Author: Ulrich Tuerke
Date: 14:01:52 05/01/02
Go up one level in this thread
On May 01, 2002 at 16:54:48, Mike Hood wrote: >On April 30, 2002 at 08:08:28, Ulrich Tuerke wrote: > >>On April 30, 2002 at 07:15:01, Uri Blass wrote: >> >>>On April 30, 2002 at 07:06:18, Ulrich Tuerke wrote: >>> >>>>On April 30, 2002 at 06:42:55, Uri Blass wrote: >>>> >>>>>On April 30, 2002 at 02:48:17, Ulrich Tuerke wrote: >>>>> >>>>>>On April 29, 2002 at 23:55:11, Christophe Theron wrote: >>>>>> >>>>>>>On April 29, 2002 at 21:28:21, Jorge Pichard wrote: >>>>>>> >>>>>>>>Interesting Fritz 1 on a Celeron 433 Mhz search up to depth 13, but Comet B42 on >>>>>>>>the same Celeron 433 Mhz can only search up to depth 12. This is when I set it >>>>>>>>up to game in 30 minutes. Of course that Comet B42 is about 350 Elo points >>>>>>>>higher than Fritz 1, but Comet B42 could be optimized to search higher than 12. >>>>>>>> >>>>>>>>Pichard. >>>>>>> >>>>>>> >>>>>>> >>>>>>>Where have you seen that Comet is 350 elo points higher than Fritz 1.0??? >>>>>> >>>>>>That's certainly overestimated. I 'd expect Fritz1 still being a tough opponent. >>>>>> >>>>>>Uli >>>>> >>>>>I do not know. >>>>> >>>>>I guess that Fritz1 was never tested on the new hardware >>>>>and it is possible that some bug that is relevant only at slower >>>>>time control than tournament time control on 386 prevents >>>>>it to play well. >>>>> >>>>>I cannot know based on the data if the claim that >>>>>Comet is 350 elo better than Fritz1 is correct or not correct >>>>>because we cannot make conclusions based on the rating >>>>>of Fritz1 on slow hardware. >>>>> >>>>>I know that programs may have bugs at long time control >>>>>if they are not designed for it. >>>>> >>>>>One possible problem is that the order of moves can be wrong >>>>>at long time control because the program gives score to moves >>>>>based on some assumptions that are simply wrong >>>>>at long time control. >>>>> >>>>>Uri >>>> >>>>I also think, that Fritz1 can be beaten by many of today's amateurs. However, a >>>>difference of 350 ELO means that Fritz1 would score essentially 0 % ? >>> >>>350 elo difference means more than 0%. >>> >>>I can avoid losing rating based on the israeli formula >>>for rating if I lose 22.5-1.5 against an opponent that is >>>350 elo better than me. >> >>Agreed. 1.5-22.5 is "essentially 0" for me. -:) >> >>Uli >> >> >>> >>>The simple linear formula that is used for calculating >>>the israeli rating is biased when the rating difference is big >>>so the expected result for the weaker player >>>is even better than 22.5-1.5. >>> >>>Uri > >Fritz 1 seems strong to me, but it has serious problems in its time management >algorithm. I've played several 60 minute Blitz games against it where it was in >a clearly winning position, but it lost on time. The time management of the Fritz-1.20 engine under the Fritz5-GUI seemed quite reasonable to me. It played a bit slower than Comet, but didn't loose on time any of the games. Regarding the old DOS-exe, you may be right. Uli
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.