Stockfish 18 for STC Rating List

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

User avatar
Rebel
Posts: 7493
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Stockfish 18 for STC Rating List

Post by Rebel »

jorose wrote: Fri Feb 06, 2026 7:00 am I don't think there is a fair configuration.
Correct, it's like about most TCEC finals, who had the better hardware, Stockfish or Leela ?

This rating list of mine is different than others on several aspects as doing the same as others does not make any sense. And one of these different aspects is to include Lc0 / Leela the next engine after Stockfish, neglected by other rating lists.
jorose wrote: Fri Feb 06, 2026 7:00 am That being said, I am a bit confused, on your site you write you are running the list with 16 cores and a 4080. So is Jouni mistaken and you are running these matches on 16 cores or is the information on your website misleading?
I am using -concurrency 16
jorose wrote: Fri Feb 06, 2026 7:00 am Regarding matches between CPU and GPU based engines, I would also appreciate if you would be transparent in your posts exactly which hardware and settings you are running in these matches. I shouldn't have to go to your website to find that information if you are already sharing numbers here. You are not the only person guilty of this, but as the topic arose and I respect you a lot I feel it is a good time to point it out.
What exactly is unclear ?
90% of coding is debugging, the other 10% is writing bugs.
jorose
Posts: 383
Joined: Thu Jan 22, 2015 3:21 pm
Location: Zurich, Switzerland
Full name: Jonathan Rosenthal

Re: Stockfish 18 for STC Rating List

Post by jorose »

Rebel wrote: Fri Feb 06, 2026 11:29 pm
jorose wrote: Fri Feb 06, 2026 7:00 am That being said, I am a bit confused, on your site you write you are running the list with 16 cores and a 4080. So is Jouni mistaken and you are running these matches on 16 cores or is the information on your website misleading?
I am using -concurrency 16
So.. Does this mean you are running SF on one thread? You have 16 Leela instances running at once sharing the GPU? Based on your post my understanding was you first ran a gauntlet where probably less than 16 Leela's would be running at once, but then you ran a match between SF and Leela, where based on this I presume you had 16 Leela's running at once.
Rebel wrote: Fri Feb 06, 2026 11:29 pm
jorose wrote: Fri Feb 06, 2026 7:00 am Regarding matches between CPU and GPU based engines, I would also appreciate if you would be transparent in your posts exactly which hardware and settings you are running in these matches. I shouldn't have to go to your website to find that information if you are already sharing numbers here. You are not the only person guilty of this, but as the topic arose and I respect you a lot I feel it is a good time to point it out.
What exactly is unclear ?
The settings you ran your experiments with. I don't see any of the information in your post on this site and even in the link you shared to your site all I see is the physical hardware the games were run on, not how much of that hardware you let the engines utilize.
-Jonathan
User avatar
Rebel
Posts: 7493
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Stockfish 18 for STC Rating List

Post by Rebel »

jorose wrote: Sat Feb 07, 2026 5:53 am
Rebel wrote: Fri Feb 06, 2026 11:29 pm
jorose wrote: Fri Feb 06, 2026 7:00 am That being said, I am a bit confused, on your site you write you are running the list with 16 cores and a 4080. So is Jouni mistaken and you are running these matches on 16 cores or is the information on your website misleading?
I am using -concurrency 16
So.. Does this mean you are running SF on one thread? You have 16 Leela instances running at once sharing the GPU? Based on your post my understanding was you first ran a gauntlet where probably less than 16 Leela's would be running at once, but then you ran a match between SF and Leela, where based on this I presume you had 16 Leela's running at once.
Of course running Leela -concurrency 1

You are the author of Winter and you don't understand?

Rebel wrote: Fri Feb 06, 2026 11:29 pm
jorose wrote: Fri Feb 06, 2026 7:00 am Regarding matches between CPU and GPU based engines, I would also appreciate if you would be transparent in your posts exactly which hardware and settings you are running in these matches. I shouldn't have to go to your website to find that information if you are already sharing numbers here. You are not the only person guilty of this, but as the topic arose and I respect you a lot I feel it is a good time to point it out.
What exactly is unclear ?
jorose wrote: Sat Feb 07, 2026 5:53 am The settings you ran your experiments with. I don't see any of the information in your post on this site and even in the link you shared to your site all I see is the physical hardware the games were run on, not how much of that hardware you let the engines utilize.
Time control : 40m10s
Hash : 128 Mb

When something is unclear people ask and I answer, like you now. I hardly have to because folks here are quite educated.
90% of coding is debugging, the other 10% is writing bugs.
jorose
Posts: 383
Joined: Thu Jan 22, 2015 3:21 pm
Location: Zurich, Switzerland
Full name: Jonathan Rosenthal

Re: Stockfish 18 for STC Rating List

Post by jorose »

Rebel wrote: Sat Feb 07, 2026 6:36 am You are the author of Winter and you don't understand?
I am unsure why this is your reaction. Is this your way of telling me my respect for you is misplaced?
Rebel wrote: Sat Feb 07, 2026 6:36 am When something is unclear people ask and I answer, like you now. I hardly have to because folks here are quite educated.
I understand enough to know it has nothing to do with their education.
-Jonathan
User avatar
Rebel
Posts: 7493
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Stockfish 18 for STC Rating List

Post by Rebel »

jorose wrote: Sat Feb 07, 2026 7:26 am
Rebel wrote: Sat Feb 07, 2026 6:36 am You are the author of Winter and you don't understand?
I am unsure why this is your reaction. Is this your way of telling me my respect for you is misplaced?
Think about it, in your initial post you accused me of misleading posts, so the light went out here.
90% of coding is debugging, the other 10% is writing bugs.
Modern Times
Posts: 3807
Joined: Thu Jun 07, 2012 11:02 pm

Re: Stockfish 18 for STC Rating List

Post by Modern Times »

Rebel wrote: Sat Feb 07, 2026 6:36 am Time control : 40m10s
Hash : 128 Mb
I read that as game in 40 minutes plus 10 second increment - impossible given the very large number of games you've played.

I looked back at previous posts - it is 40 moves in 10 seconds. That could be made a bit clearer I think. 40/10 as per the website can be interpreted a few different ways.
User avatar
Rebel
Posts: 7493
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Stockfish 18 for STC Rating List

Post by Rebel »

Modern Times wrote: Sat Feb 07, 2026 8:36 pm
Rebel wrote: Sat Feb 07, 2026 6:36 am Time control : 40m10s
Hash : 128 Mb
I read that as game in 40 minutes plus 10 second increment - impossible given the very large number of games you've played.

I looked back at previous posts - it is 40 moves in 10 seconds. That could be made a bit clearer I think. 40/10 as per the website can be interpreted a few different ways.
Yep, will change it into 40m/10s
90% of coding is debugging, the other 10% is writing bugs.