Author: James T. Walker
Date: 13:02:01 10/31/99
Go up one level in this thread
On October 31, 1999 at 02:28:55, Ed Schröder wrote: >>Posted by Eelco de Groot on October 30, 1999 at 19:07:01: >> >>>I have mixed feelings on the Tiger option. I have mixed results too, from >>>extreme good to bad results. Christophe and I discussed this during Paderborn >>>and came to the conclusion his technique needs some final tuning before >>running >>>optimal in Rebel, a matter of time, I hope. Of course the TIGER technique >>>itself gives Chess Tiger a real push as it is fits in Christophe's approach >>>and way of doing things. >>> >>>Christophe and I often come to the conclusion our programs are totally >>>different in approach and we learn a lot from each other. Lots of fun too. >>> >>>To avoid the fluctuations in match scores the Tiger option isn't rewarded in >>>Rebel's default settings. >>> >>>Hope this answers your question. >>> >>>Ed >> >>Ed , could you maybe say something more about which kind of positions the >>Tiger technique for Rebel handles well and which not so well? > >The Tiger technique as used in Rebel is a highly selective search. > >>The Winhelp manual for Century recommends System Tiger especially for older >>machines, I suppose because then its high selectivity gives better results. If >>that is the case it should also do well in comp-comp blitz-games I would >>think, so for instance as an opponent for the BLITZ_PERSONALITIES? Can this >>be done? > >Sure, just mention the engine settings (menu OPTIONS) together with your >created personality when you contribute for the contest. There will be an >entry form soon on my pages that will cover it all. > >>I don't believe there is a way to use System Tiger inside a personality? > >No. > >>On the other hand the Tiger technique was also used in the special Rebel >>ECTool engine and I think it worked fine there with marathonsize analysis- times. Of >>course I used a rather slow PC for that too... Is the problem mainly one of >>reducing selectivity because Rebel itself already prunes/gives less time to a >>lot of moves at the root with different techniques than Tiger? > >I think it's more because I haven't programmed it in the EXACT way >Christophe told me. I had to change the algorithm a bit to ensure it >would not hurt Rebel's other stuff because Rebel and Chess Tiger >are two different animals. Christophe's idea is a powerful but it >need some final touch to make it work in a balanced way. > >Ed > >>Sorry Ed, I haven't gotten around to much testing of Rebel Century yet. On the >>386 I have two very slow harddisks so I will see if I experience some of Jim >>Walkers problems there, with the harddisk activity under DOS. Couldn't it >>be the result of AUTO232? Jim, maybe if you are reading this, do you still see the >>same effects under DOS if you don't use the 'A' parameter anymore in the >>command-line? >> >>Regards,Eelco Hello Eelco, Yes. The "A" parameter fixed the "Autoplay Series" problem when it was removed but nothing so far has affected the HD driving in DOS. It drives about 2-3 seconds after every move for each side. This is anoying in Blitz but probably would not bother you in Standard time control games if you don't mind your HD grinding after every move. I haven't tried "Howard's" DOS fixes yet. Jim
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.