Parallel Performance Studies for a 3-D Elliptic Test Problem on the 2018 Portion of the Taki Cluster

Author/Creator ORCID

Date

Department

Program

Citation of Original Publication

Richard Ebadi, Carlos Barajas and Matthias K. Gobbert, Parallel Performance Studies for a 3-D Elliptic Test Problem on the 2018 Portion of the Taki Cluster, http://hpcf-files.umbc.edu/research/papers/PoissonHPCF201819.pdf

Rights

This item is likely protected under Title 17 of the U.S. Copyright Law. Unless on a Creative Commons license, for uses protected by Copyright Law, contact the copyright holder or the author.

Abstract

The new 2018 nodes in the CPU cluster taki in the UMBC High Performance Computing Facility contain two 18-core Intel Skylake CPUs and 384 GB of memory per node, connected by an EDR (Enhanced Data Rate) InfiniBand interconnect. The performance studies use the test problem of the Poisson equation in three spatial dimensions, discretized by the finite difference method to give a very large and sparse system of linear equations that is solved by the conjugate gradient method. The algorithm is known to be memory-bound so it is a good test for the architecture of the nodes and the parallel network connecting them. Strong scalability studies varying the number of processes per node as well as the number of compute nodes demonstrate excellent scalability when using multiple nodes as well as very good scalability using multiple cores per node.