SEVENTH FRAMEWORK PROGRAMME Research Infrastructures

Transcription

1 SEVENTH FRAMEWORK PROGRAMME Research Infrastructures INFRA Preparatory phase for 'Computer and Data Treatment' research infrastructures in the 2006 ESFRI Roadmap PRACE Partnership for Advanced Computing in Europe Grant Agreement Number: RI D3.1.7 Final Version: 1.0 Author(s): Anni Jakobsson, CSC, Naomi Messing-Klopstra, NCF Date:

2 Project and Deliverable Information Sheet PRACE Project Project Ref. : RI Project Title: Partnership for Advanced Computing in Europe Project Web Site: Deliverable ID: < D3.1.77> Deliverable Nature: <DOC_TYPE: Other> Deliverable Level: Contractual Date of Delivery: PU* 31 / October / 2009 Actual Date of Delivery: 30 / October / 2009 EC Project Officer: Maria Ramalho-Natario * - The dissemination level are indicated as follows: PU Public, PP Restricted to other participants (including the Commission Services), RE Restricted to a group specified by the consortium (including the Commission Services). CO Confidential, only for members of the consortium (including the Commission Services). Document Control Sheet Document Authorship Title: ID: Version: 1.0 <> <D3.1.7> Available at: Status: Final Software Tool: Microsoft Office Word 2007 File(s): Written by: Contributors: Reviewed by: Approved by: D3.1.7.doc Anni Jakobsson, CSC Naomi Messing, NCF Christos Kanellopoulos, GRNET; Florian Berberich, FZJ Technical Board PRACE - RI i

3 Document Status Sheet Version Date Status Comments /May/2009 Draft /September/2009 Version 0.2 Additions by Naomi Messing-Klopstra /September/2009 Version 0.3 Anni Jakobsson /October/2009 Final version PRACE - RI ii

4 Document Keywords and Abstract Keywords: PRACE, DEISA, HPC, Research Infrastructure Abstract: PRACE, the Partnership for Advanced Computing in Europe, and DEISA, the Distributed European Infrastructure for Supercomputing Applications, merged for the first time their annual science symposia into one big European HPC (High Performance Computing) event: The DEISA PRACE Symposium This symposium took place from May 11 to 13 in Amsterdam, and was hosted by SARA and NCF at the Royal Tropical Institute. The theme of the symposium was "HPC Infrastructures for Petascale Applications", it was of major interest to a broad audience: from scientific users, HPC technology experts and vendors to government representatives and industry partners. The symposium attracted almost 200 participants from more than 20 countries and four continents. Copyright notices PRACE Consortium Partners. All rights reserved. This document is a project document of the PRACE project. All contents are reserved by default and may not be disclosed to third parties without the written consent of the PRACE partners, except as mandated by the European Commission contract RI for reviewing and dissemination purposes. All trademarks and other rights on third party products mentioned in this document are acknowledged as own by the respective holders. PRACE - RI iii

5 Table of Contents Project and Deliverable Information Sheet... i Document Control Sheet... i Document Status Sheet...ii Document Keywords and Abstract...iii Table of Contents... iv List of Figures... v List of Tables... v References and Applicable Documents... v List of Acronyms and Abbreviations... vi Executive Summary Introduction Motivation and Venue Organising Committees Programme Committee Organising Committee Dissemination Invitations Press Releases Websites Flyer Symposium Programme Speakers Biographies and Abstracts Other material Exhibition at the Symposium Conference Evening Attendees Relations to Stakeholders Post-event Activities Feedback Conclusions Annex List of Attendees Evaluation form Speakers Biographies and Abstracts PRACE - RI iv

6 List of Figures Figure 1: The Royal Tropical Institute of Amsterdam... 3 Figure 2: DEISA PRACE Symposium 2009 web... 6 Figure 3: DEISA PRACE Symposium 2009 website... 7 Figure 4: DEISA PRACE Symposium flyer...8 Figure 5: DEISA PRACE Symposium flyer...8 Figure 6: DEISA PRACE Symposium 2009 audience Figure 7: Kostas Glinos, Head of Unit, Géant and e-infrastructures of the European Commission Figure 8: Maria Ramalho Natário from the European Commission Figure 9: Abani Patra from the NSF, USA Figure 10: Horst Simon from Lawrence Berkeley National Laboratory, USA Figure 11: Vladimir Voevodin from the Moscow State University Figure 12: DEISA PRACE Symposium banner Figure 13: DEISA PRACE Symposium name badges Figure 14: Exhibition table at the DEISA PRACE Symposium Figure 15: Posters at the DEISA PRACE Symposium Figure 16: Conference participants embarking the boats List of Tables Table 1: DEISA PRACE Symposium participants Table 2: Feedback form: Relevance, Agenda and Discussions Table 3: Feedback form: Organisation References and Applicable Documents [1] [2] [3] [4] [5] PRACE - RI v

7 List of Acronyms and Abbreviations ANU-SF Belief CNRS CSC CSCS DEISA The Australian National University Supercomputing Facility Bringing Europe s electronic Infrastructures to Expanding Frontiers French National Centre for Scientific Research IT Center for Science, Finland Swiss National Supercomputing Centre Distributed European Infrastructure for Supercomputing Applications. EU project by leading national HPC centres. DoE, NERSC Department of Energy, National Energy Research Scientific Computing Center, USA EGI European Grid Initiative ENES European Network for Earth System Modelling EPCC The Edinburgh Parallel Computing Centre, UK ESFRI European Strategy Forum on Research Infrastructures; created roadmap for pan-european Research Infrastructure. FZJ Forschungszentrum Jülich, Germany HPC High Performance Computing; Computing at a high performance level at any given time; often used synonym with Supercomputing. HPC-Europa Consortium of six leading (HPC) infrastructures and five centres of excellence providing transnational access; EU project. ICT Biannual research event for information and communication technologies in Europe ISC International Supercomputing Conference; European equivalent to the US based SC0x conference. Held annually in Germany. IS-ENES Infrastructure for ENES, European Network for Earth System Modelling ITER Joint international research and development project that aims to demonstrate the scientific and technical feasibility of fusion power. Also used as the name for the reactor. NCF Netherlands Computing Facilities Foundation NCSA The National Center for Supercomputing Applications (NCSA), located at the University of Illinois at Urbana-Champaign, USA NSF National Science Foundation, USA MoU Memorandum of Understanding. MSU Moscow State University PRACE Partnership for Advanced Computing in Europe; Project Acronym. RIKEN Rikagaku Kenkyūsho, a large natural sciences research institute in Japan PRACE - RI vi

