Is hardware still relevant in ELO gain?

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

Moderator: Ras

Juan P. Naar
Posts: 26
Joined: Fri May 15, 2020 3:55 pm
Full name: Juan Pablo Naar

Is hardware still relevant in ELO gain?

Post by Juan P. Naar »

As engines have been getting more and more optimized at playing chess, I was wondering if we've reached a point where no matter the hardware, most games still will end up as draws.

It would be great if anyone with a very powerful PC (16 cores or more) could test this with the most recent version of Stockfish: 1CPU vs 16CPU (or more) to see if there's an actual ELO difference between the two.
Uri Blass
Posts: 10704
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Is hardware still relevant in ELO gain?

Post by Uri Blass »

Juan P. Naar wrote: Wed Jan 15, 2025 7:54 pm As engines have been getting more and more optimized at playing chess, I was wondering if we've reached a point where no matter the hardware, most games still will end up as draws.

It would be great if anyone with a very powerful PC (16 cores or more) could test this with the most recent version of Stockfish: 1CPU vs 16CPU (or more) to see if there's an actual ELO difference between the two.
If you use normal book then hardware does not change much at long time control and I remember tests that showed that result between engines are almost always draw even when one side has more than 10:1 time advantage.

This is the reason that people use biased books in competitions like TCEC when they give engines positions that white has advantage that it is not clear if enough to win and in these cases there is still a big difference from faster hardware.
Jouni
Posts: 3546
Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski

Re: Is hardware still relevant in ELO gain?

Post by Jouni »

16 cores is not very powerful anymore. In TCEC Stockfish plays with 500 threads. And it is sure better than previous 100 threads engine. But is this enough to win with neutral book? At least you need very much games to detect possible minor gain..
Jouni
Ciekce
Posts: 174
Joined: Sun Oct 30, 2022 5:26 pm
Full name: Conor Anstey

Re: Is hardware still relevant in ELO gain?

Post by Ciekce »

Jouni wrote: Thu Jan 16, 2025 2:19 pm In TCEC Stockfish plays with 500 threads.
only for this bonus event, which is happening on the kibitzer machines
royb
Posts: 553
Joined: Thu Mar 09, 2006 12:53 am

Re: Is hardware still relevant in ELO gain?

Post by royb »

Ciekce wrote: Thu Jan 16, 2025 5:34 pm
Jouni wrote: Thu Jan 16, 2025 2:19 pm In TCEC Stockfish plays with 500 threads.
only for this bonus event, which is happening on the kibitzer machines
Since you seem to know about the current match going on at TCEC - can you please enlighten me as to what the bonus event is all about? There was already a match between Stockfish and LC0 that just completed before this bonus match. Why have a bonus match? What is the point/purpose of such a bonus match?

Thanks for your help in understanding this.
Jouni
Posts: 3546
Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski

Re: Is hardware still relevant in ELO gain?

Post by Jouni »

This bonus will be a replay of the Superfinal
Same book
Same Time Control
Same engine versions
Same configurations (except obviously adjust hardware-related options: threads, minibatch size, hash)
Ponder on
Stockfish hardware: CPU: 2x AMD EPYC 9754. RAM: 24x 32GB DDR5
LCZero hardware: CPU: 2x AMD EPYC 9274F. GPU: 8x NVIDIA RTX 4090. RAM: 24x 32GB DDR5
Administered by Windfishballad.

So that makes an order of magnitude or 6x to 8x more nodes for Leela, 10x to 12x more for SF.
Highest quality chess match ever played!
Jouni