UPDATED 00:01 EDT / DECEMBER 11 2018


IBM targets enterprise AI workloads with new Nvidia-powered converged system

Research firm International Data Corp. estimates that 75 percent of enterprise applications will use artificial intelligence by 2021. To address this growing market better, IBM Corp. today introduced a powerful AI-optimized converged system called Spectrum AI with Nvidia DGX.

The platform joins the rising number of data center platforms tailored specifically for machine learning. In September, Cisco Systems Inc. debuted an AI server that packs eight of Nvidia Corp.’s top-end Tesla V100 graphics cards. Earlier this year, NetApp Corp. and Pure Storage Inc. launched their own AI-optimized platforms powered by Nvidia’s DGX-1 appliance.

As its name indicates, Spectrum AI with Nvidia DGX is based on the same system. Like Cisco’s server, the DGX-1 includes eight Tesla V100s graphics cards with more than 45,000 processing cores between them. About 5,000 of those are specialized circuits called Tensor Cores that are tuned to run machine learning models well.

All this silicon enables the DGX-1 to deliver more than a petaflop of computational power, according to Nvidia. A petaflop equals a quadrillion floating-point operations per second, a standard unit of computer performance.

IBM’s new system pairs the DGX-1 with its own Elastic Storage Server. It’s a storage platform that offers up to a petabyte of usable flash capacity and runs the company’s Spectrum Scale management software.

The inclusion of Spectrum Scale is notable for several reasons, not least of which is the fact that it’s used in Summit, the world’s most powerful supercomputer. On top of the ability to accommodate massive amounts of data, the software also provides an extensive set of administrative capabilities. Among them is a tool that enables information technology teams to offload some of the records in an environment to the public cloud.

Another piece of software included in Spectrum AI with Nvidia DGX is the RAPIDS framework that Nvidia released two months ago. It’s a collection of libraries that can be installed together with popular AI and analytics tools to squeeze more computing power out of the chipmaker’s graphics cards.

The appliance can accommodate projects of greatly varying size, according to IBM. Companies may set up Spectrum AI with Nvidia DGX environment using a single Elastic Storage Server and a small handful of DGX-1 appliances or, on the other end of the spectrum, deploy multiple racks’ worth of gear. A single rack with nine DGX-1 machines provides data throughput of up to 120 gigabytes per second, which IBM said is more than all of the competing solutions it had tested.

Photo: IBM

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

Show your support for our mission by joining our Cube Club and Cube Event Community of experts. Join the community that includes Amazon Web Services and Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger and many more luminaries and experts.

Join Our Community 

Click here to join the free and open Startup Showcase event.

“TheCUBE is part of re:Invent, you know, you guys really are a part of the event and we really appreciate your coming here and I know people appreciate the content you create as well” – Andy Jassy

We really want to hear from you, and we’re looking forward to seeing you at the event and in theCUBE Club.

Click here to join the free and open Startup Showcase event.