HPC System Scalability

When HPC vendors mention the word scalability, not far behind comes talk of Exascale computing. Using the simple definition of the Top500 list, an Exascale system would be about 400 times more powerful than the current Top500 record holder clocking in at 2.5 Petaflops. Today HP is one of the leading US companies with demonstrated Petascale performance. In fact, at least by measure of the Top500 benchmark, HP is one of only two companies (the other being Cray) building Petascale systems using industry standard x86 processors and GPUs, the building blocks of which are shown in this systems diagram from the Keeneland system.

However, perhaps a better definition of scalability is how well your vendor’s HPC architecture scales down the Top500 list (and below) in terms of performance. I like to talk about HPC for the masses. It is great to have the technology to build a $30M supercomputer, but how many people actually get to use such a system? Some of the world’s fastest supercomputers are reserved for use by only a handful of researchers and require years of custom software development. By contrast, HP’s most powerful petaflop supercomputer scales down to entry level HPC building blocks costing just a few thousand dollars and programmable by anyone. Later this month, anyone can learn how to program a GPU powered supercomputer thanks to a workshop being offered by the Keeneland project.

On April 14 and 15, the Keeneland project and the Georgia Tech NVIDIA CUDA Center of Excellence will present a two day tutorial on GPU heterogeneous processing for computational science. If you can get to Atlanta this month, the $100 registration fee has to be one of the best price/performance bargains in the education world.

Of course, you don’t have to be a programmer to benefit from the advanced performance of HP’s GPU powered SL390. With a quick visit to Nvidia’s Cuda Zone you can find 100’s of software applications, libraries, and tools that are already available for GPU computing.

So while researchers at HP Labs Intelligent Infrastructure project work on technologies likely to find their way into future Exascale systems, you don’t have to work at HP labs to experience the power of HP supercomputing today, just get to Atlanta on April 14, or better yet, buy your own SL390 GPU-powered system.

About Marc Hamilton

Marc Hamilton – Vice President, Solutions Architecture and Engineering, NVIDIA. At NVIDIA, the Visual Computing Company, Marc leads the worldwide Solutions Architecture and Engineering team, responsible for working with NVIDIA’s customers and partners to deliver the world’s best end to end solutions for professional visualization and design, high performance computing, and big data analytics. Prior to NVIDIA, Marc worked in the Hyperscale Business Unit within HP’s Enterprise Group where he led the HPC team for the Americas region. Marc spent 16 years at Sun Microsystems in HPC and other sales and marketing executive management roles. Marc also 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.
This entry was posted in Uncategorized. Bookmark the permalink.