GOVERNMENT

UCL (University College London) and Dell have unveiled plans for the installation of Legion, a supercomputer that will enable UCL researchers to tackle problems of unprecedented size and complexity, and enable a step change in research possibilities. Michael Dell, Chairman and CEO of Dell, and Professor Malcolm Grant, President and Provost of UCL, also signed a Memorandum of Understanding (MOU), cementing plans to pursue a series of collaborative research activities based on UCL’s new Dell high performance computing (HPC) solution. Legion will be built using cutting-edge computer cluster technology to create a single computational machine with an expected peak performance of 42.9 TeraFlops per second (TF/s), representing the equivalent strength of approximately 2,700 desktop systems. For comparison, the new National Supercomputing service HECToR (www.hector.ac.uk/) will deliver approximately 60 TF/s. Legion is expected to enter production service in December 2007. Once complete it will weigh 21 metric tonnes, 7km of cabling will be used in its construction, and it will consume 19,000 litres of air per second. Dell now supplies 13 of the world’s fastest supercomputers. “Using industry standards-based technology, rather than high-priced, proprietary systems, researchers have access to previously unavailable levels of computing power, at a remarkably lower cost,” said Mr. Dell “UCL researchers are conducting research with high performance computing that will accrue benefits for society for years to come.” New possibilities which Legion opens up for research at UCL include:

• Researchers in the Department of Chemistry, working with clinicians at the National Hospital for Neurology and Neurosurgery, will use fluid simulation applications and patient-specific MRI data to model blood flow through the brain of stroke victims. This will enable consultants to make changes to the newly created patient-specific model of the vascular system in real time, so as to test treatments and then alter their prognosis of the condition;
• The Department of Physiology will perform simulations of key area of the brain to the level of individual cells that make up neural networks, thus reproducing real nerve activity;
• Researchers in the Department of Physics and Astronomy will perform the most detailed simulations ever conducted of cold dark matter structure formation in the Universe. This will test our understanding of the origin of galaxies and of gravity itself.

Both Dell and UCL recognise that in the 21st century, the rate of development of science, engineering and biomedicine as well as progress in the arts, will depend on powerful, high-quality computing capabilities. The provision and development of advanced computing facilities for research are therefore of critical importance. “We are excited to be working with Dell to implement a world-class supercomputing research infrastructure,” said Professor David Price, Executive Dean of Mathematical and Physical Sciences and Chair of the UCL Research Computing Sub-Committee. “High-end supercomputing used to be the preserve of an elite few in the academic world, but not anymore. It is our goal to create a central HPC resource from which our 16,000 researchers across all disciplines at UCL can benefit, especially in the Biomedical Science area where we are seeing increasing uptake. Research Computing is now firmly embedded in UCL’s overall information strategy.” With £3.85 million funding from the Science Research Infrastructure Fund (SRIF), UCL will implement a main distributed memory HPC cluster consisting of 2560 processor cores based on Intel Dual-core technology, an SMP cluster comprising 96 processor cores and 192TB high-performance storage. Working closely with Dell’s Supercomputing Practice in the UK, UCL’s Information Systems staff will install the main cluster configured as 10 inter-connected computational units, offering UCL researchers the ability to tackle ever more complex problems, carry out more detailed analysis and address larger computational challenges. Dell’s Supercomputing Practice helps both academic and commercial organisations design, deploy and run high performance clusters for complex or compute intensive operations. The group shares best practice with customers across Europe, ensuring they can take advantage of Dell’s deep knowledge in efficiently building, powering and cooling standards-based computing grids and data centres. For more information about Dell’s HPC and Supercomputing Solutions, go to its Web site.