PIConGPU - Bringing a GPU-driven PIC code to large machines

Bussmann, M.; Burau, H.; Widera, R.; Hübl, A.; Debus, A.; Kluge, T.; Schramm, U.; Cowan, T. E.; Schmitt, F.; Juckeland, G.; Nagel, W.; D'Humières, E.; Kilian, P.; Ganse, U.; Siegel, S.; Spanier, F.

PIConGPU - Bringing a GPU-driven PIC code to large machines

Bussmann, M.; Burau, H.; Widera, R.; Hübl, A.; Debus, A.; Kluge, T.; Schramm, U.; Cowan, T. E.; Schmitt, F.; Juckeland, G.; Nagel, W.; D'Humières, E.; Kilian, P.; Ganse, U.; Siegel, S.; Spanier, F.

We present recent results on PIConGPU, a charge-conserving 3D relativistic particle-in-cell code running on graphic processors. We discuss best practices on how to implement the particle-in-cell algorithm on this new hardware and run it on large GPU clusters. We show that these codes have become mature enough to be considered for real life applications, delivering fast response time even for large physical problems.

Keywords: particle in cell; simulation; gpu; relativistic; parallel computing; laser; plasma

Lecture (Conference)
DPG-Frühjahrstagung der Sektion AMOP (SAMOP), 12.-16.3.2012, Stuttgart, Deutschland

Permalink: https://www.hzdr.de/publications/Publ-17380