ABSTRACT
In many fields of plasma physics where the particle-in-cell (PIC) method is used; the progress of experimental technology has paved the way for new physical processes and discoveries, which requires considerably more computing power and memory. PIC codes are now used as a simulation tool in a wide range of applications such as astrophysical and space plasma physics, nuclear fusion, electronics, particle acceleration, and high-intensity laser—matter interaction. Exascale machines are needed to simulate large-scale and accurate 3D physical problems with PIC codes. On former scalar computer architectures, a common way of storing particle quantities in memory was to use array of structures, where each structure contained the different properties of one particle contiguously in memory. The roofline performance model provides a visual analysis of the computational constraining resources of every system from single-core to much cores architecture. Codes using the PIC method are examples of codes that may perform more poorly without adaptations on Many Integrated Core than on previous architectures.
