Author: Uri Blass
Date: 16:40:58 10/19/02
Go up one level in this thread
On October 19, 2002 at 19:16:04, Vincent Diepeveen wrote: >On October 19, 2002 at 17:59:05, Jesper Antonsson wrote: > >I am under the clear belief that Fritz eval is much >better than that of deep blue. Of course it's not a fair >compare. Fritz is made in 2002. Deep blue in 1996-1997. > >It is comparing an airplane from 1910 with an airplane 2010. > >computerchess progresses so much and so fast. I am sure >that programs from 2002 soon will be having major problems >with generations from 2007 too. > >the best example is programs from 1997 running them on >todays hardware. for example this tournament ZZZZZZ from 1997 >is joining. > >when reaching 9 ply, then getting to 10 and 11 ply is a major >problem. branching factors from 10 are pretty normal for it. I suggest you try Genius3 This program is from 1994 and it has not the problem that you describe. I ran it for many hours on p100 in the past and it seems to have a stable branching factor of less than 4 in the middle game. I think that programs with big branching factors of 10 simply had bugs because the programmers did not test them at long time control. For example if you use 16 bit numbers for history table then the order of moves by history table can be wrong after you search enough nodes. 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.