1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
|
/*
* $Id: gwyomp.h 22348 2019-07-31 11:04:26Z yeti-dn $
* Copyright (C) 2018-2019 David Necas (Yeti).
* E-mail: yeti@gwyddion.net.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor,
* Boston, MA 02110-1301, USA.
*/
/*< private_header >*/
#ifndef __GWY_OMP_H__
#define __GWY_OMP_H__
#include "config.h"
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
#include <glib.h>
#include <libgwyddion/gwymacros.h>
#include <libgwyddion/gwythreads.h>
#include <libgwyddion/gwyutils.h>
G_BEGIN_DECLS
#ifdef _OPENMP
#include <omp.h>
#define gwy_omp_max_threads() \
(gwy_threads_are_enabled() ? omp_get_max_threads() : 1)
#define gwy_omp_num_threads() \
(gwy_threads_are_enabled() ? omp_get_num_threads() : 1)
#define gwy_omp_thread_num omp_get_thread_num
#define gwy_omp_chunk_start(size) \
(omp_get_thread_num()*(size)/gwy_omp_num_threads())
#define gwy_omp_chunk_end(size) \
((omp_get_thread_num() + 1)*(size)/gwy_omp_num_threads())
#else
#define gwy_omp_max_threads() 1
#define gwy_omp_num_threads() 1
#define gwy_omp_thread_num() 0
#define gwy_omp_chunk_start(size) 0
#define gwy_omp_chunk_end(size) (size)
#endif
/*
* Helpers for common array reductions -- which we do not use from OpenMP
* because, first, they require OpenMP 4.5 and, second, I observed mixed
* performance results when I tried them, so I do not trust the compiler
* support much so far.
*
* The allocator reduces to no-op in the single-thread case. We might want to
* use an aligned memory allocator to avoid false sharing (cache line
* cross-talk)...
*
* The reduction helpers should be paired with it and free the array the
* allocator created, making the code transparent.
*/
#define gwy_omp_if_threads_new0(a,n) \
(gwy_omp_num_threads() > 1 ? g_malloc0(sizeof((a)[0])*(n)) : (a))
#define gwy_omp_if_threads_new(a,n) \
(gwy_omp_num_threads() > 1 ? g_malloc(sizeof((a)[0])*(n)) : (a))
#define gwy_omp_if_threads_dup(a,n) \
(gwy_omp_num_threads() > 1 ? g_memdup((a), sizeof((a)[0])*(n)) : (a))
#define gwy_omp_if_threads_free(a,ta) \
do { \
if ((ta) != (a)) \
g_free(ta); \
} while (0)
static inline void
gwy_omp_if_threads_sum_double(G_GNUC_UNUSED gdouble *a,
G_GNUC_UNUSED gdouble *ta,
G_GNUC_UNUSED guint n)
{
#ifdef _OPENMP
guint i;
if (ta == a)
return;
#pragma omp critical
for (i = 0; i < n; i++)
a[i] += ta[i];
g_free(ta);
#endif
}
static inline void
gwy_omp_if_threads_sum_uint(G_GNUC_UNUSED guint *a,
G_GNUC_UNUSED guint *ta,
G_GNUC_UNUSED guint n)
{
#ifdef _OPENMP
guint i;
if (ta == a)
return;
#pragma omp critical
for (i = 0; i < n; i++)
a[i] += ta[i];
g_free(ta);
#endif
}
/* Atomic value helpers, used namely for cancellation flags. */
static inline gboolean
gwy_omp_atomic_read_boolean(gboolean *ptr)
{
gboolean value;
#ifdef _OPENMP
#pragma omp atomic read
value = *ptr;
#else
value = *ptr;
#endif
return value;
}
/* XXX XXX XXX: GCC prints
* warning: parameter ‘value’ set but not used
* without the G_GNUC_UNUSED silencer. No idea why. Is this the correct
* way to do this? But I essentially replicated atomic2.c from OpenMP
* examples. */
static inline void
gwy_omp_atomic_write_boolean(gboolean *ptr, G_GNUC_UNUSED gboolean value)
{
#ifdef _OPENMP
#pragma omp atomic write
*ptr = value;
#else
*ptr = value;
#endif
}
static inline gint
gwy_omp_atomic_increment_int(gint *ptr)
{
gint value;
#ifdef _OPENMP
#pragma omp atomic capture
value = (*ptr)++;
#else
value = (*ptr)++;
#endif
return value;
}
static inline guint
gwy_omp_atomic_increment_uint(guint *ptr)
{
guint value;
#ifdef _OPENMP
#pragma omp atomic capture
value = (*ptr)++;
#else
value = (*ptr)++;
#endif
return value;
}
static inline gboolean
gwy_omp_set_fraction_check_cancel(GwySetFractionFunc set_fraction,
gint i, gint ifrom, gint ito,
gboolean *pcancelled)
{
/* Cannot cancel without set_fraction() function. */
if (!set_fraction)
return FALSE;
/* Update progress and check cancellation. Do this only in the master
* thread. We do not use pragma master because of nesting limitations.
* Thread numbers can be used freely. */
if (!gwy_omp_thread_num() && !set_fraction((i-ifrom + 1.0)/(ito - ifrom)))
gwy_omp_atomic_write_boolean(pcancelled, TRUE);
/* In all threads, report the cancellation flag. */
return gwy_omp_atomic_read_boolean(pcancelled);
}
G_END_DECLS
#endif /* __GWY_OMP_H__ */
/* vim: set cin et ts=4 sw=4 cino=>1s,e0,n0,f0,{0,}0,^0,\:1s,=0,g1s,h0,t0,+1s,c3,(0,u0 : */
|
