Author: Mike S.
Date: 18:59:05 11/16/02
Go up one level in this thread
On November 16, 2002 at 20:28:52, Mike Byrne wrote: >On November 16, 2002 at 19:34:04, Anthony Martini wrote: >(...) >> http://www.uib.no/people/pfvaf/chesslib/ >> >> ... but they are 7 3/4 GB in total !!! >> >> (don't ask how long they took me to D/L - & I have broadband) >I have 56K and it took me abour 3-4 weeks - you need to use an FTP clinet that >can resume where it left off when you get disconnected ..and you will get >disconnected on 56K ... It seems reasonable to me in case of a 56K modem only (which I have too), to download just the small tablebase generator program and generate the complete 3- and 4-piece tbs subset only. These are already very useful for the engines, and I don't expect that many games will be lost just for the difference of 4-piece to 5-piece tbs., and the slowdown effect will also be smaller (which mean engines may even win a few games which they'd have lost with the 5-piece tbs., although I don't doubt that the 5-piece tbs. give a slightly better score in general). The generation process will take a few hours only, depending on cpu speed, for the 3- and 4-piece tablebases (~ 30 MB). The best compromise is probably these + all KR#-KR (or at least KRN-KR, KRB-KR to cover 2 important draws and their exceptions) from the 5-piece range. But AFAIK a lot of RAM is necessary to generate them, so maybe those must be downloaded. I don't know if the URL above has the generator; alternatively try: ftp://ftp.cis.uab.edu/pub/hyatt/TB/ (retry later if unreachable) Regards, M.Scheidl
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.