The Tokyo Institute of Technology announced last week that they have upgraded their latest cluster supercomputer, Tsubame, with NVIDIA Tesla GPU accelerators. Specifically, the group added 170 NVIDIA S1070 1U nodes to the cluster. Wowzers.
It took about a week to add the (Tesla S1070) units in mid-October,” said Satoshi Matsuoka, a professor at the Global Scientific Information and Computing Center of the institute. “I knew we could do it.”
If one considers the 4.1Tflops [32bit, single precision] peak performance of the S1070, Tokyo Tech just bought themselves an additional 697 Tflops. If they combine the original peak performance of Tsubame, they have a whopping 910 Tflops of single precision peak compute. The team was still able to hit 77.48 Tflops [measured] double precision for their Top500 run. Not bad considering the S1070’s are still brand spanking new. They’re hoping for more, given time.
We could achieve about 90TFLOPS even with Linpack if we adjusted the system properly, but this application is rather disadvantageous for our system,” [Matsuoka] said. “With some tweaking, even single precision arithmetic can achieve good results in most of scientific calculations. With our latest efforts to enhance performance, our system acquired computing power that is virtually on par with the top-ranking systems in the world.”
For more info on the upgrades to Tsubame, read the full article here.