Finite strip method construction analysis program execution speed improvement on an MPI cluster by using multiple network links

The execution of the Finite strip method construction analysis program on an MPI cluster has bottlenecks. In this paper we are addressing the problem of communication speed by using multiple network links. The comparison of the program execution times is given in cases of cluster nodes using one and two network interfaces, varying the number of nodes, types of algorithms and inputs. At the end analysis is provided, showing advantages and shortcomings of this solution.