Section: New Results

High-Performance Particle-in-Cell Simulations

Participants : Arthur Charguéraud, Yann Barsamian, Alain Ketterlin.

Yann Barsamian's PhD thesis focuses on the development of efficient programs for Particle-in-Cell (PIC) simulations, with application to plasma physics. Typically, a simulation involves a cluster of machines, each machine hosting several cores, and each core being able to execute vectorized instructions (SIMD). The challenge is to efficiently exploit these three levels of parallelism. Regarding the processing on one given multicore machine, existing algorithms either suffer from suboptimal execution time, due to sorting operations or use of atomic instructions, or suffer from suboptimal space usage. We have developed a novel parallel algorithm for PIC simulations on multicore hardware that features asymptotically-optimal memory consumption, and that does not perform unnecessary accesses to the main memory. The algorithm relies on the use of chunk bags, i.e., linked lists of fixed-capacity arrays, for storing particles and allowing to process them efficiently using SIMD instructions. Practical results show excellent scalability on the classical Landau damping and two-stream instability test cases. A paper was published at PPAM [12].