Impact of on-demand connection management in MPI over VIA

Designing scalable and efficient Message Passing Interface (MPI) implementations for emerging cluster interconnects such as VIA-based networks and InfiniBand is important for building next generation clusters. In this paper, we address the scalability issue in implementation of MPI over VIA by an on-demand connection management mechanism. On-demand connection management is designed to limit the use of resources by applications that absolutely require them. We address 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 applications. It also shows that the on-demand mechanism delivers comparable or better performance as the static mechanism in which a fully-connected process model usually exists in the MPI implementations. 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.

[1]  Eleanor Chu,et al.  Minimizing Communication Penalty of Triangular Solvers by Runtime Mesh Configuration and Workload Redistribution , 2004, The Journal of Supercomputing.

[2]  David E. Culler,et al.  An analysis of VI Architecture primitives in support of parallel and distributed communication , 2002, Concurr. Comput. Pract. Exp..

[3]  Greg J. Regnier,et al.  The Virtual Interface Architecture , 2002, IEEE Micro.

[4]  Remzi H. Arpaci-Dusseau,et al.  Architectural Requirements and Scalability of the NAS Parallel Benchmarks , 1999, ACM/IEEE SC 1999 Conference (SC'99).

[5]  Anthony Skjellum,et al.  A High-Performance, Portable Implementation of the MPI Message Passing Interface Standard , 1996, Parallel Comput..

[6]  Wolfgang Rehm,et al.  Comparing MPI Performance of SCI and VIA , 2000 .

[7]  Reza Rooholamini,et al.  Architectural and Performance Evaluation of GigaNet and Myrinet Interconnects on Clusters of Small-Scale SMP Servers , 2000, ACM/IEEE SC 2000 Conference (SC'00).

[8]  Jeffrey S. Vetter,et al.  Communication characteristics of large-scale scientific applications for contemporary cluster architectures , 2002, Proceedings 16th International Parallel and Distributed Processing Symposium.

[9]  Monica Reggiani,et al.  Design of a VIA based communication protocol for LAM/MPI suite , 2001, Proceedings Ninth Euromicro Workshop on Parallel and Distributed Processing.

[10]  Ronald B. Brightwell,et al.  Scalability limitations of VIA-based technologies in supporting MPI , 2000 .

[11]  David E. Culler,et al.  An Implementation and Analysis of the Virtual Interface Architecture , 1998, Proceedings of the IEEE/ACM SC98 Conference.

[12]  Message P Forum,et al.  MPI: A Message-Passing Interface Standard , 1994 .

[13]  Message Passing Interface Forum MPI: A message - passing interface standard , 1994 .

[14]  Dhabaleswar K. Panda,et al.  VIBe: a micro-benchmark suite for evaluating virtual interface architecture (VIA) implementations , 2001, Proceedings 15th International Parallel and Distributed Processing Symposium. IPDPS 2001.

[15]  Hong Ong,et al.  Performance Comparison of LAM/MPI, MPICH, and MVICH on a Linux Cluster Connected by a Gigabit Ethernet Network , 2000, Annual Linux Showcase & Conference.