Adventures in Mineral Oil: Deep-Fried Longhorns

Let’s say that you and your pals need to build a high performance cluster to run a number of HPC apps. It has to be fast. Damned fast. Fast enough to beat the very best student-built systems from seven other…

Read More

Supersized Science

We spent a few minutes at the Oak Ridge SC11 booth talking with Jack Wells who is, not coincidentally, with Oak Ridge.  He’s looking forward to the newest addition to the Oak Ridge supercomputer family, the Titan. This new system…

Read More

Biggest Computer Upgrade – Ever

Peg Williams, Cray SVP of High Performance Computing, is working on the biggest computer upgrade in history – transforming the 200-cabinet, 42,000-processor Oak Ridge Jaguar supercomputer into its new Titan form. When it’s finished, Titan will have 21,000 16-core AMD…

Read More

SCC Final: Taiwan Repeats; the Rise of GPUs

The final results of the 2011 Student Cluster Competition are in the books and can be revealed. Well, they should have been revealed last night, but a series of meetings, bad hotel Wi-Fi, and perhaps alcohol played a role in the delay. In upcoming articles, you’ll see more videos and analysis of the results. But for now, I’ll sketch out the big picture.

In the closest imaginable finish, Team Taiwan from National Tsing Hua University took the overall student clustering crown. They did it on a hybrid Acer system consisting of 72 Xeon CPU nodes and six NVIDIA Tesla C2070 GPU accelerators. On the memory side they had 48 GB/node, a bit low compared to other teams.

Just barely behind Team Taiwan was China’s National University of Defense Technology (NUDT). The SC committee hasn’t (ever) released actual figures, but I was skulking around while they were being compiled and can say that the Taiwan-China scores were so close that literally hours of checking, re-checking, and then checking once again were required to make sure the results were correct.

Team China also ran a hybrid CPU-GPU system. But the Chinese system, while sporting the same six NVIDIA C2070 Tesla accelerators as Team Taiwan, had only 24 Xeon cores. China had more memory per node at 96 GB each, but less overall memory at 192 GB total vs. 288 for Taiwan. (Read more below…)

Read More

SCC – The Final Hours: Texas

The Texas team captured a lot of attention at the Student Cluster Competition this year. Their mineral oil deep-fried cluster was certainly a hit with the crowd, but did it pay computing dividends? They put in the best non-GPU fueled…

Read More

SCC – The Final Hours: Taiwan

We had a chance to talk to Team Taiwan as the Student Cluster Competition came down to the wire. Although they were their typical friendly selves, there was an air of tension and anticipation in their booth. They have a…

Read More

SCC – The Final Hours: Russia

We dropped by the Team Russia booth to congratulate them on their record-setting LINPACK victory and to get their thoughts on SCC11. Is their LINPACK win a sign of things to come? Can they take the whole ball of wax?…

Read More

SCC – The Final Hours: Purdue

As can be seen by their LINPACK results on Monday (only .233 TFlop), Purdue either completely melted down or had something up their collective sleeves. It turns out that they had a plan – a cunning plan. One that might…

Read More

SCC – The Final Hours: Costa Rica

The Costa Rica team gave it their all at their first Student Cluster Competition. We caught up with them just a few minutes before they turned in their final results files to the judges…

Read More

SCC – The Final Hours: Colorado

We grabbed Team Colorado for a few final thoughts before the end of the Student Cluster Competition. Spirits were good, despite a few problems with their hardware and memory usage.

Read More