A Parallel Algorithm to Test Chordality of Graphs

TL;DR

This paper introduces a GPU-accelerated parallel algorithm for testing graph chordality, leveraging parallel Lexicographical BFS, achieving efficient performance on multi-core GPU hardware.

Contribution

The paper presents a novel parallel algorithm for chordality testing based on parallel Lexicographical BFS, implemented on CUDA GPU, with performance analysis and comparison to sequential methods.

Findings

Algorithm runs in O(N) time on N-threads machine

Work complexity is O(N^2)

GPU implementation outperforms sequential algorithm

Abstract

We present a simple parallel algorithm to test chordality of graphs which is based on the parallel Lexicographical Breadth-First Search algorithm. In total, the algorithm takes time O(N ) on N-threads machine and it performs work O(N 2 ) , where N is the number of vertices in a graph. Our implementation of the algorithm uses a GPU environment Nvidia CUDA C. The algorithm is implemented in CUDA 4.2 and it has been tested on Nvidia GeForce GTX 560 Ti of compute capability 2.1. At the end of the thesis we present the results achieved by our implementation and compare them with the results achieved by the sequential algorithm