CEGT - rating lists March 10th 2024

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

40 / 20:
New games: 3.117 with 20 different engines
Total:         1.775.568     games

NEW Engines
116 Black Marlin 9.0 x64 1CPU 3434 +17 -17 1106 games (+79 to v. 8.0)
25 Obsidian 11.0 x64 1CPU 3537 +17 -17 1026 games (+ 17 to v. 10.0 )
144 Velvet 7.0.0 x64 1CPU 3412 +17 -17 1229 games (+ 39 to v. 6.0.0 )

UPDATES
131 Lizard 10.2 x64 1CPU 3422 +16 -16 1210 games (-1)
1 Stockfish 16.1 x64 1CPU 3605 +14 -14 1512 games (-7)


40 / 4
last update was February 21th 2024
New games: 18.100 ; total now: 3.760.630  games (3174 engines, +12)

we are testing:
Obsidian 11.0NN x64 1CPU
Viridithas 12.0.0NN x64 1CPU
Clarity 5.1.0NN x64
Testing Patricia 1.0NN: ~2858/1200
Expositor 4WR02 x64 1CPU:~ELO 3266 / 2100 games =>  -new-

Black Marlin 9.0NN x64 1CPU: ~ELO 3462 / 2500 games => +50 to v. 8.0.0 (3412)
Starzix 4.0NN x64: ~3439 / 1000 / +49
Obsidian 10.0NN x64 1CPU: ~3570 / 2000 / +34
Stockfish 16.1NN x64 1CPU: ~ELO 3641 /  2200 games => +1 to v.16.0 and -17 to Dev240214 (3640/3658)
Velvet 7.0.0NN x64 1CPU: ~ELO 3457 / 2100 games => +40 to v. 6.0.0 (3417)
Stockfish 20240214NN x64 1CPU ~3658 / 2000 (+18 to SF 16.0)
and: https://cegt.forumieren.com/t2113-tournaments-with-new-engines-versions
Altair 7.0.0NN x64 1CPU ~ELO 3451 / 2000 games => +54 to v. 6.0.0 (3397))
Hiarcs 15.3 x64 1CPU: ~ ELO 3196 / 2100 games => -13 to 15.2 (3209)
Odonata 0.8.0 x64 1CPU: ~ 2635/1300
Caissa 1.17NN x64 1CPU: ~ ELO 3563 / 2000 games => +16 to v. 1.15 (3547)
Leorik 3.0NN x64: ~ELO 3126 /  1800 games => +360 to v.2.5 (2766)
Reckless 0.5.0 x64: ~ 3103/1200 (Reckless 0.3.0 = 2479)
Willow 4.0NN: ~ ELO 3407 / 1800 games => +138 to v. 3.1 (3269)
Winter 3.0NN x64 1CPU: ~ ELO 3232 / 2200 games => +117 (!) to v.2.0 (3115)
Alexandria 6.0.0NN x64 1CPU: ~ ELO 3527 / 1600 games => +24 to v. 5.1.0 (3504)
Lizard 10.1NN (NEW) ~3370 / 2000
Midnight 9NN x64: ~3357 / 1200 / +104
Booot 7.3NN x64 1CPU: ~ ELO 3448 / 800 games => +21 to v. 7.2 (3427)
Tucano 11.00.1NN x64 1CPU: ~ 3264 / 2000 games => -14 to v10.00

5'+3'' pb=on
last update is from February 29th with 14.100 new games, total now 517.350    with 389 Engines/Versions (+7)

we are testing:
Caissa 1.17 NN x64
Velvet 7.0.0NN x64
Tucano 11.00.1NN (Best Versions List)


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%20List/Coordination/gow.jpg

Werner
CEGT-Team

The hiarcs versions do not really make any progress.

Yes, I think he needs some help from Chris

According to Mark, the engine is about 3 Elo stronger than Hiarcs 15.2, tested with 140,000 games. There is a new engine Hiarcs 15.3.1, but it shouldn't be stronger than 15.3.

I dont think NN is the way to go for Mark.

Mclane wrote:

I dont think NN is the way to go for Mark.

Is it Mark who is still working on Hiarcs?

Yes, according to Mark, he has been working on a NNUE version of Hiarcs for about 1.5 years. I can't understand why this shouldn't be a progress.

Stockfish and Lc0 have massive teams with gigantic hardware farms. They can throw a thousand CPU cores or a million GPU cores at testing a new feature. It is very, very hard for any small team to compete with that.

Dann Corbit wrote:

Stockfish and Lc0 have massive teams with gigantic hardware farms. They can throw a thousand CPU cores or a million GPU cores at testing a new feature. It is very, very hard for any small team to compete with that.

Yep.

Ok i downloaded latest version and try it out.

So far it was one of hiarcs strength that it has a handmade evaluation function.

A slow engine, unique and with a known style of play.