OpenChess

Homayoun wrote: Sat Jun 15, 2024 10:20 am Hi Sedat
Thank you very much for tests. New strong engines ( Raptora and killifish ) also arrived. I expected so much from obsidian 12.29. I found it very strong .Let’s see what happens in continue.
Best regards

Sedat Canbaz wrote: Sat Jun 15, 2024 10:45 am

Homayoun wrote: Sat Jun 15, 2024 10:20 am Hi Sedat
Thank you very much for tests. New strong engines ( Raptora and killifish ) also arrived. I expected so much from obsidian 12.29. I found it very strong .Let’s see what happens in continue.
Best regards

Dear Homayoun,

Not at all...

Yes I wonder too about what will RapTora + Killfish's performance..
Because both engines (sure previous vers) managed to be Champs...
But you know, latest CS is one of hardest..so no any idea what will
Be next results...but I think that it is worth to give a try...
In any case, FUN comes as 1st..rest does not matter much, right ?)

About Obsidian 12.29 engine,
Frankly the latest Obsidian is doing good (under these cond.)
As we see, about 40 Elo behind than Stockfish plus it's based...
Anyhow..it's very hard to be a competitor vs many SF authors...
That's why..since more than 10 years: SF leads in major ratings!!
And to be honest, I can't remember any better/greater record !

Actually once Stockfish 8 has been defeated (vs AlphaZero) e.g
With clear points difference, but the match cond were not fair...

Greetings )

Homayoun wrote: Sat Jun 15, 2024 10:20 am I expected so much from obsidian 12.29.
Best regards

Homayoun wrote: Sat Jun 15, 2024 1:36 pm Yes. This is the fact that I personally don’t like too. But I can’t understand. For example, Stockfish is an engine with alpha-Beta search , while Obsidian is an engine using nets provided from leela data. And we know that leela is using Mcts search method. I think in different positions they should select different moves and different ways. In this case the similarity should be less than this. I am using obsidian 12.29 in some special lines in openings that I know most engines have difficulties to play and evaluate, for example some lines of Budapest defence or Benoni defence for my personal practice. And I am satisfied form obsidian more than stockfish 16.1 official in these variants. It plays and evaluates these lines better. Obsidian 12.29 shows great positional understanding although it’s weaker than Stockfish 16.1. It is very reliable engine for this purpose. The only defect of this version of Obsidian (12.29) is that engine is not very aggressive but in positional playing it’s fantastic. I would like to know your idea about Titan and Uralochka engines too. (Their strength, similarity tests and may be other parameters you want to mention to. )
Best regards

OrgZ wrote: Sat Jun 15, 2024 8:43 pm Thanks Sedat
For private19sf you currently have Private19sf_002b, the other stronger variant is the publicly released one, when i gave you that one, Tom and Alex were working on some crazy parameters. at that moment when i gave you, your variant was stronger and better than its father. but later on new stronger parameters were tuned .
buts its ok Friend. lets give others a chance

Code: Select all

2m+0.20s
Score of Private19Sf_002b vs Private19Sf_sse41: 14 - 17 - 219 [0.494]
...      Private19Sf_002b playing White: 8 - 6 - 111  [0.508] 125
...      Private19Sf_002b playing Black: 6 - 11 - 108  [0.480] 125
...      White vs Black: 19 - 12 - 219  [0.514] 250
Elo difference: -4.2 +/- 15.1, LOS: 29.5 %, DrawRatio: 87.6 %
250 of 250 games finished.

looking forward to a big competition from Killfish

Stay well
TR