Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!csulb.edu!hammer.uoregon.edu!hunter.premier.net!news.nl.innet.net!INnl.net!feed1.news.erols.com!howland.erols.net!newsfeed.internetmci.com!in2.uu.net!scnews.sc.intel.com!ilnews.iil.intel.com!actcom!news
From: bruck@actcom.co.il (Uri Bruck)
Subject: Re: Programming a computer to play
X-Nntp-Posting-Host: p5.negev1.actcom.co.il
Message-ID: <DzsupJ.G1F@actcom.co.il>
Sender: news@actcom.co.il (News)
Reply-To: bruck@actcom.co.il
Organization: ACTCOM - Internet Services in Israel
X-Newsreader: Forte Free Agent 1.0.82
References: <50cveb$rlf@hil-news-svc-6.compuserve.com> <533d0p$83a@prometheus.localnet.com> <32588e7e.1496486@nntp.hip.cam.org> <53dhgb$qqc@news.ld.centuryinter.net> <53o2b4$7v9@gyda.ifi.uio.no> <53v768$pe8@news.ld.centuryinter.net> <5404gc$s0u@delling.ifi.uio.no>
Date: Thu, 24 Oct 1996 20:36:03 GMT
Lines: 21


>I don't know any estimates of the number of legal chess positions, but your
>upper bound calculation seems correct, and you lower bound guesses seem
>reasonable. However, there is one problem: You will need a very large
>transposition table to store all these positions. Let us assume that 
>you need a table containing 1e20 positions, and that we use 32 bytes to
>store each position. We would then need about 3e12 gigabytes of memory to
>store all positions! Also, with a parallel search, all processors would 
>need to be able to access the entire transposition table.

>Tord

32 bytes for a chess board is too much.
I've seen a scheme once which goes almost as low as 100 bits per
position. I don't remember the details, but it was partialy based on a
"20-questions" style storage, where some of the bits represented
yes/no replies to questions, about the board position.

Uri


