WarpX is a general purpose electromagnetic particle-in-cell code that was originally designed to run on many-core CPU architectures. We describe the strategy followed to allow WarpX to use the GPU-accelerated nodes on OLCF's Summit supercomputer, a strategy we believe will extend to the upcoming machines Frontier and Aurora. We summarize the challenges encountered, lessons learned, and give current performance results on a series of relevant benchmark problems.
@article{arxiv.2101.12149,
title = {Porting WarpX to GPU-accelerated platforms},
author = {A. Myers and A. Almgren and L. D. Amorim and J. Bell and L. Fedeli and L. Ge and K. Gott and D. P. Grote and M. Hogan and A. Huebl and R. Jambunathan and R. Lehe and C. Ng and M. Rowan and O. Shapoval and M. Thévenet and J. -L. Vay and H. Vincenti and E. Yang and N. Zaïm and W. Zhang and Y. Zhao and E. Zoni},
journal= {arXiv preprint arXiv:2101.12149},
year = {2021}
}
Comments
11 pages, 5 figures, accepted by Parallel Computing. Minor revisions, results unchanged