This reminds of of a similar calculation I did in college when thinking about big computing problems.. I called it "hurry up and wait".
Assuming Moore's law continues (computing power/$ doubles every 18 months), and you have a big computation to do, should you start now or should you wait?
If you have $1000 to spend on a computer and a problem that's going to take 3 years for it to compute, you can actually wait 1.5 years before you start. The computer you buy then will be twice as fast, and you'll finish at the same time.
On any problem that takes more than 1.5 years, the fastest thing to do is just wait until you can afford a computer that can complete it in 1.5 years.
Assuming Moore's law continues (computing power/$ doubles every 18 months), and you have a big computation to do, should you start now or should you wait?
If you have $1000 to spend on a computer and a problem that's going to take 3 years for it to compute, you can actually wait 1.5 years before you start. The computer you buy then will be twice as fast, and you'll finish at the same time.
On any problem that takes more than 1.5 years, the fastest thing to do is just wait until you can afford a computer that can complete it in 1.5 years.