8 RZG SARA SC Tier-0 TeraGrid VIRGO Rechenzentrum Garching, Germany Computing and Networking Services, Amsterdam, the Netherlands Annual Supercomputing Conference arranged in the USA Denotes the apex of a conceptual pyramid of HPC systems. In this context the Supercomputing Research Infrastructure would host the tier-0 systems; national or topical HPC centres would constitute tier-1. is an open scientific discovery infrastructure combining leadership class resources at eleven partner sites to create an integrated, persistent computational resource in the USA. The TeraGrid project is funded by the National Science Foundation. The Virgo Consortium for Cosmological Supercomputer Simulations PRACE - RI vii

9 Executive Summary PRACE, the Partnership for Advanced Computing in Europe, and DEISA, the Distributed European Infrastructure for Supercomputing Applications, merged for the first time their annual science symposia into one big European HPC (High Performance Computing) event: The DEISA PRACE Symposium This symposium took place from May 11 to 13 in Amsterdam, and was hosted by SARA and NCF at the Royal Tropical Institute. The theme of the symposium was "HPC Infrastructures for Petascale Applications", it was of major interest to a broad audience: from scientific users, HPC technology experts and vendors to government representatives and industry partners. The symposium attracted almost 200 participants from more than 20 countries and four continents. Prominent keynote speakers from all over the world gave their global perspectives of High Performance Computing in the Petascale era, the new generation of supercomputers for scientific research. The symposium also featured speakers from different science communities. PRACE provided its perspectives on HPC architectures, Applications, Training and Education. DECI, the DEISA Extreme Computing Initiative presented ten computational science grand challenge projects from all over Europe covering many science areas. The DEISA PRACE symposium was very successful regarding to the number of participants, the relations to the European scientific communities established, and the overall feedback received from the participants. The symposium was a follow-up event to the first PRACE Scientific Conference held on 26 November 2008 in Lyon, organised alongside ICT PRACE - RI

10 1 Introduction The mission of PRACE is the preparation of a persistent pan-european high performance computing (HPC) service to meet the needs of academia, industry and society. Supercomputers enable scientists and engineers to solve today s problems and to develop the new technology for tomorrow s industry, affecting national employment patterns and national wealth. To further the aim of enabling the academic and industrial research in Europe to develop and use world class HPC, it is important to strengthen contacts with and between scientific communities; in particular those who are involved in helping to develop the very science case on which this effort is based. To this end, PRACE was responsible for organising two scientific conferences during the course of the project. The first of these was held on November 26, 2008 in Lyon, organised alongside ICT 2008, the largest research event for information and communication technologies in Europe. The second one, the DEISA PRACE Symposium, was held on May 11 13, 2009 in Amsterdam, the Netherlands. This event is the focus of the present report. 2 Motivation and Venue The first PRACE scientific conference was held on November 26, 2008 in Lyon, organised alongside ICT 2008, the largest research event for information and communication technologies in Europe. The previous DEISA Symposia have been arranged in Paris (2005), Bologna (2006), Munich (2007) and Edinburgh (2008). This year, DEISA and PRACE for the first time merged their annual science symposia into one big European HPC event: The DEISA PRACE Symposium The symposium took place at the Royal Tropical Institute (KIT) in the heart of Amsterdam and it was jointly hosted by SARA and NCF. The location of the venue had been predetermined by DEISA before the two projects aggreed on a joint event. The aim of the symposium was to reach the scientific users, HPC technology experts and vendors, government representatives and industry partners that are likely to contribute or benefit from the future HPC infrastructure deployed by PRACE and DEISA. The theme of the symposium was HPC Infrastructures for Petascale Applications. This is what PRACE and DEISA are jointly creating for Europe and what other organisations and projects are building in the USA and Asia. Prominent keynote speakers from all over the world gave global perspectives of High Performance Computing (HPC) in the Petascale era providing an overview of the new generation of supercomputers for scientific research. Speakers on the first day included Kostas Glinos of the European Commission, Abani Patra from the National Science Foundation (USA), Ryutaro Himeno from RIKEN (Japan), Horst Simon from the Department of Energy (USA), Ben Evans from the Australian National University (Australia) and Vladimir Voevodin from the Moscow State University (Russia). The symposium also featured speakers from different science communities: cosmology, fusion research, climatology and biosciences. PRACE additionally provided its perspectives on HPC architectures, Applications, Training and Education. PRACE - RI

11 From DEISA, the Extreme Computing Initiative presented ten computational science grand challenge projects from all over Europe covering many science areas. The programme is detailed in this report in section 4.5. Figure 1: The Royal Tropical Institute of Amsterdam PRACE - RI

12 3 Organising Committees The organisation of this symposium was divided into two groups: the programme and the organising committee. Some of the members of the programme committee had an initial meeting about the symposium during SC08 in December 2008 in Austin, Texas. The first teleconference took place on 9 January, This teleconference was a start-up teleconference in which everyone of the DEISA PRACE organisation participated. Afterwards the communication between all organisers took place via and therefore two aliases were created: deisa-prace-symposium@fz-juelich.de reaching all organisers deisa-prace-symposium-contact@fz-juelich.de reaching only the organising committee. In addition, the organising committee had weekly teleconferences: in total 18 teleconferences were arranged regarding the DEISA PRACE symposium. 3.1 Programme Committee The programme committee was in charge of the symposium programme, the overall coordination of the symposium including political and financial issues. The programme committee consisted of the following people: Axel Berg, SARA, host site representative Thomas Eickermann, FZJ, PRACE project manager Hermann Lederer, RZG, DEISA dissemination leader Achim Streit, FZJ Andreas Schott, RZG Ari Turunen, CSC, PRACE dissemination leader The Programme committee invited the speakers on behalf of PRACE project coordinator, Prof. Dr. Achim Bachem and DEISA project coordinator, Dr. Stefan Heinzel. 3.2 Organising Committee The organising committee consisted of a local organisers group with people from the Netherlands: Marina den Hartog, SARA Naomi Messing, NCF Dionne Verhoeven, SARA In addition to the local organisers, CSC was involved in the organising committee: Anni Jakobsson (PRACE dissemination officer) and Heli Autere (DEISA representative) CSC took part of the dissemination, website and graphical design of the symposium s material. The DEISA website was updated by Markus Rampp and Johannes Reez from RZG. PRACE - RI

