partial_sum.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/partial_sum.h
00026  *  @brief Parallel implementation of std::partial_sum(), i. e. prefix
00027  *  sums.
00028  *  This file is a GNU parallel extension to the Standard C++ Library.
00029  */
00030 
00031 // Written by Johannes Singler.
00032 
00033 #ifndef _GLIBCXX_PARALLEL_PARTIAL_SUM_H
00034 #define _GLIBCXX_PARALLEL_PARTIAL_SUM_H 1
00035 
00036 #include <omp.h>
00037 #include <new>
00038 #include <bits/stl_algobase.h>
00039 #include <parallel/parallel.h>
00040 #include <parallel/numericfwd.h>
00041 
00042 namespace __gnu_parallel
00043 {
00044   // Problem: there is no 0-element given.
00045 
00046 /** @brief Base case prefix sum routine.
00047   *  @param begin Begin iterator of input sequence.
00048   *  @param end End iterator of input sequence.
00049   *  @param result Begin iterator of output sequence.
00050   *  @param bin_op Associative binary function.
00051   *  @param value Start value. Must be passed since the neutral
00052   *  element is unknown in general.
00053   *  @return End iterator of output sequence. */
00054 template<typename InputIterator,
00055      typename OutputIterator,
00056      typename BinaryOperation>
00057   OutputIterator
00058   parallel_partial_sum_basecase(InputIterator begin, InputIterator end,
00059                 OutputIterator result, BinaryOperation bin_op,
00060                 typename std::iterator_traits
00061                 <InputIterator>::value_type value)
00062   {
00063     if (begin == end)
00064       return result;
00065 
00066     while (begin != end)
00067       {
00068         value = bin_op(value, *begin);
00069         *result = value;
00070         ++result;
00071         ++begin;
00072       }
00073     return result;
00074   }
00075 
00076 /** @brief Parallel partial sum implementation, two-phase approach,
00077     no recursion.
00078     *  @param begin Begin iterator of input sequence.
00079     *  @param end End iterator of input sequence.
00080     *  @param result Begin iterator of output sequence.
00081     *  @param bin_op Associative binary function.
00082     *  @param n Length of sequence.
00083     *  @param num_threads Number of threads to use.
00084     *  @return End iterator of output sequence.
00085     */
00086 template<typename InputIterator,
00087      typename OutputIterator,
00088      typename BinaryOperation>
00089   OutputIterator
00090   parallel_partial_sum_linear(InputIterator begin, InputIterator end,
00091                   OutputIterator result, BinaryOperation bin_op,
00092                   typename std::iterator_traits
00093                   <InputIterator>::difference_type n)
00094   {
00095     typedef std::iterator_traits<InputIterator> traits_type;
00096     typedef typename traits_type::value_type value_type;
00097     typedef typename traits_type::difference_type difference_type;
00098 
00099     if (begin == end)
00100       return result;
00101 
00102     thread_index_t num_threads =
00103         std::min<difference_type>(get_max_threads(), n - 1);
00104 
00105     if (num_threads < 2)
00106       {
00107         *result = *begin;
00108         return parallel_partial_sum_basecase(
00109             begin + 1, end, result + 1, bin_op, *begin);
00110       }
00111 
00112     difference_type* borders;
00113     value_type* sums;
00114 
00115     const _Settings& __s = _Settings::get();
00116 
00117 #   pragma omp parallel num_threads(num_threads)
00118       {
00119 #       pragma omp single
00120           {
00121             num_threads = omp_get_num_threads();
00122 
00123             borders = new difference_type[num_threads + 2];
00124 
00125             if (__s.partial_sum_dilation == 1.0f)
00126               equally_split(n, num_threads + 1, borders);
00127             else
00128               {
00129                 difference_type chunk_length =
00130                     ((double)n
00131              / ((double)num_threads + __s.partial_sum_dilation)),
00132           borderstart = n - num_threads * chunk_length;
00133                 borders[0] = 0;
00134                 for (int i = 1; i < (num_threads + 1); ++i)
00135                   {
00136                     borders[i] = borderstart;
00137                     borderstart += chunk_length;
00138                   }
00139                 borders[num_threads + 1] = n;
00140               }
00141 
00142             sums = static_cast<value_type*>(::operator new(sizeof(value_type)
00143                                * num_threads));
00144             OutputIterator target_end;
00145           } //single
00146 
00147         thread_index_t iam = omp_get_thread_num();
00148         if (iam == 0)
00149           {
00150             *result = *begin;
00151             parallel_partial_sum_basecase(begin + 1, begin + borders[1],
00152                       result + 1, bin_op, *begin);
00153             ::new(&(sums[iam])) value_type(*(result + borders[1] - 1));
00154           }
00155         else
00156           {
00157             ::new(&(sums[iam]))
00158           value_type(__gnu_parallel::accumulate(begin + borders[iam] + 1,
00159                      begin + borders[iam + 1],
00160                      *(begin + borders[iam]),
00161                      bin_op,
00162                      __gnu_parallel::sequential_tag()));
00163           }
00164 
00165 #       pragma omp barrier
00166 
00167 #       pragma omp single
00168           parallel_partial_sum_basecase(
00169               sums + 1, sums + num_threads, sums + 1, bin_op, sums[0]);
00170 
00171 #       pragma omp barrier
00172 
00173         // Still same team.
00174         parallel_partial_sum_basecase(begin + borders[iam + 1],
00175                       begin + borders[iam + 2],
00176                       result + borders[iam + 1], bin_op,
00177                       sums[iam]);
00178       } //parallel
00179 
00180     ::operator delete(sums);
00181     delete[] borders;
00182 
00183     return result + n;
00184   }
00185 
00186 /** @brief Parallel partial sum front-end.
00187   *  @param begin Begin iterator of input sequence.
00188   *  @param end End iterator of input sequence.
00189   *  @param result Begin iterator of output sequence.
00190   *  @param bin_op Associative binary function.
00191   *  @return End iterator of output sequence. */
00192 template<typename InputIterator,
00193      typename OutputIterator,
00194      typename BinaryOperation>
00195   OutputIterator
00196   parallel_partial_sum(InputIterator begin, InputIterator end,
00197                        OutputIterator result, BinaryOperation bin_op)
00198   {
00199     _GLIBCXX_CALL(begin - end)
00200 
00201     typedef std::iterator_traits<InputIterator> traits_type;
00202     typedef typename traits_type::value_type value_type;
00203     typedef typename traits_type::difference_type difference_type;
00204 
00205     difference_type n = end - begin;
00206 
00207     switch (_Settings::get().partial_sum_algorithm)
00208       {
00209       case LINEAR:
00210         // Need an initial offset.
00211         return parallel_partial_sum_linear(begin, end, result, bin_op, n);
00212       default:
00213     // Partial_sum algorithm not implemented.
00214         _GLIBCXX_PARALLEL_ASSERT(0);
00215         return result + n;
00216       }
00217   }
00218 }
00219 
00220 #endif /* _GLIBCXX_PARALLEL_PARTIAL_SUM_H */

Generated on 19 Jun 2018 for libstdc++ by  doxygen 1.6.1