Optimising data-parallel programs using the BSP cost model
We describe the use of the BSP cost model to optimise programs, based on skeletons or data-parallel operations, in which program components may have multiple implementations. BSP’s view of communication transforms the problem of finding the best implementation choice for each component into a one-dimensional minimisation problem. A shortest-path algorithm that finds optimal implementations in time linear in the number of operations of the program is given.
Unable to display preview. Download preview PDF.
- 1.B. Bacci, B. Cantalupo, M. Danelutto, S. Orlando, D. Pasetto, S. Pelagatti, and M. Vanneschi. An environment for structured parallel programming. In L. Grandinetti, M. Kowalick, and M. Vaitersic, editors, Advances in High Performance Computing, pages 219–234. Kluwer, Dordrecht, The Netherlands, 1997.Google Scholar
- 3.S. Ciarpaglini, M. Danelutto, L. Folchi, C. Manconi, and S. Pelagatti. ANACLETO: a template-based p3l compiler. In Proceedings of the Seventh Parallel Computing Workshop (PCW ’97), Australian National University, Canberra, 1997.Google Scholar
- 4.M. Danelutto, F. Pasqualetti, and S. Pelagatti. Skeletons for data parallelism in p3l. In C. Lengauer, M. Griebl, and S. Gorlatch, editors, Proc. of EURO-PAR ’97, Passau, Germany, volume 1300 of LNCS, pages 619–628. Springer-Verlag, August 1997.Google Scholar
- 5.S. Pelagatti. Structured development of parallel programs. Taylor&Francis, London, 1997.Google Scholar
- 6.T. Rauber and G. Rünger. Deriving structured parallel implementations for numerical methods. The Euromicro Journal, (41):589–608, 1996.Google Scholar
- 7.D.B. Skillicorn, J.M.D. Hill, and W.F. McColl. Questions and answers about BSP. Scientific Programming, 6(3):249–274, 1997.Google Scholar
- 8.H.W. To. Optimising the Parallel Behaviour of Combinations of Program Components. PhD thesis, Imperial College, 1995.Google Scholar