An Evaluation of Parallel Graph Partitioning and Ordering Software on a Massively Parallel Computer

We empirically study state-of-the-art parallel graph partitioning and sparse matrix ordering software packages. We compare their speed, quality, and robustness. For a model case, in which good partitionings (even optimal partitionings, in some cases) can be constructed manually, we compare the size of the edge cuts of the manual partitions with that of the partitions generated by the multilevel heuristics that are at the heart of modern graph partitioning software packages. We show that the quality of the partitions generated by the software is only slightly worse than that of the manual partition for this class of model graphs. We discuss the shortcomings of the current ordering software and argue that there is an urgent need for more robust, scalable, and high-quality software for sparse matrix ordering to support scalable solution of sparse linear systems by direct methods on massively parallel computers.

By: Anshul Gupta

Published in: RC25008 in 2010


This Research Report is available. This report has been submitted for publication outside of IBM and will probably be copyrighted if accepted for publication. It has been issued as a Research Report for early dissemination of its contents. In view of the transfer of copyright to the outside publisher, its distribution outside of IBM prior to publication should be limited to peer communications and specific requests. After outside publication, requests should be filled only by reprints or legally obtained copies of the article (e.g., payment of royalties). I have read and understand this notice and am a member of the scientific community outside or inside of IBM seeking a single copy only.


Questions about this service can be mailed to .