Build your Own Supercomputer

You must’ve some time looked up at the night sky and wondered whether there’s life elsewhere. You are not alone in that. The search for extraterrestrial intelligence (SETI) is as old as mankind itself. The modern approach to this is to analyze electromagnetic radiation from the stars for non-random patterns, like with TV or radio broadcasts. As modern as starting in 1959!

The Tech behind Supercomputing Today, supercomputers are built by combining many machines together to work as one Grandmaster? Deep Blue is the first (and only) computer to defeat a reigning World Chess Champion7 Password Cracking Distributed.net is a coordinated effort over the Net to crack a series of encrypted messages, to prove the need for stronger encryption Designed to Fly Made-in-India supercomputers are deployed in areas as diverse as missile design, weather forecasting and research The Champs When we started writing this piece, the fastest supercomputer was the ASCI White, which has since been overtaken by the Earth Simulator A Brief History of Time Stephen Hawking has been using a supercomputer to unravel the mysteries of the universe Future Champs How Multi-processing Works The previous articles dealt with the applications of supercomputers. With this article, we start our story on building a supercomputer. But first the basics Understanding Clustering Clustering many machines lets you increase computing power many fold Build a Windows Supercomputer Building a supercomputing cluster on the Windows platform is a matter of interconnecting existing Microsoft solutions with third-party tools Supercomputing in Linux A step-by-step guide on how to set up a cluster of PCQLinux machines for supercomputing

Now you can collect a humungous amount of data using radio telescopes. But to analyze all of it would require supercomputers of a magnitude never built before. Of the many SETI programs running, SETI@home, a project run by the University of Berkely

(http://setiathome.ssl.berkeley.edu) found an innovative way out–to use idle computers connected to the Internet to crunch bits of data! So, instead of running any screensaver, you run the SETI@ home screensaver, which analyzes the data in the background, while showing graphs of the data as the screensaver itself. The central computer sends packets of data to the individual machines, which after processing them send the results back to the central server.

The data comes from a radio telescope at the Arecibo Observatory at Puerto Rico, run by the Cornell University. Data is collected in high-density tapes at the rate of about 35 GB per day, and mailed to Berkeley, where it is divided into .25 MB chunks, to be sent to the screensavers.

So, what does SETI@home look like from the computing angle? Very impressive, to say the least. Just look at the vital statistics. As of 15 April, there were 36,60,853 users participating in the effort, who had totally given CPU time equivalent to 946673 years (1107 years in the previous 24 hours alone)!

67 different types of processors, ranging from SPARC to Alpha to MIPS to X86 and PowerPC, have contributed to the effort. The OS count, including different versions and builds and flavors, has over time added up to a whopping 121. And these machines are spread across 226 countries.

Does all this make it into a supercomputer? Well, this is what the SETI@home site has to say, “The most powerful computer, IBM’s ASCI White, is rated at 12 TeraFLOPS and costs $110 million. SETI@home currently gets about 15 TeraFLOPs and has cost $500K so far.”

Krishna Kumar