An Auto-Parallelizer for Distributed Computing in Haskell

TL;DR

This paper presents a Haskell auto-parallelizer that simplifies the development of scalable, fault-tolerant distributed applications by leveraging Haskell's purity, demonstrated through benchmarking on various examples.

Contribution

It introduces a novel auto-parallelizer for Haskell that provides an easy-to-use interface for distributed computing, exploiting the language's functional purity.

Findings

Effective parallelization of Haskell functions demonstrated

Benchmarks show promising performance improvements

Potential for future enhancements in distributed Haskell applications

Abstract

One of the main challenges in distributed computing is building interfaces and APIs that allow programmers with limited background in distributed systems to write scalable, performant, and fault-tolerant applications on large clusters. In this demonstration, we designed and implemented a Haskell auto-parallelizer with a simple yet powerful interface by taking advantage of the default purity of Haskell functions. Finally, we benchmarked our implementation on a set of examples to illustrate the potential for future work in this direction.