Impact of On-Demand Connection Management in MPI over VIA
Jiesheng Wu, Jiuxing Liu, Pete Wyckoff, Dhabaleswar Panda

Designing scalable and efficient Message Passing Interface (MPI)implementations for emerging cluster interconnects such as VIA-based networks and InfiniBand are important building next generation clusters. In this paper, we address the scalability issue in the implementation of MPI over VIA by on-demand connection management mechanism. The on-demand connection management is designed to limit the use of resources to what applications absolutely require. Thus, the MPI implementation ensures that resource usage scales only as demanded by the application itself, not the underlying system.

We address the design issues of incorporating the on-demand connection mechanism into an implementation of MPI over VIA. A complete implementation was done for MVICH over both cLAN VIA and Berkeley VIA. Performance evaluation on a set of microbenchmarks and NAS parallel benchmarks demonstrates that the on-demand mechanism can increase the scalability of MPI implementations by limiting the use of resources as needed by application.

Furthermore, performance evaluation also shows that the on-demand mechanism delivers comparable performance as the static mechanism in which a fully-connected process model exists in MVICH over cLAN VIA. It even performs better for certain applications. It performs better in MVICH over Berkeley VIA on Myrinet. These results demonstrate that the on-demand connection mechanism is a feasible solution to increase the scalability of MPI implementations over VIA- and InfiniBand-based networks.