Parallel ocean modeling using Glenda

ABSTRACT

This paper will discuss our use of Glenda to parallelize the ocean modeling program SWEM. Following an introduction to both SWEM and Glenda, we will describe the details of the parallelization process. We also discuss many of the problems we encountered, as well as the debugging methods we used to resolve them. Finally, we describe our test results and discuss how our model compared with another parallel version of SWEM.

REFERENCES

• 1: Robert D. Bjornson, Linda on distributed memory multiprocessors, Yale University, New Haven, CT, 1993
• 2: Adam Beguelin, Jack Dongarra, Al Geist, Robert Manchek and Vaidy Sunderam, "A User's Guide to PVM", Oak Ridge National Laboratory, Oak Ridge, TN, Jul. 1991.
• 3: Nicholas John Carriero, Jr., Implementation of tuple space machines, 1987
• 4: Chris Hogue, "POWER Fortran Accelerator#8482; User's Guide", Silicon Graphics, Inc., Mountain View, CA, Jan. 1992.
• 5: Germana Peggion, University of Southern Mississippi Center for Ocean and Atmospheric Modeling, personal communications, 1994.
• 6: B. R. Seyfarth, J. L. Bickham and M. R. Fernandez, "Glenda: An Environment for Easy Parallel Programming", Scalable High Performance Computing Conference, Knoxville, TN, May. 1994.
• 7: Nicholas Carriero , David Gelernter, How to write parallel programs: a first course, MIT Press, Cambridge, MA, 1990

INDEX TERMS

Keywords:
PVM, SWEM, array dependency, glenda, linda, ocean dynamics, parallel, tuple