13 4 Dissemination One of the aims of the symposium was to provide an opportunity for PRACE to strengthen its links with HPC initiatives in the US, Japan, and other parts of the world in order to get their experience and practice of managing, running and scaling applications to Petaflop/s levels. This valuable information is to be used in assisting the decision making process in PRACE. The links and relationships created in the symposium are described in this document chapter 5.1. The symposium was meant for a broader audience: from scientific users, HPC technology experts and vendors to government representatives and industry partners. This had to be taken into account when inviting possible participants. The economical recession in the world also affected the symposium as many companies and institutes had travel restrictions at the time. Despite of this, the organisers were very successful getting people to this seminar. Other European projects, such as HPC-Europa2 and Belief-II, were taken into account in dissemination as well: the symposium was advertised at the Belief-II site, and HPC-Europa2 dissemination material was available at the symposium exhibition (see more from chapter 4.8. of this document). 4.1 Invitations The programme committee took care of inviting the speakers to the seminar on behalf of PRACE project coordinator, Prof. Dr. Achim Bachem and DEISA project coordinator, Dr. Stefan Heinzel. The programme consisted of a total of 30 speakers. The details of the programme are reported in section 4.5. The potential participants were invited by the organising committee. The DEISA and PRACE stakeholders were invited with the help of the DEISA and PRACE partners by sending invitations by and advertising the event at DEISA and PRACE partners websites. There was no attendance fee for the event for the participants. A total of 182 participants took part to the event. More detailed summary of the participants in this report is stated in section Press Releases A press release HPC Infrastructures for Petascale Applications DEISA PRACE Symposium 2009 was published before the event on 25 February 2009 on both PRACE and DEISA websites. In parallel it was also sent to various online HPC magazines, such as HPCWire, International Science Grid This Week, Inside HPC, Scientific Computing World and to the Primeur magazine. The press release was also distributed via the AlphaGalileo [2] service the world's leading source of European research news which reaches over 6,000 journalists worldwide. URL to the press release on the PRACE website: Another press release was published after the event, covering the outcome of the symposium. This press release was published on 22 May, 2009 using the same distribution channels as the first press release. URL to the second press release on the PRACE website: PRACE - RI

14 participants-from-more-than-20-countries-and-four-continents The press releases gained press cuttings all across Europe, for example in HPCWire. The Primeur magazine made feature articles about the event. One of them was entitled Amsterdam was supercomputing capital of Europe for three days [3]. 4.3 Websites Figure 2: DEISA PRACE Symposium 2009 web (After the event the site was moved to the section of past events on the PRACE website: The PRACE [1] and DEISA [4] websites featured specific sections for the event. The URLs of the websites are: PRACE website: amsterdam-the-netherlands/) PRACE - RI

15 DEISA website: Figure 3: DEISA PRACE Symposium 2009 website The flyer and programme of the symposium were uploaded to both websites. Also, useful information about the means of transportation in Amsterdam was uploaded to DEISA and PRACE websites. Registration to the event was conducted via the PRACE website. Also, dedicated pages for hotel information and conference evening were created to the PRACE website and links to these sites were provided from the DEISA website. After the symposium all presentations from the event were uploaded to both websites. Also, follow-up s were sent to all speakers and attendees including links to the presentations and the list of participants. This way the symposium attendees were able to download presentations from the event. 4.4 Flyer A flyer about the symposium (Figures 4 and 5) was created before the symposium. The format of the flyer was triple folded A4. The flyer was sent as a pdf attachment with invitations to the possible participants. Also, the flyer was part of the symposium kit that was distributed at the venue. The flyer included information about the symposium, the programme along with general information about PRACE and DEISA. PRACE - RI

16 Figure 4: DEISA PRACE Symposium flyer Figure 5: DEISA PRACE Symposium flyer PRACE - RI

17 4.5 Symposium Programme The final programme was printed on the flyer of the symposium (Figures 4 and 5). The flyer was available on the PRACE and DEISA websites as well as in the conference kit which was distributed to every participant. In addition to the final programme and the list of participants, the conference kit included the speakers biographies and abstracts, an evaluation form, a map of Amsterdam, information about the venue of the conference diner, a timetable of public transportation and details on how to access the wireless network that was available at the venue. Programme: Monday, May 11 12:00 18:00 hrs 12:00 Registration & Lunch in the Marble Hall of the Royal Tropical Institute 13:00 13:05 Welcome Anwar Osseyran (SARA) 13:05 13:10 Opening of the Symposium Paul Doop (Vice-President of the Board of University of Amsterdam, Member of the Board of SARA) 13:10 14:55 Global Perspectives Session chair Patrick Aerts (NCF) Kostas Glinos (EC): e-infrastructure: EC perspective Abani K. Patra (NSF, USA): Supporting Computational Discovery at Scale (Managing Complexity) Ryutaro Himeno (RIKEN, Japan): Japan s Next-Generation Supercomputer R&D Project and Grand Challenges in Life Science 14:55 15:25 Coffee break 15:25 17:10 Global Perspectives Session chair Stefan Heinzel (RZG) Horst Simon (DoE, NERSC, USA): Science Drivers, Current HPC Software Development, and Platform Deployment Plans for the USA Ben Evans (ANU-SF and NCI National Facility, Australia): Toward Petascale Computing in Australia Vladimir Voevodin (MSU, Russia): MSU Petascale Facilities and Perspectives for HPC Infrastructure in Russia 17:10 17:45 Science Communities Peter Coveney (Life Sciences, UK): DEISA-PRACE & The Virtual Physiological Human Tuesday, May 12 09:00 18:00 hrs 09:00 10:40 Global Perspectives Session chair Anwar Osseyran (SARA) PRACE - RI

