Acceleration of Full-PIC simulation on a CPU-FPGA tightly coupled environment

Hall thruster is a sort of electric propulsion and has been studied in many research institutes. In the design process of Hall thruster, a numerical simulation called Full-PIC (Particle-In- Cell) method is used. Although this simulation provides high accurate result, it is known as a very time consuming job. In this paper, we present a study of acceleration of Full-PIC simulation on a CPU-FPGA tightly coupled environment. A high-load part is selected and off-loaded to an FPGA. Zynq-7000 and Vivado HLS are used for implementation. To optimize the implemented design, every target process was divided into some parts for pipelining and adjustment interval. Three off-loaded subroutines named "field source", "particle att ion" and "particle att ele" achieved 8.53 times, 12.78 times and 14.95 times performance compared with the software execution, respectively. The total execution time of target part is sped up 5.17 times compared with Cortex-A9 667MHz in Zynq.