Performance Analysis of Computational Fluid Dynamics Workloads on HPC in the Cloud
Mandeep Kumar · 2023
Computational Fluid Dynamics (CFD) simulations are computationally intensive and often require High Performance Computing (HPC) resources to complete within a reasonable timeframe. Traditionally, on-premise HPC clusters have been employed for running CFD simulations. However, with the emergence of cloud computing, the option to utilize HPC in the cloud for CFD simulations has become available. This work presents an architecture for HPC cluster configuration in the cloud, which incorporates essential components such as an HPC job scheduler and a license server. Additionally, we carefully select and integrate various resources and services to optimize scalability and enhance performance specifically for CFD workloads. We perform a comprehensive performance evaluation of commercially available core-based floating license CFD applications on HPC in the cloud. By thoroughly assessing the performance metrics and capabilities of these applications, we contribute to a deeper understanding of their suitability and effectiveness of HPC in the cloud for CFD workloads.