LAM/MPI is a cluster-friendly implementation of MPI parallel standard for efficient message passing among nodes.
It is important to note that LAM/MPI is currently in maintenance mode, which means that while critical bug fixes and patches are still available, no new "real" work is being conducted on this software. Instead, the team is dedicating their time primarily to the development of our next-generation MPI implementation - Open MPI. However, LAM/MPI will still continue to operate indefinitely, and all related resources such as web pages and user lists will still be accessible to users.
We would like to encourage all LAM/MPI users to consider migrating to Open MPI. This process should be relatively simple, as Open MPI is also an MPI implementation; thus, users can recompile and re-link their applications to Open MPI and expect them to function as usual. In fact, Open MPI boasts additional features and performance enhancements that are not available in LAM/MPI.
The most recent release of LAM/MPI includes numerous improvements, including man page cleanups, tkill fixes, and a resolution to hangs in 64-bit builds on Mac OS X systems. Additionally, some enhancements have been added to support Fortran types MPI_REAL{4,8,16} for predefined floating point operations, and various errors have been resolved including an "ib" RPI compile error and a problem with MPI_COMM_ACCEPT on Fedora Core 4. Lastly, some checkpoint/restart documentation has been added, and an update to the hostfile comments has been made.
Overall, LAM/MPI is a reliable and efficient MPI implementation that is designed specifically for clusters. While development and progress in LAM/MPI have slowed down, users can still benefit from it, and the transition to Open MPI is highly recommended.
Version 7.1.4: N/A