Author: Uri Blass
Date: 15:21:15 02/02/02
Go up one level in this thread
On February 02, 2002 at 18:02:59, Otello Gnaramori wrote: >On February 02, 2002 at 17:14:51, Uri Blass wrote: > >>On February 02, 2002 at 15:33:19, Otello Gnaramori wrote: >> >>>On February 02, 2002 at 14:35:09, Jose Menendez wrote: >>> >>>>On February 02, 2002 at 13:21:54, Otello Gnaramori wrote: >>>> >>>>>I ran an 8 games blitz tournament between Chess Tiger 14 and Fritz7 (latest >>>>>release). >>>>>Here follows the results quite surprising (it's a crush victory for Tiger!) : >>>>> >>>>>Engine Tournament, Blitz:25' 0 >>>>> >>>>> 12345678 >>>>>1 Chess Tiger 14.0 11½10111 6.5/8 >>>>>2 Fritz 7 00½01000 1.5/8 >>>>> >>>>> >>>>>Let me know if you are interested in the list of the games. >>>>> >>>>>w.b.r. >>>>>Otello >>>> >>>> my results were just the opposite, i ran a 9 game match between the two with >>>>Fritz winning 7.0 to 2.0 >>> >>>Which kind of settings do you have ? >>>I have set 24 Mb of cache for both, no tablebase and no book , and which version >>>of Tiger you have ? >>>Mine is 14.0.0.1 reading the engine file Tiger.eng, while the fritz7 is 7.0.0.6 >>> >>>Thanks in advance, >>>Otello >> >>No book? >>I am surprised that you did not get the same game again and again. >> >>It is not the right way to test programs. >> >>I have no problem with testing with no book but it should be mentioned >>otherwise the results may be misleading. >> >>If the target is to test the engines and not the opening book because the >>engines may be used for analysis then it is more logical to use a book that is >>not the default book for both engines and not to use no book. > >Thanks for your feedback Uri. >I did the match intentionally without books to test the engines as you said. >Sorry to not have included in my first post this setting of "no book". >I will repeat the test using a book "not default" as you suggest, but I'm >wondering why the results can be misleading...and I can tell you that the games >are always differents also without books. > >w.b.r. >Otello Most engines are deterministic so I find it surprising if the games are not the same. I also know that Fritz and some other programs had no problem to repeat wins in the ssdf games against programs with no learning function like Rebel8. It is obvious from the result that the games are not the same but I wonder if the programs really play in their real strength. Maybe Fritz automatically is using some random evaluation when you tell it not to use book in order to prevent playing the same game twice and the result is that the engine plays weaker. Maybe other application cause the engines to change their speed during the games and it cause different games I do not know but I find it strange that the games or at least part of them are not the same when I know that repeating the same games happened in the ssdf 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.