I note maths performed within LDmicro is nominally 16 bit integer work.

While this is true claimed I also see that we can cast a variable to be a longer than 16bits integer value.

Good for multiplication say. When invoking MUL instruction and looking for a 16 bit by 8 bit product how does MUL handle that?

Is it necesary to cast the variable whihc will hold the product into a 3 byte variable before invoking MUL? or is the product automatically cast to the correct size variable?