Home // AP2PS 2011, The Third International Conference on Advances in P2P Systems // View article

An Empirical Study of MPI over PC Clusters

Authors:
Fazal Noor
Majed Alhaisoni
Antonio Liotta

Keywords: MPI benchmark, Homogeneous PC Cluster, Ethernet, Collective Communictions, Latency, Bandwidth, All-to-all, Gather, Scatter, Broadcast.

Abstract:
Message Passing Interface (MPI) is an important mechanism in P2P. Herein we assess how different types of MPI collective communication functions perform on a Gigabit Ethernet Homogeneous Beowulf PC cluster. In this way we provide an insight on the factors that affect P2P applications over an enterprise context such as the emerging Cloud-based services. By contrast to the literature, which includes mostly theoretical studies, we carry out an empirical study. We show that in case of gather and scatter, the message size is decreased among the nodes where a minimum round time (MRT) is achieved. However, further decreasing the message among increasing number of nodes will again start increasing the MRT

Pages: 65 to 70

Copyright: Copyright (c) IARIA, 2011

Publication date: November 20, 2011

Published in: conference

ISBN: 978-1-61208-173-1

Location: Lisbon, Portugal

Dates: from November 20, 2011 to November 25, 2011