Investigating performance portability of a highly scalable particle-in-cell simulation code on various multi-core architectures