GCS @ ISC'13 in Leipzig
This year, the international HPC community came together in Leipzig/Germany to participate in the 2013 International Supercomputing Conference (ISC’13) taking place for the first time at the Leipzig Congress Center. The Gauss Centre for Supercomputing participated in the exhibition yet again with a 64m2 booth on which its three member centres HLRS (High Performance Computing Center Stuttgart), LRZ (Leibniz Supercomputing Centre) and JSC (Jülich Supercomputing Centre) showcased their latest research projects and presented sophisticated HPC tools. Countless like-minded HPC users, researchers, technology leaders, scientists, IT-decision makers as well as high tech media representatives stopped by to meet and talk with the directors of the three GCS centres Prof. Bode (LRZ), Prof. Lippert (JSC), the newly elected GCS Chairman of the Board, Prof. Resch of HLRS, and GCS Managing Director Dr. Mu?ller, as well as with many other scientists of the three GCS centres.
GCS Panel Session: GCS to Establish HPC Centre of Excellence
Representatives of the three GCS centres hosted and participated in a number of workshops, Birds-of-a-feather sessions, tutorials and other ISC events. Well received was the GCS panel session “GCS to Establish HPC Centre of Excellence” in which the GCS Managing Director and the directors of the three GCS member centres unveiled their plans and strategies about the future of GCS to a very attentive audience.
Two GCS HPC systems amongst Top Ten of TOP500
The newly released TOP500 overview listing the most powerful supercomputers in the world once again delivered proof for Germany being a world-wide key player in HPC. GCS features the two most powerful supercomputers in all of Europe with JSC’s HPC system JUQUEEN and LRZ’s SuperMUC capturing positions in the Top Ten of said list. The newly coronated Number One is TH-IVB-FEP Cluster “Tianhe-2” (peak performance: 55 Petaflops) which is to be deployed at the National Supercomputer Center in Guangzho, China.
The 41th edition of the TOP500 revealed the following rankings for the three HPC systems hosted by the GCS member centres:
TOP500 Ranking (06/2013) - GCS Systems
• #7: JUQUEEN of JSC (architecture: IBM Blue Gene/Q system / peak performance: 5.9 Petaflops),
• #9: SuperMUC of LRZ (IBM iDataplex system / 3.2 Petaflops),
• #32: Hermit of HLRS (Cray XE6 system / 1.04 Petaflops).
Hermit furthermore holds second place on the list of most powerful supercomputers world wide used for industrial science and research activities (TOP500, sub-list Industry).
GCS Booth Highlights
On the GCS booth, the three supercomputing centres presented their wide-ranging HPC activities. HLRS concentrated on interactive 3D visualizations of simulation results on a tiled 3D-display wall. Its highlight was the visualization of the deformations in crystal structures on a high-resolution tiled 3D-display.
JSC exhibited a wide spectrum of scientific results obtained with its supercomputers JUGENE, JUQUEEN and JUROPA.
In particular, JSC showcased LLview, the comprehensive interactive monitoring software for supercomputers developed in-house, in live demonstrations on supercomputers worldwide, while LRZ focused on highlighting their HPC system SuperMUC and presented current science projects and research activities in 2D-videos.
Lots of attention enjoyed the Augmented Reality demo of science project CoolEmAll which aims at optimizing the energy efficiency of data centres. By moving a camera around the server, visitors could observe the temperature distribution and airflow right within a high density Christmann RECS server. They were able to see immediately where hot spots appear in this operating condition and how these could be avoided through either geometry modification or adapted scheduling of workload, and they realized how this technology could be used to optimize their servers and data centres.
ISC’13 Gauss Award Winner
Each year at ISC, GCS presents the Gauss Award to recognize the most outstanding paper in the field of scalable supercomputing from all papers accepted for the ISC’13 Research Paper Sessions. This year, the award honoured the paper titled "TUE, a new energy-efficiency metric applied at ORNL's Jaguar".
Dr. Michael K. Patterson, Principal Engineer in the Intel Architecture Group of Intel Corp., accepted the 2013 Gauss Award in the name of the Energy Efficient High Performance Computing Working Group (EE HPC WG), which was conceived and is led by Lawrence Berkeley National Laboratory to promote energy-efficient green computing best practices.
GCS provided the platform for a team of scientists to win another ISC’13 award. The vast computing power of GCS system SuperMUC, installed at LRZ Garching, was used by a team of scientists from the LRZ, the Technische Universität München (TUM) and other institutions to simulate the behaviour of more than four trillion particles, which is a four-fold increase over the previous record.
The related paper "591TFLOPS Multi-Trillion Particles Simulation on SuperMUC", submitted to the ISC Research Paper Sessions, was honoured with this years ISC award presented by PRACE (Partnership for Advanced Computing in Europe), which GCS is a hosting member of.
(Full article in inSiDe Spring 2013, Vol. 11, No. 1)