Abstract：Generalized inverse matrix(GIM) is a new force method based on the generalized inverse matrix theory，which is an iteration method for solving material nonlinear problems. Unlike the classical force method，GIM does not need to consider the basic structure，so it is also feasible for computer to calculate. This method brings the force method a new light in the computer calculation field. GIM has some natural parallel-calculating characteristics，which are different from the classical substructural algorithm. In GIM，the system control equations are divided into a linear group，which includes the equilibrium equation and the compatibility equation，and into a nonlinear group，which includes the constitutive equation. The iteration procedure starts from a special solution of the equilibrium equation which is achieved by using the GIM theory. The whole iteration procedure can be divided into the global stage and the local stage. In the global stage，the linear group is used，while in the local stage，the nonlinear group must be considered. The calculation can be perform in a parallel way according to the time in the global stage and the space in the local stage. The parallelization process of GIM is present. Several different methods of parallelization are given and compared. The parallelization is considered in the cluster system with message passing interface(MPI). MPI is a kind of parallel environment which is widely used nowadays. Cluster system is a system of many computers linked by high-speed network. The parallelization of GIM in such an environment works well，which proves the parallel-calculating characteristics of GIM to have better performance. The ratio of the time between communication and calculation is an important key during the parallelization. And the parallel program must be compatible with both the hardware and software environment of parallelization.
