I hope the test data is the same when comparing the different runs. So the big o notation should be the same across different runs.