CEGT - rating lists April 06th 2014

[ernest]

Rating dropped a bit

Going in the right direction...

[ernest]

Rating dropped a bit from game 600 to game 900, but still +118 to 1CPU version...

As of today, in this 40/4 rating list, the 1-CPU => 4-CPU jumped again to 2832 => 2958, that is to +126

But more interesting, looking at the 40/20 rating list, the 1-CPU => 4-CPU is only 2832 => 2899, that is to +67
(to be confirmed, since Senpei 4CPU has only played 200 games)

Maybe this can be related to the discussion in the
http://www.talkchess.com/forum/viewtopic.php?t=52219
Threads-Test thread.

[jdart]

I notice you do not seem to have ever tested GNU Chess 6, which is the current version. (Although my understanding is it is probably weaker than 5.50).

--Jon

[Werner]

Yes,
we tested the Versions 5.50 and 5.07 from Michael.
I think, GNU Chess 6 is a different engine and I am not sure how to continue. Which version is still under devellopment? It seems to be a bit confusing....

best wishes
Werner

[Graham Banks]

Yes,
we tested the Versions 5.50 and 5.07 from Michael.
I think, GNU Chess 6 is a different engine and I am not sure how to continue. Which version is still under devellopment? It seems to be a bit confusing....

best wishes
Werner

Agreed.
I'm sticking with Michael's versions too for now.

[jdart]

GNU Chess 6.x is a Fruit derivative (based on Fruit 2.1). Fabien assigned the Fruit copyright to FSF:

https://www.gnu.org/software/chess/manu ... ound-of-v6

It is still being maintained (Antonio Ceballos is apparently the lead developer now) but there have been few significant changes lately except bug fixes.

--Jon

[Michel]

Which version is still under devellopment? It seems to be a bit confusing....

You may consider GNU Chess 5.50 as being no longer under development. I have no time to work on it anymore.

I have a development version which is about 45 elo stronger. I will release that some time in the future as GNU Chess 5.55 but that will be it. This last version has been playing on FICS for 6 months or so.

My aim starting this project was to modernize GNU Chess 5.07 and to make it user friendly. I have done that.

I found out that GNU's evaluation function was surprisingly good and that it was mainly being dragged down by bugs. So I got carried away a little and made GNU several 100's elo stronger. However after the phenomenal success of SF's industrial development model it seems rather silly to pursue this path further.