If it's reasonable to backport the kie-benchmarks for optaplanner with drools score calculation to older release branches, let's do that to immediately get a sense on how much faster/slower optaplanner has gotten in reality.