For many years, the majority of servers in the world, from those used in supercomputers on the Top500 List to the servers filling Google and Facebook’s massive data centers, started with a motherboard containing two x86 processors. Server manufacturers then piled on one or more network interface cards, disk drives, memory, and a host of other support chips, then packaged up with fans and power supplies in a rack mount or blade form factor. Over the last several years, however, customers workloads, especially in HPC and giant web data centers, increasingly started to look at non-x86 processors with different performance/watt characteristics. A great example in the HPC space is Nvidia’s Tesla series of GPU, with server’s like HP’s ProLiant SL390s adding from 1 to 8 Tesla GPUs to a 2-socket x86 base. Nvidia painted one picture of the future of HPC earlier this year when they announced Project Denver which ultimately will free the GPU from being tied to an x86 processor.
Today, HP announced the next step in the rethinking of server design, Project Moonshot.
The first 3 parts of Project Moonshot are:
The HP Redstone server development platform, designed for benchmarking, testing, and proof of concept.
The HP Discovery Lab, to experiment, test, and benchmark applications
The HP Pathfinder Program, an ecosystem of industry leading partners
Using ARM System On a Chip (SOC) technology, the Redstone platform packs 288 single socket servers into a single 4RU chassis, delivering extreme energy efficiency for a wide range of web workloads. While the fact that the Redstone development platform is based on ARM processor technology is interesting, the focus of Moonshot is not just about ARM, it is about rethinking server design from the ground up, for scalability, for performance, and for energy efficiency.
Visit the Project Moonshot page on HP’s web site for more info and to keep up to date as HP rethinks server design, again!
Marc Hamilton – Hyperscale Business Unit, HP Enterprise Group.
Marc works in the Hyperscale Business Unit within HP's Enterprise Group where he leads the HPC team for the Americas region. He brings more than 26 years of global engineering, sales and executive management experience to HP. Marc’s team works across HP engineering, marketing, and sales organizations as well as HP Labs to design, develop, and deliver world class HPC systems, ranging from some of the world’s fastest supercomputers installed at national research labs and leading universities to commercial HPC systems across a variety of industries including energy, manufacturing, financial services, and life sciences.
Prior to joining HP in October 2010, Marc spent 16 years at Sun Microsystems and Oracle in HPC and other sales and marketing executive management roles. At Sun, his team built a number of systems that placed in the top 10 of the Top500 list of the world’s fastest supercomputers, including systems at Sandia National Labs, Texas Advanced Computing Center (TACC), Germany’s Juelich supercomputing center, and the Tokyo Institute of Technology.
Prior to Sun, Marc worked at TRW developing HPC applications for the US aerospace and defense industry. He has published a number of technical articles and is the author of the book, “Software Development, Building Reliable Systems”. Marc holds a BS degree in Math and Computer Science from UCLA, an MS degree in Electrical Engineering from USC, and is a graduate of the UCLA Executive Management program.
Pingback: Video: HP’s Project Moonshot to Leverage Low-Power Redstone Processors | insideHPC.com