Page 1 of 1

Krill, What do you think of te new endgames for Shredder 10?

Posted: Mon May 15, 2006 4:49 pm
by gigabyte137
Can we convert what we have? Can we use them in any chess engine or Just Shredder 10? Any info you have is greatly appreciated. I see that the 345 tables are like 135 mb's compared to 7.1 Gigs on old system. Any thoughts?

Re: Krill, what is with the new endgames for Shredder 10?

Posted: Mon May 15, 2006 5:12 pm
by Kirill Kryukov
gigabyte137 wrote:Can we convert what we have? Can we use them in any chess engine or Just Shredder 10? Any info you have is greatly appreciated. I see that the 345 tables are like 135 mb's compared to 7.1 Gigs on old system. Any thoughts?
Shredder bases are some kind of WDL, Nalimov are DTM. So Nalimov bases have more information. That's why shredder bases are so small. Note that this case is similar to existing examples of The King bases, Scorpio bases, Diep bases, etc.. Shredder bases can be used by Shredder only, and there are no 6-men bases yet. So no revolution so far.

Nalimov bases are open, since the probing code is abailable (more or less). This is critical condition for any new format to rival Nalimov, at least to me. I don't want to lock my experience to just one engine.

I don't have Deep Shredder 10 yet. Comments from anyone who tried it are most welcome!

Open BB format?

Posted: Tue May 16, 2006 6:52 am
by jshriver
Kirill Kryukov wrote: Shredder bases are some kind of WDL, Nalimov are DTM. So Nalimov bases have more information. That's why shredder bases are so small. Note that this case is similar to existing examples of The King bases, Scorpio bases, Diep bases, etc.. Shredder bases can be used by Shredder only, and there are no 6-men bases yet. So no revolution so far.
!
I have a suggestion for a possible project. The generation and creation of a BSD licensed or public domain WDL dataset. That way there would never be an issue over using the datasets or the code.

I've read on other forums were people working on bitbases have just taken the data out of nalimov egtb's for generating bitbases. Though the whole legal issues surrounding whether the data itself is licensed or just the code is unclear. If this is a problem, I wonder how hard it would be to modify H.G. Muller's egtb generator for generating 1-6men bitbases.

I'm not great at chess programming (still a lot to learn and trying to digest his generator) but I'd love to work on creating an open egtb or egbb format that's tight and fast.

Comments, suggestions, oppinions?

Mr. Muller would you be interested in working with me on this?

-Josh

Posted: Wed May 17, 2006 12:53 pm
by Arpad Rusz
Can a WDL base, like the new Shredderbase, use the 50 moves rule?
If I understand corectily, Shredder 10 UCI uses for playing the Nalimov databases and only for analysis the Shredderbase, exactly for that reason.