The PRACE e-Infrastructure: European High Performance Computing for Science

Gheller, Claudio

In the past years the European Union has identified High Performance Computing as a strategic priority for Europe. The size and the cost of the related research infrastructures is steadily increasing insomuch as, today, the most advanced computing facilities can be conceived only on trans-national or even continental basis, with a joint contribution of a number of different member states. A first effort in this direction was represented by DEISA, the Distributed European Infrastructure for Supercomputing Appllications (http://www.deisa.eu/). Today, the European path toward petascale and, in a further perspective, exascale computing is represented by the PRACE project (Partnership for Advanced Computing in Europe, http://www.prace-project.eu/). PRACE's goal is to provide Europe with world-class systems for high-end science and to strengthen European scientific and industrial competitiveness. PRACE will maintain a pan-European HPC service consisting of up to six top of the line systems (Tier-0) well integrated into the European HPC ecosystem. Top class scientific projects, based on numerical applications, will be supported by periodic calls. In the first call (june 2010), ten research projects have been awarded access to the PRACE infrastructure. In total 321.4 Million compute core hours were granted.

We will provide an overview of the roadmap and the main features of the PRACE e-infrastructure. In particular, we will highlight the perspectives for science and for a new generation of large scale research cases. As an example, we will focus on cosmology and its challenges, providing an example of state-of-the-art numerical application, the computational requirements and the solutions adopted for the huge data volume produced, which led to the realization of one of the largest simulated galaxy cluster public archives: the IRA-CINECA archive (http://data.cineca.it). Finally, the possible developments (and issues) on petascale computing systems will be presented.

Return to poster list