Using a novel "gender-based" genetic algorithm, GGA is a state-of-the-art parameter tuner for the automatic configuration of solvers. GGA can handle discrete, continuous and real parameters and has been used to tune solvers for a variety of problems, such as satisfiability (SAT), set covering and mixed integer programs (MIPs). GGA represents the parameters of a target solver as a "parameter tree", a type of and-or tree that expresses the relations of parameters, and exploits the tree structure during optimization.
Source code is not currently available, but binary copies of GGA can potentially be distributed. Please contact me for more information. Alternatively, you can write your own parameter tuning using the algorithm given in our 2009 CP paper.