[drpixie]

Remember the Cray-1. It had no division instruction but could perform reciprocals. Division was performed as 1 reciprocal and 2 or 3 multiplications (depending on the precision needed). Thus division was only a little more complex than multiplication. (A reciprocal instruction took about twice the time of a multiply.)

See https://www.ed-thelen.org/comp-hist/CRAY-1-HardRefMan/CRAY-1...

So really, it's tradition that division (which is needed relatively rarely) is not performed quickly in hardware. The old time versus silicon tradeoff strikes again - you can have it quick but it will be expensive, how much do you want it?

Re the original question - (almost) anything can be done in enough silicon. Both multiplication and division could be done as very fast "ROM" lookups, but that would use an enormous amount of silicon.