The Particle-In-Cell (PIC) simulation method is a modern technique in studies of collisionless plasmas in applications to astrophysics and laboratory plasma physics. Inherent to this method is its parallel nature, which enables massively parallel MPI applications which can use thousands of CPU-cores on HPC systems. In order to achieve a good performance of a PIC code several techniques are available. In this work we study the impact of particle sorting on the performance of the PIC code THISMPI. We compare dual-pivot five-way quicksort with the standard quicksort. We focus on finding optimum sorting frequency.
CITATION STYLE
Dorobisz, A., Kotwica, M., Niemiec, J., Kobzar, O., Bohdan, A., & Wiatr, K. (2018). The impact of particle sorting on particle-in-cell simulation performance. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10777 LNCS, pp. 156–165). Springer Verlag. https://doi.org/10.1007/978-3-319-78024-5_15
Mendeley helps you to discover research relevant for your work.