Yes, you are seeing the inherent limitations of 64-bit floating point. If you really need more than two decimal places on a number as big as 123,456,789,201,234, you need some sort of extended arithmetic library. This question has some suggestions and discussion.