https://nextchessmove.com/dev-builds/sf14 .
Seriously this VVLTC search tune should be tested only with VVLTC? What is definition of VVLTC actually? 40 moves in 2 hours ?
Stockfish 5.5. has now reached SF 16.1 level
Moderators: hgm, Rebel, chrisw
-
- Posts: 3316
- Joined: Wed Mar 08, 2006 8:15 pm
-
- Posts: 605
- Joined: Fri Mar 30, 2018 7:20 am
- Full name: Andreas Matthies
Re: Stockfish 5.5. has now reached SF 16.1 level
Just read what is written by SF developers: https://github.com/official-stockfish/S ... /pull/5216Jouni wrote: ↑Tue May 07, 2024 4:57 pm https://nextchessmove.com/dev-builds/sf14 .
Seriously this VVLTC search tune should be tested only with VVLTC? What is definition of VVLTC actually? 40 moves in 2 hours ?
So VVLTC is 60+0.6 on 8 threads (or 70+0.7 on 7 threads or 80+0.8 on 6 threads cause they had lots of workers offering only 7 threads at time of testing).
And what do you mean by Stockfish 5.5 in this context?
And is this really worth a new thread?
-
- Posts: 77
- Joined: Tue Apr 09, 2024 6:24 am
- Full name: Michael Chaly
Re: Stockfish 5.5. has now reached SF 16.1 level
my wild guess would be that 5.5. is a 05.05 date.
But I guess some people are not smart enough to check this - https://github.com/official-stockfish/S ... sion-Tests
where you can see that at 60+0.6 1 thread indeed master is not that far ahead of 16.1 and effective NCM TC is lower than fishtest 60+0.6.
But with 60+0.6 8 threads it is 10 elo ahead of 16.1 - because of bigger and slower net and various search changes. In fact this progress is really big for less than 2 months.
But I guess some people are not smart enough to check this - https://github.com/official-stockfish/S ... sion-Tests
where you can see that at 60+0.6 1 thread indeed master is not that far ahead of 16.1 and effective NCM TC is lower than fishtest 60+0.6.
But with 60+0.6 8 threads it is 10 elo ahead of 16.1 - because of bigger and slower net and various search changes. In fact this progress is really big for less than 2 months.