Article View: gnu.chess
Article #4520Killer moves
From: Kong Sian
Date: Tue, 21 Mar 1995 04:07
Date: Tue, 21 Mar 1995 04:07
21 lines
1538 bytes
1538 bytes
> > BTW, yesterday I played around with the part of search.c where the > history and killr are saved. My preliminary conclusion is if we save > them only when (best > beta) it saves 30% less redundant moves and the > program plays better. Of course I removed killr3 and Swag3. I can't > play with it again for the next 3 weeks so you guys might want to test > that idea. > > Johanes Suhardjo (johanes@farida.cc.nd.edu) I've played around with this, i.e. not keeping the move as a killer unless best > beta. On a few positions, the newer approach tends to show about 5-10% improvement. This looks like a good thing, so it will go into pl75. Regards. -- Chua Kong Sian <kongsian@nsrc.nus.sg> National Supercomputing Research Centre Singapore
Technical Details
Message-ID:
<9503210410.AA24368@aba.nsrc.nus.sg>Path:
arc1-ca.pugleaf.net!nntp.gmd.de!stern.fokus.gmd.de!ceres.fokus.gmd.de!zib-berlin.de!news.cs.uni-magdeburg.de!RRZ.Uni-Koeln.DE!news.dfn.de!swiss.ans.net!newsjunkie.ans.net!gatech!swrinde!cs.utexas.edu!math.ohio-state.edu!cis.ohio-state.edu!nsrc.nus.SG!kongsianReferences:
<9503161507.AA24739@rice-chex>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 10d71d,fb62cff6dea1bf09,start X-Google-Attributes: gid10d71d,public X-Google-ArrivalTime: 1995-03-20 20:07:19 PST Path: arc1-ca.pugleaf.net!nntp.gmd.de!stern.fokus.gmd.de!ceres.fokus.gmd.de!zib-berlin.de!news.cs.uni-magdeburg.de!RRZ.Uni-Koeln.DE!news.dfn.de!swiss.ans.net!newsjunkie.ans.net!gatech!swrinde!cs.utexas.edu!math.ohio-state.edu!cis.ohio-state.edu!nsrc.nus.SG!kongsian From: kongsian@nsrc.nus.SG (Kong Sian) Newsgroups: gnu.chess Subject: Killer moves Date: 20 Mar 1995 23:07:19 -0500 Organization: GNUs Not Usenet Lines: 21 Sender: daemon@cis.ohio-state.edu Distribution: gnu Message-ID: <9503210410.AA24368@aba.nsrc.nus.sg> References: <9503161507.AA24739@rice-chex>