Abstract | ||
---|---|---|
Nearly all implementations of the Message Passing Interface (MPI) employ a two-level protocol for point-to-point messages. Short messages are sent eagerly to optimize for latency, and long messages are typically implemented using a rendezvous mechanism. In a rendezvous implementation, the sender must first send a request and receive an acknowledgment before the data can be transferred. While there are several possible reasons for using this strategy for long messages, most implementations are forced to use a rendezvous strategy due to operating system and/or network limitations. In this paper, we compare an implementation that uses a rendezvous protocol for long messages with an implementation that adds an eager optimization for long messages. We discuss implementation issues and provide a performance comparison for several microbenchmarks. We also present a new micro-benchmark that may provide better insight into how these different protocols effect application performance. Results for this new benchmark indicate that, for larger messages, a significant number of receives must be pre-posted in order for an eager protocol optimization to outperform a rendezvous protocol. |
Year | DOI | Venue |
---|---|---|
2003 | 10.1007/978-3-540-39924-7_46 | Lecture Notes in Computer Science |
Keywords | Field | DocType |
point to point,message passing interface,operating system | Latency (engineering),Computer science,Communication source,Implementation,Message Passing Interface,Rendezvous,Transmission protocol,Message passing,Message size,Distributed computing | Conference |
Volume | ISSN | Citations |
2840 | 0302-9743 | 20 |
PageRank | References | Authors |
1.34 | 4 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ron Brightwell | 1 | 1060 | 94.72 |
Keith D. Underwood | 2 | 847 | 77.39 |