Author: Dann Corbit
Date: 10:31:34 08/29/05
Go up one level in this thread
On August 28, 2005 at 12:25:30, Ricardo Gibert wrote: >On August 28, 2005 at 09:39:03, Gian-Carlo Pascutto wrote: > >>On August 28, 2005 at 09:27:21, Roy Brunjes wrote: >> >>>It seems about 2 years ago or >>>so that there was beginning to develop a community consensus that tablebases >>>were essential to strong endgame play by an engine. >> >>Huh? I don't believe this is true. It has always been said that tablebases are >>worth at most 20 ELO or so. > >I would be surprised if it amounted to 5 elo. Wasn't some of the testing >reported finding no significant difference in overall playing strength? The >percentage of the games that would practically benefit from the additional >knowledge EGTBs would provide is surely miniscule. I have seen three tests and all of them showed no Elo difference for using Tablebase files verses not using them. The biggest gain of tablebase files is pretty (ideal) play in the late endgame. Bitbase files tests have all shown modest gains in Elo. Now, if you happen to have an array of fast ultra320 scsi disks at 15K RPM, it might be that you will see a big gain. But most people don't have this and a configuration like that has never been tested to the best of my knowledge. >>Realistically only a few tables like KRPKR and so will actually win or draw a >>game that otherwhise wouldn't. >> >>I wonder about the impact of 6 man tables. It could be bigger. >> >>-- >>GCP
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.