Information für die Wissenschaft Nr. 57 | 1. Oktober 2014
Priority Programme “Software for Exascale Computing” (SPPEXA, SPP 1648)
Trilateral Call for Proposals – 2nd funding period
The Senate of the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) has established the Priority Programme "Software for Exascale Computing“ (SPPEXA, SPP 1648) in 2011. The programme is designed to run for six years; the present call invites proposals for the second three-year funding period 2016 to 2018 and is now opening up to foster international collaboration. The call is intended to support collaborative projects of bi- or trilateral research teams, bringing together researchers from France (ANR), Germany (DFG), and Japan (JST).
Idea and Objectives
The Priority Programme addresses fundamental research on the various aspects of High-Performance Computing (HPC) software, which is particularly urgent against the background that we are currently entering the era of ubiquitous massive parallelism. Exascale performance, i.e. 1018 FLOP/s, and beyond will require massive parallelism realised by many-core processors assembled to systems beyond 107 processing units. With this, several challenges come along, such as efficient treatment of I/O, resilience or load balancing. Mastering the various challenges related to this paradigm shift from sequential or just moderately parallel to massively parallel processing will be the key to any future capability computing application – e.g., large-scale simulations – at exascale, but it will also be crucial for learning how to effectively and efficiently deal with commodity systems of the day after tomorrow for smaller-scale or capacity computing tasks – and it is the overall scientific objective of SPPEXA.
To this end, SPPEXA re-connects several relevant sub-fields of computer science with the needs of Computational Science and Engineering (CSE) and HPC. In the first funding period, SPPEXA established the framework for a much closer cooperation and a much more co-design-driven approach – instead of a merely service-driven collaboration of groups focussing on fundamental HPC methodology (computer science or mathematics) on the one hand with those working on science applications and providing the large codes (science and engineering) on the other hand. Topically, SPPEXA will continue to drive research towards extreme-scale computing in six areas or research directions:
- computational algorithms
- system software and runtime libraries
- application software
- data management and exploration
- software tools
Hardware peak performance is ever increasing, exascale systems are currently predicted for around 2020, and there is a world-wide consensus that a “racks without brains” strategy cannot allow the science communities to exploit the huge potential of the computational approach in a massively parallel world. Against this background, SPPEXA provides an ideal environment for bundling research activities on a national as well as, via intra-project or inter-project collaboration, an international level and for enabling the participating groups to significantly advance the state of the art in HPC software technology.
SPPEXA has been implemented as a DFG Priority Programme, but it is different from others in funding source and volume, in the range of disciplines involved (informatics, mathematics, natural sciences, and engineering sciences), in the clear orientation towards a set of time-critical objectives, in both more and more intense structural/governance measures taken, and in an explicit internationalisation step taken for the second funding period. The latter is considered to be especially promising and important against the background of the current world-wide activities in high-performance computing shaping the science of tomorrow.
SPPEXA’s projects are small interdisciplinary research consortia, currently 13, typically consisting of three to five groups or researchers, respectively. These researchers must represent different scientific fields and contributions to CSE and HPC, and they typically (but not necessarily) come from different institutions. The involvement of international research partners has already been possible in the first period and has been effected on an individual, non-coordinated basis. For the second period, there is now a joint initiative of the DFG with JST (Japan) and ANR (France) to particularly encourage and foster consortia with research groups from France, Germany, and Japan.
Each SPPEXA project must cover at least two and focus on at most three of the above six sub-topics, thus incorporating not only the networking of disciplines, but also the networking of HPC-relevant issues, while nevertheless ensuring depth. Most importantly, all proposals have to clearly address SPPEXA’s overall scientific goal: to master the various challenges related to this paradigm shift from sequential or just moderately parallel to massively parallel processing. Although this is no restriction to extreme-scale applications that need and are able to benefit from cutting-edge core numbers, it does exclude moderately parallel scenarios that do not contribute to advance the frontier of parallel computing.
As a DFG programme, SPPEXA aims at fundamental research and offers space for both more evolutionary developments (e.g., to build a basis for future definition of standards) as well as completely new, revolutionary approaches. This clearly distinguishes it from other related activities. Since SPPEXA aims at including the relevant expertise in a broad way, we expect individual researchers or research groups to participate in more than three proposals only in exceptional cases.
