libstdc++
par_loop.h
Go to the documentation of this file.
00001 // -*- C++ -*-
00002 
00003 // Copyright (C) 2007, 2008, 2009 Free Software Foundation, Inc.
00004 //
00005 // This file is part of the GNU ISO C++ Library.  This library is free
00006 // software; you can redistribute it and/or modify it under the terms
00007 // of the GNU General Public License as published by the Free Software
00008 // Foundation; either version 3, or (at your option) any later
00009 // version.
00010 
00011 // This library is distributed in the hope that it will be useful, but
00012 // WITHOUT ANY WARRANTY; without even the implied warranty of
00013 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00014 // General Public License for more details.
00015 
00016 // Under Section 7 of GPL version 3, you are granted additional
00017 // permissions described in the GCC Runtime Library Exception, version
00018 // 3.1, as published by the Free Software Foundation.
00019 
00020 // You should have received a copy of the GNU General Public License and
00021 // a copy of the GCC Runtime Library Exception along with this program;
00022 // see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
00023 // <http://www.gnu.org/licenses/>.
00024 
00025 /** @file parallel/par_loop.h
00026  *  @brief Parallelization of embarrassingly parallel execution by
00027  *  means of equal splitting.
00028  *  This file is a GNU parallel extension to the Standard C++ Library.
00029  */
00030 
00031 // Written by Felix Putze.
00032 
00033 #ifndef _GLIBCXX_PARALLEL_PAR_LOOP_H
00034 #define _GLIBCXX_PARALLEL_PAR_LOOP_H 1
00035 
00036 #include <omp.h>
00037 #include <parallel/settings.h>
00038 #include <parallel/base.h>
00039 #include <parallel/equally_split.h>
00040 
00041 namespace __gnu_parallel
00042 {
00043   /** @brief Embarrassingly parallel algorithm for random access
00044    * iterators, using hand-crafted parallelization by equal splitting
00045    * the work.
00046    *
00047    *  @param __begin Begin iterator of element sequence.
00048    *  @param __end End iterator of element sequence.
00049    *  @param __o User-supplied functor (comparator, predicate, adding
00050    *  functor, ...)
00051    *  @param __f Functor to "process" an element with __op (depends on
00052    *  desired functionality, e. g. for std::for_each(), ...).
00053    *  @param __r Functor to "add" a single __result to the already
00054    *  processed elements (depends on functionality).
00055    *  @param __base Base value for reduction.
00056    *  @param __output Pointer to position where final result is written to
00057    *  @param __bound Maximum number of elements processed (e. g. for
00058    *  std::count_n()).
00059    *  @return User-supplied functor (that may contain a part of the result).
00060    */
00061   template<typename _RAIter,
00062        typename _Op,
00063        typename _Fu,
00064        typename _Red,
00065        typename _Result>
00066     _Op
00067     __for_each_template_random_access_ed(_RAIter __begin, _RAIter __end,
00068                      _Op __o, _Fu& __f, _Red __r,
00069                      _Result __base, _Result& __output,
00070       typename std::iterator_traits<_RAIter>::difference_type __bound)
00071     {
00072       typedef std::iterator_traits<_RAIter> _TraitsType;
00073       typedef typename _TraitsType::difference_type _DifferenceType;
00074       const _DifferenceType __length = __end - __begin;
00075       _Result *__thread_results;
00076       bool* __constructed;
00077 
00078       _ThreadIndex __num_threads = __gnu_parallel::min<_DifferenceType>
00079     (__get_max_threads(), __length);
00080 
00081 #     pragma omp parallel num_threads(__num_threads)
00082       {
00083 #       pragma omp single
00084     {
00085       __num_threads = omp_get_num_threads();
00086       __thread_results = static_cast<_Result*>
00087         (::operator new(__num_threads * sizeof(_Result)));
00088       __constructed = new bool[__num_threads];
00089     }
00090 
00091     _ThreadIndex __iam = omp_get_thread_num();
00092 
00093     // Neutral element.
00094     _Result* __reduct;
00095 
00096     _DifferenceType
00097       __start = equally_split_point(__length, __num_threads, __iam),
00098       __stop = equally_split_point(__length, __num_threads, __iam + 1);
00099 
00100     if (__start < __stop)
00101       {
00102         __reduct = new _Result(__f(__o, __begin + __start));
00103         ++__start;
00104         __constructed[__iam] = true;
00105       }
00106     else
00107       __constructed[__iam] = false;
00108 
00109     for (; __start < __stop; ++__start)
00110       *__reduct = __r(*__reduct, __f(__o, __begin + __start));
00111 
00112     if (__constructed[__iam])
00113       {
00114         ::new(&__thread_results[__iam]) _Result(*__reduct);
00115         delete __reduct;
00116       }
00117       } //parallel
00118 
00119       for (_ThreadIndex __i = 0; __i < __num_threads; ++__i)
00120     if (__constructed[__i])
00121       {
00122         __output = __r(__output, __thread_results[__i]);
00123         __thread_results[__i].~_Result();
00124       }
00125 
00126       // Points to last element processed (needed as return value for
00127       // some algorithms like transform).
00128       __f._M_finish_iterator = __begin + __length;
00129 
00130       ::operator delete(__thread_results);
00131 
00132       delete[] __constructed;
00133 
00134       return __o;
00135     }
00136 
00137 } // end namespace
00138 
00139 #endif /* _GLIBCXX_PARALLEL_PAR_LOOP_H */