gnugo-devel
[Top][All Lists]
Advanced

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

Re: [gnugo-devel] gnugo lost


From: Arend Bayer
Subject: Re: [gnugo-devel] gnugo lost
Date: Sun, 24 Mar 2002 15:34:04 +0100 (CET)

On Sun, 24 Mar 2002, Morten Gulbrandsen wrote:

> How many handicap stones does gnugo need,
> in order to beat any professional playing 9 x 9 ?
> 
> How many minutes, (hours)  per move does gnugo need with 
> 4 handicap stones playing black  9 x 9 in order to 
> firmly beat any dan ranked player ?
I cannot beat GNU Go at 4 stones with no time limit (I am EGF 3D). GNU Go
needs around 20 min for such a game (on my K6-II 400 MHz).
I don't think GNU Go would get much stronger if you would make it think
longer (by increasing reading depths etc.). I don't think anyone has
managed to write a go program that scales well with increased computing
power (this is exactly opposite to the case of chess programs AFAIK).

> Attached is a gnugo game, with me.
> main time 5 minutes. What time does gnugo need 
> in order to give me good moves ? 
> 
> 5 minutes main time and 20 stones in 5 minutes Byo yomi 
> sometimes can be a problem for gnugo.
> 
> If gnugo does not find a reasonable good move, 
> it is no shame to resign. Or just try a randome move.

The autolevel feature is supposed to make GNU Go hurry if time is running
out. However, you have to add "--autolevel" in the command line when you
run it from cgoban via Go Modem.

Hmm, maybe autolevel should be triggered by default if time limits are
given? This would definitely be more intuitive.

Arend





reply via email to

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