FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Discussion of computer chess matches and engine tournaments.

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Frank Quisinsky
Posts: 5162
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Frank Quisinsky » Thu Aug 20, 2020 3:18 pm

Hi there,

the results from round 19 of 50 are online!

Best
Frank
I like computer chess!

Frank Quisinsky
Posts: 5162
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Frank Quisinsky » Thu Aug 20, 2020 7:03 pm

Hi Jonathan,

at first ...
10-15 Elo + contempt parameter = in my humble opinion ... reason enough for an update!

What you wrote to the pawns / Senpai!
That is exactly my observation!

Wasp have a great king safty but if you are looking in detail:
Average of pawn moves below move 60 ... Wasp is to place 2 in the group of 41 engines (not public statistic).
Means, only one engine plays more pawn moves below move 60 (average).

Wasp searching very often very aggressive the pawn moves and lost vs. stronger engines later in transposition into endgame with an inferiority pawn structure.

OK, but I will not missing the strength Wasp have, because games are clearly more nice for looking.
In my opinion is such a style with aggressive pawns much more the human-like-playing-style.

Style is for myself more important as Elo!

Wasp have a clearly better king safty as Senpai!
So the bad pawns are not good for Senpai, Wasp compensated the weaknesses with the good king-safty, Senpai not have.

I hope that Fabien will work on Senpai and can make to the points "king-safty" and "pawn-structures" improvements. Fabien have many great ideas in the time he sent me Senpai 2.0. Sure, that Fabien is working on Senpai. Fabien is to 100% chess programmer and like to search new ways.

King safty for Fruit and Senpai isn't good. Same for engines used Fruit or Senpai ideas.

BTW: Nice to read that you are using Wasp for your own test-games.
At the moment SlowChess is around 160 + 15 (your improvements) stronger as Wasp.
For Wasp is SlowChess a good test-opponent ...
For SlowChess is Wasp ... for the moment ... maybe not good enough as test-opponent.

I am thinking that in a test-field (round-robin) engines should not 200 Elo stronger or 100 Elo weaker.

Thanks for your great engine and the very interesting information you gave.
Best and good luck for your development SlowChess!!

Best
Frank

PS: I am most of time in September and Oktober not at home. Have a lot of private work. I must reduced the quantity of rounds from 50 to 25. But in January 2021 I will start the same tourney with current engine versions again. 25 or 50 rounds again!
I like computer chess!

Frank Quisinsky
Posts: 5162
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Frank Quisinsky » Sun Aug 23, 2020 7:59 am

Hi there,

the final results from round 20 out of 25 are online.
Now 800 out of 1.000 games played.

Frank's Chess Page
https://www.amateurschach.de

Best
Frank
I like computer chess!

Alayan
Posts: 429
Joined: Tue Nov 19, 2019 7:48 pm
Full name: Alayan Feh

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Alayan » Sun Aug 23, 2020 2:30 pm

25 rounds isn't nice, engines will have played each other an odd number of times so it won't feature proper reverse games and will have uneven white/black pieces chances. 24, 26, 30 would be better in this regard if you must shorten the tournament.

Frank Quisinsky
Posts: 5162
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Frank Quisinsky » Sun Aug 23, 2020 3:29 pm

Hi Alayan,

I am for around 6 Wochen not at home (from Sept. 09th).
Do you think it would be better to restart the tourney for statistic reasons after 6 weeks?
Below Sept. 09th, 2020 25 rounds can be play.

Or do you think I should stopped the tourney?
That is the questions.

I have in my *.zip file the results round by round (inside are all statistics, round by round).
From this point of view it's not important to stopped the tourney after round 24 or later.

Best
Frank
I like computer chess!

Alayan
Posts: 429
Joined: Tue Nov 19, 2019 7:48 pm
Full name: Alayan Feh

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Alayan » Mon Aug 24, 2020 9:16 pm

I don't know, 24 rounds would feel cleaner than 25 in regards to h2h. 26 too, but time may be short and 24 is nicer-looking.

In any case I like your tournament and the side-stats a lot, a new tournament once you're back sounds good.

Frank Quisinsky
Posts: 5162
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Frank Quisinsky » Tue Aug 25, 2020 6:58 am

Hi Alayan,

thank you!

Main idea was ...
To play 50 rounds with 20 minutes game + 5 seconds per move and to replay the complete 41.000 games tourney with 20% of time (means, 4 minutes game + 1 second per move, 256Mb for hashtables). So we can looking for the differents for the complete group of 41 engines and I have enough time to make the opening book stronger.

I think I will hold the main idea. For that reason I purchase my second i9-10900.
New engine updates can be collect for FCP Tourney-2021 with a complete ready Excel file by Klaus for statistics and an opening book with better (near to the perfection) priorities.

Best
Frank

PS: I have here a lot of fun with the tourney. Looking often in still running games. It seems that after 800 games per engine Chiron have to bad results and Fritz 16 have to strong results. Wasp is playing around 10 Elo over the performance I await. From Wasp view the World is fully OK. For the moment I have most fun with the playing style comes from Fizbo. Only vs. clearly stronger (Stockfish, Komodo & Ethereal) Fizbo have no chance with the speculative style. Hope that the programmer of Fizbo are working a long time on that program because the biggest fun-factor comes after all I saw from Fizbo and ... quit clear ... from Stockfish.
I like computer chess!

Frank Quisinsky
Posts: 5162
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Frank Quisinsky » Tue Aug 25, 2020 12:16 pm

Changed back to 50 rounds!
With a break from September 09th - October 31st, 2020
I like computer chess!

Frank Quisinsky
Posts: 5162
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Frank Quisinsky » Wed Aug 26, 2020 7:17 am

Hi there,

the results from round 21 of 50 are online.
I updated the FEOBOS 20.2 Contempt 3 book for Shredder GUI.

FIX-3 is now available.
I changed the priorities for around 50 ECO codes.
Nice to see in the Excel file by Klaus Wlotzka from which codes I have not enough or to many games if I compare with quantity of ECO codes to the current GM-Theory of best 860 grandmasters.

Frank's Chess Page
http://www.amateurschach.de

Best
Frank
I like computer chess!

Jouni
Posts: 2227
Joined: Wed Mar 08, 2006 7:15 pm

Re: FCP Tourney-2020: 41.000 games 20+5 to play on Intel i9-10900k

Post by Jouni » Wed Aug 26, 2020 12:23 pm

BTW have You missed, that Houdini 6.03 is actually Stockfish 8.5?
Jouni

Post Reply