Scalability of Transient CFD on Large-Scale Linux Clusters

April 7th, 2009 4:40 pm
Posted by Stan Posey
Tags: , , , , , ,

[Excerpt] This work examines the parallel scalability characteristics of commercial CFD software FLUENT and STAR-CD for up to 256 processing cores, and research CFD software CDP from Stanford University for up to 512 cores – for transient CFD simulations that heavy in IO relative to numerical operations. In three independent studies conducted with engineering contributions from the University of Cambridge, Intel, SGI, and Panasas, each Linux HPC environment combined Intel Xeon clusters with a Panasas parallel file system and shared storage. The motivation for these studies was to quantify the performance and scalability benefits of parallel IO in CFD software on a parallel file system (PanFS) verses a conventional serial NFS file system for a variety of transient CFD cases.

Click here to download "Scalability of Transient CFD on Large-Scale Linux Clusters"


You must be a Registered Member in order to comment on Cluster Connection posts.

Members enjoy the ability to take an active role in the conversations that are shaping the HPC community. Members can participate in forum discussions and post comments to a wide range of HPC-related topics. Share your challenges, insights and ideas right now.

Login     Register Now

Author Info

Stan Posey currently manages the Panasas strategy of HPC Applications and Industry Development for a variety of target vertical markets. Mr. Posey is responsible for company guidance and consultation on Panasas advancement of HPC technology and solutions, with emphasis on applications in computational mechanics and sciences. Prior to joining Panasas in 2007, Mr. Posey has contributed for more than 20 years in various roles of HPC applications development including applications engineering and market development roles at SGI, research and engineering positions with the US Department of Energy, Oak Ridge National Laboratory, consulting and software vendor CD-adapco Group, and systems vendor Control Data Corporation. Mr. Posey has co-authored several papers for HPC research journals and technical conferences, and his education includes a M.Sc. in Mechanical Engineering from the University of Tennessee, Knoxville, TN, USA.