Solving partial differential equations in a data-driven multiprocessor environment

  • Authors:

  • J. L. Gaudiot;C. M. Lin;M. Hosseiniyar

  • Affiliations:

  • Univ. of Southern California, Los Angeles;Univ. of Southern California, Los Angeles;Univ. of Southern California, Los Angeles

  • Venue:
  • ISCA '88 Proceedings of the 15th Annual International Symposium on Computer architecture
  • Year:
  • 1988

Quantified Score

Hi-index 0.00

Visualization

Abstract

Partial differential equations can be found in a host of engineering and scientific problems. The emergence of new parallel architectures has spurred research in the definition of parallel PDE solvers. Concurrently, highly programmable systems such as data-flow architectures have been proposed for the exploitation of large scale parallelism. The implementation of some Partial Differential Equation solvers (such as the Jacobi method) on a tagged token data-flow graph is demonstrated here. Asynchronous methods (chaotic relaxation) are studied and new scheduling approaches (the Token No-Labeling scheme) are introduced in order to support the implementation of the asynchronous methods in a data-driven environment. New high-level data-flow language program constructs are introduced in order to handle chaotic operations. Finally, the performance of the program graphs is demonstrated by a deterministic simulation of a message passing data-flow multiprocessor. An analysis of the overhead in the data-flow graphs is undertaken to demonstrate the limits of parallel operations in data-flow PDE program graphs.