I'm not sure I exactly understand what exactly you are trying to do, but why couldn't you just divide your Y's by your X's. That's what you do in math and division is relatively quick. If you need decimal places, you can just tack on an extra byte to your Y values and your final result will have 1 extra byte of decimals.
It is relatively quick, but I was hoping for something slightly faster. Still, I guess I don't have to be 100% optimized