Parallel BGL logo

The Parallel BGL Software

  |   Home   |   Documentation   |  

The Parallel BGL source code is available as Open Source software under the Boost Software License (details). We do not distribute binary packages at this time.

To build and use the Parallel BGL, you will need:

  • A modern compiler. We have tested the Parallel BGL on GCC versions 3.3, 4.2 (from Apple), and 4.1 on Intel x86, Apple G5, and AMD's Opteron processors as well as version 10.1 of the Intel compiler.
  • A recent copy of Boost. Boost 1.43.0 or newer is required (this is currently the trunk as of 3/8/10).
  • An implementation of MPI-1. We typically test with OpenMPI 1.4, but we have also used MPICH.
  • If you want to build documentation, you will need a Python interpreter and reStructuredText. This step is optional: all documentation is available online.
Experimental releases of the Parallel BGL can be downloaded below as well as being available in Boost 1.40 and later. The source code repository is also open to the public for anonymous, read-only access. See here for more information. The source code repository may contain significant additional functionality relative to released versions of the code, once features in the repository are stable, tested, and documented they will be released.

Version Source code Comments
0.7.0 beta parallel-bgl-0.7.0.tar.gz This updated Parallel BGL release includes a much improved process group, additional algorithms, and improved data structures. Documentation is available on the vast majority of features and examples are included though no tutorials exist as of yet.
0.50.0 alpha parallel-bgl-0.5.0.tar.gz This initial release of the Parallel BGL should be considered experimental. While we have done a significant amount of testing, this is the first public release of this code. In particular, this release lacks tutorials and documentation that would help users to grow accustomed to the Parallel BGL. Please feel free to contact us with any questions or for help getting started.

Experimental active messages implementation.

For message-based PBGL2, see http://www.crest.iu.edu/projects/pbgl2.