A second show was held last week concurrently with NVIDIA’s GTC and ECS shows: the OpenPOWER Foundation’s Summit 2016. What makes this event interesting is that it showcases IBM’s move against Intel’s dominance using open source methods and a social community approach to next-generation server processor architectures. Realizing it couldn’t take on a power like Intel alone, it effectively gave away its technology in exchange for making it more relevant, and the effort appears to be working.
Let’s talk about some of the big takeaways from the event.
Massive Google Support for OpenPOWER
Perhaps the biggest indicator that OpenPOWER is making headway is that Google now stands solidly behind the OpenPOWER effort along with development partner Rackspace. Apparently, Google has co-designed a high capacity 48V rack with Facebook to specifically address both companies’ unique, very high-capacity needs, and the open nature of the OpenPOWER platform has allowed them to create a solution specifically for these new racks.
This is a huge endorsement because companies like Google and Facebook push the envelope for cloud server loading, performance and cost. It is unlikely that any other type of company would need to have a level of sustained performance, suggesting that OpenPOWER solutions can scale to almost any cloud load.
And to further exemplify Google’s unique load requirements, it pointed out that in 2008, it only had to track 1 trillion addresses, but today that number is 60 trillion (no wonder web searches seem so much slower now).
University of Michigan to Collaborate with IBM
Speaking of scale out, the University of Michigan announced at the event that it was collaborating with IBM to build a new class of HPC systems specifically designed to increase the pace of scientific discovery. IBM is one of the few companies that maintains a relatively large scientific R&D function, indicating that this firm not only would be interested in selling such a system but it could actually use it as well. Often a vendor that has a common use with a customer on a new design more quickly grasps unique aspects of the shared problem it is being designed to correct, resulting in a better solution faster.
The combined effort has resulted in a new resource called ConFlux, a very high-speed communications protocol connecting high-performance computing clusters directly. The resulting systems are specifically targeting large-scale, data-driven modeling of complex physical problems. The example given was to analyze an aircraft engine at a molecular level in order to increase its output and efficiency. This requires a massive parallel effort to emulate all of the physical and chemical interactions and then be able to adjust the virtual parameters in real time and at high speed to determine the optimal design.
Thought to be the foundational element to ending most diseases, genomics research has been a target for HPC efforts. Requiring massive levels of compute power, this research area is not only one of the most important for fighting disease but also an excellent showcase for applied performance. OpenPOWER added to existing efforts with LSU and tranSMART with the Dragen Genomics platform, developed in conjunction with Edico Genome and Xilinx. This new platform is designed to enable ultra-rapid analysis. Historically, it has taken hours to analyze an entire genome, but with this new technology the time has been cut to minutes. Currently, this technology is being used to identify patients at higher risk for cancer before conditions worsen. Customers for the resulting solution include Rady Children’s Institute for Genomic Medicine of San Diego.
Wrapping Up: Creative Competition
While this approach to hardware competition is still relatively new, the progress that the OpenPOWER foundation is making in what is a very competitive market is impressive. This is showcasing that thinking out of the box with regard to technologies like this and giving up control can result in a far better outcome than the more typical head-to-head competition more commonly used. The end result is that OpenPOWER is advancing far better as a consortium than it did when it was wholly owned by IBM and demonstrating yet another creative path to competing with an otherwise unbeatable foe.
Rob Enderle is President and Principal Analyst of the Enderle Group, a forward-looking emerging technology advisory firm. With over 30 years’ experience in emerging technologies, he has provided regional and global companies with guidance in how to better target customer needs; create new business opportunities; anticipate technology changes; select vendors and products; and present their products in the best possible light. Rob covers the technology industry broadly. Before founding the Enderle Group, Rob was the Senior Research Fellow for Forrester Research and the Giga Information Group, and held senior positions at IBM and ROLM. Follow Rob on Twitter @enderle, on Facebook and on Google+