January 02, 2013
China has made no secret about becoming a world player in the supercomputing arena. With 72 machines on the latest TOP500 list, the Asian giant is second only to the United States in the number of top systems. Along the same line, China also vying to host the world's first exascale computer.
The country may be making a big step in that direction if the latest report from VR-Zone's Theo Valich turns out to be true. According to him, the Chinese government is planning to deploy a 100-petaflop supercomputer within the next 18 months, which would put the country on a extra-fast trajectory to exascale computing.
In fact, if China manages to field such a system in 2014, that would put it a year ahead of the TOP500's performance projection for a 100-petaflop machine, as well as a year ahead of China's original plans for a such a system. Last October, the Guangzhou Supercomputing Center was talking about the Tianhe-2 system, a 100-petaflop machine that would succeed China's current number one system, the Tianhe-1A. The Tianhe-2 was slated to be deployed in 2015 by China's National University of Defense Technology, but it's not clear from the VR-Zone piece if this new supercomputer is simply that system on a faster timeline or an entirely different machine.
Supposedly, this 2014 system is going to be based on Intel parts -- specifically 100,000 Xeon Ivy Bridge-EP CPUs paired with 100,000 Xeon Phi coprocessors. The coprocessors alone should be enough to supply all those FLOPS, given that even the 2013-era Phi parts deliver more than a teraflop apiece. The 100,000 Ivy Bridge Xeons would just be petaflop gravy.
According to Valich's (unnamed) source, the project's processors are expected to cost around $100 million. As he noted, that's likely to be well below the retail price for this hardware, given that both the high-end Xeon and Xeon Phi parts on the market today both run well above $1,000. As a result, Valich estimates the retail price for all this processing power would normally total over $500 million. Assuming this is all true, that would mean Intel won't even cover its cost for the silicon on the deal.
When asked by HPCwire to confirm or deny the report, Intel responded with "We don't comment on any rumors and speculations."
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?