128 CPU Stockfish Cluster Comments.

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

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Leo
Posts: 858
Joined: Fri Sep 16, 2016 4:55 pm
Location: USA/Minnesota
Full name: Leo Anger

Re: 128 CPU Stockfish Cluster Comments.

Post by Leo » Wed Feb 06, 2019 4:15 pm

Jouni wrote:
Tue Feb 05, 2019 9:08 pm
This +130 gives nice 3700 ELO! Maybe the upper limit for chess?
If you ad a book like Brainfish or Rebelfish or something similar the Elo keeps getting bigger. Hardware will keep getting better. If Lzero keeps improving that will work too. I think Elo will keep growing indefinitely. Way past 3700.
Advanced Micro Devices fan.

Werewolf
Posts: 1199
Joined: Thu Sep 18, 2008 8:24 pm

Re: 128 CPU Stockfish Cluster Comments.

Post by Werewolf » Wed Feb 06, 2019 5:28 pm

Leo wrote:
Wed Feb 06, 2019 4:15 pm
Jouni wrote:
Tue Feb 05, 2019 9:08 pm
This +130 gives nice 3700 ELO! Maybe the upper limit for chess?
If you ad a book like Brainfish or Rebelfish or something similar the Elo keeps getting bigger. Hardware will keep getting better. If Lzero keeps improving that will work too. I think Elo will keep growing indefinitely.
Of course it won't.

For exactly the same reason that the elo of a simpler game (naughts and crosses, connect 4, draughts / checkers etc) doesn't go up forever. As soon as we get anywhere near perfect chess the best engines won't lose anymore and the elo of everyone will be capped.

jefk
Posts: 394
Joined: Sun Jul 25, 2010 8:07 pm
Location: the Netherlands
Contact:

Re: 128 CPU Stockfish Cluster Comments.

Post by jefk » Fri Feb 08, 2019 5:08 pm

werewolf, that's my hunch as well.
If properly calculated, the engines will not go much beyond 4000 i suspect;
actually we could make graph, and see/estimate where it might end.
(there are some of such graphs, but just from memory i did not see
yet a clear asympotical end level; yet 5000 seems too high imho)
jefk

Uri Blass
Posts: 8650
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: 128 CPU Stockfish Cluster Comments.

Post by Uri Blass » Fri Feb 08, 2019 5:17 pm

Werewolf wrote:
Tue Feb 05, 2019 9:51 pm
Dann Corbit wrote:
Tue Feb 05, 2019 9:47 pm
Jouni wrote:
Tue Feb 05, 2019 9:08 pm
This +130 gives nice 3700 ELO! Maybe the upper limit for chess?
I doubt it, but possibly an upper limit for shared hash table alpha-beta.
On the other hand, the evaluation of LC0 is clearly vastly superior to the evaluation of Stockfish (due to less than 0.1% as many nodes during search giving an answer that is about equal in quality).
So suppose that the evaluation of Stockfish were tuned to equal the evaluation of LC0.
Perhaps the limit is 5000 Elo.
No, I doubt any entity will get over 4000 elo. I think we'll start to see the draw rate dramatically rise as this level of play is reached. But I'd love to be proved wrong.

I think that we need the concept of anti-engine to increase elo more than we can do only by using faster hardware in a normal way.
The idea is that we can use significantly faster hardware to prepare against specific opponents when we know also their weaker hardware.

The anti-stockfish engine may find by using stockfish that
after 22.Rad1 stockfish is going to make a losing blunder when
after 22.Rac1 stockfish is not going to make a losing blunder.

In theory both 22.Rad1 and 22.Rac1 are drawing moves against perfect play so better hardware with no anti-stockfish algorithm may choose 22.Rac1 and only get a draw.

Leo
Posts: 858
Joined: Fri Sep 16, 2016 4:55 pm
Location: USA/Minnesota
Full name: Leo Anger

Re: 128 CPU Stockfish Cluster Comments.

Post by Leo » Fri Feb 08, 2019 6:34 pm

Werewolf wrote:
Wed Feb 06, 2019 5:28 pm
Leo wrote:
Wed Feb 06, 2019 4:15 pm
Jouni wrote:
Tue Feb 05, 2019 9:08 pm
This +130 gives nice 3700 ELO! Maybe the upper limit for chess?
If you ad a book like Brainfish or Rebelfish or something similar the Elo keeps getting bigger. Hardware will keep getting better. If Lzero keeps improving that will work too. I think Elo will keep growing indefinitely.
Of course it won't.

