LS-ratinglist is finished...

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

User avatar
pohl4711
Posts: 2447
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

LS-ratinglist is finished...

Post by pohl4711 »

I finished the LS-ratinglist for several reasons. See:

http://ls-ratinglist.beepworld.de


Stefan
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: LS-ratinglist is finished...

Post by Laskos »

pohl4711 wrote:I finished the LS-ratinglist for several reasons. See:

http://ls-ratinglist.beepworld.de


Stefan
Don't use the latest Cutechess for Windows in gauntlets or RR, it's buggy. It works well only in a match of two engines. Use the previous one.
User avatar
velmarin
Posts: 1600
Joined: Mon Feb 21, 2011 9:48 am

Re: LS-ratinglist is finished...

Post by velmarin »

Sometimes it is best to rest, and perhaps,
back with improved ideas.

Thank you very much for everything done.

Rest and back with force.

Gracias Stefan. :D
Adam Hair
Posts: 3226
Joined: Wed May 06, 2009 10:31 pm
Location: Fuquay-Varina, North Carolina

Re: LS-ratinglist is finished...

Post by Adam Hair »

Laskos wrote:
pohl4711 wrote:I finished the LS-ratinglist for several reasons. See:

http://ls-ratinglist.beepworld.de


Stefan
Don't use the latest Cutechess for Windows in gauntlets or RR, it's buggy. It works well only in a match of two engines. Use the previous one.
If it works well with matches, then I would do the testing as a series of matches. This can be set up in a batch file.
User avatar
Ajedrecista
Posts: 1972
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.

Re: LS-ratinglist is finished...

Post by Ajedrecista »

Hello Stefan:
pohl4711 wrote:I finished the LS-ratinglist for several reasons. See:

http://ls-ratinglist.beepworld.de


Stefan
I am sorry to read this. Some people refer to cutechess-cli 0.5.1 instead of 0.6.0 as a solution, others say direct matches instead RR (this approach needs more effort, I know). Both ideas could save LS rating list.

I read that results will be online up to 1st of May. Is it possible to maintain them online forever or there is a money reason with the host of the site? Anyway, thank you very much for your time! :) Things are perceived different after a break.

Regards from Spain.

Ajedrecista.
Ferdy
Posts: 4834
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: LS-ratinglist is finished...

Post by Ferdy »

Laskos wrote:
pohl4711 wrote:I finished the LS-ratinglist for several reasons. See:

http://ls-ratinglist.beepworld.de


Stefan
Don't use the latest Cutechess for Windows in gauntlets or RR, it's buggy. It works well only in a match of two engines. Use the previous one.
On desktop with windows 7 prof, with or without sp1, I got crashes when using version 0.6.0, as a result I use version 0.5.1. However on laptop with windows 8.1, version 0.6.0 is stable.
User avatar
hgm
Posts: 27836
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: LS-ratinglist is finished...

Post by hgm »

Of course you could also use WinBoard, if you want to do gauntlets, round-robins or Swiss! :idea:
User avatar
pohl4711
Posts: 2447
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

Re: LS-ratinglist is finished...

Post by pohl4711 »

velmarin wrote:Sometimes it is best to rest, and perhaps,
back with improved ideas.

Thank you very much for everything done.

Rest and back with force.

Gracias Stefan. :D

There will be more testing by me. For sure.
But I see no more need for the LS-ratinglist. Andreas Strangmuellers Bullet-list is very good and bigger than the LS-list.
And the problems with cutechess and Littleblitzer are really not nice.
And I know, that the biggest interest of the community are my Stockfish-Tests. So I will continue testing Stockfish-dev-versions. But with other testing conditions.
I will use LittleBlitzerGuI again, but with draw-by-GUI-decision at 120 moves to avoid the buggy 50move-rule of the LBG. And i will play 1000 games (my 500 LS-opening-position-set) against Houdini, Komodo and Gull. So only 3000 games instead of 10000, but with longer thinking time (3'+500ms per game/engine).
So a testrun of Stockfish should take 3-4 days. On one Quadcore-notebook.
On my other notebook I will play a neverending RoundRobin-Tournament with the 3-5 top engines (that depends on the further development in computerchess) with all 4 cores for one engine and a long (15'+3'') thinking time (and 512MB Hash). Ponder off, no Tablebases, Shredder Classic GUI. And each engine gets a private copy of a neutral 8move-book, which allows the engines to learn and avoid bad opening-moves.
At the moment the tournament started with Stockfish 140423, Houdini 4, Komodo TCECr and Gull 3. It will run until a new official release of one engine (I have to think about Stockfish...).Then the new version will replace the old version and the RR-tournament goes on. I will create a ctosstable with percent-values, not points, to avoid distortions in that crosstable...

I will create a new website for publishing the results of both projects in some weeks...

So the testing will go on.

Sorry for my awful english.
Stay tuned.

Stefan
gladius
Posts: 568
Joined: Tue Dec 12, 2006 10:10 am
Full name: Gary Linscott

Re: LS-ratinglist is finished...

Post by gladius »

pohl4711 wrote:I finished the LS-ratinglist for several reasons. See:

http://ls-ratinglist.beepworld.de


Stefan
Sorry to hear about the crashes Stefan! It was always a lot of fun to watch your matches - I see you are still thinking of continuing other types of tests, that would be great to see :).
yanquis1972
Posts: 1766
Joined: Wed Jun 03, 2009 12:14 am

Re: LS-ratinglist is finished...

Post by yanquis1972 »

agreed, saw this linked @ TCEC & thought it was a great idea. real shame but thanks, hope you carry on in some capacity!