When people consider float vs int, performance is seldom a relevant factor. You have to ask, what do you even mean if you use floating point for representing m

When people consider float vs int, performance is seldom a relevant factor. You have to ask, what do you even mean if you use floating point for representing money? If you use floating point, the whole point is that you want the behaviour 1e100 = 1e100 + 1 This is exactly what you want for many analysis and decision making problems, but it is a dangerous default.