hmm well what do you call "more efficient" watcom already has assembler for 64bit multiply + divide of uint32_t.. and it works on 8086 too... and in 32bit watcom I dont see the point of trying to optimise printf.

unlike 8bit home machines like the c64, the pc doesnt really have routines in rom per-se, it as the bios and really only uses the disk int13h stuff....

Okay. Thank you.

On 8086, you can divide a 32-bit quantity in DX:AX by a 16-bit quantity, but only if both the remainder and quotient both fit in 16 bits (otherwise it triggers an exception).

You might read up on how compilers might optimize a divide by 5, from which you could derive divide (and remainder) by 10: https://stackoverflow.com/questions/41183935/ … ng-integer-divi

Why don't just use x87? The poor fellows without an NPU can run EMUL87.