Hewlett-Packard, Intel joining forces to create world’s most energy-efficient supercomputer

Written By: Thatcher Michelsen September 27, 2012 0
The Department of Energy's National Renewable Energy Laboratory has requested an ultra-efficient supercomputer from Hewlett-Packard and Intel.
The Department of Energy's National Renewable Energy Laboratory has requested an ultra-efficient supercomputer from Hewlett-Packard and Intel.

Computer hobbyists pride themselves on being able to build powerful systems that minimize energy expenditure. However, they are about to be outclassed on a professional scale as a U.S. Department of Energy (DoE) initiative is set to begin work on the world's "greenest" computer by collaborating with top information technology companies Hewlett-Packard (HP) and Intel.

The long-time tech giants will be working together with the DoE's National Renewable Energy Laboratory (NREL) to create a data-processing center that can supply some of its own power. The project is expected to cost around $10 million, and though designs have yet to be finalized, there are several components that government officials expect to see once it is completed.

Most notably, the proposed supercomputer may be able to compute data by the petaflop, or one million billion computations per second. According to industry source HPCwire, a system capable of processing at this magnitude requires around a megawatt of consistent energy to operate at peak capacity.

The system will incorporate a network of warm water coolers designed to maintain a certain temperature, which would prevent the supercomputer from burning too much excess energy. This design, reportedly still in development, is meant to help create the "perfect" operating temperature. Heat created by the computers will be collected and siphoned into a generator that would then be used to support climate control mechanisms in other parts of the NREL.

NREL officials told HPCwire that the Intel/HP collaboration is part of a broader strategy to reign in government energy expenses. Steve Hammond, a computation director for the laboratory, said that the DoE initiatives seek to cut future costs in every possible way.

"We've taken a chips-to-bricks approach to data centers," Hammond told HPCwire. "We're managing both the bytes and the BTUs."

The new supercomputer is expected to be brought online sometime in 2013, and could be the first of many energy-efficient data-crunching systems to come.

Leave A Response »