Directions
Facebook
Twitter
Instagram
YouTube
Japanese
Contact Us
TOP
Research
Operations and Computer Technologies Division
System Operations and Development Unit
System Operations and Development Unit
Japanese
Unit Leader Yuji Iguchi
yuji.iguchi[at]riken.jp (Lab location: Kobe)- Please change [at] to @
- 2023
- Deputy Director, Operations and Computer Technologies Division, Unit Leader, System Operations and Development Unit, RIKEN R-CCS (-present)
- 1981
- Fujitsu Limited
Keyword
- High Performance Computing
- Power Efficiency
- Job Scheduling
Research summary
The system operations and development unit has conducted operations and maintenance management, user management, and user support of super computer systems operated by R-CCS in addition to the research and development on advanced management and operations of them. We have made improvements of the job scheduler, file system, and MPI library (communications library) while analyzing the operational statistics of systems.
Main research results
Improvement of System operations
We have analyzed the execution status of jobs and optimized some parameters of the scheduler, etc. We have also resolved various problems on the system operations, and have optimized the management and operations of systems and the system maintenance. On the K computer, we improved about 20% of the system utilization by optimizing the scheduling parameters.
Improvement of System Software and Development of Tools
We have improved the system software (MPI library, compliers and etc.) and they have enabled users to improve the performance of the user’s application without any code changes. And we have also developed tools that support the use of computer systems.
User Support
We have conducted the user support, such as the user management and the consulting services.
Representative papers
- Keiji Yamamoto, Yuichi Tsujita, and Atsuya Uno.:
"Classifying Jobs and Predicting Applications in HPC systems"
ISC High Performance 2018, Lecture Notes in Computer Science, Vol. 10876, pp.81-99 (2018).
- Y. Tsujita, T. Yoshizaki, K. Yamamoto, F. Sueyasu, R. Miyazaki, and A. Uno.:
"Alleviating I/O Interference Through Workload-Aware Striping and Load-Balancing on Parallel File Systems"
LNCS 10266, pp. 315-333 (2017).
- Atsuya UNO, Ryuichi SEKIZAWA.:
"A Study of Job Scheduling Performance focusing on Compute Node Maintenance"
HPCS2016, Miyagi, Japan (2016).
- Atsuya UNO, Hajime HIDA, Fumio INOUE, Naoki IKEDA, Toshiyuki TSUKAMOTO, Fumichika SUEYASU, Satoshi MATSUSHITA, Fumiyoshi SHOJI.:
"Operation of the K computer Focusing on System Power Consumption"
IPSJ Transactions on Advanced Computing Systems Vol.8, No.4, pp.13-25 (2015).
- Fumiyoshi Shoji, Shuji Matsui, Mitsuo Okamoto, Fumichika Sueyasu, Toshiyuki Tsukamoto, Atsuya Uno, Keiji Yamamoto.:
"Long term failure analysis of 10 Petascale supercomputer"
ISC2015 HPC IN ASIA POSTER AWARD, Frankfurt, Germany (2015).
- Keiji Yamamoto, Atsuya Uno, Hitoshi Murai, Toshiyuki Tsukamoto, Fumiyoshi Shoji, Shuji Matsui, Ryuichi Sekizawa, Fumichika Sueyasu, Hiroshi Uchiyama, Mitsuo Okamoto, Nobuo Ohgushi, Katsutoshi Takashina, Daisuke Wakabayashi, Yuki Taguchi and Mitsuo Yokokawa.:
"The K computer Operations: Experiences and Statistics"
ICCS2014, Cairns, Australia (2014).
