New developments continue to surface in the supercomputing sector, with announcements from the public sector, education and manufacturing all making headlines. Here is Cray’s roundup of the biggest stories from the last week.
Cray charts path to the future with new offerings
Seattle Business Magazine highlights Cray and its new strategies in a piece for the August 2013 issue. The article chiefly focuses on new offerings beyond the high-end, multi-million dollar supercomputers for government and research applications. One of Cray’s most important platforms in this arena is Urika, which is playing a key role in the company’s internal development.
Urika works by unearthing the hidden connections between various data and constructing patterns that tie them together. Combined with the ability to quickly process new data as it enters the equation, the analytics system produced by Cray-owned YarcData is perfect for its current primary application, the detection of Medicare fraud. The federal government faces over $60 billion in fraud out of the approximately $500 billion of payments it processes annually. Detecting these losses demands incredibly fast, responsive and scalable computing with the ability to filter through hundreds of millions of transactions, and Urika aims precisely at that very target.
Texas researchers seek to build more resilient supercomputers
Researchers at the University of Texas at Austin are now working with Cray to solve one of the biggest problems facing the supercomputer industry – resilience – reports a new article from the U.S. Department of Energy.
Anyone who’s worked in a computing- or tech-heavy field knows how important it is for computers to be able to bounce back from various fail-states. Inevitably, supercomputers will miss calculations or encounter errors that occur as a result of highly miniaturized components. Traditionally, the article points out, this problem is combated by saving the system’s internal state, so that it can return to that if a problem arises. However, as the size of supercomputers reaches the petaflop and exaflop scale, the resource cost of saving an entire system state becomes prohibitive.
Using a strategy called a containment domain, Cray and the UT researchers hope to reduce the scale of these save and check processes to smaller, specific algorithms. Using containment domains, the system will run and check its operations on a smaller level, and the problem will be detected at a lower level, decreasing the necessary time to roll back and correct a mistake. The containment domain approach also offers several advantages over other ways of dealing with the problem, including improved levels of isolation from other parts of the system and more tools for programmers to customize the reaction to a particular type of error.
In Australia, supercomputers help astronomers search the skies
A new radio telescope in Western Australia – dubbed the Murchison Widefield Array – began sending a flood of data to one of the country’s premier supercomputing centers, Phys.org reported this week.
The Pawsey Center in Kensington is a multi-stage project, and its second stage includes a major CPU-based Cray supercomputer dedicated to storing and processing data from radio astronomy arrays. The center, which has been under construction since 2009, officially opens later this year, but it’s already begun supporting many of the research projects it is slated to work with
“We now have more than 400 megabytes per second of MWA data streaming along the National Broadband Network from the desert 800km away,” Professor Andreas Wicenec of The University of Western Australia told the website.
Wicenec described the sophisticated distribution and filtering of the data as well, explaining how the flow of radio data can be specified to the needs of particular institutions around the world, preventing wasted resources spent transferring data to partners who don’t actually need it.
“We’re really impressed with iVEC and the staff from Cray and SGI,” said Wicenec. “We’ve been pushing to have resources ready as soon as possible so we could take advantage of Pawsey’s capabilities for the MWA. They have provided us with very early access to the facility, which has let us interact directly with the experts and optimize the integration and setup of complex hardware and software systems.”
Check back with us next week for more developments in high-performance computing sector.