CPU vs GPU Which will dominate in the future?

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

Moderators: hgm, Rebel, chrisw

Chessqueen
Posts: 5685
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

CPU vs GPU Which will dominate in the future?

Post by Chessqueen »

Russian believe that they are special, that they can Kill thousands of Ukranian civilians, but cry like babies when a few Russian...
Werewolf
Posts: 1847
Joined: Thu Sep 18, 2008 10:24 pm

Re: CPU vs GPU Which will dominate in the future?

Post by Werewolf »

Obviously neither will "dominate" because there's not enough head-room left in chess for the deficiencies of MCTS or AB to be exposed. All you'll see is an increasing draw rate by the top engines and more and more desperate openings to try to keep the game alive.
chessica
Posts: 708
Joined: Thu Aug 11, 2022 11:30 pm
Full name: Esmeralda Pinto

Re: CPU vs GPU Which will dominate in the future?

Post by chessica »

Chessqueen wrote: Mon May 13, 2024 5:16 am https://tcec-chess.com/
In a market economy, price is crucial.
AndrewGrant
Posts: 1819
Joined: Tue Apr 19, 2016 6:08 am
Location: U.S.A
Full name: Andrew Grant

Re: CPU vs GPU Which will dominate in the future?

Post by AndrewGrant »

Werewolf wrote: Mon May 13, 2024 9:56 am Obviously neither will "dominate" because there's not enough head-room left in chess for the deficiencies of MCTS or AB to be exposed. All you'll see is an increasing draw rate by the top engines and more and more desperate openings to try to keep the game alive.
Well Stockfish manages 70 to 100 elo margins over Leela at some CCC events. That is certainly quite a bit of head-room.
Friendly reminder that stealing is a crime, is wrong, and makes you a thief.
"Those who can't do, clone instead" - Eduard ( A real life friend, not this forum's Eduard )
User avatar
towforce
Posts: 11751
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK

Re: CPU vs GPU Which will dominate in the future?

Post by towforce »

chessica wrote: Mon May 13, 2024 10:36 amIn a market economy, price is crucial.

Rule doesn't apply to luxury goods! :)

However, in a "normal" market, the following cycle tends to occur (there are other types of disruption, but this is the most common):

1. Market becomes dominated by small number of players

2. The products gradually become more "feature rich" (more elaborate)

3. The disruptor comes in with a new product that's "good enough" and costs less

4. This becomes the new standard. Loop back to step 1

Unfortunately, this hasn't happened to Windows. The Chromebook should have removed the Windows laptop from most homes, but people are attached to Windows for reasons that don't stand up to scrutiny (I accept that there are genuine use-cases where a Chromebook would be unsuitable).

In chess, though, the process is happening: I personally find that Chess.com meets my chess requirements handily, even though I know that its bots/engines/analysis tools aren't as good as Stockfish on an expensive graphics card.

Going forward, once CPUs are able to play unbeatable chess quickly, there will be no need to buy a GPU - so in the chess world, the CPU will eventually be the winner.
The simple reveals itself after the complex has been exhausted.
Werewolf
Posts: 1847
Joined: Thu Sep 18, 2008 10:24 pm

Re: CPU vs GPU Which will dominate in the future?

Post by Werewolf »

AndrewGrant wrote: Mon May 13, 2024 10:51 am
Werewolf wrote: Mon May 13, 2024 9:56 am Obviously neither will "dominate" because there's not enough head-room left in chess for the deficiencies of MCTS or AB to be exposed. All you'll see is an increasing draw rate by the top engines and more and more desperate openings to try to keep the game alive.
Well Stockfish manages 70 to 100 elo margins over Leela at some CCC events. That is certainly quite a bit of head-room.
That is quite a bit. But as time goes on and the engines are improved, as hardware gets faster, as TC gets longer - especially if all 3 are applied - then the Elo difference will drop.

I don't know if Chessbase were right to say no engine will ever surpass 4000 Elo, but that estimate must be true in principle.