Abstract: Here we use the CUDD package as our main method to solve the floorplan problem. After surveying some of the functions and implementation of the CUDD package, we choose the functions in the basic CUDD library instead of the procedure library provided by nanotrov, which we need to get the parser and do much more complicated implementation to transform .blif files into the form we want. We use the fast construction of BDD provided by CUDD package and trace them using the test cases, which we generate all possible solutions and then feed into the BDD to find the feasible ones. Since we do the exhaustive search, we confirm that we can get the optimal solution. We implement two project flows, one with exhaustive search and the other with random generation. In the flow 1 we can get the optimum solutions successfully, but the constraints, memory usage and running time, of the CUDD limit the number of modules which we can handle. In the flow 2 we can get the local optimum solutions successfully, but the quality of solutions are depended on the random solutions.