Designing a parallel suffix sort

TL;DR

This paper introduces a parallel suffix sorting method leveraging generalized polynomial factorization, enabling efficient sorting of suffixes with many repeated characters by using parallel threads and shared memory.

Contribution

It proposes a novel parallel suffix sorting algorithm based on polynomial factorization, improving efficiency in handling repeated characters in strings.

Findings

Efficient parallel implementation using shared memory.

Predefined factors facilitate polynomial sorting.

Applicable to genomics and software applications.

Abstract

Suffix sort plays a critical role in various computational algorithms including genomics as well as in frequently used day to day software applications. The sorting algorithm becomes tricky when we have lot of repeated characters in the string for a given radix. Various innovative implementations are available in this area e.g., Manber Myers. We present here an analysis that uses a concept around generalized polynomial factorization to sort these suffixes. The initial generation of these substring specific polynomial can be efficiently done using parallel threads and shared memory. The set of distinct factors and their order are known beforehand, and this helps us to sort the polynomials (equivalent of strings) accordingly.