Interview: Serge Bogaerts

Serge Bogaerts was appointed in February 2017 as Managing Director of PRACE. With an already extensive knowledge of the inner workings of PRACE gained in his role as the Secretary of the PRACE Council, Bogaerts is looking to help the organisation evolve in the coming years. We spoke to him about his new role and how he envisions PRACE meeting the needs of Europe.

Congratulations on your new role as Managing Director of PRACE. What does the role entail, and what experience do you bring?

Before starting as PRACE Managing Director, I was involved in the operation of a Tier-1 supercomputer in Belgium, and was the Belgian delegate on the PRACE Council, so I already had a good understanding of how PRACE works. The research centre I was working for was Cenaero, an applied research centre in which we were trying to increase the industrial uptake of High Performance Computing (HPC) globally, helping companies to invent and design more competitive products through the provision of high fidelity numerical simulations.

For several years, PRACE had no permanent Managing Director. I think the most important part of the role is to provide permanent and local management to the organisation. A lot of this will involve day to day management, and replacing the members of the Board of Directors who were having to support the organisation remotely. They were doing a great job, and I am humbly trying to take over their responsibilities.

On a higher level, the Managing Director will provide an interface between the operation of the organisation and other bodies. For instance, I will be reporting directly to and receiving recommendations from the PRACE Council, the body that decides which programmes and projects we implement. There are also several other bodies, such as the Strategy Working Group (SWG), which is a subset of the Council that prepares the perspective and the decisions of the Council. Then, we have the Scientific Steering Committee (SSC) and the Industrial Advisory Committee (IAC), which respectively advise the organisation in terms of scientific strategy and policy, and work out ways to serve industry. Last but not least, I am chairing the Board of Directors, the executive body of the organisation.

Another aspect of my role is to interface with the outside world. I will, together with the Board of the Council, be involved in discussions with the European Commission, the scientific community, and other HPC related organisations, stakeholders, and associations.

You are currently overseeing the transition from PRACE 1 to PRACE 2. How is this going, and what changes will we see?

PRACE 2 is a transition with respect to PRACE 1, and the two models are now occurring in parallel. We are already allocating computing resources on machines under the PRACE 2 model, while “using up” the resources that were earmarked for PRACE 1.

PRACE has been quite successful in distributing resources to excellent science, to ensure that the projects selected focused on the impact on science. The PRACE Access Committee (AC), the body that selects the projects to be awarded, has been doing a tremendous job to bring higher standards to this process. The most recent goal of the AC has been to ensure that all of the hundreds of projects submitted to the most recent Call for Proposals for Project Access were reviewed on an equal basis. The projects are rated purely on their scientific excellence, and I am very pleased that this improvement was implemented as part of the PRACE 2 programme. We have learned lessons from the first edition of this exercise and we are already trying to put in place further improvements. We will keep doing this from Call to Call.

Another aspect is that, as part of the PRACE 2 programme, we have what we call High Level Support Teams (HLST). These teams are there to support scientists to improve the performance of their codes. Because it involves some legal and contractual work, effort is needed to put this in place and it is now being progressively implemented. We hope that the scientists will see the benefit of this the coming year.

At the recent PRACEdays17 conference, PRACE signed a Memorandum of Understanding (MoU) with XSEDE and RIST to strengthen ties and improve international cooperation. What will be the nature of these relationships, and how else is PRACE working to broaden its influence globally?

The focus of Memorandums of Understanding such as this one, is to exchange information, depending on the interests of the different parties. For instance, PRACE is trying to get industry to be aware of the advantages HPC. I met delegations from RIST soon after I started and we had very interesting exchanges on how to get that into operation. How do we get industry to use the infrastructure, and how do we make this compatible with our mission to support science, are questions of great interest. We also have to make sure that we are not competing inappropriately against commercial actors in the HPC business. So we exchange our experiences and best practices on implementing and running programmes, and that gives us an idea of the evolution of the HPC landscape globally. It is very important for us to continue this exchange of information. I imagine that in the near future we will exchange information on how we train people in HPC. We are collaborating on this already, but I would like us to go a step further and deepen our collaboration with other groups, depending on the history and the context of the different organisations. We are always looking for ways to enhance international collaboration, mainly to ensure that scientists have easy access to different resources. We try to align our different programmes or at least reduce the trouble that scientists have to exchange and collaborate across the different continents.

