NVIDIA CUDA Center of Excellence (CUDA COE) at Tokyo Institute of Technology is a collaborative effort to educate students and scientists on GPU-based accelerated computing using the NVIDIA CUDA technology and to explore the future GPU computing direction by researching on large-scale GPU-accelerated HPC system software and applications. Both the education and research are conducted on the TSUBAME supercomputer at the Global Scientific Computing and Information Center (GSIC), one of the national major computing centers.

TSUBAME is our current generation supercomputer with 680 NVIDIA Tesla GPUs (170 Tesla S1070 boards) hosted by more than 600 Opteron-based nodes, which is currently ranked at the 41st in the June '09 TOP500 list. The GSIC is currently undertaking to investigate and review the proposals for the next generation TSUBAME 2.0, which is expected to be delivered in 2010. The TSUBAME 2.0 will further enhance GPU-based HPC acceleration with much improved intra-node and inter-node communication bandwidths. The CUDA COE at Tokyo Tech will continue to play a key role in research and education with the next generation of our TSUBAME supercomputer.

The GSIC has been providing students and researchers at Tokyo Tech with not only computing infrastructure but also educational services to utilize such advanced resources. One of the recent major focuses is GPU computing, attracting many participants to our tutorial series, which motivated us to organize the GPU Computing Study Group, a GSIC organization dedicated to education and knowledge sharing on GPU computing. It is open to anyone interested in GPU computing, including those who are not affiliated with Tokyo Tech, and currently consists of more than three hundreds of members from a wide variety of academic and industrial fields. More information on future tutorials and talks can be found at its web page.