Author: Uri Blass
Date: 06:18:08 12/06/00
Go up one level in this thread
On December 06, 2000 at 08:31:06, Frank Quisinsky wrote: >On December 06, 2000 at 06:07:42, Thorsten Czub wrote: > >>On December 05, 2000 at 17:29:20, Christian Koch wrote: >> >>>Hi Frank, >>> >>>a result 29 - 9 is not a difference by 300 ELO points!! >> >>look: on 1000 Mhz machine, playing 40/120 games, you would maybe >>get 50%. >> >>you play half speed (500 Mhz) and at least 3 times faster time-controls. >>in this mode gandalf cannot show its strength and gets outsearched. >>the result would change much if you would give faster machines or >>longer time controls. > >When a chess programm make with longer time controls over 50% against the best >chess programs in the world but maybe 35% - 40% with blitz level is my comment >clear. I think the program has at the moment the best chance for the future. > >And this is important. > >Best >Frank Unfortunately we cannot know from the information that gandalf gets 30-40% at blitz and 50% at tournament time control that if it can get more than 50% at longer time control. I hope that it is better but it is also possible that the same program does 30-40% in blitz,50% at tournament time control on 1000 mhz and again 30-40% at 20 hours/40 moves. The only way to know if Gandalf is better at longer time control is to test it in hundreds of games and getting the games at 30 minutes per move time control can take a lot of time unless there are a lot of testers who test in this way when everyone of them give few 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.