Author: Uri Blass
Date: 13:07:54 02/25/06
Go up one level in this thread
On February 25, 2006 at 14:01:02, Tansel Turgut wrote: >I would appreciate your experience and recommendations: > >It looks like Rybka seems to be leading all the programs currently. > >1)I am interested to see what happens on longer time controls (infinite >analysis) (ie: if a couple hours/move is allowed) > >is there any data on this? You already asked similiar question here http://www.talkchess.com/forums/1/message.html?488941 there cannot be meaningful data on this unless you use hundreds of computers to test. Assume time control of one hour per move and assume average length of 60 moves. average time per game is 120 hours=5 days. Assume you want to get conclusion in 3 months(after more than it the programs may be old and the data not relevant espacially today when there is huge improvement in computer chess). You can play only 18 games with one computer. If you want 20 programs to play 300 games you need to play 3000 games(20*300/2) and it means that you need 3000/18=150 computers to get something important. > >2) Do the leading programs superiority continue, or do some other programs do >better when couple hours of time given for each move? I do not know but my assumption is that the leading programs are still superior unless I see huge branching factor or mistakes that happen only at long time control when programs can easily refute them. > >3)I am also interested to see how Hiarcs 10 does if all engines are give long >time to analyze. It sees alot, but it is very slow and do not >see many plies deep compared to other programs. I do not have hiarcs10 but I read that it is very good program in solving test suites at short time control and it is even better than rybka in this task. This strongly suggest that hiarcs is a very fast engine. number of plies may be misleading because there are extensions and hiarcs may see deeper in the important lines. 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.