3 ms·

Nice! If I'm not wrong, under many conditions this derivation gives even faster divisibility test then the one presented in the article we comment.