Using AMD EPYC™ processors in an optimised infrastructure for running computational fluid dynamic workloads

The Introduction:

The use of Computation Fluid Dynamics (CFD) has widespread applications. From Aerospace to Meteorology to Chemical Engineering,  Computational Fluid Dynamics is a tool found in any industry where the flow of liquids or gasses over surfaces needs to be analyzed. CFD has become an integral tool in the industrial research & design processes – replacing expensive physical modeling in driving the pace and lowering the cost of innovation. A High-Performance Computing (HPC) cluster can often be the critical ingredient for industrial enterprises where time and cost are the difference between success and failure against their competitors.

The Business Case:

Although Computational Fluid Dynamics is designed to replace costly and time-consuming physical modeling, many companies fail to realize its full potential due to substandard implementation. CFD simulation and modelling software require the correct blend of computational resources to power the numerical analysis and algorithms that simulate, analyze and solve problems in the fluid flow. These algorithms are extremely complex in themselves, and, depending on the desired complexity and accuracy of the model, can often end up taking days or weeks, even with the most powerful HPC clusters. It makes sense that a well-designed and administered HPC
cluster can make the difference between a successful and optimized CFD solution and an unnecessarily expensive, inefficient one that hampers the design process, time to market, and ultimately, an organization’s edge over its competitors. ClusterVision provides a range of services and hardware solutions that are tailor-made to match an organization’s CFD workload and provide the best possible blend of performance, budget, and support.

Read or download the case study here.