Title :
Fast Diameter Computation of Large Sparse Graphs Using GPUs
Author :
Dal, Giso H. ; Kosters, Walter A. ; Takes, Frank W.
Author_Institution :
Inst. for Comput. & Inf. Sci., Radboud Univ. Nijmegen, Nijmegen, Netherlands
Abstract :
In this paper we propose a highly parallel GPU-based bounding algorithm for computing the exact diameter of large real-world sparse graphs. The diameter is defined as the length of the longest shortest path between vertices in the graph, and serves as a relevant property of all types of graphs that are nowadays frequently studied. Examples include social networks, webgraphs and routing networks. We verify the performance of our parallel approach on a set of large graphs comprised of millions of vertices, and using a CUDA GPU observe an increase in performance of up to 21.1x compared to a CPU algorithm using the same strategy. Based on these results, we provide a characterization of the types of graphs that are well-suited for traversal by means of our parallel diameter algorithm. We furthermore include a comparison of different GPU algorithms for single-source shortest path computations, which is not only a crucial step in computing the diameter, but also relevant in many other distance and neighborhood-based algorithms.
Keywords :
graph theory; graphics processing units; mathematics computing; parallel algorithms; parallel architectures; CUDA GPU; compute unified device architecture; fast diameter computation; graph vertices; highly parallel GPU-based bounding algorithm; large real-world sparse graph diameter computation; longest shortest path length; parallel diameter algorithm; performance verification; single-source shortest path computations; Arrays; Graphics processing units; Instruction sets; Kernel; Optimization; Programming; CUDA; diameter; eccentricity; graph traversal; sparse graphs;
Conference_Titel :
Parallel, Distributed and Network-Based Processing (PDP), 2014 22nd Euromicro International Conference on
Conference_Location :
Torino
DOI :
10.1109/PDP.2014.17