Editorial & Analysis
Also by this author
12 Sep 2012
The US Department of Energy’s National Renewable Energy Laboratory (NREL) expects delivery late summer 2012.
When researchers at the US Department of Energy’s National Renewable Energy Laboratory (NREL) get their hands on a new, $10 million supercomputer designed for them by HP and Intel in around a year from now, they won’t just have the world’s largest system for researching renewable energy and energy efficiency, but also an installation capable of making some pretty spectacular energy-efficiency boasts of its own.
The power consumption of supercomputers is one of the industry’s most pressing issues. It’s a major drag on boosting computer performance by clustering multiple powerful systems together. Several innovations in this new supercomputer, however, are expected to result in the NREL facility becoming the world’s most efficient data centre, with a power usage effectiveness (PUE) rating of 1.06 or better.
PUE ratings are based on a ratio of the energy consumption of the servers, storage and networking equipment in a data centre to the energy consumption of ‘overhead’ – the lights, cooling equipment, air conditioning and other items not directly active in the processing and storing of data. Ideally, a data centre’s PUE rating should be as close to 1.0 as possible, but on average, most data centres achieve a 1.8 ratio, according to a survey of 500 data centres conducted last year by research firm, the Uptime Institute. That means that they use almost twice as much electricity on overhead as they do on their computing systems.
At NREL, the new system will reside in a data centre specifically designed to be compact, resulting in short runs for both electrical and plumbing components. It will do without traditional cooling systems, in favour of a design cooled by water. This will maximise the reuse of ‘waste heat’ generated by the supercomputer, pumping it into nearby office and laboratory space at NREL’s campus in Golden, Colorado, cutting down on the organisation’s heating costs.
The system itself, meanwhile, will be powered by a combination of current 32nm Xeon E5 processors and forthcoming 22nm Ivy Bridge processors from Intel, together with around 600 of the company’s Xeon Phi co-processors, all running inside HP ProLiant Gen 8 servers.
The total peak performance of this massive system is expected to exceed 1 Petaflop – the equivalent to a thousand trillion floating point operations per second.
According to an NREL statement, the new system will “greatly expand its modeling and simulation capabilities, including advancing materials research and developing a deeper understanding of biological and chemical processes. It will also support research into fully integrated energy systems that would otherwise be too expensive, or even impossible, to study directly.”
Said NREL Computational Science Centre director, Steve Hammond: “The industry is more and more cognizant of the amount of energy being used in our nation’s data centres. NREL’s new HPC data centre in the ESIF will set standard for sustainable and energy-efficient computing. The data centre will have a world-leading PUE and reuse nearly all waste heat generated. Most data centres do only one or the other, not both.”