I don't want a textual representation of the number I want:
- For ints: typical bit vector representation of fixed width integers
- For floats: IEEE representation of single-precision floating point number
Outputting it as text bloats the size of the data. For example, 50000 in text takes up 5 bytes, as a 32-bit machine representation it is only 4 bytes. It is even worse for floating point, where the text representation can be very long if the entire precision is output as decimal. Something like 1.4534636 (made up number) takes up 9 bytes as opposed to 4 bytes in machine format, bloating it by more than 100%.