Optimising Data-Parallel Programs Using the BSP Cost Model

David Skillicorn and Marco Danelutto and Susanna Pelagatti and Andrea Zavanella

Abstract
We describe the use of the BSP cost model to optimiseprograms using skeletons or data-parallel operations, in whichprogram components may have multiple implementations. The use of BSPtransforms the problem of finding the best implementation choice foreach component that minimises overall execution time into a one-dimensionalminimisation problem. An algorithm which finds optimal implementationsin time linear in the length of the program is given.
Contact
David Skillicorn
Computing and Information Science,Queen's University,Kingston Ontario,CANADA K7L 3N6
skill@qucis.queensu.ca