Cluster Computing 6, 299–313, 2003
2003 Kluwer Academic Publishers. Manufactured in The Netherlands.
Performance Analysis of a Myrinet-Based Cluster
TEDDY SURYA GUNAWAN and WENTONG CAI
Parallel and Distributed Processing Laboratory, School of Computer Engineering, Nanyang Technological University, Singapore 639798
In recent years, there has been a growing interest in the cluster system as an accepted form of supercomputing, due to its high
performance at an affordable cost. This paper attempts to elaborate performance analysis of Myrinet-based cluster. The communication
performance and effect of background load on parallel applications were analyzed. For point-to-point communication, it was found that an
extension to the Hockney’s model was required to estimate the performance. The proposed model suggested that there should be two ranges
to be used for the performance metrics to cope with the cache effect. Moreover, based on the extension of the point-to-point communication
model, the Xu and Hwang’s model for collective communication performance was also extended. Results showed that our models can make
better estimation of the communication performance than the previous models. Finally, the interference of other user processes to the cluster
system is evaluated by using synthetic background load generation programs.
Myrinet, cluster system, communication performance, user interference, synthetic background load
In recent years, cluster system has gained its popularity as
one of the most powerful computing resources [1–4]. Over
100000 computer clusters are in use worldwide, as estimated
by P±ster . A cluster is a collection of complete computers
(nodes) interconnected by a high-speed network. Typically,
each node is a workstation, PC, or symmetric multiproces-
sor (SMP). Clusters are being used for climate modeling,
synthetic aperture radar processing, environment modeling,
astronomy database, space-²ight mission simulation, image
processing, and antenna design using genetic algorithm .
Now that cluster computing is an accepted form of super-
computing because of its high performance at an affordable
Networked clusters of workstations using communication
environment such as PVM  and MPI [6,7] are becom-
ing faster and cost-effective. MPI has become a commonly
accepted communication standard for specifying message-
passing functions in programming multicomputers or cluster
system . There have been many researches on performance
analysis for communication operations in parallel computing.
Design and implementation of MPI point-to-point and collec-
tive communications has been reported in [9–11] which tar-
geted at Intel Paragon system. The collective communication
operations on the IBM SP2, Cray T3D, and Intel Paragon has
been evaluated in . The communication behavior of In-
tel Paragon for point-to-point and collective communication
operations has been modeled in . In , the MPI per-
formance on the SGI Power Challenge has been described.