Notes from the Expo Floor - Home
Industry and Academia at Supercomputing
Brian Kirk
DEC 13, 2012 13:38 PM
A+ A A-

 

Supercomputing 2012 was filled with great content, a huge expo floor, and a ridiculous street-closing party complete with motorcycles, ramps, and a buffet. It was eclectic to say the least.

 

One of the more interesting aspects of the conference was interaction between different types of organizations. I have written previously about the need for supercomputing to be recognized as an enabling technology for disparate areas of study, and looking at the fields of research impacted by the vendors on the show floor, it is very apparent that many areas of science and research are reaping the benefits of the ever-growing HPC market space.

 

The conference’s annual student cluster challenge, in its seventh iteration, is exactly what the future needs for innovation in the HPC realm. As per the rules (found here), student teams must team up with a hardware vendor prior to the competition. This year’s winning team, out of the University of Texas, partnered with Dell, Intel, Nvidia, and others, to create a computationally strong cluster.

 

Creating a high-powered computing cluster is in and of itself a challenging task. Having it stay within the power constraints is a much different problem. Working closely with the vendors is a key step in the competition, and one that can provide the students with a much needed—and rare—exposure to dealing with potential post-school interactions.

 

Along those lines, I had a chance to talk to the folks at Silicon Mechanics, one of the vendor sponsors for the Massachusetts Green High-Performance Computing Consortium Team (MGHPCC), and it was personally and intellectually refreshing to hear about their partnership and some of the things that have come out of their relationship (Go here to read IEEE Spectrum's story on MGHPCC). Specifically, I spoke with Art Mann and David Manning from Silicon Mechanics and Dan Kamalic, Manager of Research Computing with Boston University about how both sides of the partnership have benefitted from the arrangement. What would seem to be a one-sided partnership (the company gets a lot of PR from this, right?) actually is a bit more nuanced.

 

Computer Science graduates (to some) may seem like code monkeys, slaving away in dark basements at institutes and IT departments, but as many of us know, that’s not the case (all the time). Dealing with constricting budgets, ever-changing tech, and pressing deadlines, computing professionals must balance technical know-how with a touch of business acumen. A typical student’s time in college is filled with lectures, labs, and homework. That model has been around for a very long time, and while it has its benefits, the winds of change are beginning to creep in (see the aforementioned bit about constricting budgets et al.). Having the students from the MGHPCC work with vendors on needs, constraints, and problems, provides them with invaluable experience prior to entering the workforce. These aren’t typical scenarios for college students, and from the conversations that I had with some of the competitors from the MGHPCC team, they understand the importance of working with vendors and seem grateful for the increased exposure to real-world problems prior to graduation.

 

Along those lines, I had a chance to stop in and talk with Mark Bookout from Missouri S&T, along with some of his students. When I first showed up at their booth, they demonstrated a 3D visualization of earthquake data, using a Wiimote as the means of manipulating and rotating the data. Looking at the data, one could see some anomalous aspects to the images, and they talked about how that particular visualization had thrown into question some of the assumptions of the traditional views of the nature of Earth’s crust. After a tangential rant about how awesome that was (where I probably talked way too much and way too excitedly), we started talking about how they handled their CS students.

 

Mark explained that their undergrad CS students often acted as consultants through their school’s Research Support Services department. More than IT staff, they assist other students and faculty in their research areas, utilizing visualization, data collection and analysis, and high-powered computing resources for researchers who aren’t necessarily equipped to handle the computational aspects of their research. The students explained that it has opened up their eyes to the research around them, and it has provided them a chance to problem-solve on a level that many other students (even graduate students) might not get at other universities.

 

Their approach is typical of the types of innovative partnerships that I witnessed throughout the entirety of the Supercomputing conference, as supercomputing isn’t necessarily an end unto itself; it is more of an enabling technology. This technology can assist in all manners of research, but you have to have the expertise of a knowledgeable problem-solver. Limited by budget constraints, these new generations of students are coming out of college with real-world experience at managing resources and looking at problems in new ways. More than anything, it suggests that continued investment in education is a must for many areas of HPC research and development, if not simply for a vibrant test-bed of students ready and willing to push the boundaries of a traditional education (and break existing systems).

 

While the future of supercomputing may seem cloudy (pardon the pun), the students who are starting out in these fields will dictate the direction. For that reason alone, we should pay attention to what they are doing (and help out in any way we can). 

 

FIRST
PREV
NEXT
LAST
Page(s):
[%= name %]
[%= createDate %]
[%= comment %]
Share this:
Please login to enter a comment:
 
RESET