[Computer-go] living by two false eyes
fastgo at gmail.com
Thu Apr 7 11:10:55 PDT 2011
Yes, I use that on 19x19 as well. It is turned on at around 320 steps. The
reason I said it was not very effective is based on comparison between
StoneGrid and MoGo 3.0. Given same time, MoGo 3.0 is much stronger than
StoneGrid on 19x19 on my home PC. As MoGo can perform a lot of more
simulations given same time.
I guess it would help if you simply compare between same engine with one
turned on and the other turned off with same simulations. I did not
thoroughly/seriously tried that, as if I turned it off, there would be a lot
of more end game blunders to be taken care of.
On Thu, Apr 7, 2011 at 6:09 AM, Petr Baudis <pasky at ucw.cz> wrote:
> On Wed, Apr 06, 2011 at 10:11:05AM -0400, fastgo wrote:
> > In StoneGrid, it has built in the Benson algorithm in the playout. It
> > not seem to help much on overall strength. But I think it should solve
> > case here.
> > I think you can implement the Benson Algorithm in the late stage of the
> > playout.
> > It has another advantage finishing the playout in shorter length. In
> > StoneGrid, the playout on average finishes in a little bit more than 70
> > stones on 9x9.
> > You need quite a lot of effort to get it fast as well as correct. So the
> > return of the effort is low.
> Do you also use this on 19x19?
> One of my long-standing worries is that especially on 19x19, the playout
> takes awfully long and the second half of the moves are almost all
> totally meaningless, but with all the expensive heuristics etc. still
> being computed. It seems to me that a lot of time is wasted this way
> and that Benson life detector would help even much more here, so I'm
> surprised you do not see it as that much of an improvement.
> Petr "Pasky" Baudis
> UNIX is user friendly, it's just picky about who its friends are.
> Computer-go mailing list
> Computer-go at dvandva.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Computer-go