18 Maria F. Ramalho (EC): e-infrastructure: The European HPC service Achim Bachem (PRACE): PRACE On the way to a persistent European high-end computing service Stefan Heinzel (DEISA): Towards a European HPC Infrastructure for Petaflop Applications DEISA and its contributions John Towns (TeraGrid, NCSA, US): TeraGrid and the Path to Petascale 10:40 11:10 Coffee break 11:10 12:55 Science Communities Session chair Hermann Lederer (RZG) Sylvie Joussaume (ENES leader, climate research, France): High-performance computing for climate modelling Frank Jenko (Fusion Research, DE): Extreme computing in support of ITER Carlos Frenk (Cosmology, UK): Simulating cosmic evolution 12:55 14:15 Lunch break 14:15 15:30 PRACE Perspectives Session chair Thomas Eickermann (FZJ) Aad van der Steen (NCF, the Netherlands): Directions in HPC Technology Mark Bull (EPCC, UK): A Survey of HPC Systems and Application in Europe Timothy Stitt (CSCS, Switzerland): Towards a European HPC Training and Education Infrastructure for Petascale Computing 15:30 16:00 Coffee break 16:00 17:15 DEISA Extreme Computing Session chair Axel Berg (SARA) Harm Jonker (Weather and Climate Research, The Netherlands): PINNACLE: Pinning down the growth-rate law of Atmospheric Convective Boundary Layers Hans-Joachim Bungartz (Engineering, Germany): Efficient Algorithmic Approaches for Flow Simulations on Cartesian Grids Hannu Häkkinen (Materials Science, Finland): Golden superatoms and quantum dots Evening Conference Evening Wednesday, May 13 09:00 14:00 hrs 09:00 09:35 QCD breakthrough 2008: Session chair Hermann Lederer (RZG) Zoltan Fodor (Germany): The origin of the mass of the visible Universe 09:35 10:50 DEISA Extreme Computing Session chair Hermann Lederer (RZG) Rosa Dominguez-Tenreiro (Astrophysics, Spain): Modelling Galaxy Formation in the Universe: the Need for Supercomputing PRACE - RI

19 Jean-Philippe Laval (Engineering, France): Direct Numerical Simulations of convergingdiverging channel flow Michele Migliore (Computational Neuro Sciences, Italy): Large-scale simulations of the olfactory bulb 10:50 11:10 Coffee break 11:10 12:50 DEISA Extreme Computing Initiative Session chair Alison Kennedy (EPCC) Matthias Krack (Materials Science, Switzerland): Extending the time and length scale of abinitio molecular dynamics simulations Luís O. Silva (Plasma Physics, Portugal): Laser Plasma Accelerators towards the energy frontier with transformative simulations Rainer Spurzem (Astrophysics, Germany): From Newton to Einstein relativistic dynamics of black holes in dense star clusters David van der Spoel (Comput. Bio Sciences, Sweden): Large Scale Biomolecular Dynamics Simulations 12:50 Closing remarks Anwar Osseyran (SARA) 13:00 Lunch in the Marble Hall of the Royal Tropical Institute Prominent keynote speakers from all over the world gave their global perspectives of HPC in the Petascale era providing an overview of the new generation of supercomputers for scientific research. Speakers on the first day included Kostas Glinos of the European Commission, Abani Patra from the National Science Foundation (USA), Ryutaro Himeno from RIKEN (Japan), Horst Simon from the Department of Energy (USA), Ben Evans from the Australian National University (Australia) and Vladimir Voevodin from the Moscow State University (Russia). Figure 6: DEISA PRACE Symposium 2009 audience PRACE - RI

20 The following notes from the DEISA PRACE Symposium are made by Wolfgang Gentzsch and published in the DEISA newsletter 3/2009. Figure 7: Kostas Glinos, Head of Unit, Géant and e-infrastructures of the European Commission Kostas Glinos, Head of Unit, Géant and e-infrastructures of the European Commission, was highlighting the ambition to make Europe a leading player in supercomputing, not only from the user perspective, but also for European suppliers of components, software, systems, and services. Glinos stated that this can be achieved when Member States and the Commission join forces and pool investments. He also noted that it is up to the scientific community to implement a coordinated plan of action. The supercomputing strategy should strengthen the European researchers in their global competitiveness. Areas such as medicine, climate and energy are of worldwide importance. The European Commission is stimulating the support for escience with e-infrastructures consisting of high-speed networking, Grid infrastructures, and data infrastructures. A relative new item is supercomputing. The goal is to increase the number of Europe-based supercomputers in the upper regions of the TOP500 list. Also the European IT industry should be encouraged. PRACE - RI

21 Figure 8: Maria Ramalho Natário from the European Commission Maria Ramalho Natário from the European Commission explained, that the focus is more on a complete European HPC ecosystem, with different machines and Grids at different levels, but accessible by a scientist through a workspace that hides as much as possible the complexity of the systems and the Grids, such that a researcher can create workflows where different steps can use different system architectures or Grids when needed. The next funding call for projects, that opens in July 2009 and closes in November 2009 will support this complete ecosystem. Figure 9: Abani Patra from the NSF, USA PRACE - RI

22 Abani Patra from the National Science Foundation (USA) in his Global perspective presentation talked about Computational Discovery at Scale. One of his key messages was about the new oracle Predictive Science. This means that data, models and people together build a systematic integration of observational data, knowledge of physics and best available computing methodology and hardware to extrapolate from available data and settings. He closed his presentation with a set of challenging questions which are worth noting: What are the new applications that are emerging or likely to emerge in the coming decade? How can we best stimulate development of software for applications and the hardware architectures to match? How can useful software that has been developed be sustained beyond the development period? What systems software will be required? Distributed systems support, programming environments, runtime support, data management user tools? What application support environments will be needed? Application packages, numeric and non-numeric library packages, problem solving environments? How can we aid or catalyze developments that make it possible to use the same tools, including compilers, debuggers and performance tools, on system scales all the way down to the typical researcher s laptop or desktop? What education and training actions should be considered to prepare researchers, students and educators for future cyber infrastructure? Figure 10: Horst Simon from Lawrence Berkeley National Laboratory, USA PRACE - RI

