Presentation on theme: "Contact: Hirofumi Amano at Kyushu 40 Years of HPC Services In this memorable year, the."— Presentation transcript:
Contact: Hirofumi Amano at Kyushu 40 Years of HPC Services In this memorable year, the R. I. I. T participated in “Joint Usage/Research Center for Interdisciplinary Large-Scale Information Infrastructures,” also known as JHPCN: Japan High Performance Computing and Networking plus Large-scale Data Analyzing and Information Systems, together with other supercomputer centers. Mission High-Performance Computing is now an essential tool for various fields of science. It shows us what we cannot see in a laboratory, such as turbulences in the supersonic stream around turbine blades of a jet engine, or the mantle convection inside our own planet. consists of scientists and supporting employees, but it is not just a "pure" institute. Though it is related to various aspects of information technology, its mission also includes the following IT- related services and studies: high-performance computing services; network services for the university; other IT-related campus-wide services; advanced studies for the future of these services. User Support Computer centers usually have professional staff supporting users in programming, debugging or performance tuning. However, they usually do not know much about various research fields such as materials engineering or molecular science. To bridge the gap between Computational Science and Computer Science, the R. I. I. T. has computational scientists and computer scientists working together. Such a unique collaboration is the key to their cutting edge to assist users in the fundamental part of research. Pioneering the Computing Services for Academic Studies in Japan The Research Institute for Information Technology (R. I. I. T.) of Kyushu University is founded in 2007, but its computing services have a long history of 40 years. It was formerly known as "Computer Center" or "Computing and Communications Center". Since the early days of the Computer Center, it has been one of the seven national supercomputer centers in Japan.
?? Current Systems Lineup (Photo by Courtesy of Fujitsu, Ltd.) Scalar-Parallel Supercomputer (SMP): Fujitsu PRIMEQUEST580 (Photo by Courtesy of Fujitsu, Ltd.) Peak Performance: 13.1 TFLOPS #Nodes: 32 #Cores: 2,048 Memory: 4.0 TB Node: Itanium2 1.6 GHz (dual core)×32, 128 GB Memory Interconnect: InfiniBand 4xDDR (20Gbps) ×4 / node Disk: 250 TB (data) ＋ 250TB (backup) #Users: 369 (as of Oct. 2010) To be replaced in Spring 2011 PC Cluster: Fujitsu PRIMERGY RX200S3 Peak Performance: 18.4 TFLOPS #Nodes: 192×2 sets #Cores: 1,536 Memory: 3.0 TB Node: Xeon 3.0 GHz (dual core)×2, 8 GB Memory Interconnect: InfiniBand 4xDDR (20Gbps) ×1 / node #Users: 169 (as of Oct. 2010) Scalar-Parallel Supercomputer (NUMA): Hitachi SR16000/L2 Peak Performance: TFLOPS #Nodes:42 #Cores: 1,344 Memory: 5.5 TB Node: POWER6 4.7 GHz (dual core)× GB Memory×41 nodes ＋ 256GB Memory×1 node Interconnect: InfiniBand 4xDDR (20Gbps) ×2 / node Disk: 200TB + Tape Library (112TB) #Users: 464 (as of Oct. 2010) Upgrade History and Future Plan (Fujitsu VPP5000) (IBM eServer p5) Fujitsu PRIMEQUEST + PRIMERGY (Hitachi SR11000) ? Hitachi SR16000 March 2005 May 2009 June 2007 Spring 2011 (Photo by Courtesy of Hitachi, Ltd.) Spring 2013 ? PGY (augmented) Storage Test Bed: IBM Bladecenter HS22 + XIV Storage System Intel Xeon 5570 ×2 (with 16GB Memory) × 8 blades Storage: 27TB with NAS Gateway : IBM N6040 Spring 2012 (See the separate poster for more details.)