What To Look For At SC12

SC12 kicks off in just under 2 weeks, November 10, 2012, in beautiful Salt Lake City. Officially the International Conference for High Performance Computing, Storage, Networking, and Analysis, but often just referred to as ‘SC for short (not to be confused with USC), this show is one of the few remaining conferences for engineers and computer scientists truly interested in hardware. Since the conference has for many years taken on a life of its own, it is easy to forget it is sponsored by the IEEE Computer Society, for which I’ll give a quick plug. Each year, before SC, I like to step back and think about what industry trends to watch for at the conference. Here are a few to think about as you plan your schedule for what is always a packed week of activities.

Of course with over 10,000 attendees present from the high performance computing community, many vendors hold user group meetings before and during the show (everyone has learned that after a week of SC, you are so information overloaded that after the show is not a good time for user groups). HP is no exception and we kick off our HP-CAST community meeting on November 9th. But I digress, so back to my “what to look for” list.

Expect many papers, presentations, and discussions to focus on performance/watt from one angle or another. But certainly performance/watt topics that will find broad coverage at the show will include:

  • New component, server, rack, and data center cooling technologies. One of the HP-CAST keynotes will be given by the National Renewable Energy Lab’s Steve Hammond, talking about NREL’s new ESIF data center and the petaflop class supercomputer they recently purchased from HP. I expect every vendor will be talking, either on the show floor or in their NDA suites, about their latest approaches to cooling, which today often accounts for 50% or more of the energy usage in a data center. While HPC data centers tend to push power and cooling technology to the limits of physics, large web and cloud data centers are not far behind and addressing data center cooling is an area of industry-wide interest far beyond HPC data centers. Ultimately, delivering the best performance/watt requires more efficient cooling technologies that today’s traditional computer room air conditioning systems.
  • GPU, APU, and co-processor technologies. I expect Intel, along with their systems partners and customers, will be talking a lot about the upcoming Intel Xeon Phi co-processor, along with Nvidia talking up GPUs and AMD their APU technology. Not to go down a comparison of these technologies, but certainly what they all have in common in the promise of delivering significantly higher performance/watt than traditional x86 processors. But the real win here for the HPC community, I believe, is that with Intel throwing their weight behind massively multi-core chips, a much broader range of computer scientists and software developers will start thinking about how to express parallelism in their algorithms and in their software. Once that happens on a broader scale, it will benefit everyone from developers simply using the latest Intel compilers for Xeon Phi to developers using Nvidia’s latest CUDA5 technology to developers writing cross-platform code with OpenACC based compilers.
  • Commercial successes of HPC. While commercial companies are sometimes reluctant to share “trade secrets” about how they are using HPC, and while previous years conferences have had lots of discussion about the “missing middle” of medium size manufacturing companies that are not able to use HPC today for various reasons, almost any electrical or mechanical product on the market today relies on HPC software during the design phase to speed time to market and improve product quality. So while “macro economic trends” may cause companies to come in under or above their financial projections in any given quarter, HPC use in manufacturing, oil & gas, media/entertainment, and other industries continues to grow and I expect to see a number of success stories discussed at the show.

    It promises to be a great week, one that I always look forward to, and I hope to see many of you in November at SC12 and HP-CAST.

  • 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.