University group deploys UK’s most powerful GPU supercomputer

The new “Emerald” system will be shared with researchers across the UK to drive discovery in astrophysics, genomics and nanotechnology.

A group of leading British universities last week deployed United Kingdom’s most powerful GPU-accelerated supercomputer. The new supercomputer “Emerald” will be shared by four  U.K.-based universities to drive research across a range of scientific and engineering fields.

A consortium of four British universities have deployed “Emerald,” the largest GPU supercomputer in the United Kingdom. (Source: Nvidia)

The STFC Rutherford Appleton Laboratory in Didcot is home for Emerald, an 84-node cluster equipped with 372 NVIDIA Tesla M2090 GPUs, delivering more than 114 teraflops of performance. The lab has also been designated a Center for Innovation in High Performance Computing (HPC) by Oxford University, one of the four universities involved.

Emerald was developed to enable scientists and engineers from across the UK to accelerate computationally intensive research in astrophysics, bioinformatics, chemistry, engineering, genomics, life sciences, nanotechnology, physics, and many other fields.

Established by the e-Infrastructure South Consortium, which includes the Universities of Oxford, Bristol, Southampton, and University College London, the Center for Innovation in High Performance Computing provides the infrastructure for the development of scientific and engineering applications, and will enable the training of HPC scientists and engineers.

The center’s new GPU-accelerated Emerald supercomputer and other computing systems were funded as part of a £3.7 million ($5.8 million USD) grant from the UK Engineering and Physical Sciences Research Council.

Nvidia has named the University of Oxford a CUDA Center of Excellence in recognition of its ongoing work in parallel computing research and education using Nvidia GPUs and the CUDA parallel programming environment.

Emerald is already hard at work, according to University of Bristol Simon McIntosh-Smith. “Applications we’ve already been running here in Bristol include drug discovery simulations with the pharmaceutical industry as well as some of our own research into the workings of emerging antiviral resistance in certain strains of influenza.” McIntosh-Smith adds that “all of the GPU codes being developed in Bristol are being written in the open industry standard for GPU computing, OpenCL, while other applications on Emerald use Nvidia’s proprietary language CUDA or even OpenACC.”
Host university Oxford is considered a world leader in genetics, mathematics, scientific computing, and the physical and life sciences; it is the 19th institution to be named a CUDA Center for Excellence. The designation provides access to equipment and grants from Nvidia to support a number of research and academic programs across its mathematics, physical and life sciences divisions, including:

Astrophysics—Real-time pulsar detection application for the forthcoming Square Kilometre Array Project to deploy the world’s most powerful radio telescope;

Bioinformatics—Analysis and statistical modeling of whole-genome sequencing data;

Chemistry—Molecular dynamics simulations of key DNA nanotechnology mechanisms.

CUDA is Nvidia’s parallel computing environment, which enables dramatic increases in computing performance by harnessing the power of GPUs. More than 580 universities and institutions worldwide teach CUDA parallel programming within their curriculum.