Is Stockfish secret simply search depth?

Discussion of chess software programming and technical issues.

Moderator: Ras

Jouni
Posts: 3621
Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski

Is Stockfish secret simply search depth?

Post by Jouni »

In opening position SF reaches 50 plies in 10 minutes with old i5 CPU. Other engines some 30-40. Berserk goes to 45, but it is number 2 free engine.
Jouni
User avatar
Ajedrecista
Posts: 2103
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.

Re: Is Stockfish secret simply search depth?

Post by Ajedrecista »

Hello Jouni:
Jouni wrote: Mon Nov 27, 2023 6:10 pm In opening position SF reaches 50 plies in 10 minutes with old i5 CPU. Other engines some 30-40. Berserk goes to 45, but it is number 2 free engine.
I am not a programmer, so please take my answer as an educated guess: I would simply say no. A good eval through a fast search helps, of course.

I remember EXchess, which is a strong engine in terms of human standards, but not a top tier engine. That engine ran into depth like there was not tomorrow. You have the opposite case some years ago: Rybka, which was one of the strongest engines or even the strongest, yet search depths were low when compared with other engines.

All of these cases happen because search depth is not universally defined for every engine: each engine counts search depths differently, as well as searched nodes. You can not compare nominal depths of different engines because it is like comparing apples and oranges, as the saying goes.

I would like to see a more formal, technical answer by anyone, if possible. Thanks in advance.

Regards from Spain.

Ajedrecista.
Jouni
Posts: 3621
Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski

Re: Is Stockfish secret simply search depth?

Post by Jouni »

Rybka was infamous to display fake nps and depth :lol: .
Jouni
User avatar
Ajedrecista
Posts: 2103
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.

Re: Is Stockfish secret simply search depth?

Post by Ajedrecista »

Hello again:
Jouni wrote: Mon Nov 27, 2023 7:38 pm Rybka was infamous to display fake nps and depth :lol: .
It is true! (I had it in mind while answering to you in my other post). We do not know about others.

I reaffirm myself about different authors counting differently, even without faking output data intentionally. Let us see what CPW says at some point:

Depth - Chessprogramming wiki
CPW wrote:Depth Comparison of different programs
Due to different implementations, the reported search depth of chess programs is not comparable in general. Programs like The King (Chessmaster), Junior and Rybka are known for interpreting depth differently for whatever reasons.
Regards from Spain.

Ajedrecista.
syzygy
Posts: 5696
Joined: Tue Feb 28, 2012 11:56 pm

Re: Is Stockfish secret simply search depth?

Post by syzygy »

Jouni wrote: Mon Nov 27, 2023 6:10 pm In opening position SF reaches 50 plies in 10 minutes with old i5 CPU. Other engines some 30-40. Berserk goes to 45, but it is number 2 free engine.
Depth is a mostly meaningless number if you compare it between engines.

It is easy to reach huge depth. Just add enough pruning and reductions without caring whether it hurts the search.
Jouni
Posts: 3621
Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski

Re: Is Stockfish secret simply search depth?

Post by Jouni »

It's stunning, that SFdev with 3500 kn/s reaches same depth than SF classic with 10 000 kn/s :o .
Jouni
Werewolf
Posts: 1996
Joined: Thu Sep 18, 2008 10:24 pm

Re: Is Stockfish secret simply search depth?

Post by Werewolf »

Jouni wrote: Mon Nov 27, 2023 7:38 pm Rybka was infamous to display fake nps and depth :lol: .
Rybka nps = 17x what you see on the display
Rybka depth = add 3 plies to the display