Some confusion
I found that there is a problem in Max gen that puzzles me, that is, many calculations have implemented security mechanisms to eliminate NaN and Denormal, such as division and modular operation. Since this is the case, what is the significance of the existence of fixnan and fixdenorm modules?
i haven tried it, but it is surely possible to get division or modulo output nan when the input was nan already.
then there can be bugs in updates... or third party objects... so always better to have filters at your disposal.