Given two elements of a vector space acted on by a reductive group, we ask whether they lie in the same orbit, and if not, whether one lies in the orbit closure of the other. We develop techniques to optimize the orbit and orbit closure algorithms and apply these to give a partial classification of orbit closure containments in the case of cubic surfaces with infinitely many singular points, which are known to fall into 13 normal forms. We also discuss the computational obstructions to completing this classification, and discuss tools for future work in this direction.