The University of Oxford and a consortium of U.K. academic institutions this week booted up the country's new most powerful GPU-based supercomputer packed with 372 Tesla M2090 graphics cards from Nvidia.
The £3.7 million, 84-node cluster, which delivers 114 teraflops of performance, has been dubbed Emerald.
Emerald was unveiled on 4 July at the new Center for Innovation in High Performance Computing (HPC) at Oxford's STFC Rutherford Appleton Laboratory in Didcot, Oxfordshire.
The supercomputer will be used for "computationally intensive research in astrophysics, bioinformatics, chemistry, engineering, genomics, life sciences, nanotechnology, physics, and many other fields," Nvidia said in a joint statement with U.K. academics and government officials.
"The Emerald supercomputer forms part of the government's £145 million investment in e-infrastructure and will be an invaluable asset to business and universities. It will drive growth and innovation, encourage inward investment in the U.K. and keep us at the very leading edge of science." Universities and Science Minister David Willetts said.
In addition to Oxford, members of the e-Infrastructure South Consortium which commissioned Emerald include the Universities of Bristol and Southampton, and University College London. The Center for Innovation in High Performance Computing will be both a resource for computationally intensive research and a training ground for HPC scientists and engineers.
Emerald uses GPUs based on Nvidia's current-generation "Fermi" architecture but if needed, it could be upgraded with next-generation "Kepler" parts when the company brings its powerful new GPU architecture to its Tesla line of products later this year, an Nvidia spokesperson said.
"The scientific applications Oxford develops for the Fermi GPUs will run great on Kepler GPUs. They just may need to update drivers and get the latest version of CUDA, which is free," he said.
Oxford was also named a CUDA Center of Excellence by Nvidia "in recognition of its ongoing work in parallel computing research and education using Nvidia GPUs and the Nvidia CUDA parallel programming environment," the company said. The university is one of just 19 global CUDA Centers of Excellence recognized by Nvidia.