Author: Vasik Rajlich
Date: 06:06:17 01/17/06
Go up one level in this thread
On January 16, 2006 at 12:33:30, Torstein Hall wrote: >On January 16, 2006 at 07:01:01, Vasik Rajlich wrote: > >>On January 16, 2006 at 03:58:38, Joseph Ciarrochi wrote: >> >>>I've been testing the heck out of the newest beta version of rybka, and I don't >>>see a big strength improvement (this was not vasik's intention anyway.). It >>>keeps wining the usual 58-60% against fritz 9 in 4 min 2 sec blitz (n = 60 at >>>present). I am still waiting on the results from a no book tournement, but so >>>far i don't see any increment. (maybe i'll see increment with bigger sample >>>size) >>> >> >>Actually, it was my intention. Beta 10 has a new search - if it's not better, I >>will have something to think about. >> >>Anyway, please post here (or email me) your results. I want to understand the >>improvement in Beta 10, and more importantly I want to understand how to >>configure the various settings there so that Rybka 1.2 has the best default >>configuration. Real work on the search has now stopped, aside from a few very >>small refinements and (mainly) the configuration testing. >> >>Also - it goes without saying, but I love it when testers play _huge_ game sets. >>No sample size is too big. :) > >I did post a 30 games blitz match here: >http://www.talkchess.com/forums/1/message.html?479708 Rybka won +13 =11 -6 Looks >pretty convincing to me. All the games are in the post by the way. > >Torstein > Thanks, I had missed that. Everything is being added up and we'll see what the data says - so thanks. Vas >> >>Vas >> >>>This is not to criticize vasek of course. He can't really be expected to provide >>>big strength improvements, in two months (especially when he has to do extra >>>work to get beta ready for full release). Indeed, Fabien i don't think made huge >>>increases in fruit in only two months. >>> >>>I could be wrong about the amount of improvement of the newest beta. Please post >>>your results (I am motivated to be wrong, as I want rybka to be stronger!!) >>> >>> >>>So......Has rybka's massive success made us overly optimistic about how much >>>engines are going to improve in the next year? I have heard people say rybka >>>will be 100 points stronger in its full release. I am now betting it is no more >>>than 20 or 30 points stronger. This is no reflection of vasek. There is only so >>>much he can do in a couple of months (and I think he is delivering a fantastic >>>product, well worth the price). >>> >>> >>>so if rybka gains 30 points, and shredder and fruit gain about 70, they will all >>>be of about equal strength and fritz will be in fourth place. >>> >>> >>>Please post evidence to prove me wrong (that rybka beta 10d has not gained >>>strength). maybe their are stronger configurations then i am using (default). >>> >>> >>>best >>>Joseph
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.