Welcome to the UMDNJ High Performance Computing and Storage Homepage!

HPC Services:

Scientific Server Accounts - Contact System Administrator via e-mail or phone: 5-4473; For more information consult the or use the request form here.

Research Consultation - Use this online form to make a request for consultation.

Internet 2 Access- Contact University Network Administrator via e-mail or phone: 2-3603

Whitepapers - For whitepapers, go to the whitepapers homepage here.

Definition of High Performance Computing

High Performance Computing (HPC) can be defined as computing devices and attendant storage which permits computational problems that would be intractible on common desktop devices for reasons that include but might not be limited to processor speed, number of processors, memory capacity, memory cache size, external storage and bandwidth of external storage to the connected device.

HPC environments exist in multiple flavors including high performance multiple processor servers with very fast processor and I/O interconnects, clusters of identical computers (typically Linux) with specialized software designed to leverage aggregate capability, computing grids comprising heterogenous computing types under the management of software capable of dispatching and controlling jobs based upon a set of parameters describing program requirements, priority and other factors.

No one environment is suitable for all types of applications.  Programs requiring ultra high connectivty between processors are generally more suitable to a single multiprocess environment or cluster with with very fast interconnects (e.g., Infiniband).  Others, which can be easily "chunked," may be more suitable to a grid environment. Available programming resources is yet another consideration.  AST can assist researchers with these decisions and suggest whether UMDNJ or external resources are better suited.

AST's HPC asset consists of a SunFire 6900 with 24 dual-core processors, 48 Gbytes of main memory and fiber interconnects to a high performance SAN (storage area network).  Cluster and grid computing environments are being evaluated for future deployment.

For additional information, please consult:

Leslie Michelson, Ph.D. at 2-4558 or via e-mail at michelso@umdnj.edu

John Kerrigan, Ph.D. at 5-4473 or via e-mail at kerrigje@umdnj.edu