A parallel approximate string matching under Levenshtein distance on graphics processing units using warp-shuffle operations

Thienluan Ho,Seung-Rohk Oh,Hyunjin Kim,Quan Zou

doi:10.1371/journal.pone.0186251

Abstract

Approximate string matching with k-differences has a number of practical applications, ranging from pattern recognition to computational biology. This paper proposes an efficient memory-access algorithm for parallel approximate string matching with k-differences on Graphics Processing Units (GPUs). In the proposed algorithm, all threads in the same GPUs warp share data using warp-shuffle operation instead of accessing the shared memory. Moreover, we implement the proposed algorithm by exploiting the memory structure of GPUs to optimize its performance. Experiment results for real DNA packages revealed that the performance of the proposed algorithm and its implementation archived up to 122.64 and 1.53 times compared to that of sequential algorithm on CPU and previous parallel approximate string matching algorithm on GPUs, respectively.

Highlights

Approximate string matching (ASM) has been widely applied in many fields, including network intrusion detection systems, voice recognition, web searching, and computational biology [1,2,3]
There is a popular method for ASM that allows three edit operations of insertion, deletion, and substitution to transform a factor of the input string into the pattern, which is called ASM with differences
This method is called as ASM with edit distances or ASM with Levenshtein distance [5]

Summary

Introduction

Approximate string matching (ASM) has been widely applied in many fields, including network intrusion detection systems, voice recognition, web searching, and computational biology [1,2,3]. The paper in [4] proposed an algorithm with a parallel scheme difference from the works in [16, 17, 19], where all elements in the same row of edit distance matrix could be calculated in parallel by eliminating data dependency This parallel scheme achieved the maximum number of threads processed in the same time up to the length of input string. Traditional sequential algorithm for solving ASM with k-differences is developed based on dynamic programming model In this case, the edit distance matrix D of input string T and pattern P is calculated, where each element contains the minimum edit operations between a factor of T and a factor of P. X1⁄2i; j À 1; otherwise: Algorithm 1 Parallel algorithms for ASM with k-differences

2: Output

Experimental results

Conclusion