I would not trust the opinion of ICOBench for this site noticed a manipulation of the results in relation to the objects of analysis.