I just tried the 64 bits compile, and noticed that Sting time management is very bad. It just lost the game on time in completely won position. Can Sting time managemnt be improved, so it doesn't lose games on time ?
I imagine it would not be too difficult.
Sting time management code is probably roughly compatible with current Stockfish, so Stockfish time management could be transplanted.
I am trying to run a match vs other engines and am not interested to use this engine for analyzing. There are many other engines who are much stronger than Sting who I use to analyze my games. I am interested to see how this engine performs vs other engines.
Thank you Dann for your efforts.
Indeed now Sting is able to find the best move leading to fortress, but it does not return value draw in such positions.
The exception described by you (8/p7/qkpN1p2/8/1B1P4/PP3K2/5P2/8/ w - -) regards zugzwang position and why Sting returns Value ZERO it may be only a bug. I will take a look at this.
Indded, Now I see that there is a bug, probably in very highly scored (+ or -) positions.
I wiil prepare the version.
Damir, It is a fact, that Sting does'nt like terminated time menagment.
Thank you for information.
lech wrote:I am asked to join Sting's ideas with newest Stockfish.
My answer:
Just something is valuable when was stolen.
Seriously, I respect authors of Stockfish and I am far to do it whenever.
My project should join not divide.
BTW. It is a fact, I am very tired.
According force prevailing knowledge among fans.
The rating numbers mark the way.
It is forgivable. But it is interesting to see if their progress can give increased strength, however small.
It is interesting to try.
A pity you do not document your code or distinguish your changes.
Everything can try, is very interesting.
lech wrote:I am asked to join Sting's ideas with newest Stockfish.
My answer:
Just something is valuable when was stolen.
Seriously, I respect authors of Stockfish and I am far to do it whenever.
My project should join not divide.
BTW. It is a fact, I am very tired.
According force prevailing knowledge among fans.
The rating numbers mark the way.
It is forgivable. But it is interesting to see if their progress can give increased strength, however small.
It is interesting to try.
A pity you do not document your code or distinguish your changes.
Everything can try, is very interesting.
Jose, the code of Stockfish's 2.1.1 is available on Stockfish's www page. There are many tools to display differences in failes.