Yeah :( The formula is giving a value that's below the real one, so it thinks one CSII won't finish it, but it does. A few updates ago I implemented a cache for the inverse situation, when the formula gives a value higher than the real one, it'll cache the real value obtained and use that...