Routing algorithms performance in different routing scopes

Whenever new routing algorithms are proposed, corresponding performance gains are reported. Are these gains the same for different interconnect scenarios? Do the proposed techniques always outperform the previous ones in all possible routing scopes? We found out that the answer for these questions is no. In this work we evaluate several routing algorithms under an extensive set of experiments and different interconnect scenarios. The results show that different algorithms are preferable according to the different routing scopes, process generations and net sizes.