Author: Chessfun
Date: 03:50:45 11/28/04
Go up one level in this thread
On November 27, 2004 at 13:02:14, Kurt Utzinger wrote: >On November 27, 2004 at 11:41:24, Chessfun wrote: > >>On November 27, 2004 at 11:14:02, Kurt Utzinger wrote: >> >>>On November 27, 2004 at 10:03:49, Chessfun wrote: >>> >>>>On November 27, 2004 at 04:19:37, Kurt Utzinger wrote: >>>> >>>>>On November 27, 2004 at 03:33:59, George Sobala wrote: >>>>> >>>>>>Gandalf 6 v Shredder 8 >>>>>> >>>>>>Classical time control: 40/120' + 20/60' +game/60' >>>>>>Shredder Classic GUI >>>>>>Both engines 64M hash, pondering off >>>>>>Books: Shredder using shredder8a.bkt, Gandalf using own book >>>>>>3,4 and some 5 man TBs >>>>>>Centrino 1.6GHz, Windows XP SP2 >>>>>> >>>>>>G6 00½½½½½½0½ 3 >>>>>>S8 11½½½½½½1½ 7 >>>>>> >>>>>>last 5 games PGN will follow >>>>>> >>>>>>match continues ... >>>>> >>>>> It will be interesting to see the result >>>>> of this match after 30-50 games -:) Keep >>>>> up your work. >>>>> Kurt [http://www.utzingerk.com] >>>> >>>>I tried this match as well same settings. Both programs get to 99.9% hash used >>>>in under a minute. IMO It's a waste of time, maximum with this hash setting >>>>should be probably 30 mins game. >>>> >>>>Sarah. >>> >>> Hi Sarah >>> Do you really think that more hash would >>> make a sensible influence on the outcome >>> of this match. I very much doubt. A result >>> of 3-7 can happen, the next serie may be >>> 6-4 for G60 (I hope) >>> Kind regards >>> Kurt >> >> >>I think it can from watching the engines in the game I ran. It seems to me that >>if the engine is at 100% hash then we don't really have a 40/120' + 20/60' >>game/60' time control. And if we believe that engines perform differently at >>different controls then we have a match that is in fact only 30 mins game. >> >>Sarah. > > > Hi Sarah > In my opinion your assumption is not correct. > You could test this by comparing 120'/40 64 MB hash > and 30m/game 64 MB hash and 120'/40 64 MB hash with > 120'/40 and 256/512 MB hash. Search depth of moves > with 64 MB will hardly be different to 256/512 MB > hash but in any case much higher than a 30m/game with > 512 MB hash. According to my tests, the importance > of hash size is much overestimated. > Kurt I had actually tested it on the position he had posted from game 1. However that could be that for whatever reason his PC slowed at some point. While importance of hash maybe overestimated in this case I think the size is far too low for the time controls used. Sarah.
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.