Assessing contention effects of all‐to‐all communications on clusters and grids

Abstract

PurposeThe purpose of this paper is to explain one of the most important collective communication patterns used in scientific applications which is the complete exchange, also called All‐to‐All. Although efficient algorithms have been studied for specific networks, general solutions like those available in well‐known MPI distributions (e.g. the MPI_Alltoall operation) are strongly influenced by the congestion of network resources.Design/methodology/approachIn this paper we present an integrated approach to model the performance of the All‐to‐All collective operation, which consists in identifying a contention signature that characterizes a given network environment, using it to augment a contention‐free communication model.FindingsThis approach, assessed by experimental results, allows an accurate prediction of the performance of the All‐to‐All operation over different network architectures with a small overhead.Practical implicationsThe paper discusses the problem of network contention in a grid environment, studying some strategies to minimize the impact of contention on the performance of an All‐to‐All operation.Originality/valueThe approach used, assessed by experimental results, allows an accurate prediction of the performance of the All‐to‐All operation over different network architectures with a small overhead. Also discussed is the problem of network contention in a grid environment and some strategies to minimize the impact of contention on the performance of an All‐to‐All operation.