Why an 8-bit microcontroller needs more than 8-bit arithmetic:
A wide range of control applications uses data with more than 8 bits of precision. Most 8-bit microcontrollers can handle limited, multibyte operations such as addition and subtraction through the use of subroutines and the arithmetic unit's carry flag. More complex operations can be built from these. Tradeoffs include performance and code space for the subroutines. Performance tends to be the biggest limiting factor since doubling the number of bytes in an operation typically translates to much more than doubling the execution time. Reducing execution time means moving to multibyte hardware arithmetic units. Some chips have multibyte pointer registers designed to improve memory access, but they tend to have limited arithmetic functionality.