Green500: HPC System SuperMUC Meets Unmatched Level of Energy Measurement Methodologies
PRESS RELEASE 06/2013:
LRZ Garching Adopts Pioneering Role in Data Centre Energy Efficiency
Berlin/Germany, July 9, 2013 – The Leibniz Supercomputing Centre (LRZ) pushes the envelope regarding the transparency of data centre energy efficiency. For the recently released 13th edition of the Green500 List (June 2013), the Garching-based GCS member centre submitted data which fulfilled the previously unmatched „Level 3“ criteria, a new energy measurement methodology developed in tandem with the Energy Efficient HPC Working Group, TOP500, and The Green Grid. With this move, LRZ is one of only three supercomputing centre operators world-wide who rose to the challenge to fulfil the most accurate measurement criteria for energy consumption of a data centre available to date.
Level 3 in the rules of Green500 represents the “current best” in supercomputing power measurement methodology, while Level 2 is less comprehensive thus easier to achieve. The vast majority of supercomputing centres, though, still submit Level 1-data, which is similar to the Green500 run rules to date. Due to the fact that by Level-1 standards the measurement criteria are defined more loosely this level leaves certain room for the distortion of results, making it hard to compare data of different systems. “LRZ has always been a pioneer in the field of energy efficiency and we are proud of our new power measurement capabilities for SuperMUC”, affirms Prof. Dr. Arndt Bode, Director of LRZ. “Measuring energy efficiency means measuring supercomputer performance and power consumption at the same time. While the benchmarks for measuring performance are well established and performance results in FLOPs (Floating Point Operations Per Second) are easy to compare among different machines, the methodology for measuring power consumption was missing a commonly accepted standard so far.” Energy consumption has been of ever growing concern for data centre operators as, according to statistics, up to 50 per cent of a data centre’s energy consumption is not caused by computing but by powering components such as the cooling devices.
Not only broke the Leibniz Supercomputing Centre new ground with the novel power measurement capabilities for its HPC system, with SuperMUC it also runs one of the most powerful supercomputers world-wide. SuperMUC, a System X iDataPlex from IBM, currently holds position 9 in the renowned TOP500 list of the world’s fastest supercomputers (Linpack benchmark ran on 147,456 computing cores/peak performance: 3.2 Petaflops). The other two institutions who, like LRZ, went all the way and submitted Level 3-criteria for the Green500 run systems of far less computing power and size. Supercomputer “Colosse” of Calcul Canada @ Université de Sherbrook (position # 126 on TOP500/June 2013) holds a total of 37,728 computing cores delivering a peak performance of 0.317 Petaflops, and system “Riptide” of Hawaii’s Maui High-Performance Computing Center (rank # 149) has 12,096 computing cores available and delivers a peak performance of 0.252 Petaflops.
The measurement guidelines and the definition of the requirements for Green 500 Levels 1-3 were designed under the guidance of the Energy Efficient HPC Working Group. The new methodology was tested by a small group of early adopters for the June and November 2012 Green500 list releases and subsequently refined given the feedback from the early adopters.
About LRZ and SuperMUC:
The Leibniz Supercomputing Centre in Garching near Munich is one of the three member centres of GCS. LRZ runs a System X iDataPlex from IBM code named SuperMUC, which is not only one of the fastest, but also one of the most energy efficient supercomputers in the world. With a PUE* value of 1.1, the LRZ system claims fame for being on the forefront of Green IT efforts (Greengineering) in data centres. An innovative new form of hot water cooling technology implemented by IBM takes credit for this achievement. The Intel processors and the system software running on SuperMUC offer further opportunities to save energy. Thanks to all these measures, the total energy consumption was drastically reduced to crunching 0.86 gigaflops for every watt consumed – a ratio currently unmatched by any other HPC system of comparable architecture.
*) PUE is the ratio of total amount of power used by a computer data center facility to the power delivered to computing equipment.
The Gauss Centre for Supercomputing (GCS) combines the three national supercomputing centres HLRS (High Performance Computing Center Stuttgart), JSC (Jülich Supercomputing Centre), and LRZ (Leibniz Supercomputing Centre, Garching near Munich) into Germany’s Tier-0 supercomputing institution. Concertedly, the three centres provide the largest and most powerful supercomputer infrastructure in Europe to serve a wide range of industrial and research activities in various disciplines. They also provide top-class training and education for the national as well as the European High Performance Computing (HPC) community. GCS is the German member of PRACE (Partnership for Advance Computing in Europe), an international non-profit association consisting of 25 member countries, whose representative organizations create a pan-European supercomputing infrastructure, providing access to computing and data management resources and services for large-scale scientific and engineering applications at the highest performance level.
GCS is jointly funded by the German Federal Ministry of Education and Research and the federal states of Baden-Württemberg, Bavaria and North Rhine-Westphalia.
GCS has its headquarters in Berlin/Germany.
Regina Weigand, GCS Public Relations
+49 711 685-87261
This press release as pdf file: