ibm

A new line of IBM Power servers scheduled for release in 2016 is receiving power upgrades that will make the servers faster than the company’s Watson supercomputer.

IBM (NYSE: IBM) has been touting Watson, best known for outperforming humans in the TV game show Jeopardy in 2011, as a powerful number cruncher well-suited for big data and analytics tasks. IBM’s Power servers are based on the Watson architecture.

Big Blue is gunning for the lucrative server market in China and wants to overtake Intel’s x86 server chips. After its recent sale to Lenovo of its x86 server business, Power chips have become critical to IBM’s server operations.

Historically, IBM has held tight control over Power development but now the company is working with several partners to boost the performance of Power servers, according to Brad McCrede, vice president and fellow at IBM.

There is a growing need for more bandwidth as computers become faster and IBM is aiming to speed up application performance of Power chips and resolve latency issues by adopting Nvidia Corp.’s NVLink interconnect technology, according to a report in Techworld.  Nvidia (NASDAQ: NVDA) claims NVLink can deliver five times more throughput that PCI-Express pipes in most severs today.

Nvidia also has plans to use NVLink interconnects in its graphics processors which could be embedded into the new Power servers.

IBM is collaborating with the OpenPower Foundation, which includes Google, Nvidia, Samsung, and Tyan Computer Corp. The consortium was established last year to enable other companies to develop chips, servers and components for the Power architecture.

The idea is to gain market dominance by turning Power servers into more commercially available servers.

Google has a developed a prototype server based on Power chips and Tyan is preparing to release its first non-IBM Power server.

 

 

 

 

 



Related Download
Top tips for securing big data environments Sponsor: IBM
Top tips for securing big data environments
Download this white paper to find out how your organization can improve security decision-making and monitor big data environments.
Register Now