In order to reduce the error in solving the effectiveness of DEA decision making units (DMU's) and to simplify the computation concerned, this paper applied the linear programming of traditional DEA models as the basis, and eliminated the non-Archimedean infinitesimal in the linear programming of traditional DEA models, which was converted into a linear programming without non-Archimedean infinitesimal, and then, further converted into a simple linear programming by introducing Poset set. At last, a numerical example was given and calculated with this new algorithm. The result reveals that this new algorithm eliminates the non-Archimedean infinitesimal and is able to reduce calculation errors. Meanwhile, the introduction of the Poset set reduces the operations of DEA models.