Recently I've been benchmarking optaplanner using jmh.

There are combinations of CH, LS heuristics, LS move selectors and datasets. Each test was written to end to 30 sec.

each test is repeated 100 times (there are 10 forks, in each fork 10 warming iterations and 10 measured iterations)

it seems that issue PLANNER-252 has result in slower chaining solutions. When I have results, I will provide them here as attachment