Placing processes onto nodes

The placement of processes for a parallel task within ParaStation MPI can be biased by various environment variables. Setting this variables is optional, if not set, the default behavior will make sure, that the usability of the cluster will not be jeopardized.

The user can predefine lists of nodes, can request exclusive use or overbooking of nodes, may change sorting criteria for node lists, or even may control process neighborhood on SMP nodes. Consequently, it is possible to precisely adjust the process distribution within the cluster to algorithms implemented and thus gain a maximum of performance out of the parallel task.

In addition, the administrator may define partitions reserved for users or group of users.

The manual page process_placement(7) will discuss all this parameters in detail.