Notes from the Expo Floor - Home
SuperComputing 2012 (SC12)
Brian Kirk
NOV 29, 2012 09:38 AM
A+ A A-


After a snow-filled week in Salt Lake City for SC ’12, the IEEE Computer Society and ACM co-sponsored conference on supercomputing and high-powered computing, I tried to relax in my airplane seat. When striking up the usual small-talk that accompanies flying coach, I found out that the gentleman sitting in the aisle seat was leaving the conference as well. The man in the middle? Not a techie, and he posed an interesting question to us both: “What exactly is supercomputing?”
I won’t bore you with the answer that we gave, but I will say that the question is an important one. Most of the conferences that I find myself at are focused on some narrow aspect of computing, be it mobility, graphics, or software engineering, but I was very excited to be going to my first Supercomputing conference. I like the idea of having a place where HPC, networking, data storage, and education all mix together. Most of my excitement came from curiosity about what kinds of stuff I’d find in the exhibit hall. Would there be flashy booths? Huge HPC racks? Cutting edge applications? 
The thing that was immediately apparent from my first lap of the expo floor was the size. SC is a large, large conference. Not quite the visual spectacle of CES or even Siggraph, what SC lacks in flash it more than makes up with sheer square footage of space (granted it’s not as big as CES, but still). This year’s conference boasted 334 exhibits spread out over the 125,000 square feet of exhibition space. And even with that much space, the approximately 10,000 people in attendance would occasionally cause a crowd-spike in the large lanes around the booths. Walking shoes are a good idea if you’d like to take more than one lap around the floor. 
There were a few rather large installations on the show floor (including a big rig parked in the middle, sporting a bunch of high-tech toys), and all of the usual big names were there. Those are interesting sometimes, but in my opinion, the exciting content at these sorts of events come from some of the smaller booths, booths that represent a larger proportion of operating budgets. Also, it was great to see academia sporting a fairly large presence on the floor, some of them even having displays on par with some of the big-budget mega-exhibitors.
XSEDE had a good presence on the floor, and I usually get excited about their approach to enabling researchers’ access to high-powered computing and analysis tools. We chatted mostly about their upcoming conference in San Diego, but it’s good to see how they act as an intermediary in high-level research. On those same lines, I talked with some of the people at Symmetric Computing, and their approach to HPC is pretty interesting. By using off-the-shelf server hardware, they’re able to offer people a more affordable supercomputing option. Richard Anderson, the CTO of Symmetric, hesitantly called it “supercomputing for the masses.” A distinction should be made here, as the term “masses” isn’t meant to be pejorative; instead, it means that the majority of supercomputing-use can be accomplished in a much simpler, streamlined system. Their contention is that 90% of supercomputing use can be accomplished in a much more economically-responsible fashion.
What XSEDE and Symmetric have in common is that they both work on efficiently helping scientific and other computing-intensive work get done. With all of the hoopla around the show floor, that was a theme that kept emerging as I talked to exhibitors. Supercomputing itself may be a nebulous term for people outside of the industry, but it’s an incredibly important field because of what it enables. Bioinformatics, data visualizations, and earth sciences all benefit from the power of these high-powered machines, and because many of the researchers in those fields aren’t necessarily computing experts, ease-of-use is extremely important. 
This year’s keynote address for SC’12 was from Michio Kaku, billed on the official SC’12 site as a “popularizer of science,” and while I enjoyed his presentation, I did hear from several attendees that it was a little too “fluffy.” I understand the hesitation, as I would’ve appreciated a tad more citations (or perhaps a handout with an annotated bibliography), supporting Dr. Kaku’s claims. Not because I don’t believe him per se (I’ve actually seen some of the research underpinning a few of the claims about the future), but because the research that is done is fundamental to the conclusions drawn. Supercomputing brings industry and academics together, and if Dr. Kaku is a popularizer of science, maybe he could be a popularizer of scientific research, too. Maybe then the question of “what is supercomputing?” can turn to “what else is supercomputing doing?”
I’m sure the answer to that question will be in Denver in 2013.
[%= name %]
[%= createDate %]
[%= comment %]
Share this:
Please login to enter a comment: