Strong Scalability Studies for the 2-D Poisson Equation on the Chip 2024 Cluster with Historical Comparison

Abstract

The UMBC High Performance Computing Facility (HPCF) opened the cluster chip in early 2025. The new 2024 CPU nodes in the cluster chip-cpu contain two 32-core Intel Emerald Rapids CPUs and at least 512 GB of memory per node, connected by a high-performance InfiniBand interconnect. Parallel performance studies for the memory-bound test problem of the Poisson equation in two spatial dimensions yield several conclusions: Strong scalability studies demonstrate excellent performance when using multiple nodes due to the low latency of the high-performance interconnect and good speedup when using all cores of the multi-core CPUs. Even for the largest numbers of processes per node, the runtime for the code is still significantly reduced, an excellent result for memorybound code. Comparisons to results on past clusters in HPCF bring out that core-per-core performance of serial code improvements has continued to improve, despite a stagnant clock rate, demonstrating the quality of the newest CPUs and their memory access. Node-per-node performance of parallel code continues to improve dramatically due to the larger number of cores available on a node. Tests indicate that architecture specific compile options are vital to achieve this performance, and appropriate options in the batch run script are needed.