Feeling Blue on Petascale Day?
Today, the University of Illinois’s National Center for Supercomputing Applications will be celebrating “Petascale Day,” seeing as today is October 15 and the prefix “peta-“ refers to 10 to the fifteenth power. As part of the day’s festivities, the center will be offering tours of Blue Waters, their new petascale machine.
Bill Kramer, the deputy director of the Blue Waters project, spoke about computer’s capabilities with regard to parallel processing, bandwidth, and its potential real-world applications.
“(Blue Waters) is one of the fastest computational systems if you count peak flops,” Kramer said “but it’s also one of the largest amounts of memory of any system in one place, the most data and storage capabilities, if you count the amount of capacity we have. And the amount of bandwidth is not surpassed by any other system in the world that we know of.”
Coming out of the renowned supercomputing facility at Champaign-Urbana and funded in part by a $208 million grant from the National Science Foundation, Blue Waters was expected to be one of the largest and fastest supercomputers in the world. The key to that operation, according to Kramer, might actually be its top-notch bandwidth.
For Kramer, one of the biggest challenges moving forward in the world of big data is that of bandwidth. It is no small feat to actually process and analyze a petabyte of data. But according to Kramer, transporting that data is the most limiting function.
“Bandwidth is the thing that’s becoming the most limited and the most expensive,” Kramer said. “You can build more and more processors, but what’s really limiting their ability to do sustained work is the bandwidth, which gets the data in and out of the systems.”
Kramer hopes that in time (five to ten years), smaller vendors and organizations will be able to feasibly deal with the amount of data Blue Waters plans to deal with, in part because of the extensive educational undertaking the Blue Waters is planning, particularly with regard to their advancements in parallel processing.
“The methods that we develop and our vendor partners develop to deal with highly parallel things will be commonplace not too long from now.”
Kramer echoed the notion that parallel processing gets hampered when even one out of a thousand threads is underperforming. Blue Waters could overcome that with its sheer immensity, but in order for it to be most efficient and operate at the petaflops level that they desire, those issues would have to be solved.
Of course, there are plenty of applications for the Blue Waters machine, with Kramer noting that it’s fitted for the widest possible variety of applications. Specifically with regard to vendors, Kramer noted that “five percent of our systems are also devoted to our industrial partners, ranging from validation of parallel methods to … the largest possible aircraft and its mechanical systems.”