Parallelizing Transformations for a Concurrent Rule Execution Language

  • Authors:
  • Daniel P. Miranker;Chin Kuo;James C. Browne

  • Affiliations:
  • -;-;-

  • Venue:
  • Parallelizing Transformations for a Concurrent Rule Execution Language
  • Year:
  • 1989

Quantified Score

Hi-index 0.00

Visualization

Abstract

Most work on parallelizing forward-chaining production system programs may be described as parallelizing sequential production system interpreters. We are now studying an approach that parallelizes the entire execution of a production system program. We first modify the semantics of the OPS5 production system language into a rule language suitable for parallel execution. We then define a compilation method that partitions rule systems into disjoint subsets that execute asynchronously with respect to each other and which communicate through asynchronous message passing. Within each subset rules may be fired in parallel. The approach is similar to the methods used in parallelizing compilers for block structured languages and is founded on the formalisms developed for assuring the correct operation of concurrent database systems. Our primary results to date involve the definition of the syntax and semantics of a parallel production system language and the development of an ensemble of optimizing transforms.