1. Fujitsu announced that it has received an order for a new supercomputer system from Canon.. The system will be a 96-node configuration of the Fujitsu Supercomputer PRIMEHPC FX10, and will have a theoretical peak performance of 20.2 teraflops.
The system will contribute to more sophisticated analytical simulations in Canon’s product development processes. The system is expected to begin operations in October 2013.
2. NCSA’s Blue Waters supercomputer is being upgraded with 12 additional Cray XK racks, each with 96 nodes. This brings the total number of Cray XK nodes to 4,152 and boosts the system’s peak performance to over 13 petaflops.
During the past six months, NCSA has seen more and more science and engineering teams modifying their codes to take advantage of the considerable computational power of the GPU accelerators offered by the XK nodes.
China is the current supercomputer leader with the Tianhe-2.. It is a supercomputer developed by China’s National University of Defense Technology, is the world’s new No. 1 system with a performance of 33.86 petaflop/s on the Linpack benchmark, according to the 41st edition of the twice-yearly TOP500 list of the world’s most powerful supercomputers. It has a peak speed of 55 petaflops.
There is still the possibility that the Tianhe-2 could get GPU upgrades next year or get new chips in 2015 or that a Tianhe-3 with upgraded components could be built in 2015 to achieve a 100+ petaflop supercomputer.
Tianhe-2 is powered by 32,000 of the upcoming 12-core Intel Xeon processors E5-2600 v2 based on Ivy Bridge architecture, and 48,000 Intel Xeon Phi coprocessors, with a total system power of 17.8 MW
The current “Knights Corner” Xeon Phi coprocessors have 60 active cores (on a die with 64 cores) running at 1.05GHz delivering just over 1 teraflops of oomph. To get 80 petaflops of peak number-crunching oomph, you would need 75,973 Xeon Phi cards, which would work out to around 1.5 Xeon Phi cards per node. Call it two Xeon Phi’s in the Knights Corner generation per node just for fun, and that alone gives you 105.3 petaflops with the current generation.
Xeon Phis by as much as 25 to 30 per cent without moving to 14 nanometers. So with 100,000 Xeon Phi v2 coprocessors, you’d be at somewhere around 135 petaflops on the Xeon Phis and another 21 petaflops on the server nodes. Now you are pushing up to 156 petaflops peak.
It is possible that China is working on such a machine, but it is hard to imagine that it will cost as little as $100 million for the processing elements. If you bought 100,000 Xeon Phi v1 coprocessors at list price based on 1,000-unit trays, you’d pay $265 million, and 50,000 server nodes would run you maybe another $300 million depending on the memory and networking if you bought them as one at a time online.
Assuming China is using its own proprietary interconnect, you might be able to get the base servers at $250 million list. Call it a cool $515 million at list for both the servers and the Xeon Phis, and maybe with a 45 per cent discount and some rounding, you could get it down to $285 million.