The University of Tennessee, Knoxville’s supercomputing capabilities are about to become more powerful. The UT-managed National Institute for Computational Sciences (NICS) is adding 300 teraflops to the TeraGrid’s total computational capability thanks to two awards from the National Science Foundation (NSF) which total $3.4 million.
Researchers will also have access to more than 200 million additional service units, or CPU hours, per year, bringing the total available from NICS to over 800 million and benefitting the organization’s entire user community.
The first part of the award will increase the size of Kraken, the first academic petaflop computer and currently the world’s fourth fastest machine, by 12 cabinets, adding 144 teraflops of computing power. A petaflop is equal to a quadrillion floating point operations per second.
The Cray XT5 will now total 100 cabinets and provide 1.17 petaflops of computing capability and 147 terabytes of memory. A terabyte is a trillion bytes of memory or 1,024 gigabytes. While Kraken is an ideal resource for running some of the world’s most computationally demanding simulations, the new cabinets will also assist the myriad of smaller jobs continually running on NICS’s flagship system.
“We are extremely pleased to be able to put more continually available resources at the disposal of researchers with smaller codes, while still supporting the very largest applications,” said NICS Director Phil Andrews. “The importance of a research activity cannot be defined by the size of the code involved, and we want to give all NICS users the best possible service.”
Although Kraken is the only resource in the NSF’s computing portfolio capable of running simulations at its full potential of 8,256 nodes, it also is a massive capacity resource.
Many of the more computationally demanding codes running on Kraken use the “sweet spot” of 8,192 nodes, the largest power of two that can be accommodated within the 8,256-node machine. While a code of this size is running, albeit for only part of each week, a maximum of 64 nodes remains for other users. The extension of Kraken to 9,408 nodes will increase this by a factor of 19, providing an additional 1,216 nodes for smaller jobs to run concurrently. This will greatly improve availability of the system for smaller “capacity” jobs while still allowing the extremely large “capability” jobs access to the NSF’s most powerful supercomputing system.
The second part of the award will fund the operation of Athena, a 166-teraflop Cray XT4 that is currently ranked as the TeraGrid’s third largest computational resource. Athena features 18,048 cores and 18 terabytes of memory and is an extremely reliable system, most recently used as a dedicated platform for climate, weather and quantum chromodynamics research. Athena will be available through the TeraGrid allocations process beginning October 1, 2010, and will be allocated in conjunction with Kraken. This will allow NICS to maximize the usefulness of both of these leading resources, each of which are running at over 90 percent utilization, by apportioning researchers to the most appropriate machine.
NICS is a joint effort of UT Knoxville and Oak Ridge National Laboratory. NICS was founded in 2007 and is supported by NSF and the State of Tennessee. It is a resource provider in the NSF’s TeraGrid program and is located at Oak Ridge National Laboratory, home to the world’s most powerful computing complex.
The TeraGrid, sponsored by the NSF Office of Cyberinfrastructure, is a partnership of people, resources and services that enables discovery in U.S. science and engineering. Through coordinated policy, grid software and high-performance network connections, the TeraGrid integrates a distributed set of high-capability computational, data-management and visualization resources to make research more productive. With Science Gateway collaborations and education programs, the TeraGrid also connects and broadens scientific communities.