A Hybrid MPI-OpenMP Solution for a Linear System on a Cluster of SMPs

In this paper, we apply a hybrid MPI and OpenMP parallel programming model for a cluster of SMPs to solve a random dense linear system. We compare our solution to HPL, which is a benchmark that applies a pure MPI approach for the same purpose and is widely used for evaluating high performance systems. Our evaluation results show that the proposed hybrid solution achieves better performance than the pure HPL approach in about 10% on our experimental environment.