CENTER FOR COMPUTATION & TECHNOLOGY

TigerWeekly By Brian Fontenot He is perhaps most famous for the Beowulf project, which revolutionized super computing back in 1994. The impact of his project is still felt today as even modern super computers utilize the technology developed by Beowulf, which allows off-the-shelf computers to be networked in such a way that enable them to run as a single computer. But, why was the project named “Beowulf” in the first place? Sterling spent a month trying to come up with a name for his project, but to no avail. Only when an administrator demanded the name of the project because of the immediate deadline of the paperwork did Sterling make his choice. “I was desperate. I hadn’t a clue, and in a moment of high anxiety I caught sight of my mother’s old copy of the epic of Beowulf. She had done her graduate work in Old English Literature. The print date of the book was 1923 and I still have it. Needing to respond and having no better ideas, I said into the phone, “Oh hell, just call it ‘Beowulf;’’ nobody will ever hear of it anyway,” said Sterling. A combination of things drew Sterling to LSU. LONI, the Louisiana Optical Network Initiative, is scheduled to go live this fall. LONI is basically an unprecedented, extremely fast computer network and is a lure for researchers all over the country. The publicity surrounding Sean O’Keefe’s appointment as LSU’s chancellor also played a part. “It is a coincidence that, coming from NASA, both Chancellor O’Keefe and I chose to advance our respective professional goals through positions at LSU,” said Sterling, “However, it was probably through that highly-publicized event that LSU was put on my radar screen: If O’Keefe had identified real opportunity at LSU, then something was probably happening there. I was right about that.” And a colleague of Sterling’s at the NASA JPL actually helped to play matchmaker for Sterling and the CCT. “Serendipity rather than fixed plans can often be a powerful force, just as it was in this case,” said Sterling. Sterling is currently working on his MIND (Memory, Intelligence, and Network Device) project, a unique type of multi-processor architecture. He hopes it will not only improve computing by one to three orders of magnitude, but also make computers much more energy efficient. This means taking computers beyond the Petaflop (one thousand trillion operations per second) scale. Computers currently capable of Petaflops operate in excess of 10 megawatts of power, and the annual rate for a single megawatt runs in the million-dollar range. Sterling’s MIND project could one day make science fiction into reality, which serves as an inspiration for Sterling. “I have been inspired by my recognition of the revolution that computing in the trans-Petaflops performance regime, up to and beyond an Exaflops, can bring to a wide range of fields including climate modeling, microbiology, medicine, controlled fusion energy generation, engineering and technology, materials, economics, and national security,” said Sterling. Sterling also happens to be a fan of the 1982 film, “Tron.” “While not the most popular domain of research in recent years, it is my conviction that symbolic processing will ultimately consume more cycles by society overall than numeric processing, and the requirements for this class of computing varies significantly from conventional numeric computing. Autonomous systems, those that are goal-directed rather than imperative-programmed, will require such architectures,” said Sterling, “Yes, one possible application is robots right out of science fiction. But many more pedestrian applications nearer term are likely to emerge, such as powerful user interfaces, self-programming systems, high-reliability systems, sophisticated decision making, homeland security, and my favorite: deep space autonomous unmanned spacecraft.”