Abstract
New discrete-event formulations of physics simulation models are emerging that can outperform traditional time-stepped models, especially in simulations containing multiple timescales. Detailed simulation of the Earth's magnetosphere, for example, requires execution of sub-models that operate at timescales that differ by orders of magnitude. In contrast to time-stepped simulation which requires tightly coupled updates to almost the entire system state at regular time intervals, the new discrete event simulation (DES) approaches help evolve the states of sub-models on relatively independent timescales. However, in contrast to relative ease of parallelization of time-stepped codes, the parallelization of DES-based models raises challenges with respect to their scalability and performance. One of the key challenges is to improve the computation granularity to offset synchronization and communication overheads within and across processors. Our previous work on parallelization was limited in scalability and runtime performance due to such challenges. Here we report on optimizations we performed on DES-based plasma simulation models to improve parallel execution performance. The mapping of the model to simulation processes is optimized via aggregation techniques, and the parallel runtime engine is optimized for communication and memory efficiency. The net result is the capability to simulate hybrid particle-in-cell (PIC) models with over 2 billion ion particles using 512 processors on supercomputing platforms.