I agree. It would even be more interesting or reliable if the benchmark could compare a MEMFS made by dokan author and not Winfsp author like here.