Sure, IBM’s 10 petaflop supercomputer might receptive to advice impressive, though Cray can do you 5 improved — a outfit only voiced a Cray XK6, an upgradable, hybrid supercomputing complement means of some-more than 50 petaflops of computational muscle. Powered by Cray’s Gemini interconnect, AMD Opteron 6200 processors, as well as NVIDIA Tesla 20-Series GPUs, a XK6 complement blends x86 as well as GPU environments with a firm’s own season of Linux. The folks during Cray won’t review to bragging, however — they’re humbly dogmatic a appurtenance to be a initial “general-purpose supercomputer formed upon GPU technology,” as well as not, as they put it, a attempt to place tall upon any Top 500 lists. Suggestive, aren’t they? Check out a artless press recover after a break. Show full PR textCray Unveils a Cray XK6 Supercomputer
New System Leverages a Latest Technologies From Cray, AMD as well as NVIDIA
SEATTLE, WA as well as FAIRBANKS, AK, May 24, 2011 — Global supercomputing personality Cray Inc. (NASDAQ: CRAY) currently voiced a launch of a Company’s latest prolongation hybrid supercomputing complement — a Cray XK6 supercomputer — during a 2011 Cray User Group (CUG) assembly in Fairbanks, Alaska. The rarely approaching latest supercomputer will mix Cray’s Gemini interconnect, AMD Opteron(TM) 6200 Series processors (code-named “Interlagos”) as well as NVIDIA Tesla 20-Series GPUs to emanate a tightly-integrated supercomputing complement upgradeable to some-more than 50 petaflops (quadrillions of operations/second) of discriminate power.
The Cray XK6 complement introduces an innovative proceed to augmenting programmer capability with a one x86/GPU programming sourroundings which includes tested as well as absolute tools, libraries, compilers as well as third-party software. When total with a Cray Linux Environment, a outcome is a entirely integrated Cray supercomputer which blends scalable hardware, module as well as a network. Cray XK6 business will be means to implement a capabilities of a multi-purpose supercomputer written for a next-generation of many-core, tall opening computing (HPC) applications.
The Swiss National Supercomputing Centre (CSCS) in Manno, Switzerland is Cray’s initial patron for a latest Cray XK6 system. CSCS, which develops as well as promotes technical as well as systematic services for a Swiss investigate village in a margin of HPC, sealed a stipulate with Cray to ascent a Cray XE6m system, nicknamed “Piz Palu,” to a multi-cabinet Cray XK6 supercomputer. CSCS is a long-standing Cray patron as well as supports scientists operative in different fields such as continue forecasting, climatology, chemistry, physics, element sciences, geology, biology, genetics, initial medicine, astronomy, arithmetic as well as mechanism sciences.
Professor Dr. Thomas Schulthess, Director of CSCS, said, “Given a conspicuous seductiveness in GPU record from a Swiss computational scholarship community, it is necessary which CSCS adopt this record in to a high-end prolongation systems soon. However, you have been not seeking for an additional GPU formed attempt to place tall upon any Top500 lists. The Cray XK6 promises to be a initial general-purpose supercomputer formed upon GPU technology, as well as you have been really many seeking brazen to exploring a opening as well as capability upon genuine applications applicable to a scientists.”
A absolute further to Cray’s line of highly-advanced supercomputers, a Cray XK6 complement pairs industry-leading HPC technologies from AMD as well as NVIDIA to emanate a loyal hybrid supercomputer with intra-node flexibility. By mixing AMD Opteron processors with NVIDIA GPUs, a discriminate node in a Cray XK6 complement gives users a choice to run applications with possibly scalar or accelerator components. The Cray XK6 discriminate blades can additionally be blended with Cray XE6 discriminate blades in to a single, multi-purpose supercomputing system.
“Every aspect of a Cray XK6 has been engineered to encounter a real-world opening final of researchers as well as scientists tasked with elucidate a world’s grand challenges,” pronounced Barry Bolding, clamp boss of Cray’s product division. “Cray has a prolonged story of operative with accelerators in a matrix technologies. We have been leveraging this imagination to emanate a scalable hybrid supercomputer — as well as a compared first-generation of a one x86/GPU programming sourroundings — which will concede a complement to some-more productively encounter a systematic hurdles of currently as well as tomorrow. We built a world’s initial prolongation petaflops complement with a Cray XT5 supercomputer, reinvented tall opening networking with a Gemini interconnect, as well as you have been right away redefining accelerator-based supercomputing with a one GPU as well as scalar technologies built in to a Cray XK6 system.”
Upgradeable from Cray XT4, Cray XT5, Cray XT6 or Cray XE6 systems, a Cray XK6 complement is approaching to be accessible in a second half of 2011. It can be configured in a singular cupboard with tens of discriminate nodes, to a multi-cabinet complement with tens of thousands of discriminate nodes.
“GPUs have been personification an increasingly critical purpose in tall opening computing as researchers as well as scientists demeanour to enlarge a CPU opening upon sure codes,” pronounced Earl Joseph, IDC module clamp boss for HPC. “Cray is a personality in a high-end of a supercomputing marketplace as well as has designed, built as well as commissioned a little of a world’s many absolute systems. The association has decades of knowledge in successfully integrating hybrid processor sorts in to vast scale systems. This plan of hybrid supercomputing is intensely valuable, as well as Cray is in a great on all sides to take value of it.”
View a Original article
