CEGT - rating lists January 17th 2021

[Werner]

Hi all,
our actual rating lists are online and can be found under the attached links!

40 / 20:
New games: 3.372; 37 different engines
Total: 1.430.725

NEW Engines
552 Cheng 4.40 x64 1CPU: 2866 - 1005 games (+85 to v. 4.39)
Pedone 3.0NN x64 1CPU: 3222 - 1205 games (+150 to v. 2.1)
42 Scorpio 3.0.11 Cuda (e20x256): 3400 - 300 games (< v. 3.0.10 - now using newer bugfixed version)
308 Toga III 0.2NN 1CPU: 3019 - 1000 games (+261 to v. 4.0 a/b)

UPDATES
2 KomodoDragon 1.0 x64 4CPU: 3545 - 721 games (-2)

40 / 4
last update was November 29th; next update comming soon!
we are testing:
LCZero 0.26.3 703810 (latest 70.... net) ELO 3501 out of 600 games
LCZero 0.26.3 CUDNN J92-330 = ELO 3555 out of 900 games +43
Seer 1.2.1 NNUE Perf=2943
Beef 0.3.6 Perf=2975
Cheng 4.40 dev Perf=2859
Koivisto 4.0 Perf=2928
Nirvana Chess 2.5 x64 1CPU ELO 3023 out of 1500 games (+20)
Orion 0.8 x64 NNUE ELO 2943 out of 500 games (+351 to v0.7!!)
Halogen 9.0 x64 1CPU ELO 2972 out of 1000 games +176!! to v8.1
Demolito 2020-10-10 x64 1CPU ELO 3020 out of 1100 games +27 to v2020-02-10
RubiChess 1.9.0 x64 1CPU NNUE ELO 3230 out of 1200 games => +83 to v1.8.0
Minic 3.02 NaNi x64 1CPU ELO 3033 out of 1500 games (+51 to v2.48 )
Stoofvlees II a16 ELO 3486 out of 800 games (+71)
Topple 0.8.0 x64 1CPU ELO 2881 (+52 to v0.7.5)
Arasan 22.2 x64 1CPU ELO 3077 out of 500 games +6
Wasp 4.50 x64 1CPU 3088 out of 1000 games (+73 to v4.00)
Pedone 3.0 x64 1CPU NNUE 3256 out of 1000 games (+188 to v2.1!!)
SlowChess Classic 2.5 x64 1CPU NNUE -

25'+8''
last update was November 18th with 1250 new games; we are testing
RubiChess 1.9.0 x64 NNUE (3211/900 games)
KomodoDragon 1.0 x64 1CPU ELO 3494 out of 1000 games + 146
Stoofvlees II a16 -
Igel 2.8.0 x64 NNUE ~ 3248 out of 900 games

5'+3'' pb=on
last update was January 5th:

we are testing
Pedone 3.0 x64 NNUE 1CPU 3248 out of 900 games

3'+1'' pb=on
Last update was December 18th
we are testing some engines below mainlist
https://cegt.forumieren.com/t1414-new-e ... 00-to-3100

A big „Thank you“ to all testers as usual!!

Links

40/20: http://www.cegt.net/rating.htm
Blitz: http://www.cegt.net/blitz.htm
40/120: http://www.cegt.net/rating120.htm
25+8: http://www.cegt.net/rating25plus8.htm
3+1 pb=on: http://www.cegt.net/rating3plus1pbon.htm
5+3 pb=on: http://www.cegt.net/rating5plus3pbon.htm
Tester: http://www.cegt.net/testers/testers.htm
Games of the week: http://www.cegt.net/40_40%20Rating%20Li ... on/gow.jpg

Werner Schüle
CEGT-Team

[mar]

thanks, Werner.
good result for Cheng (better than what I expected), so even if it goes down a bit with more games, I'm glad it holds even at longer TC

[jonkr]

I'm not sure if I'd consider SlowChess 2.5 NNUE. If it's meant to be a generic term for using a neural net to evaluate the positions on the CPU then it definitely fits, but It might cause some confusion since it's not based on the nodchip/Stockfish NNUE and not compatible. It's possible in future I might update my code to make a version that can use nodchip/stockfish NNUE style nets and that training code and call that SlowChess NNUE to differentiate it from my own work.

[Wolfgang]

Hi Jonathan,
did you receive my PM?
My proposal for naming your engine in our lists would be: SlowChess BlitzNN 2.5 x64.
IIRC you wanted "Classic" to be deleted?!

[jonkr]

Hi Jonathan,
did you receive my PM?
My proposal for naming your engine in our lists would be: SlowChess BlitzNN 2.5 x64.
IIRC you wanted "Classic" to be deleted?!

Just saw your PM now I don't always stay logged in. But yeah SlowChess BlitzNN 2.5 x64 is fine with me.

To clarify to anyone wondering, If I add support for the standard NNUE architecture to my inference code to release a NNUE version, I would still plan to train a net to distsribute, but using the nodchip training code with Slow Chess search scoring (and if there are some standard positions people use to train maybe that too), and the "netpath" option would then support plugging in any NNUE net for those that want to. It would be up to testers if they want to test that version, I'm guessing that would be much stronger but don't know unless I try.

[Wolfgang]

...
But yeah SlowChess BlitzNN 2.5 x64 is fine with me.
...

Thats fine!