Contents
Editorial
Tribune
Guest contributor
Hot topics
Business cases
Solutions
Experts voice
At a glance
Quick poll
Events
PDF version
 

Subscribe to
Bull Direct:


 

Archives
N°32  |  December   2008
Guest contributor
“Europe at the forefront”
Professor Thomas Lippert, Director of the Supercomputing Center in Jülich.

T. Lippert

Could you please say a few words about the Jülich computing center?
The Jülich computing center is one of the largest supercomputing centers in the world. As a national computing center, it is devoted to supporting universities, research groups and industry in Germany.
It forms an integral part of a very large-scale facility, the so called Forschungszentrum Jülich. This means Jülich is working on major challenges such as energy, the ageing population – a very important research issue these days – and materials science. All these areas naturally benefit hugely from computer simulations, or depend on them to a very large extend. Without computer simulation, there would be much slower progress in all these fields. But for the major part, we provide supercomputing time to organizations in Germany – and Europe – on some of the most powerful supercomputers.

 

Can you tell us a little about the major milestones of your JuRoPA project?
JuRoPA stands for ’Jülich Research on Petaflops Architectures‘. JuRoPA is actually a computer system delivering in excess of 200 teraflops, which can be seen as a single system capable of being dedicated to a single problem, such as the great challenges that we have been talking about. Systems or problems are translated into application code that has to run for a very long time, acting on a huge number of degrees of freedom. Which is why you need such large systems; systems that scale-up to – in this case 16,000 compute cores, or 2,000 dual-processor nodes, each of them consisting of 4 cores that all work coherently together, with a very strong interconnect network, on one and the same problem. This is the real challenge we face now, because if it succeeds – and I am sure it will – it will be the largest truly scalable system of this kind. And it is based on off-the-shelf components, in the sense that together Bull and we have chosen the best components; but we still have to integrate these components to give a system that works coherently on these kinds of problems. For me, as a physicist, complexity is the striking feature common to applications as for instance the great challenges that we are talking about, like energy research, the construction of energy power plants or the construction of ITER(*); and most of these grand challenges depend on research in materials sciences. All of this is extremely complex. We have to deal with a huge number of degrees of freedom, a lot of complicated interactions, a lot of complicated relations, and we have to perform many steps towards optimizations. All this contributes to complexity, and we have to manage that complexity. As a matter of fact, we need an extremely complex system to tackle complexity. We have very complex computing machines; we have machines that consist of a lot of components that have to interact, that have to be controlled, that have to be administered and so on... and this is what we are doing with these supercomputers. We are managing complexity, using highly complex supercomputers.

Is it the reason why you chose Bull to help tackle complexity?
Yes, we chose Bull because we believe that Bull can be a real partner for us. We are not simply a customer of Bull, we are partners. First and foremost, we believe that Bull has proved that they can achieve this, with their earlier success stories. They know how to manage large systems, and they have the skills to do so. And then there is the close proximity between France and Germany, as well as political will to do something here. I have the feeling that Bull will go in the right direction when it comes to developing interesting things, and that’s what I want: the development of technology in Europe.

I would be very happy if together, in Europe, under Bull’s leadership and with contributions from many others, we could build, let’s say, a multi-Petaflops systems by 2012. I believe this is the goal we really should be following, and we should try to be the first to achieve it. Of course, it is extraordinarily complex and difficult. It’s really challenging! But I’m sure that we have a good chance to succeed.

(*) ITER: International Thermonuclear Experimental Reactor

SEND TO A FRIEND
 
Contact  |  Site map  |  Legal  |  Privacy