AuthorsA. Thune, S. Reinemo, T. Skeie and X. Cai
TitleDetailed Modeling of Heterogeneous and Contention-Constrained Point-to-Point MPI Communication
AfilliationScientific Computing
Project(s)Department of High Performance Computing , SparCity: An Optimization and Co-design Framework for Sparse Computation
Publication TypeJournal Article
Year of Publication2023
JournalIEEE Transactions on Parallel and Distributed Systems
Pagination1580 - 1593
Date Published03/2023

The network topology of modern parallel computing systems is inherently heterogeneous, with a variety of latency and bandwidth values. Moreover, contention for the bandwidth can exist on different levels when many processes communicate with each other. Many-pair, point-to-point MPI communication is thus characterized by heterogeneity and contention, even on a cluster of homogeneous multicore CPU nodes. To get a detailed understanding of the individual communication cost per MPI process, we propose a new modeling methodology that incorporates both heterogeneity and contention. First, we improve the standard max-rate model to better quantify the actually achievable bandwidth depending on the number of MPI processes in competition. Then, we make a further extension that more detailedly models the bandwidth contention when the competing MPI processes have different numbers of neighbors, with also non-uniform message sizes. Thereafter, we include more flexibility by considering interactions between intra-socket and inter-socket messaging. Through a series of experiments done on different processor architectures, we show that the new heterogeneous and contention-constrained performance models can adequately explain the individual communication cost associated with each MPI process. The largest test of realistic point-to-point MPI communication involves 8,192 processes and in total 2,744,632 simultaneous messages over 64 dual-socket AMD Epyc Rome compute nodes connected by InfiniBand, for which the overall prediction accuracy achieved is 84%.

Citation Key43210

Contact person