Abstract | ||
---|---|---|
It is well known that traditional micro-benchmarks do not fully capture the salient architectural features that impact application performance. Even worse, micro-benchmarks that target MPI and the communications sub-system do not accurately represent the way that applications use MPI. For example, traditional MPI latency benchmarks time a ping-pong communication with one send and one receive on each of two nodes. The time to post the receive is never counted as part of the latency. This scenario is not even marginally representative of most applications. Two new micro-benchmarks are presented here that analyze network latency in a way that more realistically represents the way that MPI is typically used. These benchmarks are used to evaluate modern high-performance networks, including Quadrics, InfiniBand, and Myrinet. |
Year | DOI | Venue |
---|---|---|
2004 | 10.1109/ICPP.2004.73 | ICPP |
Keywords | Field | DocType |
new micro-benchmarks,network latency,modern high-performance network,impact application performance,ping-pong communication,traditional micro-benchmarks,traditional mpi latency benchmarks,message latency,salient architectural feature,mpi queue usage,target mpi,message passing,communication complexity,benchmark testing,queueing theory | InfiniBand,Computer science,Latency (engineering),Computer network,Queueing theory,Message passing,Benchmark (computing),Distributed computing,Parallel computing,Queue,Communication complexity,Myrinet,Operating system | Conference |
ISSN | ISBN | Citations |
0190-3918 | 0-7695-2197-5 | 27 |
PageRank | References | Authors |
2.09 | 13 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Keith D. Underwood | 1 | 847 | 77.39 |
Ron Brightwell | 2 | 1060 | 94.72 |