SCC results: Taiwan wins overall crown
Texas, LSU, Taiwan top Teraflop
SC10 The Student Cluster Competition high performance computing (HPC) marathon of 2010 is over.
The final results were turned in Wednesday evening, the data examined by expert judges, and winners were announced at the SC10 awards luncheon on Thursday afternoon. Unfortunately, I had to travel to another meeting and couldn’t attend (or film) the ceremony – which leaves quite a hole in my wall-to-wall coverage.
I’m sorry I missed it. These kids worked hard, and I would have really enjoyed seeing how it all ended and seeing the winning teams get their proper recognition. Next year, I’ll definitely be there. On to the results…
First, the highest LINPACK score was turned in by the University of Texas, Austin. They were the first team in SCC history to achieve teraflop performance with their 1.07 LINPACK result. While they’re the first team in the SCC Teraflop Club, they aren’t alone. Louisiana State and National Tsing Hua University (NTHU) from Taiwan also made the TFlop milestone.
The University of Colorado (Boulder) was the recipient of the new “Fan Favorite” award based on the quality of their interviews, their ability and willingness to explain what they’re doing, and general all-around friendliness and neighborliness. This is well-deserved in my mind. Colorado really showed the spirit of the SCC by giving equipment to FAMU when their cluster failed to make it to New Orleans.
The overall winner? National Tsing Hua University for their overall performance on the HPCC benchmark, four real-world application challenges, and their interviews. The NTHU story is compelling. While the school has competed in two previous SCC challenges, this is the first time for these individual team members. They traveled farther than any other team and had to wrestle with language and cultural barriers as well.
They also had to contend with a vendor change in mid-stream; Acer, Tatung and the NCHC stepped in a month before the competition to fill a hole left by a previous vendor. So they didn’t have much time to learn the ins and outs of their cluster hardware. It didn’t seem to make much difference as they cruised to a close, but clear, victory. (The SCC committee doesn’t reveal detailed results, or I’d share them with you.)
So what did we learn from the competition this year? To me, the biggest surprise was that GPUs didn’t seem to make much difference. I had anticipated that the addition of GPUs would have a much bigger impact on the competition, but it didn’t turn out that way. Both of the teams sporting GPUs had them working, but I’m not sure if the code they were running was truly optimized. It will be interesting to see what role, if any, GPUs play in next year’s competition. The GPU ecosystem will have advanced significantly by then, and fully optimized code should be more readily available.
Also I learned how gratifying it was to observe the SCC. I was highly impressed by the kids, the SCC organizers, and the sponsors. I’d like to see more sponsors becoming more deeply involved next year. So far, sponsors are mainly tech vendors, but this isn’t a rule – any type of organization can back a student team.
Sponsoring a team pays off in a number of ways. First, there’s the obvious advertising impact from having your name associated with a team (think NASCAR). But there’s also the goodwill that comes from helping out – which can boost recruiting and/or sales down the road.
Last, but not least, it’s a lot of fun to watch the kids learning and getting the chance to show the world what they know. It’s a very cool thing.
Sponsored: Hyper-scale data management