First NVIDIA Tesla-Based Supercomputer in TOP500

The Tokyo Institute of Technology announced today that it would use NVIDIA Tesla GPUs to leverage the computational horsepower of its TSUBAME supercomputer. According to Tokyo Tech, the TSUBAME supercomputer would be able to peak 170 TFLOPS of theoretical peak performance with the addition of 170 Tesla S1070 1U systems. At the same time, the supercomputer delivers 77.48 TFLOPS of measured Linpack performance, which places it among the best ranks in the world’s Top500 Supercomputers, namely on the 29th place.

“Tokyo Tech is constantly investigating future computing platforms and it had become clear to us that to make the next major leap in performance, TSUBAME had to adopt GPU computing technologies,” said Satoshi Matsuoka, division director of the Global Scientific Information and Computing Center at Tokyo Tech. “In testing our key applications, the Tesla GPUs delivered speed-ups that we had never seen before, sometimes even orders of magnitude - a tremendous competitive boost for our scientists and engineers in reducing their time to solution.”

Matsuoka also talked about the ease of implementation, stating that “The entire upgrade was carried out in 1 week, and the TSUBAME supercomputer remained live throughout. This is an unprecedented feat in top-level supercomputing”.

“We are honored to partner with Tokyo Tech - world famous for their supercomputing expertise and success,” said Andy Keane, general manager of the GPU Computing business at NVIDIA. “NVIDIA Tesla breaking into the Top 500 marks a milestone in supercomputing history. The massively parallel GPU is now essential for supercomputing centers worldwide.”

Tokyo Tech is the first to achieve Top500 ranking with an NVIDIA Tesla based GPU cluster, while also being one of the hundreds of distinguished universities and supercomputing centers to consider adopting GPU based solutions for their research. The list of similar centers includes the National Center of Supercomputing Applications (NCSA) at the University of Illinois, Rice University, University of Heidelberg, University of Maryland, Max Planck Institute and the University of North Carolina.

The Tesla S1070 1U GPU system is built around the NVIDIA CUDA parallel architecture. Developers and researchers are able to benefit from the parallel architecture of the GPU through an industry standard C language programming environment.