23 Horst Simon from Lawrence Berkeley National Laboratory (USA), gave an overview of the current supercomputing developments in the USA. With two supercomputers currently running at over a petaflop/s, and studies underway on exascale systems, many people believe we will reach exascale computing in a decade. Horst Simon said this is too optimistic. Exascale computer architectures necessitate radical changes to the software used to operate them and the science applications. The change is as disruptive as the shift from vector to distributed memory supercomputers 15 years ago. The current message passing coupled with sequential programming languages will be inadequate for architectures based on many-core chips. The present code development, correctness, and performance analysis tools cannot scale up to millions of threads that will run in an exaflop/s computer. With perhaps millions of cores, exaflop/s computers are bound to fail on at least small parts, very often. But the current check pointing will be inadequate for reaching fault tolerance at the exascale. The exaflop/s systems will also produce massive amounts of data. Fundamental changes are necessary to manage and extract knowledge from this tsunami of data. We certainly cannot devote all of our HPC efforts to exascale systems. Though the first supercomputers have crossed the petaflop/s barrier, we still are not at a level where day-today sustained petaflop/s performance is delivered to many types of applications. Horst Simon sees no coherent Petascale software plan across different platforms and different agencies (in the US). The USA is spending a lot of money each year on HPC. For the fiscal year 2009, Horst Simon estimates that it amounts to probably about $2B total: High End Computing Infrastructure and Applications get $1,142 M and High End Computing R&D $492 M. Figure 11: Vladimir Voevodin from the Moscow State University Vladimir Voevodin from the Moscow State University (MSU) presented Russia s way to petaflop/s. Currently the fastest system in Russia is on position 35 in the TOP500 with 70 teraflop/s performance. Vladimir Voevodin, expects that by next year a machine with 0.5 PRACE - RI

24 petaflop/s performance will be operational. The new 258,48 teraflop/s MSU supercomputer will be installed in Q3 09, and upgraded early next year to 500 teraflop/s. Other Global perspectives were presented by Ryutaro Himeno from RIKEN (Japan), Ben Evans from the Australian National University (Australia), and John Towns from TeraGrid (USA). Achim Bachem from Research Centre Juelich (FZJ) and Coordinator of PRACE, and Stefan Heinzel from Garching Computing Centre (RZG) of the Max Planck Society and Coordinator of DEISA, both stressed the importance of a joint European approach to high-performance computing, which is done in a complementary way, with PRACE focusing on new European leadership-class top-tier systems, and with DEISA operating and further improving the existing distributed supercomputing infrastructure. Achim Bachem presented a detailed map on how Europe is moving ahead towards a persistent high-end computing service for the global scientific community, as a part of the long-term roadmap detailed in the ESFRI European Strategy Forum on Research Infrastructures. Within PRACE, the partners are installing several petaflop/s systems over the next five years. Concerning HPC, a total of 20 European PRACE member states are now talking with one voice. All PRACE member states will need the skills for peta- and exascale computing. Stefan Heinzel stated that the European HPC infrastructure has been built and operated successfully over the last few years. He also noted that the European HPC infrastructure has been used intensively by the computational science community. Now the focus needs to be more on the grand-challenge applications for petascale computing, running potentially on hundreds of thousands of processor cores. Challenges of dramatic optimisation, multithreading, new parallelisation strategies, new programming methods, new algorithms, libraries and tools are faced, and new codes will have to be developed through strong science community efforts in cooperation with the supercomputing centres. The symposium also featured speakers from various science communities. "High-performance computing is crucial for climate research to understand mechanisms of climate change and predict future climate change perturbed by human activities. The powerful computing is needed to understand and to predict extreme events and assess the regional impacts of the climate change on society and economy", stated Prof. Sylvie Joussaume, researcher at CNRS and expert in climate modelling. She is also chairing the European Network for Earth System modelling (ENES). ENES has started the new FP7 Infrastructure project IS-ENES to better understand and predict future climate change by high-end simulations. In the Life Science community talk Prof. Peter Coveney from University College London presented the Virtual Physiological Human project with collaborations from seven European countries. Prof. Frank Jenko from the European Fusion Research community, researcher at the Max Planck Institute for Plasma Physics, presented an overview of the world-wide ITER project and the HPC needs for its success. Prof. Carlos Frenk from Durham University gave a brilliant insight into the world of cosmology and the challenging simulations of cosmic evolution by the VIRGO Consortium, the world-leading group in this field. All four science communities are supported by DEISA and plan to use the PRACE petaflop systems. PRACE - RI

25 PRACE also provided its perspectives on HPC architectures, Applications, Training and Education. PRACE was presented by Aad van der Steen (The Netherlands) with a presentation Directions in HPC Technology, Mark Bull from EPCC (UK) with a presentation Survey of HPC Systems and Applications in Europe, and Tim Stitt from CSCS (Switzerland) with a presentation Training and Education for Petascale Computing. From the DEISA Extreme Computing Initiative ten computational science grand-challenge projects from all over Europe were presented: Harm Jonker (Weather and Climate Research, The Netherlands & UK), Hans-Joachim Bungartz (Engineering, Germany), Hannu Häkkinen (Materials Science, Finland), Rosa Dominguez-Tenreiro (Astrophysics, Spain), Jean-Philippe Laval (Engineering, France), Michele Migliori (Computational Neuro Sciences, Italy), Matthias Krack (Materials Science, Switzerland), Luís O. Silva (Plasma Physics, Portugal), Rainer Spurzem (Astrophysics, Germany), and David van der Spoel (Computational Bio Sciences, Sweden). Last but not least, Zoltan Fodor from the University of Wuppertal reported on the QCD Breakthrough 2008: The Origin of Mass of the Visible Universe, trying to answer the question of what is the source of the mass of the ordinary matter, successfully, in layman s terms. 4.6 Speakers Biographies and Abstracts Speakers biographies and abstracts of the presentations were distributed to every participant in the conference kit. The biographies and abstracts can be found in Annex Other material A banner for the entrance of the Royal Tropical Institute was created (Figure 12) for the DEISA PRACE Symposium. Also, a foam board for the lectern and two roll-ups (Figure 6) one DEISA and one PRACE, were created. All attendees got a name badge (Figure 13). Figure 12: DEISA PRACE Symposium banner PRACE - RI

