On the Implementation of Applicative Languages on Shared-Memory, MIMD Multiprocessors

This paper presents the performance of a set of algorithms written in SISAL [MSA*85] and run on multiprocessor Sequent, DEC, and Cray computers. We describe our current runtime system and discuss its implementation on each machine. We indicate where our automatic approach to parallelization works well, as well as sources of inefficient behavior. Overall we find our systems encouraging for the first release of the native-code generating software. We suggest improvements to the compiler and runtime systems, many already developed but not yet implemented, to alleviate the inefficiencies.

[1]  Peter M. Kogge,et al.  Parallel Solution of Recurrence Problems , 1974, IBM J. Res. Dev..

[2]  Harold S. Stone,et al.  A Parallel Algorithm for the Efficient Solution of a General Class of Recurrence Equations , 1973, IEEE Transactions on Computers.

[3]  M. H. Reilly,et al.  M31: a large-scale multiprocessor VAX for parallel processing research , 1988, Digest of Papers. COMPCON Spring 88 Thirty-Third IEEE Computer Society International Conference.

[4]  R. R. Oldehoeft,et al.  DI: an interactive debugging interpreter for applicative languages , 1987, PLDI.

[5]  Arvind,et al.  Future Scientific Programming on Parallel Machines , 1988, J. Parallel Distributed Comput..

[6]  David C. Cann,et al.  Applicative parallelism on a shared-memory multiprocessor , 1988, IEEE Software.