On the Computational Complexity of the Range Computation Problem