IT: Supercomputer’s performance to get 500 times faster

21 June 2011

Silicon Graphics International (SGI), supercomputer developer, hopes by 2018 to build a supercomputer 500 times faster than the most powerful today, using specially designed accelerator chips produced by Intel. SGI’s chief technology officer Eng Lim Goh aims to bring a massive performance boost to its supercomputers through highly parallel processors based on Intel's many integrated cores (MIC) architecture. He stressed that in conjunction with Xeon server CPUs, the MIC chips will run millions of threads in parallel, which will help scale supercomputer performance.

Chips based on the MIC architecture will mix standard x86 cores with specialized cores to boost high-performance computing. On June 20 was released the latest biannual TOP-500 publication - the list of Top 500 supercomputers on the planet. According to the list, the world’s most powerful supercomputer is Japanese Fujitsu K with the performance equaling 8 petaflops, in prospect to reach 10 petaflops. But efforts to improve throughput and on-chip performance are under way. A case in point is IBM, which said it will release a supercomputer that can deliver performance of over 1000 petaflops, by 2020. Such supercomputers will have hybrid multi-core processors and high-speed hubs for data transmission. 

Intel showed its first experimental MIC chip, code-named Knights Ferry. The chip sits in a PCI-Express slot, it has 32 extra cores and combines vector processing units with standard CPU cores. It has not been commercially released yet.

Intel tells that Knights Ferry chips can deliver 7.4 teraflops of performance. However Intel boasts Knights Corner chips with more than 50 cores to be made using the 22-nanometer manufacturing process. Key server and supercomputer manufacturers like HP, Dell and IBM say they take an interest in commercial release of MIC-chips.