26 4.8 Exhibition at the Symposium Figure 13: DEISA PRACE Symposium name badges The DEISA PRACE Symposium 2009 also hosted an exhibition featuring PRACE and DEISA dissemination material. A total of 11 DECI posters were displayed in the front of the main auditorium along with one poster from PRACE and two from DEISA. PRACE and DEISA dissemination material was available, such as the PRACE brochures, newsletters, buttons, PRACE USB-keys, candy, DEISA Digest magazines etc. Also, HPC-Europa2, TERENA and ISC 09 material was available at the exhibition. Figure 14: Exhibition table at the DEISA PRACE Symposium PRACE - RI

27 4.9 Conference Evening Figure 15: Posters at the DEISA PRACE Symposium A conference evening was arranged on Tuesday, May 12. The purpose of the evening for the participants was to get to know each other better in a less formal atmosphere and to build collaborations within the HPC community. The conference evening started at 19:00 hrs with a boat trip through the canals of Amsterdam. Later, the boats arrived at Restaurant d Vijff Vlieghen where the participants disembarked for the conference dinner. The evening ended around 23:00 hrs. Figure 16: Conference participants embarking the boats PRACE - RI

28 5 Attendees DEISA PRACE Symposium 2009 attracted 182 participants from 22 countries. Table 1: DEISA PRACE Symposium participants A complete list of attendees with their names, countries and affiliations can be found on Annex 9.1 of this document. 5.1 Relations to Stakeholders One aim of the symposium was to create stronger links with initiatives in the US, Japan, and other parts of the world in order to gain experience and practice of managing, running and scaling applications to Petaflop/s levels. This information is to be used to assist the decision making process in PRACE. The symposium was aimed to a broad audience: from scientific users, HPC technology experts and vendors to government representatives and industry partners. The symposium managed to attract a wide audience covering many science fields, science communities and projects, such as CNRS, ENES and IS-ENES, ITER-project and VIRGO Consortium. The symposium offered PRACE the possiblity to establish stronger links with the aforementioned communities and projects, covering geographically various countries and continents. PRACE - RI

29 6 Post-event Activities The press release, which was published after the symposium, gained press cuttings all across Europe, for example in HPCWire. The Primeur Magazine made feature articles about the event. One of them was entitled Amsterdam was supercomputing capital of Europe for three days [3]. Feature articles about the event were published in PRACE and DEISA newsletters. After the symposium the organising committee sent follow-up s to all speakers and attendees. The included links to the presentations, the evaluation form and the list of participants. Pictures taken from the event are available on PRACE and DEISA websites. A wider picture archive is available on the Picasa picture service [5]. It is possible to get access to the picture archive by sending a request to the organisers: deisa-prace-symposium-contact@fz-juelich.de. 7 Feedback An evaluation form (Annex 9.2.) was given to every symposium participant in the conference kit which was delivered at the symposium. Attendees were reminded to fill the evaluation form and return it to the registration desk. The form was also sent to all symposium attenedees after the event via . A total of 35 forms were given back to the symposium organisers. 62% of the respondents thought that the symposium was very productive and 35% thought that the symposium was productive. 45% of the respondents gave 1 (excellent) to the relevance, agenda and discussions at the symposium; 37% of the respondents gave 2 on a scale of 1 (excellent) to 5 (poor). The organisation of the symposium at the same 1 to 5 scale got the following percentages: 1 (excellent) 72% and 2 from 22% of the respondents. Below are more comments and more detailed answers from the respondents. Feedback from participants: Do you wish to receive news from DEISA/PRACE Yes: 22 No: 6 (as some already received information) No answer: 7 Overall Impression Very Productive (1): 23 = 62% Productive (2): 13 = 35% Not Productive (3): 1 = 3% (total of very productive and productive = 97%) PRACE - RI

30 Relevance, Agenda and Discussions Excellent: (1) (2) 1) Relevance of Topics discussed 24 68% 9 2) Agenda (adherend to, allocation of time per topic) 14 40% 15 3) Quality of background documents distributed 13 41% 8 4) discussions 14 41% 12 5) speakers 14 36% 20 total 79 45% 64 Average: (3) (4) 26 % 2 6% Poor: (5) 42 % 2 6% 2 6% 2 6% 25 % 8 35 % 6 25 % 3 9% 18 % 1 3% 1 3% 51 % 3 8% 2 5% 37 % 21 Table 2: Feedback form: Relevance, Agenda and Discussions 12 % 8 5% 3 1% Organisation Excellent: (1) (2) Average: (3) (4) 6) Preparation and distribution of background documents 17 49% 14 40% 3 9% 1 3% 7) Support from organising staff 32 89% 4 11% 8) Conference facilities 28 78% 5 14% 3 8 9) Hotel (choice, convenience) 18 72% 6 24% 1 4% Poor: (5) total 95 72% 29 22% 7 5% 1 1% 0 0% Table 3: Feedback form: Organisation PRACE - RI

31 How did you hear about the symposium - invitation by internet: 11 - announcement in press: 2 - at a conference/event: 1 - from colleagues: 9 - other (please specify): phone: 1 member PRACE project: 2 member DEISA PRACE: 2 member DEISA: 1 CSC website: 1 Next meeting Do you plan to attend the follow up symposium in 2010? - Yes: 35 - Depends on the subject: 1 If yes what subjects would you like to be addressed? 1. How policy making depends on the science that hpc is used for 2. PRACE and DEISA results in terms of "sustainable development" of numerical modelling. 3. Power consumption issues 4. Future programming techniques 5. Results of WP8 6. European level cooperation of TIER1 (and possible TIER2) centers. 7. Killer application areas, technology. 8. How can small countries with limited funding for hpc equipment effectively participate to the PRACE/DEISA programs? 9. Astrophysics and cosmology 10. technology developments, software developments (tools, libraries, languages etc.) performance of systems. 11. Power and cooling of a petascale hpc 12. Discussion of future needs of research teams to prepare to support them 13. Challenges and successes to coordination effects; scaling to petascale and beyond. 14. Slightly more hardware stuff PRACE - RI

