Since 1987 - Covering the Fastest Computers in the World and the People Who Run Them

August 9, 2010

Supercomputing Access and a New Economic Landscape

Nicole Hemsoth

Supercomputing in the cloud and rent-a-cluster services could save the day for the economy as an article in the San Francisco Chronicle stated this morning. Companies and research sites with limited funds for investments in new hardware, especially the big iron required for an evolving series of HPC-type applications, are looking to rented infrastructure versus up-front investments. This is good news for the vendors, naturally, but it could also prove to be of enormous value for those who need the capacity without the initial expense.

To lend a case study, the article points to the Ohio Supercomputer Center’s (OSC) Blue Collar Computing drive, which rents supercomputer capacity to business and research shops who might have otherwise been barred from entry due to high capital investment costs. Many types of companies take advantage of the Blue Collar Computing initiative to date and several more “have also used the services by accessing them through OSC partners like the Edison Welding Institute (EWI)” which is a non-profit organization that provides access to E-Weld Predictor. This application, which is accessed via a web interface allows users access to the OSC’s 1.650-node IBM “Glenn” cluster to run simulations of complicated welding tasks—a possibility that shaves weeks off the typical time it would otherwise take using standard in-house workstations.

Outside of renting time via established supercomputing centers, there are several companies that specialize in these same services, including SGI with its Cyclone product, Cycle Computing’s cloud and cluster rental options, Penguin Computing’s POD service and those offered by Sabalcore and a handful of others. Times are good for these companies on the revenue front and if the Chronicle’s article is correct, this good news could extend to companies who now have more options for harnessing greater compute power than possible without significant upfront investment.

Full story at San Francisco Chronicle

Share This