Author: enrico carrisco
Date: 11:44:38 12/27/05
Go up one level in this thread
On December 27, 2005 at 06:54:59, George Tsavdaris wrote: >On December 27, 2005 at 02:47:11, Thorsten Czub wrote: > >>sorry. the best results i had was with a version that was never public. >>i am in the moment testing out several styles for the commercial version and >>relating my best results on that with my favourite version that was never >>commercial(see http://www.talkchess.com/forums/1/message.html?473795). >> >>this version got arround 60-65 % against fruit,fritz and shredder. > > Since this H10 version scored so good against these top programs why he didn't >made it the final H10 version? If you know of course.... Was it because it was >good only at short time controls, long TC, because it did worse with other >opponents, etc...? Because it was not stronger than later versions and is not stronger than HIARCS 10 Release. Outside of Thorsten's limited game results, real testing goes on between Harvey, Eric and Myself and it doesn't involve 20 games, personal feelings or monkey's and darts. We run over 1000 games just to isolate a strong beta from the others and then run even more extensive tests on that single version. Thorsten does not receive every beta and therefore cannot speculate on the bigger picture. HIARCS 90578 is, by rough estimate, at least 15-20 ELO behind HIARCS 10 Release (perhaps more.) It is not capable of some of the other goodies that HIARCS 10 incorporates either -- such as automatic adjustment of the forward pruning and increased time management efficiency. Regards, -elc. > >> >>those 3 programs were killed easily by 90578. >> >>maybe you would like first to try the h10 default, and when you are ready >>i will let you know my best setting for h10 and even later my best setting for >>90578.
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.