Hi,
I have released V1.32 of the CT800 engine. The most important improvement for the UCI version is that also results from an incomplete final depth iteration can be evaluated, which gave around +30 Elo in self play.
Web link: https://www.ct800.net/software.htm
regards, Rasmus
CT800 V1.32 released
Moderators: hgm, Rebel, chrisw
-
- Posts: 2488
- Joined: Tue Aug 30, 2016 8:19 pm
- Full name: Rasmus Althoff
CT800 V1.32 released
Rasmus Althoff
https://www.ct800.net
https://www.ct800.net
-
- Posts: 41473
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
-
- Posts: 1364
- Joined: Sat Jul 21, 2018 7:43 am
- Location: Szentendre, Hungary
- Full name: Gabor Szots
Re: CT800 V1.32 released
Thanks Rasmus.
I have recently tested v1.31 and it scored worse than 1.11 and v1.30 (see here: http://www.computerchess.org.uk/ccrl/40 ... +opponents). I hope I can confirm your prediction of v1.32 being significantly stronger.
I have recently tested v1.31 and it scored worse than 1.11 and v1.30 (see here: http://www.computerchess.org.uk/ccrl/40 ... +opponents). I hope I can confirm your prediction of v1.32 being significantly stronger.
Gabor Szots
CCRL testing group
CCRL testing group
-
- Posts: 2488
- Joined: Tue Aug 30, 2016 8:19 pm
- Full name: Rasmus Althoff
Re: CT800 V1.32 released
Thanks for testing! However, that's strange because the functional change from 1.30 to 1.31 is stalemate recognition in KQ:K, which is needed in drawn KQ:KP games with bishop or rook pawn. Both are quite rare endgames anyway, and the fix was just cosmetic. Other than that, GCC got an update from 6.4.0 to 7.3.0. And some code reformatting and stuff, but that doesn't change functionality.Gabor Szots wrote: ↑Tue Oct 02, 2018 9:37 amI have recently tested v1.31 and it scored worse than 1.11 and v1.30
I'm astonished how 1.11 can be that far up anyway because in my testing, it's consistently weaker. I think it's statistic variation and depends a lot on the openings that the test framework rolls out. For example, the engine can't really deal with Dutch as Black. The internal book works around this, but in CCRL, "ownbook" is usually set to false.
Rasmus Althoff
https://www.ct800.net
https://www.ct800.net
-
- Posts: 1975
- Joined: Fri Oct 05, 2012 2:35 pm
- Location: Spain
Re: CT800 V1.32 released
Thank you Rasmus
Member of the CCRL Group. Write me if you want I test your engine.