UCSD Supercomputing Center Builds Raspberry Pi Linux Cluster for Education
The San Diego Supercomputer Center has built a Linux cluster around Raspberry Pi devices as part of an effort to help kids and adults learn about parallel computing.
SDSC, part of the University of California, San Diego, debuted the 16-node Meteor cluster this fall in a UCSD course on visualization. Though its primary use is as a teaching tool, it's also used as a marketing tool to help spread the word about parallel computing and generate interest in parallel programming. Meteor was put on display publicly at the SC13 conference in mid-November, where it was connected to a tiled video wall and used to hold "a friendly gaming competition."
"The goal of Meteor is to educate kids and adults about parallel computing by providing an easy-to understand, tangible model of how computers can work together," said Rick Wagner, SDSC's manager for high-performance computing, in a prepared statement. "One way we achieve this is by using Meteor as a presentation tool for demonstrations, with all of its components laid out in front of the audience. More importantly, we present Meteor in a fun, informal learning environment where students can try their hands at gaming competition while learning about the benefits of parallel programming."
He continued: "Currently, the majority of computer science degree curriculums contain relatively few courses in parallel programming and distributed architecture at the undergraduate level and below, but the basic concepts aren't difficult. Meteor and its Raspberry Pis provide us with a platform to develop methods that prepare middle- and high-school students to use and program current hardware."
Wagner has developed a curriculum around Meteor that has high school and undergraduate students developing games and hybrid CPU/GPU applications. SDSC students (and alumni) have also released a free app designed, called KA Lite, which lets users watch Khan Academy videos offline using Raspberry Pis.
"Like [SDSC's soon-to-be-launched supercomputer] Comet, Meteor is all about high-performance computing for the 99 percent," said SDSC Director Michael Norman, also in a prepared statement. "It's about increasing computing access on a broad scale to support data-enabled science and engineering across education as well as research."
But where the Intel Xeon-based Comet, funded by a $12 million NSF grant, is designed to deliver performance in the 2 petaFLOPS range, the $25 to $35 Raspberry Pis operate on much more modest scale. They provide general purpose compute performance in the range of about 24 gigaFLOPS apiece. (The Raspberry Pi units are built around a Broadcom BCM2835 SoC, integrating an ARM processor and Videocore 4 GPU.)
"We are committed to enabling every person around the world, no matter their circumstances, to access a quality education through open educational resources and facilitating their use inside and outside of classrooms around the world," said Jamie Alexandre, the software architecture lead at the nonprofit and a cognitive science doctoral candidate at UC San Diego.
More than 2 million Raspberry Pi units have shipped since the device's introduction in 2012, according to the Raspberry Pi Foundation.