Validation and benchmarking of 3d particle-in-cell simulations using multi-device architecture in the Vorpal code

The Vorpal code allows for the simulation of plasmas over a wide variety of conditions, typically solving for the plasma dynamics using particle-in-cell (PIC) and coupled electromagnetics solutions [1]. Recent improvements have been made to flexibly utilize heterogeneous architectures - from many CPUs to many GPUs, or a mixture of both. Computation on GPUs is most effective when algorithms are single-instruction multiple-data, so efficient computing of the entire PIC cycle can be challenging and often problem-dependent. Validation of the updated code was done by comparing simulation results to generally known, standardized test cases [2]. Performance benchmarking of the full PIC cycle was done for various architectures, efficiency and potential bottlenecks will be discussed.

[1] C. Nieter and J. Cary. 2004. VORPAL: a versatile plasma simulation code. J. Comput. Phys. 196, 2, 448-473.

[2] S. O’Connor, Z. D. Crawford, J. P. Verboncoeur, J. W. Luginsland, and B Shanker. 2021. A set of benchmark tests for validation of 3-D particle in cell methods. IEEE Transactions on Plasma Science, 49, 1724-1731.