by Onur Celebioglu
For many years, high performance computing has been an integral part of manufacturing - especially among the largest and most influential companies. These companies typically employ common numerical analysis techniques such as finite element analysis (FEA) and computational fluid dynamics (CFD) to simulate their designs and shorten product development cycles. HPC infrastructure at these companies are geared to improve uptime and support production jobs. Therefore, production runs often take precedence to experimenting with new architectures and approaches, thus impeding innovation.
However, the National Center for Supercomputing Applications (NCSA) has had an exciting solution to remedy this for the past three years: iForge, "a pared down, but finely tuned cluster specifically built to address the experimental needs of the 'power users' at leading manufacturing companies," as a recent story in HPCwire about the cluster described it.
In the HPCwire story, Merle Giles, the head of the private sector program and economic development initiatives at NCSA, says the iForge cluster is designed for large-scale manufacturers using it to test and deploy important code on some of the newest hardware available. The iForge configuration includes:
iForge's constant upgrades over the past three years have been necessary to continue offering the new architecture to those manufacturers it targets. Additionally, employing blades allows for simpler upgrades since they do not require changes in the entire cabling infrastructure. While power users take advantage of new architectures offered by the iForge cluster to optimize and evaluate their codes - rather than placing additional burden on their own clusters - the users also provide important feedback that the iForge team shares with system vendors.
With the funds raised from the fees the NCSA charges for cores being used for production runs, the system can be continually refreshed and expanded. And that's a win-win for everyone.