Discussion of computer chess matches and engine tournaments.
Moderator: Ras
Jouni
Posts: 3621 Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski
Post
by Jouni » Thu Dec 31, 2009 10:11 am
I tried both versions by Dann "150" Corbit:
Nunn openings,60/3m,32b
Code: Select all
Version 1
1 Naum 4 ½01101½½01½½½½0101½1 11.0/20
2 Stockfish 1.6s ½10010½½10½½½½1010½0 9.0/20
Version 2
1 Naum 4 ½11111½1½½½½01110½1½ 14.0/20
2 Stockfish 1.6s ½00000½0½½½½10001½0½ 6.0/20
Is N4 Stockfish's Angstgegner?
Jouni
DomLeste
Posts: 221 Joined: Thu Mar 09, 2006 4:53 pm
Post
by DomLeste » Thu Dec 31, 2009 10:18 am
Interesting test perhaps test other engines as well for a better comparison
For others you havent seen my postings
Stockfish 1.6s 32bit from Dann Corbit....
1st version for 32bit version
Scales null move smoothly 328 KB (336,384 bytes)
2nd update
32bit version slightly smoother scaling 326 KB (333,824 bytes)
Insanity: doing the same thing over and over again and expecting different results.
Albert Einstein
Werner
Posts: 2976 Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle
Post
by Werner » Thu Dec 31, 2009 2:11 pm
Jouni wrote: IIs N4 Stockfish's Angstgegner?
Jouni
Hi,
I am actually running a testmatch with CEGT 40/20 conditions on a Core I7 (eng-eng, ponder off) with Danns first Compile 64bit version against Naum 4 x64 1CPU. Interim result after 24 games is
Stockfish 1.6s c64 1CPU - Naum 4 x64 1CPU +14 -1 =10
This would be an elo difference of 200 !
Stockfish 1.6 offical version scored here with +25 Elos.
I will continue the match.
Werner
Howard E
Posts: 261 Joined: Wed Mar 08, 2006 8:49 pm
Post
by Howard E » Fri Jan 01, 2010 2:44 am
I have a Nunn opening set of 20 positions. Would using this test with reverse colors not yield 40 games? Or did you separate them into two files? Thanks.
DomLeste
Posts: 221 Joined: Thu Mar 09, 2006 4:53 pm
Post
by DomLeste » Fri Jan 01, 2010 6:14 am
Theirs 2 nunn openings.
Older one Nunn openings with 10 openings=20 games
Nunn2 with 20 openings you have which with reverse colours will result in 40 games.
Insanity: doing the same thing over and over again and expecting different results.
Albert Einstein
Werner
Posts: 2976 Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle
Post
by Werner » Fri Jan 01, 2010 10:37 am
Here are my first results with Danns version:
(eng-eng, ponder off, 40/20 CEGT conditions, opening set top25.pgn from Sedat)
Code: Select all
Stockfish-16DC x64 1CPU 2200 - Naum 4 x64 1CPU 2200 34.0 - 16.0 +23/-5/=22 68.00%
Stockfish-16DC x64 1CPU 2200 - Rybka 3 x64 1CPU 2200 19.5 - 30.5 +11/-22/=17 39.00%
Werner
DomLeste
Posts: 221 Joined: Thu Mar 09, 2006 4:53 pm
Post
by DomLeste » Fri Jan 01, 2010 1:31 pm
By h1a8 from Rybka forum
http://rapidshare.com/files/328693993/S ... 4.zip.html 1st original compile from Dann the best of 1.6s 32bit version as for x64 version perhaps 1.6s JA compile is best?
Is this the version your using werner?
Insanity: doing the same thing over and over again and expecting different results.
Albert Einstein
Werner
Posts: 2976 Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle
Post
by Werner » Fri Jan 01, 2010 1:53 pm
Hi,
I am using 64bit Version from DC.
I know, JA compile is a bit faster, but I started the tests with this version.
The difference is sure not meaurable with a few 100 games...
Werner