This page contains a Flash digital edition of a book.
HPC 2013-14 | HPC strategies

Achieving HPC leadership in Europe

Jean-François Lavignon and Jean-Philippe Nominé discuss the role of ETP4HPC

Te driving force behind ETP4HPC, the new ‘European Technology Platform’1


is the promotion of European HPC technologies in order to enhance the general competitiveness of Europe across all fields. Founded in 2012 by Allinea, ARM, Bull, Caps Entreprise, Eurotech, IBM, Intel, Partec, STMicroelectronics and Xyratex, as well as the Barcelona Supercomputing Center (BSC), Te French Alternative Energies and Atomic Energy Commission (CEA), Cineca, Fraunhofer, Forschungszentrum Jülich and LRZ, ETP4HPC now has more than 40 members representing the HPC European ecosystem. Tis platform is open to any organisation conducting R&D on HPC in Europe and is an industry-led stakeholder forum that develops short- to long-term research and innovation agendas and roadmaps. Aſter issuing a ‘vision paper’2


its mission and objectives, ETP4HPC began to explore a programme to strengthen Europe in the HPC field. Aſter discussions with the European Commission DG Connect and European Commission vice president, Neelie Kroes, ETP4HPC’s work has now been set out publicly in its Strategic Research Agenda (SRA): Achieving HPC Leadership in Europe3

. Producing an SRA is effectively

the whole raison d’être for an ETP. Based on the collective vision of those involved in the sector, it sets out a number of paths that could be taken, both in terms of the research itself and how it is funded.

Supporting the ecosystem as a whole While the vision paper defines the ETP’s objectives and missions by outlining the issues to be considered in the overall R&D programme, the SRA sets out a detailed


HPC usage expansion

Extreme scale requirements

menu for the programme by drawing out the main principles for the construction and orchestration of a complete ecosystem. Te first difficulty was the definition of R&D priorities across all HPC technologies. To do this, the ETP members wanted to solicit the opinions of other experts, as well as an expression of user needs from the private sector and from computer simulation soſtware vendors. Te latter represent an essential link in the value chain between technologies and ‘business’ usages in the vast majority of economic areas. As a result, more than 100 contributors were approached in total. ETP4HPC’s interactions with users from

industry and independent soſtware vendors (ISVs) explicitly confirmed the importance of not just producing very large scale supercomputers in Europe, but also medium- sized systems that are robust, affordable, efficient and easily programmable. Te SRA is therefore structured according to a multi- dimensional model, as seen in Figure 1.

A 360-degree vision Te first of the four dimensions shown in Figure 1, HPC Stack Elements, represents the evolution and improvement of the

fundamental hardware and soſtware elements that go to make up high-performance computing systems of any size and scale. Te stack of these elements and sub-systems is mainly made up of hardware components and architectures (processors, memory, network, etc.), system and management soſtware (OS, configuration management, resources management, etc.) and programming environments (taking into account massive parallelisation and with a special focus on API standardisation). Te second dimension, Extreme Scale

Requirements, identifies the conditions needed

“ETP members wanted to solicit the opinions of other experts, as well as an expression of user needs from the private sector and from computer simulation soſtware vendors”

to successfully move to exascale and beyond. Te integration of elements from the first dimension at these levels requires a global, transverse vision of the many problems that are exacerbated at such large scales, such as energy efficiency, resilience, balancing performance between computation, communications, and storage. Te third dimension, New HPC

Deployments, encompasses new usages of HPC – big data, the cloud, embedded computing, real-time, etc. – and defines ways of stimulating and supporting them. Tis dimension is distinguished by its

Figure 1: the multidimensional vision and the technical domains

highly technical content. Various developments are needed, for example when it comes to algorithms (data process, extracting meaning and value, etc.) and system tools (managing remote access modes, virtualisation, management of workflows, etc.)

New HPC deployments

HPC stack elements

Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36