For exactly the same reason that the elo of a simpler game (naughts and crosses, connect 4, draughts / checkers etc) doesn't go up forever. As soon as we get anywhere near perfect chess the best engines won't lose anymore and the elo of everyone will be capped.
We are not anywhere near perfect chess. Elo will continue to go up for years.
Advanced Micro Devices fan.

Uri Blass
Posts: 8650
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: 128 CPU Stockfish Cluster Comments.

Post by Uri Blass » Fri Feb 08, 2019 8:37 pm

Leo wrote:
Fri Feb 08, 2019 6:34 pm
Werewolf wrote:
Wed Feb 06, 2019 5:28 pm
Leo wrote:
Wed Feb 06, 2019 4:15 pm
Jouni wrote:
Tue Feb 05, 2019 9:08 pm
This +130 gives nice 3700 ELO! Maybe the upper limit for chess?
If you ad a book like Brainfish or Rebelfish or something similar the Elo keeps getting bigger. Hardware will keep getting better. If Lzero keeps improving that will work too. I think Elo will keep growing indefinitely.
Of course it won't.

For exactly the same reason that the elo of a simpler game (naughts and crosses, connect 4, draughts / checkers etc) doesn't go up forever. As soon as we get anywhere near perfect chess the best engines won't lose anymore and the elo of everyone will be capped.
We are not anywhere near perfect chess. Elo will continue to go up for years.

How do you know?
It is clear that we are getting closer to perfect chess and you cannot know if at LTC there are 400 elo more to get or 1000 elo more to get.

Leo
Posts: 858
Joined: Fri Sep 16, 2016 4:55 pm
Location: USA/Minnesota
Full name: Leo Anger

Re: 128 CPU Stockfish Cluster Comments.

Post by Leo » Fri Feb 08, 2019 9:23 pm

Uri Blass wrote:
Fri Feb 08, 2019 8:37 pm
Leo wrote:
Fri Feb 08, 2019 6:34 pm
Werewolf wrote:
Wed Feb 06, 2019 5:28 pm
Leo wrote:
Wed Feb 06, 2019 4:15 pm
Jouni wrote:
Tue Feb 05, 2019 9:08 pm
This +130 gives nice 3700 ELO! Maybe the upper limit for chess?
If you ad a book like Brainfish or Rebelfish or something similar the Elo keeps getting bigger. Hardware will keep getting better. If Lzero keeps improving that will work too. I think Elo will keep growing indefinitely.
Of course it won't.

For exactly the same reason that the elo of a simpler game (naughts and crosses, connect 4, draughts / checkers etc) doesn't go up forever. As soon as we get anywhere near perfect chess the best engines won't lose anymore and the elo of everyone will be capped.
We are not anywhere near perfect chess. Elo will continue to go up for years.

How do you know?
It is clear that we are getting closer to perfect chess and you cannot know if at LTC there are 400 elo more to get or 1000 elo more to get.
It is not clear to me at all. When you get even a 15 man table base then I will rethink it.
Advanced Micro Devices fan.

Werewolf
Posts: 1199
Joined: Thu Sep 18, 2008 8:24 pm

Re: 128 CPU Stockfish Cluster Comments.

Post by Werewolf » Sat Feb 09, 2019 12:53 pm

Dann Corbit wrote:
Tue Feb 05, 2019 8:09 pm
henk2 wrote:
Sun Feb 03, 2019 10:15 pm
If some billionaire was willing to waste all his money on dedicated Stockfish 10 ASICs.
Then the cluster could easily be blown away.

4 billion nps is still very impressive though.
That billionaire would learn the lesson of Hydra.
A dump truck full of money will make an engine that will win for 4 years.
Ten years later, a commodity PC will give it a spanking.
I had the pleasure of testing Hydra and I think it was a worthwhile project. It gave us a glimpse of the future by showing us how strong chess machines could become. We also had a demolition of Mickey Adams which I enjoyed watching, sat only 10 feet away from him.

Finally, although it's true Hydra wasn't no.1 four years later, it was the strongest tactically (my own interest at the time was tactical test suites, I had one called "New Moon" and Chrilly let me try it out on the machine) for about 6 years. It was only when the Rybka Cluster came along that I felt Hydra had been beaten tactically.

Post Reply