HPC | News
Oxford Supercomputer Goes Heavy on Graphics Processing
- By Dian Schaffhauser
Emerald has 372 Nvidia GPUs for a total of 190,464 graphics processing cores.
A consortium led by the University of Oxford has just taken a new supercomputer online that has greater graphical processing power than any other machine in the United Kingdom. Researchers will use the new computing capacity to process data in helping understand diseases like swine flu, model climate change, simulate 3G and 4G communications networks, and develop new tools for processing medical images.
The lighting up of "Emerald," housed in the Science and Technology Facilities Council's Rutherford Appleton Laboratory, also launched the consortium to run the new system. The e-Infrastructure South Consortium encompasses Oxford as well as the Universities of Southampton and Bristol and University College London. The Consortium will share access to the systems, provide infrastructure, and deliver training.
Emerald delivers 114 teraflops of performance. According to online specs, it's built as an HP system with 372 Nvidia GPUs. It has 60 compute nodes each with two six-core X5650 Intel Xeons and three 512-core M2090 GPUs and 48 GB memory as well as 24 compute nodes with two six-core X5650 Xeons and eight 512-core M2090 GPUs and 96 GB memory. Storage is provided by a Panasas storage system with about 135 TB of usable disk space.
The supercomputer was funded in part with a £3.7 million grant from the national Engineering and Physical Sciences Research Council, part of a £145 million government investment in e-infrastructure.
"The high set-up costs both in terms of equipment and expertise can be a major barrier to SMEs expanding into newer or bigger markets. This new centre will make it easier for them to step up into the next league. In turn, the supercomputers will help university-led researchers work with industrial partners to develop and test innovative new products and technologies." said Oxford Professor Anne Trefethen, who led the collaborative bid to get the project underway.
"High performance computers based within the Consortium's research-intensive universities will enable better training and recruitment of world-class research talent, help develop new research ideas, and speed up the rate at which complex data can be processed," added Lesley Thompson, director of the Research Council.
Dian Schaffhauser is a senior contributing editor for 1105 Media's education publications THE Journal and Campus Technology. She can be reached at email@example.com or on Twitter @schaffhauser.