Hi everybody. I have released a new version of Toledo Nanochess, now is twice as fast (my estimations are better, but I used the lesser number).
The Winboard interface has been enhanced, this time it supports the "level" and "time" commands, so it will do better use of time exploring further depths, making it slightly stronger, also now there is no need for distinct versions.
Also included in the pack is a version fixed at 6-ply depth, mainly for reference/historic purposes.
By the way, now is possible to play black's in the Javascript version.
Enjoy it!. http://nanochess.110mb.com/
Thanks! An astonishing engine. I tried the normal winboard version at medium time controls, like 40/10 min., in Arena 2.0.1 and it typically reached 6 or 7 plies on my old P4-like CPU. I played some "easy" test games, but I lost all of them!
While I assume that the eval function is rather small, it seems to give pretty solid 6 plies at least under these conditions, which in blitz is a challenge for a hobby player like me.
I think I noticed a bug: In positions where the engine obviously wanted to castle short (or where it was the obvious move at least), it played Rh8-f8 or Rh1-f1 only, and let the king in the center. Looks like a castling bug to me.
I'm glad you like it. Sometimes, when it has the opportunity to castle, it moves the rook instead, I could give it more points to encourage castling, but sometimes also does the castle, so I think that it is right (maybe it has something in mind! )
Last edited by nanochess on Thu Aug 13, 2009 5:12 pm, edited 1 time in total.
The compiled executable is available inside Toledo Nanochess Pack, available from my page http://nanochess.110mb.com/chess3.html, look under the section 'Winboard Interface'.
(my estimations are better, but I used the lesser number).
)