The following attachment has the same divisions as before (along with divu24), but has been optimised to take up less flash when more than one divxx() function is being used. The modification has no impact on execution time - all functions take the same amount of clock cycles as before.
Its experimental because I have only tested it in one sketch and I am not sure if I have told the compiler everything it needs to know to be able to reproduce the same code in all situations (register choices).
FastDivision (Experimental).zip (2.18 KB)