ELO rating and thinking time.

[Henk]

Is there some kind of mapping between ELO rating and thinking time ?

For instance if your chess program has the same strength as another chess program that has an ELO of 1600 but your chess program uses only 3 seconds per move. What would be the estimated rating of your chess program if it would take 90 seconds ?

[hgm]

A rough aproximation is that Elo2 - Elo1 = 100 ln(T2/T1), which amounts to 70 Elo per doubling. So 30 times as much time (about 5 doublings) would add 350 Elo.

[Henk]

A rough aproximation is that Elo2 - Elo1 = 100 ln(T2/T1), which amounts to 70 Elo per doubling. So 30 times as much time (about 5 doublings) would add 350 Elo.

So a gain of 70 ELO points is a great achievement. I know how difficult it can be to make a program twice as fast if it is has already been optimized many times before.

In terms of speed if one wants to improve his chess program with 500 ELO he has to make it about 144 times faster. And for 1000 ELO 20589 times faster. So if I want to beat Stockfish in a match I at least have to hire a supercomputer from China.

[hgm]

So fast is no match for smart...