hPIC2: a performance-portable, hybrid plasma code for dynamic plasma-material interactions
POSTER
Abstract
High-performance computing architectures are evolving to become increasingly heterogeneous, with many of the world's fastest computers now relying on graphics processing units to accelerate computation. Device manufacturers' preferences for different programming models further complicate software development for such systems. Despite the complexity of new architectures, the vastly improved computational efficiency they provide potentially allows plasma codes to investigate more complicated multiphysics problems. To this end, we have developed hPIC2 [1], an electrostatic, hybrid fluid/particle-in-cell code written with Kokkos for performance portability on a range of heterogeneous platforms. hPIC2 supports a widening range of plasma chemistry models and optionally couples to the RustBCA binary-collision approximation code, allowing the efficient simulation of plasma-material interactions. hPIC2 can also use unstructured meshes by exploiting the MFEM finite element method library. Altogether, hPIC2 can be used to efficiently simulate multiscale and multiphysics plasma problems in arbitrary geometries on most high-performance computing systems available today. [1] https://doi.org/10.1016/j.cpc.2022.108569
*Funded through DOE Award DE-SC0018141, DOE NNSA LRGF DE-NA0003960. This research used the Delta advanced computing and data resource which is supported by the National Science Foundation (award OAC 2005572) and the State of Illinois.
Publication: Meredith, L. T., et al. "hPIC2: A hardware-accelerated, hybrid particle-in-cell code for dynamic plasma-material interactions." Computer Physics Communications 283 (2023): 108569.
Presenters
-
Logan T Meredith
- University of Illinois at Urbana-Champaign