Communications latency hiding techniques for a reconfigurable optical interconnect: Benchmark studies
Communication overhead adversely affects the performance of multi-computers. In this work, we present evidence (through the analysis of several parallel benchmarks) that there exists communications locality, and that it is “structured”. We have used this in a number of heuristics that “predict” the target of subsequent communications. This technique, can be applied directly to reconfigurable interconnects (optical or conventional) to hide the communications latency by reconfiguring the interconnect concurrently to the computation.
Unable to display preview. Download preview PDF.
- 1.A. Afsahi and N. J. Dimopoulos, “Collective Communications on a Reconfigurable Optical Interconnect,” Proceedings of the International Conference on Principles of Distributed Systems, Dec., 1997, pp. 167–181.Google Scholar
- 2.A. Afsahi and N. J. Dimopoulos “Communications Latency Hidinng Techniques for a Reconfigurable Optical Interconnect: Benchmark Studies” Technical Report ECE-98-2, Department of Electrical and Computer Engineering, University of Victoria, June 1998.Google Scholar
- 3.D. H. Bailey, et al., “NAS Parallel Benchmark Result 3-94,” Proceedings of Scalable High-Performance Computing Conference, 1994, pp. 111–120Google Scholar
- 5.B. V. Dao, S. Yalamanchili, and J. Duato, “Architectural Support for Reducing Communication Overhead in Multiprocessor Interconnection Networks” Proceedings, Third International Symposium on High Performance Computer Architecture, 1997, pp. 343–352Google Scholar
- 7.T. V. Eicken, et al., “Active Messages: A Mechanism for Integrated Communication and Computation,” Proceedings of the 19th Annual International Symposium on Computer Architecture, May 1992, pp. 256–265Google Scholar
- 8.W. Gropp and E. Lusk, “User’s Guide for MPICH, a Portable Implementation of MPI,” Argonne National Laboratory, Mathematics and Computer Science Division, ANL/MCS-TM-ANL-96/6Google Scholar
- 9.J. Kim and D. J. Lilja, “Characterization of Communication Patterns in Message-Passing Parallel Scientific Application Programs, “Workshop on Communication, Architecture, and Applications for Network-based Parallel Computing, International Symposium on High Performance Computer Architecture, February 1998, pp. 202–216Google Scholar
- 10.D. G. de Lahaut and C. Germain, “Static Communications in Parallel Scientific Programs” Proceedings of PARLE’94, Parallel Architecture and Languages, Athen, Greece, July 1994Google Scholar
- 11.Message Passing Interface Forum: MPI: A Message-Passing Interface Standard. Version 1.1 (June 1995)Google Scholar
- 13.R. A. Nordin, et al., “A System Perspective on Digital Interconnection Technology,” IEEE Journal of Lightwave Technology, Vol. 10, June 1992, pp. 801–827Google Scholar
- 14.S. Pakin, M. Lauria, and A. Chien, “High Performance Messaging on Workstation: Illinois Fast Messages (FM) for Myrinet,” Proceedings of Supercomputing’95, Nov., 1995.Google Scholar
- 16.X. Yuan, R. Melhem, and R. Gupta, “Compiled Communication for All-Optical TDM Networks,” Proceeding’s of Supercomputing’96, 1996Google Scholar