AUTOMOTIVE
Writer
DEISA Extreme Computing Initiative awards 2008
The DEISA (Distributed European Infrastructure for Supercomputing Applications) EU funded Research Infrastructure is constituted of a number of leading national supercomputers in Europe interconnected with a high bandwidth10 Gb/s point to point network provided by GEANT and the National Research Networks. Selected middleware allows the deployment and operation of a number of services enabling high performance distributed computing DEISA launched the DEISA Extreme Computing Initiative (DECI) in early 2005. “DECI was created and supported as the right instrument to enhance DEISA’s impact on Europe’s competitiveness in science and technology”, explains Victor Alessandrini from CNRS-IDRIS, coordinator of DEISA. The DECI consists of the identification, enabling, deploying and operation of “flagship” applications in selected areas of science and technology. These leading, ground breaking applications must deal with complex, demanding, innovative simulations that would not be possible without the DEISA infrastructure, and which would benefit from the exceptional resources of the Consortium. Projects supported by DECI are chosen on the basis of innovation potential, scientific excellence and relevance criteria. Multi-national proposals are especially encouraged. A European Call for Extreme Computing Proposals, published annually in spring, has lead to an overwhelming response in 2007. After receiving around 40 to 50 proposals in 2005 and 2006, respectively, over 60 proposals from the 2007 call were competing for DEISA resources, asking for more than 70 million computing hours. From the 2005 and 2006 DECI calls over 50 projects have already benefited from the DECI. As one of the highlights, results of the DECI project POLYRES from a German/British group with Principal Investigator Kurt Kremer were published as the cover story of NATURE (May 24, 2007): For almost two decades, physicists have been on the track of membrane mediated interactions. Simulations in DEISA have now revealed that curvy membranes make proteins attractive. From the DECI call 2007, 45 projects have now been retained for operation in DEISA in 2008, with a total award of around 30 million processor-core hours. An allocation of 1 million processor-core hours could in principal be spent by using 16 processor-cores for about 62,000 hours (or 7 years), or by using 1024 processor-cores for about 1000 hours (or 40 days). Since DECI is dedicated to capability computing, using large fractions of a big supercomputer (or several supercomputers) simultaneously was mandatory, and only such capability computing projects were considered for execution in DEISA. The Applications Task Force, a European team of HPC experts led by Hermann Lederer from RZG, supports the enabling of the applications in the projects to be used within the heterogeneous DEISA infrastructure and also helps to select the most suitable architecture for each project, depending on its specific requirements. In this way, DEISA is also opening up the respective most powerful HPC architectures available in Europe for the most challenging projects, mitigating the rapid performance decay of a single national supercomputer within its short lifetime cycle of typically about 5 years, as implied by Moore’s law. The 45 projects retained for operation in 2008 cover major areas of science including Materials Science (12 projects), Astro Sciences (8 projects), Engineering (8 projects), Life Sciences (8 projects), Earth Sciences (4 projects), Plasma Physics (3 projects), and Informatics (2 projects). The projects to be supported involve scientists from 14 different European countries and collaborators from three more continents. For the first time as many non-DEISA European countries (the seven countries Austria, Poland, Portugal, Romania, Sweden, Switzerland, and Ukraine) have been involved as there are countries with DEISA site(s) (the seven countries Finland, France, Germany, Italy, The Netherlands, Spain, and UK). From outside Europe, scientists from Canada, US, Brazil, Chile and Israel collaborate. This huge continental demand for capability computing resources in Europe also clearly documents the need for a persistent European HPC ecosystem.