Author: Heinz van Kempen
Date: 04:25:40 10/19/05
Go up one level in this thread
On October 19, 2005 at 07:00:43, billiau wrote: >Hi, > >Are we sure Fruit WCCC05 and Fruit 2.2 are the same ? >If not, we should contact the Fruit team to be really sure. > >It seems that CEGT used Fruit WCCC05 for the test but i am not sure. >Perhaps Heinz could confirm this. > >Did you have a response from the Fruit team about the hash bug ? >It could explain some inconsistency in the results. > >G. Billiau Hi, Joachim Rang confirmed that they are both identical regarding playing strength there is nothing altered. There will come 150 games more from Ray and also some from other testers for the setting. Concerning the setting I got some mails that the history threshold reducement works better in combination with king safety 104. We would also have tested this, but after all this nasty threads motivation from CEGT testers dropped to zero. Maybe Ray still will do it for helping Fabien. We shall see, patience is needed and no results anymore with less than 300 games. We learnt from that. It is just frustrating to do many hours of work daily for organizing all and only receive critics. I can only repeat, give to me any bigger database inside or outside of CEGT and I will always find a few games that look strange for whatever reasons, mybe bugs of the engine, maybe something with hardware or any GUI. This is unavoidable with thousands of games and statisitically over many games irrelevant. Trying to find something by all means, just because of eagerness to see a certain setting perform better, it is easy to suggest or allege dozens of reasons. I answered again, to be polite and because of your own efforts, but please give me some rest now from posting. Best Regards Heinz
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.