What is PRACE? Hank Nussbacher PRACE Winter School, Tel Aviv, Feb 10, 2014 1
Realizing the ESFRI Vision for an HPC RI European HPC-facilities at the top of an HPC provisioning pyramid Tier-0: European Centres for Petaflop/s Tier-1: National Centres Tier-2: Regional/University Centres Creation of a European HPC ecosystem HPC service providers on all tiers Scientific and industrial user communities The European HPC hard- and software industry Other e-infrastructures Tier-0 European centres Tier-1 National centres Tier-2 Regional/University centres # of systems capability 2
PRACE-3IP: History HPC part of the ESFRI Roadmap; creation of a vision involving 15 European countries Creation of the Scientific Case Signature of the MoU Creation of the PRACE Research Infrastructure HPCEUR HET PRACE Initiative PRACE RI 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 PRACE Preparatory PRACE-1IP PRACE-3IP Phase Project PRACE-2IP 3
PRACE in Winter 2014 has grown up PRACE is the European High-End HPC Infrastructure provides access to 6 Tier-0 systems, funded by 4 countries: DE, ES, FR, IT has provided 5.5 Billion Core-hours to hundreds of user projects since 2010 is providing extensive training though 6 PRACE Advanced Training Centres (PATC), seasonal schools, summer student programme, workshops, currently has 25 members, representing 25 countries has secured national contributions of 450 M and 67 M EC funding is developing its strategy for 2015-2020 PRACE is user-demand driven Scientific Steering Committee advises on all scientific matters, in particular the access to PRACE resources through Peer Review User Forum provides feedback and technical requirements from end-users Scientific Case 2012-2020: requirements and recommendations >100 leading European Scientists contributed to this update 4
April, 23 rd 2010 creation of the legal entity (AISBL) PRACE with seat location in Brussels, Belgium 25 PRACE Members 67+ Million from EC FP7 for preparatory and implementation phases Grants INFSO-RI-211528, 261557, 283493, and 312763 Complemented by ~ 50 Million from PRACE members Interest by: Latvia, Belgium 5
Tier-0 Tier-1 Tier-2 PRACE is building the top of the pyramid... Sixth production system available by January 2013: 1 Petaflop/s IBM (MareNostrum) at BSC. - #34 Upgrade: 5.87 Petaflop/s IBM Blue Gene/Q (JUQUEEN) #8 in Top500 First production system available: 1 Petaflop/s IBM BlueGene/P (JUGENE) at GCS (Gauss Centre for Supercomputing) partner FZJ (Forschungszentrum Jülich) Fifth production system available by August 2011: 2 Petaflop/s IIBM BG/Q (FERMI) at CINEC - #15. Second production system available: Bull Bullx CURIE at GENCI partner CEA. Full capacity of 1.8 Petaflop/s reached by late 2011. - #20 in Top500 Fourth production system available by mid 2012: 3.1 Petaflop/s IBM (SuperMUC) at GCS partner LRZ (Leibniz-Rechenzentrum). - #10 in Top500 Third production system available by the end of 2011: 1 Petaflop/s Cray (HERMIT) at GCS partner HLRS (High Performance Computing Center Stuttgart). - #39
The HPC European e-infrastructure (ESFRI) 25 members, AISBL since 2010 530 M for 2010-2015 (inc 70M from UE) PRACE 1.0 6 supercomputers in 4 hosting countries, different architectures research and industrial access (open R&D) for all disciplines based on excellence in science, free of charge Nearly 15 Pflop/s 5 billion hours granted since 2010 7
Name of Principal Investigator Company/ Institution of Principal Investigator Type of Access Total number of CPU core-hours allocated Project System Avishai Dekel HUJI 2nd Regular 9000000 PRACE CURIE, GENCI/CEA Daniel Ceverino HUJI 2nd Regular 5000000 PRACE CURIE, GENCI/CEA Yehuda Hoffman HUJI 5th Regular 26000000 PRACE SuperMUC (GAUSS@LRZ, Germany) Avi Cohen Bar-Ilan University Preparatory 10th 50000 PRACE Hermit, GAUSS@HLRS, Yakov Strelniker Bar-Ilan University Preparatory 11th 50000 PRACE HERMIT, GAUSS@HLRS, Germany Alexander Gelfgat Tel-Aviv University Preparatory 12th 50000 PRACE FERMI Helena Vitoshkin Tel-Aviv University Preparatory 12th 50000 PRACE HERMIT, GAUSS@HLRS, Germany Preparatory 14th 40000 PRACE Dan Mordehai Technion Pinchas Alpert Tel Aviv University 450000 Linksceeem Alexander Gelfgat Tel Aviv University 160000 Linksceeem Dan Thomas Major Bar-Ilan University 400000 Linksceeem Benjamin Svetitsky Tel Aviv University 250000 Linksceeem Yakov Strelniker Bar-Ilan University 20000 Linksceeem Hanoch Senderowitz Bar-Ilan Uneiversity 20000 Linksceem-2 Alexander Gelfgat Tel Aviv University 150000 Linksceem-2 Lucio Frydman Weizmann 3000 Linksceem-2 Avi Cohen Bar-Ilan University 10000 Linksceem-2 HERMIT,@ GAUSS-HLRS, JUQUIN@GAUSS- FZJ; SuperMUC@ GASUSS-LRZ; CURIE@GENCI, FERMI@CINECA 8
PRACE 2.0: meeting Europe s ambitions with HPC Provide an infrastructure for science and industry To maintain Europe as an agenda setting science contributor By offering access to leading edge HPC platforms opened to all disciplines and countries in Europe Attract, train and retain competences To attract, train and retain highly skilled and innovative workforce in science and engineering To share knowledge and expertise Provide an high quality service With at least one supercomputer in each major architectural class To support world-leading science Lead the integration of an highly effective HPC ecosystem Including : A) scientific and industrial communities, B) national HPC centres and their support for the PRACE systems, C) training and software development efforts 9
Ambitions Provide an infrastructure for science and industry Attract, train and retain competences Provide an high quality service Lead the integration of an highly effective HPC ecosystem PRACE 2.0 strategy : Hypothesis for the baseline scenario (825 M ) (3/5) Scenario 825 M including 275 M from the EC HPC at the agenda of the Competitiveness Council 29-30 May, 2013 alone or in a consortium 5 Tier-0 of each 50 PF for 150 M (TCO 5y) Tier-0 European centers Tier-1 services for Tier- 0 50 % PRACE membres EC funds: 25% of the contribution over 2 phases 50 % EC 725 M 100 M Unanimous approval 10
PRACE in Horizon 2020 PRACE-3IP ends in mid 2014 EC has informally indicated its intention to continue IP-type support for PRACE - independent of the on-going PRACE 2.0 discussion EC needs to act on Conclusions from the Competitiveness Council Funding level and instruments to be explored 11