French scientists are eligible for funding according to ANR rules for funding. However, their participation is only allowed in trilateral project consortia.
Japanese scientists who are Principal Investigators in one of the JST-CREST Research Areas (please refer to the link below for details on CREST) are eligible for funding according to JST rules for funding. However, their participation is only allowed in bilateral (Japanese-German) or trilateral project consortia and the research supervisors of the respective CREST projects need to consent.
Scientists working in Germany are eligible for funding according to the rules for funding in DFG Priority Programmes.
Scientists based outside France, Germany, and Japan cannot act as Leading Principal Investigators of a proposal and are required to secure their own budget. Proposals that include such applicants must contain an alternative scenario addressing the consequences of an unsecured budget, including but not limited to, the consequences of their possible withdrawal from the proposal consortium. Researchers based outside Germany, France, and Japan can exceptionally be funded by DFG if their project contribution is of added value to the Priority Programme at large. This must be explained in the proposal and approved by the SPPEXA scientific coordinators.
The duration of full research grants is to be three years for Principal Investigators funded by ANR or DFG. The duration of research grants for Principal Investigators funded by JST is limited by the duration of the associated CREST Research Area.
The proposals should conform to programme aims and designated research fields as specified in this call. Proposals should adhere to the following review criteria:
- scientific quality and innovativeness of the joint research plan
- sound project management, methodological approach, feasibility and appropriateness of the joint research plan
- added value to be expected from the research collaboration for each country involved
- competence and expertise of teams and complementarity of consortium members (interdisciplinarity/availability of necessary expertise)
- appropriateness of resources and funding requested
- expected impacts: e.g., scientific, technological, economic, societal
- opportunities for early career researchers
Submission of Proposals
Proposals for the second three-year SPPEXA funding period are now invited. All proposals must be written in English and submitted via DFG’s electronic proposal processing system “elan” by:
31 January 2015, 24:00 CET
In this system please select ”SPP 1648” when submitting your proposal (second screen after “start online form”). Please include a title page with all applicant names, their affiliations, and the project title in your application. Proposals by two applicants must not exceed 20 pages. Proposals may comprise five additional pages for each additional applicant. You are further asked to send a PDF file of the proposal to the scientific coordinators of the Priority Programme.
An international review panel will evaluate the proposals. This may include a proposal colloquium in Bonn, where applicants may present their projects. Funding decisions are expected in October 2015. Funding can be started in January 2016.
Further information is provided on SPPEXA’s homepage:
For details on JST-CREST Research Areas:
DFG’s electronic portal “elan” can be found at:
For further information on “elan” please refer to:
Please note: Those Principal Investigators applying for funds who do not have an “elan” account need to register by 29 January 2015, 14:00 CET, specifying personal and address details. The registration will be confirmed within one business day. After this, access to “elan” is permitted.
Use firstname.lastname@example.org for sending your proposal (PDF file) to the SPPEXA scientific coordinators.
For proposal guidelines and preparation instructions: Please refer to DFG forms 50.05en and 54.01en on
(the English version can be found below the German version and in “elan” as well).
Questions related to the scientific objectives and strategic intentions of SPPEXA including the requirements for funding should be directed to SPPEXA’s scientific coordinators:
- Prof. Dr. Hans-Joachim Bungartz,
Technische Universität München,
Institut für Informatik,
85748 Garching bei München, Germany,
- Prof. Dr. Wolfgang E. Nagel,
Technische Universität Dresden,
Zentrum für Informationsdienste und Hochleistungsrechnen,
01062 Dresden, Germany,
Administrative questions (formal eligibility, submission in “elan” etc.) should be directed to:
- Bettina Schilling,
- Dr. Marcus Wilms,
Deutsche Forschungsgemeinschaft (DFG),
53175 Bonn, Germany
Administrative questions regarding French participation should be directed to:
- Dr. Béatrice Arnulphy,
- Dr. Jean-Yves Berthou,
Agence Nationale de la Recherche (ANR),
50, avenue Daumesnil,
75012 Paris, France
Administrative questions regarding Japanese participation should be directed to:
- Dr. Yoshi-aki Shimada,
phone: +81 3 3512-3526,
Japan Science and Technology Agency (JST),
Tokyo Headquarters (2) K’s Gobancho,
Chiyoda-ku, Tokyo 102-0076, Japan