CEGT 40/20 update including DeepSjeng2.5+ChessTiger2007.1

[Werner]

Hi all ,

Our newest rating list 40/20 (April 22th 2007) is now online. All links at the end of the posting as usual.

This week we made another 100 games with new Chess Tiger 2007.1. After now 628 games there is nearly no difference to Chess Tiger 2007 (+3). The error bar is +- 20 so there is a chance for a better rating in the future.

The new DeepSjeng 2.5 has 174 games and a first rating of 2752. This is 84 points better than DeepSjeng 1.6! Next week we will have a more reliable rating for it. We also hope for +120 Elos .

After 522 games with Loop 13.6 2CPU the rating is 20 points behind Loop 13.5 and 25 behind Loop12.32. We hope this is only a statistical error and the engine will reach at least Loop 12.32 with more games.

Another statistical error we have with the result of Shredder 10.1 after 135 games. This is around 80 points behind Shredder 10. We will make more games with the engine and see what happens.

The 2nd match between Rybka 2.2 2CPU and some 4CPU engines is ready. As a result of these games Rybka 2.1 lost some elos and we have a bit stronger Rybka 2.2 engine in our list (+11). The 4CPU engines have now 100 games against 2 opponents. The rating at the moment has no statistical relevance. The 3rd match starts in a few weeks against Rybka 2.3.2. In the meantime we will play some games between the 4CPU engines.

A big thanks to all testers as usual.

Links:
All-Versions: http://www.husvankempen.de/nunn/40_40%2 ... liste.html
Best-Versions: http://www.husvankempen.de/nunn/40_40%2 ... liste.html
Blitz: http://www.husvankempen.de/nunn/blitz.htm (also with Best-Versions and All-Versions lists)
40/120: http://www.husvankempen.de/nunn/40_120_ ... liste.html
Replay-Zone: http://www.husvankempen.de/nunn/Replay/replay.htm
Elo comparison 40/120 and 40/20: http://www.husvankempen.de/nunn/Replay/ ... arison.htm
Testers with picture and Hardware: http://www.husvankempen.de/nunn/testers/testers.htm

Werner Schüle
CEGT-Team

[mclane]

to be honest werner, this looks a little weird.

Code: Select all

    Motor                Punkte    Ch   Hi   To   Ph   De   Na   Pr   Al   No   Gi    S-B
01: ChessTiger2007.1 UCI 24,5/31 ···· 0=1  111  0011 =111 11=  111  1011 0111 111   312,50
02: Hiarcs11.1UCI        22,0/29 1=0  ···· 11=  11=  =1== =10  11=  111  =1=1 111   278,75
03: TogaII1.2.1a         18,5/29 000  00=  ···· 10=? 111  1=== 1101 110  111  111   208,50
04: Pharaon              17,0/29 1100 00=  01=? ···· 101  101  00=  11=  0111 111   208,00
05: DeepSjeng25          16,0/29 =000 =0== 000  010  ···· 1=1  011  111  =11  111   180,25
06: Naum 2.0             14,0/29 00=  =01  0=== 010  0=0  ···· =1=  1==  =11  01=1  168,75
07: ProDeo 1.2           13,5/28 000  00=  0010 11=  100  =0=  ···· 011  111  1=1   148,25
08: Alaric0702b          10,0/28 0100 000  001  00=  000  0==  100  ···· 11=  111   105,00
09: Now_2404             5,0/30  1000 =0=0 000  1000 =00  =00  000  00=  ···· 0=0    85,75
10: Gibbon2.32a          4,5/28  000  000  000  000  000  10=0 0=0  000  1=1  ····   40,25

145 Partien von 1350 gespielt
Name des Turniers: 2007.1 match
Ort/ Land: SCW, Germany
Spielstufe: Blitz 10/10

AMD Sempron 3000+, pb off

[Werner]

Yes Thorsten, indeed!
In a short match here 40/20 Toga1.2.1a made 70% against ChessTiger2007.1
I have no results against DeepSjeng2.5 1CPU or Hiarcs11.1 1CPU.
I like the ChessTigers and would like to have such good results too )

Werner

[mclane]

are you sure your testers test 2007.1 and not 2007 version ??in which GUI do you test ? CB ?

