Don't need to see floating point errors