Floating point operations might be limited in precision (and range to some extend), but pretty much any you'll normally encounter will be O(1). Unless you require arbitrary precision.