Little thought experiment

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Werewolf
Posts: 1174
Joined: Thu Sep 18, 2008 8:24 pm

Little thought experiment

Post by Werewolf » Wed May 15, 2019 10:05 am

On another world there exists a very powerful computer using alien tech.

This chess machine does a true (as in literal) Brute Force search 15 plies deep. It uses current SF E.F

How many plies deep, approximately, would normal SF have to search to equal the strength of the 15 plies alien machine?

User avatar
xr_a_y
Posts: 535
Joined: Sat Nov 25, 2017 1:28 pm
Location: France

Re: Little thought experiment

Post by xr_a_y » Wed May 15, 2019 10:16 am

First, "brute force" can still be alpha/beta, I guess brute force means no forward pruning.

We can try that with depth 6 or 8 or maybe even 10 I guess ...

Is something like a limited depth no-forward-pruning-engine versus a limited time classic-engine seems fine ?

User avatar
Eelco de Groot
Posts: 4073
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: Little thought experiment

Post by Eelco de Groot » Wed May 15, 2019 11:53 am

SF EF does not seem to do any extensions either it seems, apart from not reducing or pruning. The search horizon is only 7½ full moves. That is not enough to overcome serious horizont effects. I think it would play weakly. It is totally a guess but at first glance I'd say would be easily crushed by the better stand alones if given enough time per move, tournament time control, or maybe something like ten minutes per move, that should be enough for my old CXG Enterprise, which is not really a strong tabletop, to beat Aliens :) Image
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

User avatar
Eelco de Groot
Posts: 4073
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: Little thought experiment

Post by Eelco de Groot » Wed May 15, 2019 12:28 pm

Mmm, if you set Stockfish to fixed depth 15 search but disable all extensions that would be close? Maybe still strong vs human GM but not really SF. 15 ply used to be the depth for the bench function I thought, although now it seems to be lower, just 13 ply, it tests most parts of the search in bench but takes only 2371 miliseconds for the 42 bench positions on my machine. A little more for 15 ply perhaps but if you disable all extensions seriously less than that 2371 miliseconds. 10 minutes for the CXG Enterprise would not be enough but it was not really strong.
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

User avatar
hgm
Posts: 23211
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: Little thought experiment

Post by hgm » Wed May 15, 2019 5:37 pm

Werewolf wrote:
Wed May 15, 2019 10:05 am
On another world there exists a very powerful computer using alien tech.

This chess machine does a true (as in literal) Brute Force search 15 plies deep. It uses current SF E.F

How many plies deep, approximately, would normal SF have to search to equal the strength of the 15 plies alien machine?
Why asking this for an impossibly large depth, instead of simply trying it out for depths that can be reached with present-day technology, and extrapolate from there?

Uri Blass
Posts: 8378
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Little thought experiment

Post by Uri Blass » Wed May 15, 2019 6:01 pm

Werewolf wrote:
Wed May 15, 2019 10:05 am
On another world there exists a very powerful computer using alien tech.

This chess machine does a true (as in literal) Brute Force search 15 plies deep. It uses current SF E.F

How many plies deep, approximately, would normal SF have to search to equal the strength of the 15 plies alien machine?
Do you mean stockfish when you also disable extensions or extensions are the same as the extension of stockfish of today?

mwyoung
Posts: 1310
Joined: Wed May 12, 2010 8:00 pm

Re: Little thought experiment

Post by mwyoung » Thu May 16, 2019 4:24 am

Werewolf wrote:
Wed May 15, 2019 10:05 am
On another world there exists a very powerful computer using alien tech.

This chess machine does a true (as in literal) Brute Force search 15 plies deep. It uses current SF E.F

How many plies deep, approximately, would normal SF have to search to equal the strength of the 15 plies alien machine?

It would take my computer 11,367,235.62 years to make one move to match the alien computer....
Professing themselves to be wise, they became fools,

Werewolf
Posts: 1174
Joined: Thu Sep 18, 2008 8:24 pm

Re: Little thought experiment

Post by Werewolf » Thu May 16, 2019 8:53 am

hgm wrote:
Wed May 15, 2019 5:37 pm
Werewolf wrote:
Wed May 15, 2019 10:05 am
On another world there exists a very powerful computer using alien tech.

This chess machine does a true (as in literal) Brute Force search 15 plies deep. It uses current SF E.F

How many plies deep, approximately, would normal SF have to search to equal the strength of the 15 plies alien machine?
Why asking this for an impossibly large depth, instead of simply trying it out for depths that can be reached with present-day technology, and extrapolate from there?
At low ply levels I imagine tactics would dominate to such an extent it might make comparison hard. Also I don't know of an engine which does a pure BF search with a SF EF. But maybe you do?

Werewolf
Posts: 1174
Joined: Thu Sep 18, 2008 8:24 pm

Re: Little thought experiment

Post by Werewolf » Thu May 16, 2019 8:56 am

mwyoung wrote:
Thu May 16, 2019 4:24 am
Werewolf wrote:
Wed May 15, 2019 10:05 am
On another world there exists a very powerful computer using alien tech.

This chess machine does a true (as in literal) Brute Force search 15 plies deep. It uses current SF E.F

How many plies deep, approximately, would normal SF have to search to equal the strength of the 15 plies alien machine?

It would take my computer 11,367,235.62 years to make one move to match the alien computer....
Yeah mine is similar, but they have great tech on planet Cybertron...

Werewolf
Posts: 1174
Joined: Thu Sep 18, 2008 8:24 pm

Re: Little thought experiment

Post by Werewolf » Thu May 16, 2019 8:58 am

Uri Blass wrote:
Wed May 15, 2019 6:01 pm
Werewolf wrote:
Wed May 15, 2019 10:05 am
On another world there exists a very powerful computer using alien tech.

This chess machine does a true (as in literal) Brute Force search 15 plies deep. It uses current SF E.F

How many plies deep, approximately, would normal SF have to search to equal the strength of the 15 plies alien machine?
Do you mean stockfish when you also disable extensions or extensions are the same as the extension of stockfish of today?
For the Stockfish engine: exactly as SF is today

For the Brute Force engine: SF EF but the search part is only BF with no extensions at all.
(Having said that another version with extensions would be interesting too)

Post Reply