Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shredder 10

Author: Mridul Muralidharan

Date: 10:49:44 02/18/06

Go up one level in this thread


On February 17, 2006 at 19:26:44, Uri Blass wrote:

>On February 17, 2006 at 16:42:01, Mridul Muralidharan wrote:
>
>>On February 17, 2006 at 14:47:01, Joseph Ciarrochi wrote:
>>
>>>right now there is over 100 point difference between shredder 9 and rybka 13b
>>>(cegt 40/40). The difference might increase another 10 to 50 points with endgame
>>>knowledge
>>>
>>>Historically, what is the largest increase a chess engine has made? I think
>>>fruit gained about 100 points at one point?
>>
>>I think it was mentioned that, among strong programs , HIARCS has been
>>consistently gaining 100 elo or so between versions. Same with fruit.
>>Among amateur programs, it is entirely different story - there have been even
>>350 elo or more improvements :-)
>>
>>Mridul
>
>This is not correct for hiarcs.
>
>Hiarcs10 is more than 100 elo better than hiarcs9 but it was not the case in the
>past.
>
>This is also misleading to say that hiarcs gain 100 elo between versions because
>no new hiarcs was released for a long time and the time between hiarcs9 and
>hiarcs10 is certainly more than one year.
>
>Fruit usually gain 100 elo in clearly less time than one year.
>
>Uri

I dont remember mentioning the time gap between the releases :-) Even if it took
10 years between releases : it is factual when you say - between major versions
, there was a 100 elo improvement.
Also , I think I was purposely fuzzy about the actual elo improvement.
I never did any measurement myself ! I was just reporting what I was informed of
long ago.
From what I am to gather (and also from the hiarcs website) , it has been quite
consistently improving in the whereabouts of 100 elo or so for most of the past
releases. Ofcourse , I could be wrong on this count.

Just as a note : tougher to improve a professional program (so always having a
high elo anyway as compared to competition) with a known history of gameplay
style consistently.
I know of quite a lot of professionals who prefer HIARCS since they do not
consider it a bean counter with a computerish style of play.
Personally , I usually ask my friends to run games of my engine against HIARCS
mostly - tactical mistakes can be corrected more easily than positional ones.

Not to say fruits achievements are small - it is quite impressive , but I want
to see how long it will continue at this level between versions :-)

Mridul



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.