Accelerated simulation of microwave breakdown in gases on Xeon Phi based cluster-application to self-organized plasma pattern formation

Abstract

Modeling and simulation�of high power microwave (HPM) breakdown, involving complex coupling between high frequency�electromagnetic wave�and plasma, is a computationally challenging and expensive problem due to the stringent numerical criteria. In this article, efficient�parallelization�and�optimization strategies�for two dimensional�fluid simulation�of microwave breakdown phenomena in air/gas on Intel�s Xeon Phi Many Integrated Core (MIC) data�parallel architecture�are being presented. The numerical model used for this study is based on�Finite Difference Time Domain�solutions of�Maxwell Equation�coupled with the plasma fluid�Continuity Equation�(Boeuf et�al., 2010). The optimized�parallel version�of this algorithm using OpenMP on Xeon Phi co-processors achieves a speedup of around 5�138 times (on Knights Corner and�Knights Landing�for different problem sizes) compared to a�serial version�(on Intel i7-4790 processor) in a much more energy efficient way. Moreover, a�hybrid strategy�based on OpenMP and MPI, involving a three-level parallelization (instruction level within SIMD VPUs, thread-level over�many cores�and accelerator level across a cluster of Xeon Phi processors), achieves a speedup of around 1400 (compared to a�serial version�on xeon-phi 7250 processor) on an HPC cluster with 24 Xeon-Phi co-processors. Several fast and accurate numerical experiments have been performed on the Xeon-Phi based system, and the results are illustrated with the example of the formation of a self-organized fishbone like plasma structure during breakdown similar to the images obtained from high power microwave experiments in air (Hidaka et�al., 2008). Numerical experiments show that a good�resource utilization�can be achieved by proper code design,�cache optimization�and good programming practices, and opens up many possibilities for HPM breakdown research which require resource intensive EM-fluid simulations as a precursor at a relatively cheap cost.