HET published its final recommendations in January 2007. Apart from general recommendations and proposals for reviewing and funding procedures, they also published a document on scientific areas in Europe that need top-of-the-line supercomputer capacity.
European computational science needs computing capacities that match the fastest systems in the USA and Japan, HET says. It is an intolerable situation for the competitiveness of European computational science that this currently is not the case. It is no just the big iron that is needed, There is also a need for petaflop software, scalable algorithms and competent scientists. And the petaflops systems need to be integrated into a HPC eco system.
The HPC ecosystem can be described in terms of a performance pyramid. At the top isa small number of top-of-the line supercomputer systems, funded through national source, with additional European funding. The middle layer of the pyramid consists of a number of national and regional supercomputers. These still should be powerful supercomputers being able to run all the load below petaflop/s level.
The bottom of the pyramid consists of local supercomputers that should enable the development of a strong competence base of computational scientists.
The HET considers a "European Research Grid" desirable that will connect the systems at all levels. The Research Grid has a broader purpose that just supporting number crunching. It is envisaged to also support data intensive application and even commercial research.
Because competition in science is international, the infrastructure available to scientist should also be of international level. Computing performance is one important factor in achieving competitiveness, thus establishing a sufficiently powerful European infrastructure is a necessity, HET says. HET sees that today there exists a gap between European resources and the advanced systems available in the USA and Japan. In addition investments in scientific software development and code optimzation, middleware development, data repositories and efficient networking are needed to stay competitive. But, if that is all done well, petascale computing could become a competitive advantage for Europe, HET thinks..
You do not build a sustainable HPC ecosystem for Europe over night. It has to grow. The Frame Work 7 (FP7) resurch funding programme from the European Commission could be one instrument that facilitates this growth.
All current and future stakeholders in HPC should gradually work more and more together. This includes existing national infrastructures, scientists using the services, nations funding computing infrastructure and computing centers providing services. HET also sees an important role for the European industry as a user and producer for parts of the supercomputing environment and for the European Union and current European Grid projects.
The HET identified, however, a set of open questions. It proposes efforts in these areas and alresy proposed a solutions that could be used as a starting point, included funding models, peer review processes and organisation of the European level resources.
To buid a sustainable HPC ecosystem, HET makes a number of observation:
o Integration and compatibility with national infrastructures
o Building on the current HPC-related work, such as a successful European research infrastructures projects
o An appropriate level of communication, experience and skill sharing between the national centres
o Co-ordination by an European group with a strong mandate
o Financial support from the European Union
o Acknowledgment of the importance and contribution of computational resources in increasing the output of a number of science areas
o Creating sustainable research groups in HPC based both on current experienced ones, and establishing new ones
o Creating sustainable application development and optimization groups based on both current experienced ones, and establishing new ones
Here we summarize the recommendations from HET. The
Recommendation for the development and operation of a “top end” infrastructure.HET recommends establishment of a small number of European HPC facilities to provide extreme computing power - exceeding petaflop capability - for the most demanding computational tasks. In addition, a funding model based on national investments with an additional European share and a possible new collaborative entity governing it is recommended. Construction principles for European facilities are proposed to include tight integration in existing infrastructure targeting sustainability, optimal synergy and efficiency in operations.
Recommendation for developing the full European ecosystemHET recommends increased emphasis on the development of the full HPC ecosystem, including the local infrastructure, national and regional facilities, top-level European computing capabilities and the interoperability of their services. HET recommends growing investments on the different levels of the computational pyramid with a target to provide services in each level in a balanced way: few facilities dedicated to providing the extreme resources for top of the pyramid with a strong base of medium-size systems and integration with local infrastructures and competencies.
Recommendation to implement an efficient and highly reliable infrastructure for storing large amounts of dataHET recommends increasing emphasis on permanent and persistent data repositories as a part of the HPC ecosystem.
Recommendation to support competence development in computational science through extensive training and education activitiesHET recommends supporting extensive training and education activities which focus on enabling more efficient and higher quality use of the top-end facilities in the long run, activities such as computer-based modeling, scalable code development, integration work for the existing national infrastructures and training effort for competence development in computational sciences in general.
Recommendation to raise the visibility of HPCHET recommends strong activities to increase the visibility and improve the publicity of computational science in order to highlight the strategic impact and need of numerical simulation for most areas of science and engineering
Recommendation to boost collaboration
</em> HET recommends support for collaborative actions with a target to link the major players in HPC Ecosystem - existing grid and HPC projects, national and regional centers, main computational research groups, funding organizations and potential new planned FP7 efforts - in order to maximize synergy for actions
Recommendation to support European industryHET recommends support for collaborative efforts with European industrial HPC users and European HPC industry at large - from hardware and software R&D to product design and manufacturing.
The HET recommendation report gives additional remarks for each recommendation.
<Science that could benefit
Is it all worth the effort? Whil science benefit from European HPC ecosystem. The HET identifies several scientific fields where that could be the case:
- Weather, climatology and earth sciences
- Astrophysics, HEP and plasma physics
- Materials science, chemistry and nanoscience
- Life sciences
As an example, the report"Scientific case for a European HPC infrastructure identifies a number of the codes dealing with weather, climatology and earth sciences that are well positioned to make efficient use of a world-class supercomputer. Thes eapplications have already been tested on the Japanese Earth Simulator or on BlueGene computers with thousands of processors. It is estimated that it would take about 6 months time to adapt these applications to a new system. If a small system of the same architecture as the new supercomputer is available in 2008, the applications can be ready for the big system in 2009.
So before the end of the decade Europe could have a number 1 supercomputer in the world and having doing it useful scientific work.