Dragon 3.1 Released at KomodoChess.com

[Frank Quisinsky]

I am not speaking from half-moves ... speaking from moves.
37 and 41 moves (average per draws) more ...

Will not give a red card because I can changed that but please thinking about cost for electricity.
That is much more important as +2.7 Elo, because I am very sure that most customers will not changed standard settings!

Best
Frank

[Frank Quisinsky]

Hi there,

shortly:
In the test-tourney are 75% attackers, so the move average for won games is 82 only.
Normaly Dragon 3 NN (Komodo) is much better and can make 77er average if ... all the engines inside like
to produce a high quantity of fast lost games.

Attacker vs. Attacker:
Results is a higher move average!
82er average is really super in the field I created.

And to contempt:
Time for contempt is over since we have Neural-Network.
Before Neural-Network contempt can give stronger engines like Komodo 14.1 around +15-18 Elo.
If Dragon 3 NN (Komodo) is playing vs. NN network engines only the advantage is around +3 Elo.
Same for most other engines.

I kill every engine here produced a high draw move-average with secret contempt settings!
In case of Dragon I can changed that but for many other engines I cann't changed that.
Much more nice for looking if the games are running.

Move average for draws should be not higher as max. 90-96 moves!!

Best
Frank

[CornfedForever]

Don't hold back, Frank - tell everyone what you really think....

I've no dog in the hunt currently...and maybe I should feel bad about your power consumption and cost; but maybe Larry and team Komodo/Dragon have even more stats than yours regarding their product.

[lkaufman]

Don't hold back, Frank - tell everyone what you really think....

I've no dog in the hunt currently...and maybe I should feel bad about your power consumption and cost; but maybe Larry and team Komodo/Dragon have even more stats than yours regarding their product.

With the ccrl rapid elo gap between sf15 and dragon3 only seven elo and likely to shrink more with dragon 3.1, 3 elo is a big deal. For testing vs other NN engines I don’t object to Contemt 0, but many testers still test vs much weaker engines. Giving them draws in equal endings will hurt elo of course. In my view, testing with increment rather than with repeating time controls is the way to save electricity and testing time. Repeat time controls are obsolete in both human and engine play. No one likes to waste time playing out dead draws at a slow pace.

[BrendanJNorman]

I am posting this for Larry Kaufma, who is traveling now:

" KomodoChess.com has released Dragon 3.1, which today won the World Computer Chess Championship after a tiebreaking match with
Lc0. It is essentially the same version that won the WCCC, except for adjustments to limited Elo levels. It fixes some issues in
Dragon 3 (occasional time forfeits in standard mode and crashes in MCTS mode), and so is free for anyone who bought Dragon 3 (or got
it by subscription). It has a newer net and search improvements compared to Dragon 3, which make it about ten elo stronger with
normal openings (at CCRL blitz tc, one thread), twelve elo stronger at FRC (chess960), and twenty elo stronger with "unbalanced human
openings". It also has a user-settable minimum time per move called "Skill Time" when using the LimitStrength option, and now allows for Elo setting up
to 3499 (3500 means full strength). As usual, it is free for subscribers, and 20% discounted for purchasers of Dragon 2 thru 2.6. No
price changes from Dragon 3 "

Note I will be sending out copies to rating list testers starting tomorrow.

Hi!

I purchased Dragon 3 (and 2.5 and 1.0 and Komodo 13.2 and 12.1 lol) and I do not see Dragon 3.1 in my account downloads area.

It should be there, right?

[Modern Times]

Hi!

I purchased Dragon 3 (and 2.5 and 1.0 and Komodo 13.2 and 12.1 lol) and I do not see Dragon 3.1 in my account downloads area.

It should be there, right?

You did read the prominent notice on the web page ?

Disclaimer: the download files for dragon 3 have been updated to dragon 3.1. The dragon 3 download may still be called dragon 3.zip, but it will contain the dragon 3.1 files. (There should also be a disclaimer.txt file in the folder explaining this; The name will be fixed in time.) If you have already purchased Dragon 3, you do not need to purchase Dragon 3.1.

[BrendanJNorman]

Hi!

I purchased Dragon 3 (and 2.5 and 1.0 and Komodo 13.2 and 12.1 lol) and I do not see Dragon 3.1 in my account downloads area.

It should be there, right?

You did read the prominent notice on the web page ?

Disclaimer: the download files for dragon 3 have been updated to dragon 3.1. The dragon 3 download may still be called dragon 3.zip, but it will contain the dragon 3.1 files. (There should also be a disclaimer.txt file in the folder explaining this; The name will be fixed in time.) If you have already purchased Dragon 3, you do not need to purchase Dragon 3.1.

Nope I didn't. Thanks for pointing it out

[Andrew]

Thanks very much for the latest Komodo update! Please everyone, let's keep things on topic.
Makes it much easier to find the important information!

Andrew

[mehmet123]

According to Mark Young's test the elo difference between Dragon and Stockfish Dev is only -6 elo. (600 games 0-10 590)
It won't be surprising if Dragon enters the year 2023 as the number one chess engine. All it takes is a much stronger NNUE network.

[pohl4711]

According to Mark Young's test the elo difference between Dragon and Stockfish Dev is only -6 elo. (600 games 0-10 590)
It won't be surprising if Dragon enters the year 2023 as the number one chess engine. All it takes is a much stronger NNUE network.

LOL

The only thing, you can learn from a match with a draw rate of 98.3% is, that such matches are a waste of time and energy. Statistical complete useless...because so many draws make playing strength-differences indistinguishable.


Better look here (high draw-rate vs SF dev here, too (87-88%), but not ridiculous high (98.3%):

Ratinglist-testruns of KomodoDragon 3.1 and KomodoDragon 3.1 MCTS finished.

https://www.sp-cc.de

Code: Select all

4 KomodoDragon 3.1 avx2  : 3769 12000 (+3897,=7708,-395), 64.6 %

Stockfish 220724 avx2    : 1000 (+  7,=869,-124), 44.1 %
Stockfish 220713 avx2    : 1000 (+  7,=880,-113), 44.7 %
Revenge 3.0 avx2         : 1000 (+350,=642,-  8), 67.1 %
Ethereal 13.75 nnue      : 1000 (+390,=598,- 12), 68.9 %
Berserk 9 avx2           : 1000 (+354,=638,-  8), 67.3 %
Koivisto 8.13 avx2       : 1000 (+343,=652,-  5), 66.9 %
Slow Chess 2.9 avx2      : 1000 (+471,=523,-  6), 73.3 %
Stockfish 15 220418      : 1000 (+  9,=896,- 95), 45.7 %
Fire 8.NN MCTS avx2      : 1000 (+494,=502,-  4), 74.5 %
RubiChess 220223 avx2    : 1000 (+552,=448,-  0), 77.6 %
Stockfish final HCE      : 1000 (+507,=477,- 16), 74.5 %
Fire 8.NN avx2           : 1000 (+413,=583,-  4), 70.5 %