32 15. Competitive centralized project/proposal evaluator 16. cost-benefit analysis of results and code development 17. What is peta-scale-adequate; how to support the rest? 18. The experience with accelerators 19. User experience with PGAS languages 20. hpc simulation in industry applications and physics 22. Experiences from emerging architectures (GPU's, IPGA's, ) 23. Excellent 8 Conclusions With nearly 200 participants from 22 countries and four continents, the symposium reached its aim to create relationships between PRACE, DEISA and their various stakeholder groups most importantly with different scientific domains. The three-day programme consisting of 30 speakers from around the world was an achievement of itself. The feedback from the symposium was overall very positive. 97 per cent of the respondents regarded the symposium very productive (62%) or productive (35%). A total of 45 per cent of the respondents gave 1 (excellent) to the relevance, agenda and discussions at the symposium; 37 per cent of the respondents gave 2 on a scale of 1 (excellent) to 5 (poor). The organisation of the symposium at the same 1 to 5 scale got the following percentages: 1 (excellent) 72 % and 2 22% of respondents. This information was gathered from an evaluation form which was given to every symposium participant. A total of 35 forms were given back to the symposium organisers. The organising committee recommends that in the future at similar kind of events the evaluation form should also be available online. This way people might leave more feedback. The Primeur Magazine stated Amsterdam was supercomputing capital of Europe for three days. PRACE continues the engagement with different science fields at different science symposia until the remaining of the PRACE project. This year, PRACE has taken part to the following science symposia: OGF25 / EGEE User Forum, 2-6 March, Catania, Italy Scientific POWER Meeting, March, 2009, Mazurian Lakes, Poland 24th Forum ORAP, 26 March, Lille, France NAFEMS NORDIC Seminar NOTUR2009, May 2009, Trondheim, Norway ScicomP 15, May 18-22, Barcelona, Spain SERI 2009, June 3-5, Paris, France HPDC 2009, June 11-13, Munich, Germany PRACE is planning to attend the following science symposia during 2009: PRACE - RI

33 DFT09, August 31 - September 4, Lyon, France ParCo 2009, 1 4 September, Lyon, France NEERI 09, September 30 - October 2, Helsinki, Finland PPAM 2009, September, Wroclaw, Poland Bio IT World Conference & Expo 09, October 6-8, Hannover, Germany ICNSP 6 9 October, Lisbon, Portugal HiBi, October, Trento, Italy I3 Conference, November 4-6, Poznan, Poland PRACE - RI

34 9 Annex 9.1 List of Attendees First name Last name Institution Country Patrick Aerts NCF Netherlands Pedro Alberto University of Coimbra Portugal Victor Alessandrini IDRIS-CNRS France Lilit Axner SARA Netherlands Achim Bachem Forschungszentrum Juelich Sanzio Bassini CINECA Italy Germany Kyriakos Baxevanidis European Commission Belgium Florian Berberich Forschungszentrum Juelich Germany Axel Berg SARA Netherlands Kamen Beronov Leibniz-Rechenzentrum der Bayerischen Akademie der Wissenschaften Germany Wolfgang Bez NEC HPC Europe Germany Joachim Biercamp German Climate Computing Centre Germany Marjan Boer, de Silicon Graphics Netherlands Thomas Bönisch HLRS Germany Marina Bouianov CSC-IT Center for Science Finland Luigi Brochard IBM France Michael Browne ICHEC Ireland Rob Bruin, de RUG Netherlands Reinhard Budich Max-Planck-Institut Meteorologie Mark Bull EPCC UK Hans-Joachim Iris Bungartz Christadler für TUM, Dept. of Informatics, and Leibniz Supercomputing Centre Leibniz Centre Supercomputing Germany Germany Germany Elena Churkova T-Platforms Russia David Thijmen Peter Colignon Collignon Coveney Fund for Scientific Research & University of Liege Delft University of Technology University London College Belgium Netherlands Xavier Delaruelle CEA France UK PRACE - RI

35 Lukas Demovic Comenius University Slovak Republic Ana Bela Dias NCF Netherlands Johannes Diemer Hewlett Packard GmbH Germany Hans Dijkman University of Amsterdam Netherlands Rosa Dominguez-Tenreiro Universidad Autónoma de Madrid Spain John Donners SARA Netherlands Paul Doop UvA, HvA, SARA Netherlands Leo Martin Thomas Oskar Duc, Le Duda Eickermann Elmgren Ministry of Education Culture and Science VSB-Technical University of Ostrava Forschungszentrum Juelich Elmer Technology Development Netherlands Czech Republic Germany Finland Ad Emmen AlmereGrid Netherlands Ben Evans ANU-SF Australia Bernhard Fabianek European Commission Belgium Juha Fagerholm CSC Finland Manuel Fiolhais University of Coimbra Portugal Zoltan Fodor University of Wuppertal Germany Anton Frank Leibniz-Rechenzentrum der Bayerischen Akademie der Wissenschaften Carlos Frenk Durham University UK Germany Jean-Pierre Genin France Wolfgang Gentzsch DEISA Germany Martin Gijzen, van Deft University of Technology Sergi Girona BSC Spain Netherlands Denis Girou CNRS-IDRIS France Kostas Glinos European Commission Belgium Hannu Häkkinen University of Jyväskylä Finland Marina Hartog, den SARA Computing and Networking Services Netherlands Philippe Haye Fujitsu France Stefan Heinzel DEISA Germany Peter Joerg Hermann Heydemueller Computer Center of Slovak Technical University MEGAWARE GmbH Computer Slovak Republic Germany Martin Hilgeman SGI Netherlands Ryutaro Himeno RIKEN Japan Peter Hinrich SURFnet Netherlands PRACE - RI

