IBM today revealed that as far as the Power series processor architecture is concerned, there is strength in numbers.
In partnership with Google, NVIDIA, Mellanox Technologies and Tylan, IBM announced it has formed the OpenPOWER Consortium.
According to Brad McCredie, IBM vice president and chief technology officer for the IBM Systems and Technology Group, the alliance members will collaborate on system-on-chip designs that will drive the development of next-generation servers based on a forthcoming Power8 processor design.
The goal, says McCredie, is to increase the number of companies producing data center technologies based on Power8 to help reduce costs while simultaneously taking advantage of the skillsets of engineers from multiple companies to drive multiple SOC designs.
Google, meanwhile, plans to build its own custom servers using Power processors in addition to the custom servers it builds using x86 servers. NVIDIA, meanwhile, committed to integrate its graphics processing units (GPUs) with Power processors. Tylan plans to build motherboards for Power8 processors.
In the last few months, IBM has significantly stepped up its effort to make sure that Power processors can compete against x86 processors manufactured either by IBM or Advanced Micro Devices (AMD).
As a precursor to the development of Power8 processor, McCredie says IBM will shortly showcase new interconnects that IBM expects will drive much of the I/O performance for servers based on Power8 processors.
IBM and its partners don’t expect Power processors to usurp x86 processors entirely. But as density issues continue to challenge IT organizations of all sizes, IBM is betting that a much higher percentage of compute-intensive workloads are going to move to higher-end processor platforms.