However, complicated calculations or anything involving angles or other math functions quickly becomes more convenient when expressed as a Fix64, which is more or less a drop in replacement for float.
I would ideally use Fix64 everywhere, but given the torturous route the C# takes to be transformed into something that's executed on the client machines, my faith in the compiler's ability to generate good code for that is basically zero. I mentally treat long + long as a single instruction, but Fix64 + Fix64 as a function call.
