Nonmonotonic description logic programs are a major formalism for a loose coupling of rules and ontologies, formalized in logic programming and description logics, respectively. While this approach is attractive for combining systems, the impedance mismatch between different reasoning engines and the API-style interfacing are an obstacle to efficient evaluation of dl-programs in general. Uniform evaluation circumvents this by transforming programs into a single formalism, which can be evaluated on a single reasoning engine. In this paper, we consider recent and ongoing work on this approach which uses relational first-order logic (and thus relational database engines) and datalog with negation as target formalisms. Experimental data show that significant performance gains are possible and suggest the potential of this approach.