Organization
Tokyo Institute of Technology
Presentations
Panel
TP
Panel
W
TUT
TP
EX
EXH
Workshop
Education
Scientific Computing
Training
Scalable and Sustainable Approaches for HPC Training and Education
Emerging Tools enabling HPC Training and Education
Lightning Talks and Demos
W
Paper
GPUs
Memory
NVRAM
Performance
System Software
Tools
TP
Birds of a Feather
Machine Learning
TP
EX
EXH
Workshop
Architectures
Heterogeneous Systems
Quantum Computing
W
Paper
Networks
Resource Management
Scheduling
State of the Practice
System Software
TP
BSP Finalist
Panel
TP
Booth
427
Exhibitor Description
Tokyo Tech has started the operation of brand-new TSUBAME3.0 supercomputer, which got World No. 1 in the Green500 in June 2017. It provides the peak performance of 12PFlops with 2,160 NVIDIA Tesla P100 GPUs. Not only as a "post-peta" supercomputer, it has been designed as a "big-data" supercomputer with 16PB shared storage and high-speed NVMe Flash SSDs, whose total performance exceeds 1PB in capacity and 1TB/s in access speed. We will demonstrate a TSUBAME3.0 compute node, in addition to R&D efforts including big-data architecture and software, ultra-green cloud system operation, advanced cooling and post-petascale applications. Additionally in this year, we will share our experience of VDI system connected to TSUBAME3.0.
Products / Services
Cloud, Heterogeneous Computing, Storage, Supercomputing
Website