Author: Uri Blass
Date: 21:47:40 07/25/02
Go up one level in this thread
On July 25, 2002 at 23:29:51, Dan Andersson wrote: >The possible speed loss from probing EGTBs also factor in. Thus it is non >trivial to show that the net effect is positive. Observing games is not a valid >experiment. You actually need a lot of games played with and without EGTBs to >validate the hypothesis. > >MvH Dan Andersson There is a speed gain and not a speed loss from tablebases if you probe them only when the remaining depth is big enough(it means probing them only when the expected time of the search is bigger than the expected time of the probing). The only speed loss that you can be sure about is in positions when there are no tablebases hits because the program need to check if the number of pieces is at most 5 and the answer is always no but I believe that this speed loss is less than 1%. Uri
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.