Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: New shredder 9/fruit test starting

Author: Uri Blass

Date: 03:12:36 06/18/05

Go up one level in this thread


On June 18, 2005 at 04:07:23, Fabien Letouzey wrote:

>On June 17, 2005 at 20:11:29, George Tsavdaris wrote:
>
>>On June 17, 2005 at 18:57:27, Dann Corbit wrote:
>>
>>>This is a first pass test (high speed, 5'+1") -- just testing the interface now.
>>>
>>>This is done under the Shredder 9 classic GUI.
>>>
>>>  Program                                   Elo    +   -  Games Score  Av.Op.
>>>Draws
>>>1 Fruit 2.1 (Brian's Build; Dann's Bk)    : 2782  511 462    4  62.5 %  2694
>>>25.0 %
>>>2 Fruit 2.1 (Brian's Build; Gordon's Bk)  : 2782  511 102    4  62.5 %  2694
>>>75.0 %
>>>3 Fruit 2.1 (Gordon's Build; Gordon's Bk) : 2782  511 462    4  62.5 %  2694
>>>25.0 %
>>>4 Shredder 9 UCI                          : 2711  600 600    4  50.0 %  2711
>>>0.0 %
>>>5 Fruit 2.1  (Gordon's Build; Dann's Bk)  : 2441  379 262    4  12.5 %  2779
>>>25.0 %
>>>
>>>I will run several hundred games over the weekend.
>>
>> I hope it will not be at this time control, right.......:-)
>> My wish would be for classic time controls, but i know time is never enough for
>>this to hit the one hundred limit of games...........So a factor of 20 seconds
>>per move would make me happy......The final choice is yours of course..........!
>>
>> This damn Fruit(I observe this especially with TOGA II) seems to be the first
>>free engine that threats the great and living at another galaxy Shredder 9! I'm
>>sure that Shredder will show who is the boss...........In order for this to
>>happen, i expect a 60% or more score..........
>>
>>>
>>>My earlier test was definitely flawed, Shredder was using the default Arena book
>
>Hi,
>
>If you look at the per-engine details from Dann's previous tournament, you see
>indeed Shredder scoring 60% vs. Fruit.
>
>However most people will only look at the first ten lines it seems.
>
>Fabien.

I think that positional learning may help shredder unless fruit have also
positional learning.

I think that testing both shredder and fruit in tournaments against many engines
may be more interesting because positional learning will be less efective.

For me the interesting results are results of CEGT when shredder does not use
it's own book.
I think that it gives better picture of the potential of fruit relative to
shredder.

Uri

Uri



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.