Hi, I’m Jean-Luc Vay from Lawrence Berkeley National Laboratory. I’m the PI [principal investigator] of one of the exascale computing projects, and we develop code to simulate a new type of particle accelerators that have the potential to be much shorter and cheaper than conventional accelerators.
What is very exciting to us about the Exascale [Computing] Project is that it enables us to develop new tools on the latest computer architectures and the GPUs at unprecedented scale that will really enable us to compute these complicated particle accelerators in a way that absolutely was not possible before.
The type of particle accelerators—called plasma accelerators—that we simulate are very hungry into computer power. They’re very complex. And looking forward to developing this technology for many applications to society, there’s just no way that we can develop these applications in a reasonable time without large computers. GPUs are essential because one needs to be fast. Large computers, supercomputers, are essential because some of the simulations we have to do have to be very precise and large. In addition, we need to run ensembles, not just large simulations, but many smaller simulations in order to design these accelerators efficiently. The Exascale [Computing] Project definitely gives us the tools, the hardware tools, and also enables us to develop the software that we need. And this is a good stepping-stone toward ever-larger supercomputers that activity needs to go forward for these particle accelerators.