Teaching parallel algorithm with process topologies

  • Authors:

  • Chris McDonald;Kamran Kazemi

  • Affiliations:

  • Programming, Languages and Systems Group, Department of Computer Science, The University of Western Australia, Nedlands, Western Australia, 6907;Programming, Languages and Systems Group, Department of Computer Science, The University of Western Australia, Nedlands, Western Australia, 6907

  • Venue:
  • Proceedings of the thirty-first SIGCSE technical symposium on Computer science education
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

Parallel algorithms are often introduced to students by describing the geometric topologies formed by communicating processes and often the geographic relationships between them. However, the two most common message passing environments used in teaching, PVM and MPI, each provide only rudimentary support for the specification and execution of process topologies. There is a strong need for better syntactic and semantic support for process topologies in these environments, so that students may concentrate on the algorithms being studied, and not have to wrestle with the environments' infrastructure. This paper first motivates, and then describes the use of additional support within PVM and MPI which addresses this need.