Benchmarking IBM SP1 system for SPMD programming

The IBM SP1 is the first member of the IBM Scalable POWERparallel series, a distributed memory parallel computer based on RISC System/6000 processing element. In this paper, the benchmarking exercise of two message passing libraries, MPL and PVM, on the IBM SP1 for SPMD programming is described. We will discuss the benchmarks used in our experiment, and present the results we obtained. Our results indicate that to achieve performance improvement and to reduce the communication overhead, the use of the high performance switch is essential on the IBM SP1 machine.

[1]  W. Daniel Hillis,et al.  Data parallel algorithms , 1986, CACM.

[2]  Jack J. Dongarra,et al.  Performance of various computers using standard linear equations software in a FORTRAN environment , 1988, CARN.

[3]  Vaidy S. Sunderam,et al.  Performance of the NAS Parallel Benchmarks on PVM-Based Networks , 1995, J. Parallel Distributed Comput..

[4]  Philip J. Hatcher,et al.  Data-Parallel Programming on MIMD Computers , 1991, IEEE Trans. Parallel Distributed Syst..

[5]  William Gropp,et al.  Early experiences with the IBM SP1 and the high-performance switch , 1993 .

[6]  Philip J. Hatcher,et al.  Data-parallel programming on multicomputers , 1990, IEEE Software.

[7]  David H. Bailey,et al.  The Nas Parallel Benchmarks , 1991, Int. J. High Perform. Comput. Appl..

[8]  Craig B. Stunkel,et al.  The SP1 high-performance switch , 1994, Proceedings of IEEE Scalable High Performance Computing Conference.