UPDATED 11:46 EDT / NOVEMBER 14 2016

INFRA

New Cray supercomputer offers petaflop computing power in one cabinet

More than 30 years after selling its first supercomputer, Cray Inc. continues to push the envelope on processing power.

The company unveiled a new data crunching system at the SC16 computing event in Utah today that is the first of its kind to incorporate Nvidia Corp.’s latest-generation P100 accelerators. Based on the chip maker’s 16-nanometer Pascal architecture, the series is touted as the fastest in the world. Each unit can provide 4.7 teraflops of performance when carrying out calculations with double-precision numbers (numerical values that take up eight bytes) and more than twice that for single-precision operations.

The result is that a single cabinet in the XC50, as Cray calls its new platform, is capable of delivering more than a petaflop of computational capacity when operating at peak speed. For perspective, it would take about 93 such cabinets to match the speed of the world’s fastest supercomputer, China’s Sunway TaihuLight system. Supercomputer ranking site TOP500 states that the XC50 has a maximum capacity of 500 petaflops.

So-called petascale computing is intended to tackle the most demanding computational tasks, such as modeling weather. Cray said the system also will be useful in artificial intelligence, in particular the branch of AI called deep learning that is responsible for big advances recently in speech and image recognition. Nvidia’s graphics processing units that are used in the new machine have become the standard chips in servers used to do deep learning.

The system’s scalability is owed largely to homegrown Aries router chips that Cray has been shipping with its XC systems since the series launched in 2012. They enable cabinets to communicate at the rapid speed that high-performance computing workloads require. The new XC50 also inherits the company’s DataWarp storage acceleration technology, its management software and support for Intel Corp.’s latest-generation Broadwell processors. Organizations that aren’t satisfied with Nvidia will have the option of fitting their deployments with the chip giant’s Xeon Phi accelerators instead of P100s.

The first adopter of the XC50 is the Swiss National Supercomputing Centre, which is currently in the process of converting its existing XC30 deployment into the new platform. Dr. Thomas C. Shulthess, the director of the institute, said in a prepared statement that the upgrade will “significantly accelerate our computational research capabilities, allowing our users to perform more advanced, data-intensive simulations, visualizations and data analyses across a wide array of scientific studies.”

Image courtesy of Cray

A message from John Furrier, co-founder of SiliconANGLE:

Your vote of support is important to us and it helps us keep the content FREE.

One click below supports our mission to provide free, deep, and relevant content.  

Join our community on YouTube

Join the community that includes more than 15,000 #CubeAlumni experts, including Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger, and many more luminaries and experts.

“TheCUBE is an important partner to the industry. You guys really are a part of our events and we really appreciate you coming and I know people appreciate the content you create as well” – Andy Jassy

THANK YOU