Compiling APL for parallel and vector execution

  • Authors:
  • Timothy A. Budd;Rajeev K. Pandey

  • Affiliations:
  • Department of Computer Science, Oregon State University, Corvallis, Oregon;Department of Computer Science, Oregon State University, Corvallis, Oregon

  • Venue:
  • APL '91 Proceedings of the international conference on APL '91
  • Year:
  • 1991

Quantified Score

Hi-index 0.00

Visualization

Abstract

The inherent parallelism of applicative languages such as APL and functional languages such as FP present a little-exploited and somewhat unorthodox means of parallel programming. Here we summarize our investigation of a new approach to compiling such programs for execution on various types of parallel hardware. Our method centers around an intermediate form that is an extension of the lambda calculus. We present evidence that APL programs are easily translated into this intermediate form, and that this intermediate representation lends itself readily to code generation for a variety of parallel hardware.