Cray enters GPU supercomputing space with XK6

Company ties Nvidia chips into supercomputer

Cray has become the latest company to tie graphics chips into its high-performance computing platform.

The forthcoming XK6 supercomputer is powered by a combination of AMD Opteron CPUs and Nvidia Tesla GPU processors. At its top configuration, Cray estimates that the XK6 will reach speeds of 50 petaflops.

The supercomputer will use the GPU processors to help handle parallel processing operations, a system known as general processing over GPU (GPGPU).

GPU chips can be especially well-suited for traditional supercomputing tasks such as modelling and simulations because graphics hardware is specially designed for multi-threading operations.

"Cray has a long history of working with accelerators in our vector technologies," said Cray product division vice president Barry Bolding.

"We are leveraging this expertise to create a scalable hybrid supercomputer, and the associated first-generation of a unified x86/GPU programming environment, that will allow the system to more productively meet the scientific challenges of today and tomorrow."

Cray said that the XK6 is slated to hit the market in the second half of the year. The first system will be installed at the Swiss National Supercomputing Centre.

Cray is not the only company looking to combine GPU chips with traditional central processors. Chipmaker AMD has used hardware from its ATI graphics branch to create the Stream line of GPU cards.

Other GPGPU vendors such as NextIO have estimated that the hardware can boost performance as much as 30-fold.