Depending on the involvement of different organisations, we would like to develop ideas for the future on how to treat data and how to set up services to process data. Upcoming state-ofthe-art experiments such as the Square Kilometre Array (SKA) telescope will be generating enormous amounts of data, and HPC resources will be needed to support the processing of this. It will be very important for us to make sure that we contribute adequately with appropriate services within the international frameworks. So this is another topic that we will discuss with other organisations.

What sort of training does PRACE offer to those interested in working with HPC?

PRACE has been very successful in this respect through its six Advanced Training Centres (PATC), which offer a full range of training to scientists and industrial actors. We want to extend the basis of this training by getting more local activities started and so we are now in the process of implementing four PRACE Training Centres (PTC), which will aim to help people with their first steps in high performance computing before potentially later going to the PATC courses.

But training is just one aspect of our activities in this area. We want to be able to support scientists in doing their science and without the need to invest too much time in the computational aspects. HPC is often about setting up multi-disciplinary teams to actually solve problems. We have teams that provide focused support to enhance the performance of computer codes to solve specific problems. Essentially, we want to improve the usability of these infrastructures for scientists. In an ideal future, the situation would be that the scientists don’t actually have to care about how the problems are computed, so they can get on with the science. But we may be a decade away from that at the moment.

Has the SHAPE programme had the impact on SMEs that was hoped for so far? How will this continue into the future?

The SME HPC Adoption Programme in Europe – SHAPE – is one of the programmes we have implemented to answer the need to support HPC for industry in Europe, and so far we have supported 33 SMEs via this programme. The focus of SHAPE is to give SMEs a chance to test ideas and to see whether HPC can give them a competitive advantage in their area or with their products. We can help them in a number of ways: to develop new products, to reduce their time and cost of getting to market, and to improve the safety of their products and services.

We are considering the possibility of adapting this model, pushing it further to make sure that we improve the penetration of this programme towards different actors. This will be part of PRACE’s efforts to contribute to a better HPC ecosystem. Although it has already had some success, we believe we can scale it up further and bring it to another level.

“We want to provide scientists in Europe and the rest of the world with a world-class high performance computing infrastructure. PRACE has just entered its second phase but this is just a small step before a significant leap to take over the coming years”

To conclude, can you summarise your main aspirations for PRACE as the new Managing Director?

To start, I would say that I want to build on the current success of PRACE in supporting excellence in science. We want to provide scientists in Europe and the rest of the world with a world-class high performance computing infrastructure. PRACE has just entered its second phase but this is just a small step before a significant leap to take over the coming years.

I would also like to see PRACE to go a step further, contributing to the development of a fertile HPC ecosystem in Europe. We have been focusing a lot on excellence in science. I think that we have room to offer additional support to the HPC ecosystem and I will certainly be attentive to that in terms of the industrial aspect: I think we need to enhance our connection to industry extensively and continuously.

Here, I would like to include the operational performance of the organisation. Without a permanent, local manager, the organisation has been doing its best to be operational. I hope that my presence will improve the efficiency of PRACE, by putting in place better processes to manage day to day operations.

Generally speaking, I would like PRACE to consider how it carries out its activities. I think that we should be more agile in our capacity to provide services adapted to communities, rather than have a single model and be too focused on that. This is actually already under discussion with the European Commission. PRACE is seen as one of the main actors of the European Data Infrastructure (EDI), and I want to ensure that it retains its function as a cornerstone of EDI and that the European Commission and other European actors recognise PRACE as such.