1// -*- C++ -*-
2
3// Copyright (C) 2007-2015 Free Software Foundation, Inc.
4//
5// This file is part of the GNU ISO C++ Library.  This library is free
6// software; you can redistribute it and/or modify it under the terms
7// of the GNU General Public License as published by the Free Software
8// Foundation; either version 3, or (at your option) any later
9// version.
10
11// This library is distributed in the hope that it will be useful, but
12// WITHOUT ANY WARRANTY; without even the implied warranty of
13// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14// General Public License for more details.
15
16// Under Section 7 of GPL version 3, you are granted additional
17// permissions described in the GCC Runtime Library Exception, version
18// 3.1, as published by the Free Software Foundation.
19
20// You should have received a copy of the GNU General Public License and
21// a copy of the GCC Runtime Library Exception along with this program;
22// see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
23// <http://www.gnu.org/licenses/>.
24
25/** @file parallel/omp_loop_static.h
26 *  @brief Parallelization of embarrassingly parallel execution by
27 *  means of an OpenMP for loop with static scheduling.
28 *  This file is a GNU parallel extension to the Standard C++ Library.
29 */
30
31// Written by Felix Putze.
32
33#ifndef _GLIBCXX_PARALLEL_OMP_LOOP_STATIC_H
34#define _GLIBCXX_PARALLEL_OMP_LOOP_STATIC_H 1
35
36#include <omp.h>
37
38#include <parallel/settings.h>
39#include <parallel/basic_iterator.h>
40
41namespace __gnu_parallel
42{
43  /** @brief Embarrassingly parallel algorithm for random access
44   * iterators, using an OpenMP for loop with static scheduling.
45   *
46   *  @param __begin Begin iterator of element sequence.
47   *  @param __end End iterator of element sequence.
48   *  @param __o User-supplied functor (comparator, predicate, adding
49   *  functor, ...).
50   *  @param __f Functor to @a process an element with __op (depends on
51   *  desired functionality, e. g. for std::for_each(), ...).
52   *  @param __r Functor to @a add a single __result to the already processed
53   *  __elements (depends on functionality).
54   *  @param __base Base value for reduction.
55   *  @param __output Pointer to position where final result is written to
56   *  @param __bound Maximum number of elements processed (e. g. for
57   *  std::count_n()).
58   *  @return User-supplied functor (that may contain a part of the result).
59   */
60  template<typename _RAIter,
61	   typename _Op,
62	   typename _Fu,
63	   typename _Red,
64	   typename _Result>
65    _Op
66    __for_each_template_random_access_omp_loop_static(_RAIter __begin,
67						      _RAIter __end, _Op __o,
68						      _Fu& __f, _Red __r,
69						      _Result __base,
70						      _Result& __output,
71      typename std::iterator_traits<_RAIter>::difference_type __bound)
72    {
73      typedef typename std::iterator_traits<_RAIter>::difference_type
74	_DifferenceType;
75
76      _DifferenceType __length = __end - __begin;
77      _ThreadIndex __num_threads = std::min<_DifferenceType>
78	(__get_max_threads(), __length);
79
80      _Result *__thread_results;
81
82#     pragma omp parallel num_threads(__num_threads)
83      {
84#       pragma omp single
85	{
86	  __num_threads = omp_get_num_threads();
87	  __thread_results = new _Result[__num_threads];
88
89	  for (_ThreadIndex __i = 0; __i < __num_threads; ++__i)
90	    __thread_results[__i] = _Result();
91	}
92
93        _ThreadIndex __iam = omp_get_thread_num();
94
95#pragma omp for schedule(static, _Settings::get().workstealing_chunk_size)
96        for (_DifferenceType __pos = 0; __pos < __length; ++__pos)
97          __thread_results[__iam] = __r(__thread_results[__iam],
98                                        __f(__o, __begin+__pos));
99      } //parallel
100
101      for (_ThreadIndex __i = 0; __i < __num_threads; ++__i)
102	__output = __r(__output, __thread_results[__i]);
103
104      delete [] __thread_results;
105
106      // Points to last element processed (needed as return value for
107      // some algorithms like transform).
108      __f.finish_iterator = __begin + __length;
109
110      return __o;
111    }
112
113} // end namespace
114
115#endif /* _GLIBCXX_PARALLEL_OMP_LOOP_STATIC_H */
116