Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Sandro scores - another view

Author: Henrik Dinesen

Date: 14:10:48 09/15/05

Go up one level in this thread


On September 15, 2005 at 16:42:29, Marc Lacrosse wrote:

>On September 15, 2005 at 13:58:54, Thomas Logan wrote:
>
>>That is Sandro setting columbus egg vs Fruit
>>
>>last 3 games 2.5 points
>>
>
>>score thus far
>>
>>THOMAS-KBZT8WLT, 120'/40+60'/20+30'  0
>>
>>                123456
>>1   Shredder 9  011½11  4.5/6
>>2   Fruit 2.1   100½00  1.5/6
>>
>>classical time control 256m hash 3,4,5 piece endgame tables - 16m table hash
>>shredder 9.ctg and Fruit_ASv3.ctg & small_book.bin, learning, no pondering
>>optomised books,windows xp home a64 3400
>>
>>Tom
>
>Hi Tom,
>
>I have had a lot of tests with Sandro's Columbus Egg (at blitz time control) and
>I repeat I find it to be a very interesting setting.
>
>But i am less convinced than ever regarding the eventual overall superiority of
>Columbus Egg over shredder default settings.
>
>Presenting only a few games against Fruit default configuration gives only a
>partial evaluation.
>
>See what happened in my recent games (all involve Fruit WCC'2005)
>
>Both playing with the same neutral book, I had a first match between S9CE and
>Fruit default config.
>
>Result was +11 =10 -9 for Fruit (53.3%).
>
>I then adapted fruit's UCI parameters to build an anti-Sandro config.
>
>This Fruit anti-Sandro won +14 =7 -9 against S9 columbus egg (58.3%).
>
>But this anti-sandro config was much worse against shredder 9 default config and
>also against other opponents.
>
>And the same was true for Sandro's config against other opponents (among them :
>CM10 Berean, Ktulu 7a, Ruffian 210, Ruffian 105, Spike 1.0a, SlowChess blitz
>WV2)
>
>After 850 games I have :
>Shredder 9 : 2718 +/- 41
>Fruit WCCC : 2717 +/- 42
>Fruit WCCC anti-CE : 2703 +/- 59
>Shredder Columbus Egg : 2674 +/- 58
>
>So my results do not confirm yours in my testing conditions (same book for all
>and  short timing).
>
>Have a look at CEGT tests tomorrow : you will have a big surprise at
>intermediate time controls ...
>
>Regards
>
>Marc


Hi Marcs,

When Sandro first posted his settings, he clearly mentioned longer TC's for
testing. If they're equaly good - or close - at all TC's, I don't know.
I've only done some testing at 25"-15' (rapid) against different engines, and
those have so far provided me with the impression of a slightly more successful
Shredder. More games is as "always" needed :)

Also, it would appaer that there's used different books, as well as I suspect
the CEGT testers uses Nunn. Still, will be interesting to see your results!

Testings with the TC Thomas use is probably more in accordance with the original
thoughts of Sandro.

BTW: We, unlike you, have no opputunity to test against the Fruit you mention.

Regards
Henrik



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.