News dalla rete ITA

19 Marzo 2019

Stati Uniti


Argonne National Laboratory plans to build the world’s fastest supercomputer, with help from chipmaker Intel, computer-maker Cray and more than $500 million from the U.S. Department of Energy. The machine, called Aurora, will be the first exascale computer in the U.S., capable of 1 quintillion—or 1 billion-billion—calculations per second. It will put Argonne, and Chicago, at the bleeding edge of computing. It will be nearly 100 times faster than the newest supercomputer at Argonne in Lemont, a machine called Theta, which debuted in 2017. Theta also was built by Intel and Cray, a pioneer in supercomputers. When completed in two years, Aurora also will take its place alongside Theta and Mira, a machine built by IBM in 2012, at Argonne’s Leadership Computing Facility. What is exascale? It means 1,000,000,000,000,000,000 calculations per second—a computer fast enough to begin approximating the human brain, said Bobby Kasthuri, a neurobiology researcher at Argonne and University of Chicago, which manages the lab. Exascale computers are 1,000 times faster than petascale machines, which emerged about a decade ago. In more practical terms, it means a computer fast enough—and big enough—to take artificial intelligence to the next level, said Rajeeb Hazra, who leads Intel’s data center group. “It’s the convergence of high-performance computing with machine learning, data analytics and AI, which will accelerate advancement in computing, networking and storage to provide capability never seen before.” AI is used in everything from voice-activated computing, such as Amazon’s Alexa, to autonomous driving. After years of rudimentary functionality, AI is advancing rapidly. Researchers are likely to use the new machine for cancer research, as well as traditional science such as astronomy and physics. Aurora also will be able to analyze the massive amount of data coming from particle accelerators and telescopes in real time, said Rick Stevens, a University of Chicago professor who is Argonne’s associate lab director for computing, environment and life sciences. Supercomputers are the workhorses of modeling and simulations, using massive amounts of data and mathematical formulas, or algorithms, to test and develop everything from new materials used in solar-energy cells or batteries to processes used in next-generation nuclear power plants. “High-performance computing is one of three legs we rely on to drive energy discovery and analysis,” said Paul Dabbar, undersecretary for science at the Department of Energy. Link: (ICE CHICAGO)

Fonte notizia: