_size) + block_size; | |
65 if (d.max > loopmax) | |
66 d.max = loopmax; | |
67 d.thr_num = i; | |
68 d.data = data; | |
69 proc(&d); | |
70 } | |
71 THREAD_OFF; /* prevent debugging mode from failing under threads */ | |
72 } | |
73 | |
74 void X(threads_cleanup)(void) | |
75 { | |
76 } |