Intel Corp. has patiently built its low-latency, high-speed computing fabric over the past six years and the results, in terms of market penetration, are starting to show. After acquiring Fulcrum Microsystems (technology for Ethernet switches) in 2011, the company also added the InfiniBand switch product line from QLogic in 2012.

Less than six years later, Intel now has server penetration in more than half of the market’s advanced, hyperscale large-gigabit platforms. “We made a couple of strategic acquisitions, combined that with our own [intellectual property], and came up with Omni-Path,” said Susan Bobholz (pictured), vice president of the Omni-Path Group at Intel. “Sixty percent of the servers that are running 100 gigabit fabrics in the Top 500 list are running and connected via Omni-Path.”

Bobholz stopped by theCUBE, SiliconANGLE Media’s mobile livestreaming studio, and spoke with host Jeff Frick (@JeffFrick) at Supercomputing 2017 in Denver, Colorado. They discussed customer demand for scalable high performance computing systems and the future use of Omni-Path’s technology for deep leaning and artificial intelligence. (* Disclosure below.)

Solution remains compatible with InfiniBand

Intel’s acquisitions, followed by the release of Omni-Path eighteen months ago, were made in response to customer concerns that existing high performance computing (HPC) systems were too expensive and could not scale to meet future needs. The Omni-Path solution provides the ability to scale tens of thousands of nodes while remaining compatible with InfiniBand APIs.

“It allows you to connect systems and make big, huge supercomputers,” Bobholz explained. “That means you can build denser clusters in less space and cables with lower power, and the total cost of ownership goes down.”

As organizations move more strongly into developing artificial intelligence and deep learning applications, Intel expects to see a growing demand for systems which can reduce time to train models while maintaining accuracy. Much like high performance computing, AI demands low latency and high bandwidth.

“Our target market is high performance computing, but we’re also seeing a lot of deployments in artificial intelligence now,” Bobholz said. “They have the same exact needs. Do it fast and do it quick.”

Watch the complete video interview below, and be sure to check out more of SiliconANGLE’s and theCUBE’s coverage of the Supercomputing 2017 conference. (* Disclosure: TheCUBE is a paid media partner for the Super Computing 2017 conference. Neither Intel, the event sponsor, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE