Author: Joseph Ciarrochi
Date: 03:38:12 01/10/06
Go up one level in this thread
On January 10, 2006 at 06:32:02, Thorsten Czub wrote: >On January 10, 2006 at 06:10:34, Joseph Ciarrochi wrote: > >>But hiarcs 10 seems to get fairly killed on the cegt list, as does toga1.1 >>(http://www.husvankempen.de/nunn/cegtrating4040all.html#3). >> >>what do you think are the optimal conditions for toga and hiarcs to defeat >>rybka? >> >>(i am so tempted to buy hiarcs now) > >from my machines i can say that rybka had problems with "styled" hiarcs10, >and with togaII 1.1a. > >While hiarcs10 (in default) had problems with rybka, but beated >fruit, f9 and s9 (60-65%). > >since rybka had no book i used j9 for rybka. all others used their own book. >i do test 40/40 or 25+10. > >I have no idea how the CEGT testing is different from mine. > >IMO in the moment it makes not much difference what to buy. >fruit, rybka, hiarcs. they are all good programs. >it seems to be more a question about taste. >if i were you e.g. i would not buy f9 because it is still the same 0.00 ... fail >low - oh oh program. >of course with more knowledge over all. but the sero score and suddenly fail low >behaviour is still in it. >this is not useful for me. >fruit on the other hand is still (IMO) to much search based program. i would >apreciate if fruit would develop into a less search based program. > >it seems hiarcs and rybka are knowledged based programs. i would say rybka is >more radical in this then hiarcs is (so far). >On the other hand rybka still has many lacks of knowledge that hiarcs does not >have. >A problem is that hiarcs plays silly book moves in wrong CB-GUI's. i read in >another german forum (kurt utzinger remarked this) that HIS hiarcs10 played >1.e4 e5 2.a3?! although a3 was marked as weak move. >When others tried to replay this, it came out that kurt used an older version of >the GUI. >this older version of the gui (i think it was october 2005 GUI) is still very >common and used by many people. with e4 e5 a3 the "bug" is very easily to see. >but i doubt that you will see it very good in OTHER more deeper opening >situations. >This weakens hiarcs10 depending on the GUI the people have when using the >original h10 book. >for rybka i guess there exist many different versions. >and i think rybka is not very happy when running on slow machines or machines >with SMALL cache such as celeron. > >what to buy ?! >depends on what you have so far. >and what you do with the program. i found hiarcs a very good program for >analysis. >when i replay chess games of all kind of humans and programs, hiarcs gives me a >very precise insight of the game.
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.