Linux testing pool

Discussion of chess software programming and technical issues.

Moderators: hgm, Rebel, chrisw

crybotmark
Posts: 37
Joined: Thu May 09, 2013 9:06 pm

Linux testing pool

Post by crybotmark »

Hi all, I was wondering, since I returned to computer chess, if you guys could help me finding a set of engines for linux in the range 1950-2450 elo in order to test my engine Napoleon. In the past I had acces to another machine running windows, while my main working pc was running linux, but now I can only work on linux and running a virtual machine is not an option here...
If any of you had a similar issue in the past, or rather has an engine that meets the requirements above, would really help me a lot.
User avatar
Werner Taelemans
Posts: 119
Joined: Mon Feb 03, 2014 11:57 am
Location: Belgium
Full name: Werner Taelemans

Re: Linux testing pool

Post by Werner Taelemans »

Napoleon is one of my favourite sparring partners for Galjoen. :wink:
ZirconiumX
Posts: 1334
Joined: Sun Jul 17, 2011 11:14 am

Re: Linux testing pool

Post by ZirconiumX »

crybotmark wrote:Hi all, I was wondering, since I returned to computer chess, if you guys could help me finding a set of engines for linux in the range 1950-2450 elo in order to test my engine Napoleon. In the past I had acces to another machine running windows, while my main working pc was running linux, but now I can only work on linux and running a virtual machine is not an option here...
If any of you had a similar issue in the past, or rather has an engine that meets the requirements above, would really help me a lot.
I've been in that very situation, and have some suggestions.

- Fairy-Max is CCRL 1946, which is close enough, and
- I've heard good things about Cinnamon for testing, it's CCRL 2030.
- Dorpsgek (full disclosure: I wrote it) is CCRL 2158 for the previous version (latest is currently untested). It's written on Linux and handles just about everything I throw at it at incredibly small TCs. Should be a good sparring partner.
- Sungorus is CCRL 2307 and pretty reliable in my testing.
- Diablo is CCRL 2387. I haven't tried it personally, but I have heard good things.

Try those.
Some believe in the almighty dollar.

I believe in the almighty printf statement.
crybotmark
Posts: 37
Joined: Thu May 09, 2013 9:06 pm

Re: Linux testing pool

Post by crybotmark »

Thank you! That's very helpful.
User avatar
hgm
Posts: 27790
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: Linux testing pool

Post by hgm »

KingSlayer should also work under Linux. It is about 200 Elo stronger than Fairy-Max. Source code is at http://hgm.nubati.net/cgi-bin/gitweb.cg ... ;a=summary . (It still calls itself 'Simple', but that name turned out to be already taken.)
User avatar
Evert
Posts: 2929
Joined: Sat Jan 22, 2011 12:42 am
Location: NL

Re: Linux testing pool

Post by Evert »

Jazz (and SjaakII, for that matter) should be in that ballpark.
An alternative is to play a stronger engine at time-odds. This has the advantage of spending more wall (and CPU) time on your own engine.
jorose
Posts: 358
Joined: Thu Jan 22, 2015 3:21 pm
Location: Zurich, Switzerland
Full name: Jonathan Rosenthal

Re: Linux testing pool

Post by jorose »

I have been using SjaakII 1.3.1 to test my engine as I (incorrectly) assumed that to be the one tested on CCRL (which seems to be 1.3.0). My engine is slowly improving and I was wondering if you knew exactly how much stronger SjaakII 1.4.1 is?

Thanks again for publishing such a great engine!
jdart
Posts: 4366
Joined: Fri Mar 10, 2006 5:23 am
Location: http://www.arasanchess.org

Re: Linux testing pool

Post by jdart »

You could always run Stockfish in reduced strength mode (UCI_Elo option). Several other engines support that too.

--Jon
User avatar
Evert
Posts: 2929
Joined: Sat Jan 22, 2011 12:42 am
Location: NL

Re: Linux testing pool

Post by Evert »

I actually don't know exactly how much stronger Sjaak II 1.4.1 is. While I do tune things and try to make it stronger, most of the changes between versions go into the ability to play new variants. If anything, those reduce the strength.
Even then, adding something that increases strength may not have the same affect in all variants. I try to test a couple of different ones, and make sure that a change is overall good or neutral, while not disadvantaging any of the variants I tested...

I suspect it is not much stronger or weaker than 1.3.0 for regular (FIDE) chess.
crybotmark
Posts: 37
Joined: Thu May 09, 2013 9:06 pm

Re: Linux testing pool

Post by crybotmark »

Just to let you guys know, both Diablo 0.5.1 and Cinnamon 2.0 give illegal pv moves when playing at not so small time controls with cutechess, although it seems that the actual moves they play are fine.