Author: Uri Blass
Date: 16:36:13 10/12/05
Go up one level in this thread
On October 12, 2005 at 17:26:33, Heinz van Kempen wrote: >Hi all :-), > >we have some problems in CEGT with Deep Junior 9.0.0.3 and apparently it slows >down with 512 MB hash, although using two processors, while Deep Fritz produces >good results with this hashsize using two CPU´s. > >Anyone has results for Deep Junior 9.0.0.3 or other versions? > >Did someone detect bugs and not reproducable moves? Any other observations? > >Best Regards >Heinz Why do you give Deep versions more hash than other versions. I understand giving deep versions 2 processors against 1 because deep version can use 2 processors when not deep version cannot use them but I think that it is better to ask the programmers how much hash to give to their program in relevant time control. There is no reason to give deep version more hash unless the programmers want it(otherwise you do not test only the difference between deep Junior and Junior but also the difference between 512 mbytes and 256 mbytes) I also think that if you cannot use 512 mbytes hash for every programmer who like it then it may be better not to use 512 mbytes hash at all. Note that for Movei 64 mbytes are probably enough and I guess that more hash may give inferior results(this is only a guess and I did not test it in games). 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.