gnugo-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnugo-devel] handicap problem


From: Gunnar Farneback
Subject: Re: [gnugo-devel] handicap problem
Date: Sat, 23 Mar 2002 21:13:30 +0100
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode)

Morten wrote:
> I'd like to test gnugo  with a board number from 2 through 38,

Officially GNU Go only supports board sizes from 9 to 19. Other sizes
may or may not work. Sizes larger than 21x21 (for 3.0.0) requires
changing the MAX_BOARD constant and recompiling. Sizes larger than
32x32 definitely don't work properly with default settings but may
have a chance to survive if the caching of read results is turned off.
I'm not aware that anyone has actually tested how far the limits can
be pushed in either direction.

The ascii interface and GTP should be able to handle up to 25x25
boards. I don't think we have any interface that can handle larger
boards. 

> With handicap stones . scaling from 0 to full board. 

In GTP mode and with the ascii interface it's possible to set up an
arbitrary handicap position before starting the play, but GNU Go may
or may not handle it intelligently. The go modem protocol doesn't
specify any fixed handicap placements larger than nine stones and for
even sizes or boards smaller than 9x9 none at all.

> cgoban says it can handle boerd sizes as large as 38 x 38,  
> a cgoban error says it can permit games as large as 22 x 22,

That may depend on which mode it's working in. With the go modem
protocol it's technically impossible to play on anything larger than
22x22. (Well, it might actually be possible to play on larger boards
as long as you don't feel any need to play stones above the first 512
vertices. :-)

> I can't start a 22 x 22 game at all.

It might work if you increase MAX_BOARD to 22, as commented above, but
I don't think anyone has tried that. Playing against GNU Go on
unusually small or large boards will require a lot of investigative
work, but please report any conclusions you reach. This is pretty much
uncharted territory.

/Gunnar



reply via email to

[Prev in Thread] Current Thread [Next in Thread]