36 Jaap Hollenberg NCF Netherlands Sverker Holmgren SNIC Sweden Gyöngyi Horvath Terena Netherlands Christopher Huggins Clustervision Netherlands Paul Huygen Vrije Universiteit Netherlands Sara Ibanez Barcelona Supercomputing Center Spain Janne Ignatius CSC Finland Kimberly Ondrej Anni Florian Frank Gary Iles Jakl Jakobsson Janetzko Jenko Johnson In Theory Communications, LLC Academy of Science of the Czech Republic CSC - IT Center for Science Ltd Forschungszentrum Juelich Max Planck Institute for Plasma Physics Computational Solution Science USA Czech Republic Finland Germany Germany USA Lennart Johnsson U of H USA Bob Jones CERN Switzerland Harm Jonker Delft University Netherlands Sylvie Joussaume CNRS, IPSL/LSCE France Lori Kaldenberg NCSA/University Illinois Dick Kamphuis HP Netherlands Janne Kardaun Statistics Netherlands Netherlands of USA Daniel S. Katz University of Chicago USA Richard Keijzer Automatisering Gids Netherlands Ron Keijzer Slotervaart Ziekenhuis Netherlands John Alison Kennedy Kennedy Rechenzentrum Garching (RZG) EPCC University of Edinburgh Germany Ron Keyzer Slotervaart hospital Netherlands Kimmo Koski CSC-IT Center for Science UK Finland Jacko Koster UNINETT Sigma Norway Mikhail Kozhevnikov T-Platforms Russia Matthias Krack Paul Scherrer Institute Switzerland Stefan Kraemer Sun Microsystems Germany Leif Tomas Laaksonen Lacko CSC - IT Center for Science Ltd Computer Center of Slovak Technical University Finland Slovak Republic PRACE - RI

37 Scott Lathrop University of Chicago USA Erwin Laure KTH Sweden Jean-Philippe Laval CNRS France Jean-Francois Lavignon Bull France Hermann Lederer RZG, Max Planck Society Germany Sik Craig Pekka Michael Alexander (Lex) Lee Lee Lehtovuori Levine Levisson KISTI Center Supercomputing Open Grid Forum /The Aerospace Corporation CSC - IT Center for Science Ltd PITTSBURGH SUPERCOMPUTING CENTER Ministry of Economic Affairs Daejeon USA Finland USA Netherlands Claire Levy CNRS France Alain Lichnewsky GENCI France Rene Lier, van KNMI Netherlands Walter Lioen SARA Netherlands Thomas Lippert Jülich Centre Supercomputing Germany Rossend Llurba NCF Netherlands Thomas Kurt Ludwig Lust German Compute Centre Climate VSC - Flemish Supercomputer Centre Germany Belgium Fabrizio Magugliani SiCortex, Inc USA Tamas Maray NIIF/Hungarnet Hungary Naomi Messing NCF Netherlands Peter Messmer Tech-X Corporation USA Karl-Heinz Meurer IBM Germany Germany Hans Meurer Prometeus GmbH Germany Annejet Meyler TUE Netherlands Peter Michielse NCF Netherlands Michele Migliore Institute of Biophysics, National Research Council Riccado Murri CSCS Switzerland Jorge Naranjo BSC Spain Hans Nelemans SGI Netherlands Jane Nicholson EPSRC UK Wouter Nieuwenhuizen KNMI Netherlands Ross Nobes Fujitsu Laboratories of Europe Jean-Philippe Nomine CEA/DIF France Italy UK PRACE - RI

38 Luís O.Silva IST Portugal Anwar Osseyran SARA Computing and Networking Services Mark Parsons EPCC UK Abani Patra National Science Foundation (NSF) Netherlands Ronald Pol, van der SARA Netherlands Ronan Prendergast University College Dublin Ireland Gavin Spurzem Pringle Rainer EPCC University of Edinburgh Astronomisches Rechen- Institut, Zentrum für Astronomie, Universität Heidelberg PRACE - RI USA UK Germany Maria Ramalho Natario European Commission Belgium Markus Rampp RZG, Max Planck Society Germany Johannes Reetz RZG, Max Planck Society Germany Stephane Requena GENCI France Wim Rijks SARA Netherlands Orlando Rivera Leibniz-Rechenzentrum der Bayerischen Akademie der Wissenschaften Germany Francois Robin GENCI France Ralph Roskies PITTSBURGH SUPERCOMPUTING CENTER Eckhard Schaumann Sun Microsystems Germany Klaus Schilling CaSToRC, THE CYPRUS INSTITUTE USA Cyprus Michael Scgliephake HLRS Germany Andreas Schmidt RZG, Max Planck Society Germany Alban Schmutz OXALYA France Andreas Schott MPG/RZG Germany Ari Paavo Seitsonen CNRS France Eric Sennema University of Amsterdam Netherlands Camiel Severijns KNMI Netherlands Robert Simek IBM Slovakia Horst Simon Doe, NERSC USA Alan Simpson EPCC UK Andrey Slepuhin T-Platforms Russia David Spoel, van der Uppsala University Sweden Joerg Stadler IBM Germany Germany Aad Steen, van der NCF Netherlands Horst-Dieter Steinhöfer Leibniz-Rechenzentrum der Bayerischen Akademie der Wissenschaften Germany

39 Timothy Achim Stitt Streit Swiss National Supercomputing Centre (CSCS) Forschungszentrum Juelich Switzerland Germany Ulla Thiel Cray Germany John Towns TeraGrid USA Ari Turunen CSC-IT Center for Science Finland Robert Uebelmesser SGI EMEA Germany Damir Uzbekov T-Platforms Russia Dany Vandromme RENATER France Andrea Vanni CINECA Italy Dionne Verhoeven SARA Computing and Networking Services Netherlands Willem Vermin SARA Netherlands Roel Verstappen University of Groningen Netherlands Kees Verstoep VU, University of Amsterdam David Vicente BSC Spain Vladimir Voevodin Research Computing Center, Moscow State University Isabella Weger ECMWF UK Michele Weiland EPCC University of Edinburgh Netherlands Russia Paul Wielinga SARA Netherlands Jules Wolfrat SARA Netherlands Gustavo Yepes Universidad Autonoma de Madrid UK Spain Herbert Zwartscholten Betagraphics Netherlands PRACE - RI

40 9.2 Evaluation form This evaluation form was delivered to the symposium attendees and later mailed to every participant. Dear Participant, Thank you for attending the DEISA PRACE Symposium 2009, we hope you have enjoyed this event so far. Please take a few moments to complete this short questionnaire. Please return it to the registration desk. FEEDBACK FORM Name: Institution / dept.: Do you wish to receive news from DEISA and PRACE projects: YES NO Contact information will only be used for the projects. I. OVERALL IMPRESSION What is your overall impression of the event? Very productive 1 Productive 2 Not productive 3 PRACE - RI