[mclane]

Code: Select all

    Motor                Punkte  Ry Hi Lo Sp To Ch De To Na Bo Ph No    S-B
01: Rybka v2.3.lk.x64    17,0/20 ·· 01 =1 11 =1 =1 1  1  11 11 11 1=  147,50
02: Hiarcs11.1UCI        16,0/21 10 ·· 1= 10 11 1= == =  =1 11 11 11  136,25
03: LoopMP 12.32         11,5/20 =0 0= ·· == == == =  0  11 11 == 11   92,75
04: Spike1.2             11,0/20 00 01 == ·· =1 == 0  1  1= 1= =1 ==   94,25
05: TogaII1.3x4          11,0/20 =0 00 == =0 ·· =0 0  =  11 11 11 11   79,00
06: ChessTiger2007.1 UCI 10,0/22 =0 0= == == =1 ·· 0= =0 01 10 == =1   88,50
07: DeepSjeng25          9,5/14  0  == =  1  1  1= ·· =  =  =  1  11   81,75
08: TogaII1.2.1a         9,0/14  0  =  1  0  =  =1 =  ·· 11 == 1  1    74,75
09: Naum2.0              8,0/21  00 =0 00 0= 00 10 =  00 ·· =1 11 11   53,25
10: Booot413             6,0/21  00 00 00 0= 00 01 =  == =0 ·· 1= ==   43,50
11: Pharaon              4,5/20  00 00 == =0 00 == 0  0  00 0= ·· =1   35,25
12: Now2404              3,5/21  0= 00 00 == 00 =0 00 0  00 == =0 ··   32,75
117 Partien von 132 gespielt
Name des Turniers: neu
Ort/ Land: SCW, Deutschland
Spielstufe: Turnier 40/40

   Motor                Punkte                   Ch
1: ChessTiger2007.1 UCI 46,5/91 ··················· 
2: Hiarcs11.1UCI        12,0/19 101=1=01=0=111=10=1 
3: TogaII1.3x4          10,0/18 =10111000==1==1=01  
4: DeepSjeng25          9,5/18  01==0=111==0=1=0==  
5: Glaurung121-EM64T    9,0/18  0====00=1==11=01==  
6: Now2404              4,0/18  001=000=0000==0100  
91 Partien von 150 gespielt
Name des Turniers: chesstiger2007.1
Ort/ Land: SCW, Germany
Spielstufe: Turnier 40/40

Intel 6300, pb off,

[Daniel Mehrmann]

You're claming CEGT is doing something wrong based on your 147 games tournament with a completly different timecontrol ??



Best,
Daniel

[Werner]

Hi Thorsten,
yes I am quite sure its the new version - and I used ert with native engine and Chessbase with uci-version: I saw no differences.
Werner

[mclane]

i have much more than 147 games.
these tournaments i posted are from TODAY. those were actual tournaments.
i had ct2007.1 as beta (before it was released). long time ago.

i wonder how those different results happen.
i do mainly test in ARENA.

since the name of the engine-executable of tiger2007 is similar to that of tiger2007.1 it could be possible that someone tests e.g. the old version but believes he is testing the new version.

as you maybe know, CEGT list has also problems to differenciate different Loop versions.
so CT is not the only case where the list cannot find out about different versions. i wonder why.

there is another strange thing in the CEGT list. that shredder10.1 is behind.

so we have in fact 3 cases where something very strange is going on. i wonder if there is a hidden parameter in testing methods that produces wrong data or that produces wrong interpretation.

[mclane]

thank you werner.
i just search a reason why the results in your list seem to be equal.

i do not want to imply that the GUI has influence on the strength.

[Werner]

Hallo Thorsten,
ich glaube das ist ein etwas unfairer Beitrag. Hast du dir gut überlegt, was du da schreibst?
Wir haben alle Spiele mit allen Bewertungen zum downloaden und wir schauen uns die Ergebnisse auch an und reagieren auf Ausreisser, die auch mal rein statistisch entstehen können.
Du darfst die Spiele gern mal anschauen und dann aufzeigen, wo du was seltsames (verstecktes) entdeckst.
Gruss
Werner