OSC 154 Teraflops Supercomputer Uses Intel Xeon CPUs and Nvidia Tesla GPUs

The Ohio Supercomputer Center (OSC) has announced a short while ago that researcher’s using its computing resources can now access a new HPC system that utilizes GPU acceleration to deliver a total peak performance of 154 teraflops.

Already available for general use, OSC's supercomputer was built by HP and is based on Intel’s Xeon processors and Nvidia Tesla graphic processing units (GPUs).

Dubbed the Oakley Cluster, the machine features more cores (8,328) on half as many nodes (694) as the center's most recent flagship system, the IBM Opteron 1350 Glenn Cluster.

Furthermore, the new system provides nearly twice the memory per core (namely 4GB) and three times the number of graphic processing units (128 to be exact). 

The Oakley Cluster also promises to deliver one and a half times the performance of the Glenn Cluster at just 60 percent of Glenn's power consumption and will expand OSC storage to nearly two petabytes with the addition of 600 terabytes of new DataDirect Lustre storage.

OSC is a state-funded high performance computing center that provides Ohio's universities, industries and other clients with computation, software, storage and support services.

"This strategic investment in a powerful, state-of-the-art computational system will boost the vital research being conducted at industrial facilities and in university laboratories across Ohio," said Chancellor Jim Petro of the Ohio Board of Regents. 

"The Oakley Cluster also will help Ohio colleges and universities in their development of a highly skilled workforce by invigorating the state's innovative computational science education programs," concluded Mr. Petro.

The new supercomputer arrived on December 28 on the west campus of The Ohio State University, where engineers had been working to assemble the complex system, install various software packages and test-run the system by allowing access to a few selected "power users."

Posted in: Hardware, News