Los Alamos National Lab fires up 750-node RPi cluster
It's an exercise in learning to cluster, and could soon scale to 10,000 nodes
Posted in HPC, 13th November 2017 06:55 GMT
The Los Alamos National Laboratory will this week reveal its latest "High-performance computer" - a cluster of 750 Raspberry Pis.
The Lab's Gary Grider had the idea of trying to package up serious clusters, not to replace its supers – the three fastest at the moment are the Grizzly, Fire, and Ice systems, all in the 1-1.5 Petaflop range – but to provide a software development environment that didn't involve feeding electrons to the big iron.
In a media release issued ahead of the SC17 supercomputing conference, Grider said: “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”.
He notes that the RPi pulls just two or three watts per node, making it cheap to run, and has hopes of scaling the system to thousands of nodes.
Grider wanted a system that wasn't “Tinker Toys and LEGOs”, so the packaging the Lab settled on for its RPi cluster came from BitScope Designs, a box called the Blade that stuffs 150 Pis into a module. Five such Blades, powered by 48V power-over-Ethernet and with integrated switches, comprise the cluster.
The Lab leads a three-university consortium, the New Mexico Consortium, that will use the cluster.
The Australian-headquartered BitScope was brought into the project by HPC consultancy SICORP.
The 750 nodes puts the LANL cluster well ahead of other efforts we've heard of, like the 32-way Beowulf cluster put together in 2012, or the 2012 64-node effort at the University of Southampton that possibly inspired Grider's crack about LEGO.
In September, a group of Bolzano University researchers published their work on a 300-node RPicluster. ®