Author: Peter Skinner
Date: 19:57:38 08/12/05
Go up one level in this thread
On August 12, 2005 at 20:20:55, Roger Brown wrote: >I will yield to the experts here but I do not think that tablebases are going to >make a huge difference here. I have seen experiments posted here that suggest >that tablebases are not going to make a meaningful elo difference (particularly >in the context of a basement tournament - that is not an insult, I run them all >the while!) so go ahead with or without. It is up to you which way to go but >installing tablebases for use with chessbase products is simple and fast. I have been running a little experiment over the last few months regarding the importance of tablebases with several engines that use them, and ones that do not. I am currently testing: Crafty Fruit Pro Deo Chess Tiger 15.0 Hiarcs Fritz Of the 6, the only engine that significantly plays endgames better is Crafty. This is probably due to Robert's choice to use the tb's instead of programming vast amounts of knowledge. Of course Pro Deo and Fruit don't use them, but frankly I don't see how using them would make a huge difference. Both play endgames very well, and time management is never a problem. CT, Hiarcs, and Fritz seem to play just as good with them as without them. They might find a win or two faster, but would have found it regardless. Overall my individual testing is only showing a slight elo improvement (20-30elo) when using them. If that is significant to you, then use them. For the everyday average user playing without them is fine. Only when trying to squeeze out every single elo possible would I recommend them. Speaking with several users on ICC over the past few weeks, I was surprised at the amount of operators that _do not_ use them. For the little elo they provide, most are not willing to part with the disk space. Others though (like me) like to have as many as possible to search and grind through. I might not always use them all (most of the time I only use the 3/4/5 man) but I like to have them handy when analyzing games. There have been many people that have done similiar testing, and almost all came up with the same conclusions. The main thing to consider is the amount of endgame knowledge that a specific engine has. If it has very little then use them, but if it does, then save the disk trashing and ram and just fly "solo". Peter
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.