On Feb 23, 4:36 pm, yearlypap08 <yearlypap08.2062...@[EMAIL PROTECTED]
>
wrote:
> could try Rybka on an easier level, or else try the GNU engine.
> The GNU chess engine is much stronger than Jester,
> Try a game with GNU vs. GetClub and see what I mean.
> > http://chess.delorie.com/
> > The default is 1 second think time per move. (Its range is 1/10th sec
> to 2 min think time per move)
ok, i've done that, getclub at easy level, and delorie/gnu at
approx 1 min/move but in fact it moved much faster.
even although GNU was choosing a risky (Cochrane) gambit
sacrificing a knight in the opening (never do that against
rybka i would say..), Getclub was easily beaten
white: GNUchess/delorie
black: getclub
1.e4 e5
2,Nf3 Nf6
3.Nxe5 d6 (Petrov)
4.Nxf7?! Kxf7 (Cochrane gambit)
5.d4 Qe7
6.Nc3 Ke8
7.Bg5 Nc6
8.Bc4 a6
9.0-0 Qd7
10.Rfe1 Be7
11.e5 dxe5
12.dxe5 Ng4?
13.Nd5 Bg5
14.e6! Qd8
15.Qg4 b5
16.Qf3 Rhf8
17.Qh5 g6
18.Qxh7 Nce7
19.Nde7 Rf2
20.Nd5 Qd6
21.Qg8 Rf8
22.Qg6 Kd8
23.Qg5 Ke8
24.Qh5 Kd8 (black, i.e. getclub no says 'i'm winning boy.. :- )
25.e7 Qe7
26.Ne7 c6
27.Qc5 Kd7
28.Qc6 Kc6 and resigns
1-0 (sorry for the imperfect notation, entered it
by hand, so it might have some little mistakes
while getclub seems to have some potential to become
a better engine, i would advise Sanny to develop
a standalone version in c or so, compatible with UCI
or winboard, and start some testing against other engines.
after move 20/25 or so kings attack seems to understood
quite bad by the engine (also seen this in one test game
against rybka).
and while the java interface isnt really bad, some
features could be improved of course, especially
the notation (like eg. P-to-e4 for pawn move,
why not make it compatible with pgn notation)
PS but personally i wouldnt be surprised if
the chess game is just a test case and Sanny
would be more interested to get participants
in his clubs to sell it later for a few millions
to google or so; certainly a more sensible
business strategy then trying to build
yet another best chess or Go engine..
:)
|
