Implementation of graph algorithms over GPU: A comparative analysis

Abstract

GPU (Graphics Processing Unit) provides high computational speed at a very low cost as compared to high end systems. The field of parallel processing using GPU is advancing very fast with a new technology being introduced in the field every day. With such advancements, it is necessary to review the major works done in this field. Graph traversal is one of the major challenges in this field. So far comprehensive review work covering major areas of the graph traversal techniques have not been reported much. This paper aims at surveying the implementation of different graph traversal techniques (viz. BFS, SSSP, APSP) using CUDA architecture and GPU.