Admittedly, that box is a 35U rack server case, but the computing density is “five to ten times denser than anything before” according to BitScope CEO Bruce Tulloch.
Bruce reveals that LANL works with “clusters of 20 000+ nodes capable of doing the sort of thing that would otherwise require millions of normal PCs to achieve.”
Gary Grider, leader of the High-Performance Computing Division at Los Alamos National Laboratory and home of the Trinity supercomputer said:
“It’s not like you can keep a petascale machine around for R&D work in scalable systems software. The Raspberry Pi modules let developers figure out how to write this software and get it to work reliably without having a dedicated testbed of the same size, which would cost a quarter billion dollars and use 25 megawatts of electricity.”
BitScope 3000-core cluster computer built with Raspberry Pi boards
LANL’s challenge to BitScope was to build a 3000-core ‘pilot cluster’ as “a testbed for Los Alamos researchers to use to develop their own next-generation computers,” Bruce explains.
However, it’s “non-trivial to get 750 Raspberry Pis (or any type of computer, really) to work reliably at such high density,” continues Bruce: the problems are “power, mounting, and cooling.” Bruce confirms that the 750-node Clusters project “needs less than 4kW at full tilt”, over ten times lower than a conventional air-cooled setup.
Having worked with Raspberry Pi for quite some time, I’ve long thought it the ideal candidate to build low-cost cloud and cluster computing solutions for research and education. When SICORP approached us with Gary’s plans, we jumped at the opportunity to prove the concept.
Comments
Post a Comment