Speaker
Description
Furthering our understanding of many processes in plasma physics, including laser-wakefield acceleration and laser-driven ion acceleration, requires large-scale kinetic simulations using particle-in-cell (PIC) codes. However, these simulations are extremely demanding, requiring that contemporary PIC codes be designed to efficiently use a new fleet of exascale computing architectures, which are increasingly GPU based. We discuss a GPU algorithm for PIC codes which we implemented on the code OSIRIS [1]. A limited-feature production code based on CUDA C is complete. Our implementation features dynamic GPU-GPU load balancing and a custom memory-management scheme which enables safe utilization of maximal device memory.
[1] Fonseca et. al, 2331 LNCS (2002)
[2] Miller et al., Computer Physics Communications, 259, 107633 (2021)
Acknowledgments
Work conducted under the auspices of DOE, LLNL, NSF, LLE, and DEPS