AB-testrun of Komodo 14 finished. Finally Komodo could pass Houdini 6 and is now the world's #2 AB-engine. Komodo 14 MCTS testrun will follow after the testrun of Stockfish 200519.
Not only, Komodo 14 passed Houdini 6 in my ratinglist, but Komodo could win vs. Houdini 6 the first time in the head-to-head:
Komodo 14 vs. Houdini 6 pext : 1000 (+199,=644,-157), 52.1 %
Komodo 13.3 vs. Houdini 6 pext : 1000 (+182,=632,-186), 49.8 %
https://www.sp-cc.de
(Perhaps you have to clear your browsercache or reload the website)
SPCC: Testrun of Komodo 14 finished
Moderators: hgm, Rebel, chrisw
-
- Posts: 2434
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
-
- Posts: 670
- Joined: Sun Jan 26, 2020 10:38 pm
- Location: Turkey
- Full name: Mehmet Karaman
Re: SPCC: Testrun of Komodo 14 finished
It seems impossible to catch Stockfish for Komodo now. The difference between Stockfish Dev and Komodo is a huge number like 131 elo.
This situation will negatively affect the motivation of Komodo authors. Because the development of Houdini engine has stopped and Ethereal and Xiphos engines are far behind of Komodo.
This situation will negatively affect the motivation of Komodo authors. Because the development of Houdini engine has stopped and Ethereal and Xiphos engines are far behind of Komodo.
-
- Posts: 476
- Joined: Sun Mar 17, 2019 12:00 pm
- Full name: Henk Drost
Re: SPCC: Testrun of Komodo 14 finished
There's more to chess engines than just having the most ELO.mehmet123 wrote: ↑Wed May 20, 2020 2:40 pm It seems impossible to catch Stockfish for Komodo now. The difference between Stockfish Dev and Komodo is a huge number like 131 elo.
This situation will negatively affect the motivation of Komodo authors. Because the development of Houdini engine has stopped and Ethereal and Xiphos engines are far behind of Komodo.
-
- Posts: 5960
- Joined: Sun Jan 10, 2010 6:15 am
- Location: Maryland USA
Re: SPCC: Testrun of Komodo 14 finished
We are dealing with this in three ways:mehmet123 wrote: ↑Wed May 20, 2020 2:40 pm It seems impossible to catch Stockfish for Komodo now. The difference between Stockfish Dev and Komodo is a huge number like 131 elo.
This situation will negatively affect the motivation of Komodo authors. Because the development of Houdini engine has stopped and Ethereal and Xiphos engines are far behind of Komodo.
1. Adding new features to Komodo
2. Developing the MCTS version, which has a good claim to being the top engine vs. humans (at least with handicaps, the only way they have a chance), and also a good claim to being the top CPU based engine when they are using Multi-PV of 5 or more.
3. Working on NN hybrid version, which could realistically pass Stockfish in the next year if we're lucky, although this is partly a chess.com project so it's not under our sole control.
Komodo rules!
-
- Posts: 670
- Joined: Sun Jan 26, 2020 10:38 pm
- Location: Turkey
- Full name: Mehmet Karaman
Re: SPCC: Testrun of Komodo 14 finished
Is it Lc0 (0 to X move) + Komodo (X to Y move) engine . If it is so we will see another official NN hybrid Lc0 (0 to X move) + Stockfish (X to Y move)
I think in the near future it can come a new Lc0 version developed by running the end of the games.
-
- Posts: 5960
- Joined: Sun Jan 10, 2010 6:15 am
- Location: Maryland USA
Re: SPCC: Testrun of Komodo 14 finished
It won't be anything that simple, and we hope it will use a chess.com NN rather than Lc0, although that remains to be seen.
Komodo rules!
-
- Posts: 670
- Joined: Sun Jan 26, 2020 10:38 pm
- Location: Turkey
- Full name: Mehmet Karaman
Re: SPCC: Testrun of Komodo 14 finished
A new NN engine sounds good. Tough competition will come back. Maybe we will see some new strong engines in near future.