Modelling of Communication Contention in Multiprocessors

This paper deals with performance evaluation and modelling of point-to-point communications in parallel machines with distributed memory. Existing models of point-to-point communications in contention-free networks are presented. A major drawback of these models is that they do not reflect the behavior of the network during the execution of a real parallel application. A general methodology for studying the effect of contention on point-to-point communications under “reasonable” load is presented here. This methodology has been applied on the software router VCR on a Meganode.

[1]  Roger W. Hockney,et al.  Performance parameters and benchmarking of supercomputers , 1991, Parallel Comput..

[2]  Michel Syska.  Communication dans les architectures à mémoire distribuée , 1992 .

[3]  Anthony J. G. Hey,et al.  The Genesis distributed memory benchmarks , 1991, Parallel Comput..

[4]  Jack Dongarra,et al.  Environments and Tools for Parallel Scientific Computing , 1993 .

[5]  Lionel M. Ni,et al.  A survey of wormhole routing techniques in direct networks , 1993, Computer.

[6]  David H. C. Du,et al.  Performance evaluation of the CM-5 interconnection network , 1993, Digest of Papers. Compcon Spring.

[7]  Geoffrey C. Fox,et al.  Benchmarking the CM-5 multicomputer , 1992, [Proceedings 1992] The Fourth Symposium on the Frontiers of Massively Parallel Computation.

[8]  F. H. Mcmahon,et al.  The Livermore Fortran Kernels: A Computer Test of the Numerical Performance Range , 1986 .

[9]  Denis A. Nicole,et al.  Global communications on locally connected message-passing parallel computers , 1993, Concurr. Pract. Exp..

[10]  Jack J. Dongarra,et al.  Performance of various computers using standard linear equations software in a Fortran environment , 1987, SGNM.

[11]  Brigitte Plateau,et al.  ALPES: a tool for the performance evaluation of parallel programs , 1993 .

[12]  Leonard Kleinrock,et al.  Virtual Cut-Through: A New Computer Communication Switching Technique , 1979, Comput. Networks.

[13]  Geoffrey C. Fox,et al.  Communication overhead on the CM5: an experimental performance evaluation , 1992, [Proceedings 1992] The Fourth Symposium on the Frontiers of Massively Parallel Computation.

[14]  J. W Cohen.  Introduction aux réseaux de files d'attente: E. GELENBE and G. PUJOLLE Collection Techniques et Scientifique des Télécommunications, Eyrolles, Paris, 1982, ix + 192 pages , 1983 .

[15]  Edward S. Davidson,et al.  Evaluating the communication performance of MPPs using synthetic sparse matrix multiplication workloads , 1993, ICS '93.

[16]  Dirk Roose,et al.  Benchmarking the iPSC/2 Hypercube Multiprocessor , 1989, Concurr. Pract. Exp..