You are out a bit late with this one. Today most of the professionals (as far as I know) write everything but the really time critical stuff in C.
My projects vary from 100% C to 65% C and 35% assembler. 16 bit multiply can not be done in the time critical stuff anyhow, so an asm function for that would not be interesting to me.
In the olden days I had a library of asm routines; however, I rarely used it as a library, most often as a "coding template" since with 12 12MHz clocks you usually had to "tune to the application".

Erik