This study presents a detailed performance analysis of a 24-node Beowulf cluster built with Raspberry Pi 4B devices, equipped with 8GB of RAM, running a 64-bit operating system utilizing the STREAM Benchmark which is a widely recognized tool for evaluating memory bandwidth performance in high-performance computing (HPC) environments. Unlike typical processor benchmarks that focus on computing power, STREAM a memory bandwidth benchmark focuses on how quickly data can be transferred between the memory and the processor, which is a critical performance factor in high-performance computing (HPC) systems like Beowulf clusters. Fundamental memory operations Copy, Scale, Add, and Triad, are utilized to assess how efficiently the cluster handles memory-intensive workloads across increasing MPI process counts. Additionally, MPI-based communication benchmarks assess the inter-node message-passing performance, providing deeper insights into memory bandwidth utilization under distributed computing conditions. The findings offer valuable insights on the perspectives of using Raspberry Pi clusters for HPC applications in education, research, and prototyping. Furthermore, recommendations for performance optimizations and system enhancements are proposed to improve scalability, efficiency, and communication overhead in such low-cost HPC clusters.

Raspberry Pi 4 Beowulf cluster Cluster Message Passing Interface (MPI) MPICH Memory Performance Low-cost Clusters Parallel Computing ARM Architecture STREAM Benchmark