Skip to Main Content
High-performance processor architectures are moving toward the integration of more processing cores on a single chip. PCs and parallel computer systems will have higher computing performance. It is very meaningful for power system simulation, online stability assessment and control. But traditional serial and parallel software can not fully exploit multi-core's capability without parallelizing restructure. In this paper, parallel simulation algorithms are analyzed and software restructure strategies are studied for multi-core processor platforms. The research indicates that: (1) For PCs, implicit parallel strategy with OpenMP is easier to implement and explicit parallel strategy with MPI or Pthreads is more adaptable to large-scale system simulation. (2) For clusters and distributed systems, MPI+OpenMP hybrid programming is convenient and multilevel partitioning is potential. Both methods provide better speedup.