CEGT - rating lists February 10th 2008

[Werner]

Hi all ,

our updated rating lists are online and can be found under the attached links.

40 / 120:
Our 40/120-Quad list is updated with the finished Bright 0.2c match. Bright has 2840 elos after 450 games.
A short info about Marathon Match II will follow in a few hours. Replay zone and downloads are up to date too.

40 / 20:
This week we added more than 1800 games to our list. See more in our list "Games of the week". In total our 40/20 list is based now on 220.567 games! A few older games have been deleted after cleaning the database. We included one new engine in our list: Romichess P3k.

New engines:
The only new entry this week is Romichess P3k with an elo of 2361 after 166 games. This is stronger than P2i - but NG5 version is still better in our list. We started to test new Frenzee and new Rotor versions too but we have not enough games to list these engines today.

Updated engines:
Last week a lot of games have been played with Zappa II x64 2CPU. The engine lost 7 points but is still 20 stronger than previous version. It is of course the strongest 2CPU-engine after Rybka, but to this day 88 points away from the Champion.

2 Rybka 2.3.2a x64 2CPU WM-2007 3055
11 Rybka 2.3.2a w32 1CPU 2968
13 Zappa Mexico II x64 2CPU 2967
17 Zappa Mexico x64 2CPU 2947

Toga II 1.4 Beta5c 4CPU is now in front of Hiarcs 11.1 4CPU and Naum 2.1 x64 4CPU.

One of the other updated engines in our list is Alfil 8.11 with now 341 games, 2537 elos and a plus of 52 since last week. The same happend to Popochin 3.1 after now 304 games: +52 elo-points and now with a rating of 2297.

40 / 4:
After 6946 new games the list was updated again. Here we have listed the new Zappa version too with 2960 elo. The engine is also very close to Rybka 2.32a w32 1CPU as in our 40/20 list!

17 Rybka 2.3.2a w32 1CPU 2961
19 Zappa Mexico II x64 2CPU 2961

Other new engines are:
282 Alfil 8.1 2615 (much better than in 40/20)
217 E.T. Chess 13.01.2008 2664
227 Booot 4.14.0 26

A big „Thank you“ to all testers as usual!

40/20: http://www.husvankempen.de/nunn/rating.htm
Blitz: http://www.husvankempen.de/nunn/blitz.htm
40/120: http://www.husvankempen.de/nunn/rating120.htm
Tester: http://www.husvankempen.de/nunn/testers/testers.htm
Games of the week: http://www.husvankempen.de/nunn/40_40%2 ... on/gow.JPG
Elo-comparison: http://www.husvankempen.de/nunn/Replay/ ... arison.htm

Werner
CEGT Team

[Jouni]

Toga seems to gain a lot with longer time control in single list.

40/4:

1 Rybka 2.3.2a x64 1CPU 3021
2 Fritz 11 2908
3 Deep Shredder 11 x64 1CPU 2894
4 Fruit 2.3.3f Test Beta 2859
5 Hiarcs 11.1 1CPU 2855
6 Zappa Mexico x64 1CPU 2849
7 Loop 13.6 w32 1CPU 2846
8 Toga II 1.4 beta5c 1CPU 2843

Difference to Rybka = 178 points.

40/20:

1 Rybka 2.3.2a w32 1CPU 2968
2 Fritz 11 2919
3 Deep Shredder 11 w32 1CPU
4 Toga II 1.4 Beta5c 1CPU 2861

Difference to Rybka = 107 points (998 games played at least)!

Jouni

[Uri Blass]

Toga seems to gain a lot with longer time control in single list.

40/4:

1 Rybka 2.3.2a x64 1CPU 3021
2 Fritz 11 2908
3 Deep Shredder 11 x64 1CPU 2894
4 Fruit 2.3.3f Test Beta 2859
5 Hiarcs 11.1 1CPU 2855
6 Zappa Mexico x64 1CPU 2849
7 Loop 13.6 w32 1CPU 2846
8 Toga II 1.4 beta5c 1CPU 2843

Difference to Rybka = 178 points.

40/20:

1 Rybka 2.3.2a w32 1CPU 2968
2 Fritz 11 2919
3 Deep Shredder 11 w32 1CPU
4 Toga II 1.4 Beta5c 1CPU 2861

Difference to Rybka = 107 points (998 games played at least)!

Jouni

Note that part of the reason is that rybka is losing 32 bits when it plays 40/20 in the CEGT games

The top single version in 40/4 is
Rybka 2.3.2a x64 1CPU when the top single version in 40/20 list is
Rybka 2.3.2a w32 1CPU (64 bits 1 cpu is untested for 2.3.2a at 40/20).

Uri

[Yarget]

Hello Werner!

Thanks for the update, as always you and the CEGT-team deliver first-class testwork. I have studied your Zappa Mexico II results at 40/4 and at least one of them is quite unexpected:

Zappa Mexico II w32 2CPU 2813
Zap!Chess Zanzibar w32 2CPU 2858

Needless to say, Zap(pa) is optimized for big 64-bit Hardware and longer timecontrols but I still find this performance by Mexico bad. At least I expected Mexico II to equalize the score of Zanzibar.

Regards
Per

[ernest]

Zappa Mexico II w32 2CPU 2813
Zap!Chess Zanzibar w32 2CPU 2858

Zappa Mexico II w32 2CPU 2813 is clearly ... fishy!

Too few games, and Zappa Mexico II x64 2CPU is 2960 (+137 Elo!!!)

[Heinz Van Kempen]

Hi all ,

the report below is also available in German language for the many German readers here:

http://husvankempen.de/nunn/phpBB2/view ... =5788#5788


waiting for the new Naum 3 for including it in the Quad list I could partly use more machines for a new marathon match and already collected 20 games so far. You have to bear in mind that one game averagely lasts 24 hours.

Code: Select all

CEGT Quad Extreme 40/400 repeated  2008

1   Rybka 2.3.2a X64 4CPU      1½½0½½½½½½½11½½½1½1½ 12.0/20
2   Zappa Mexico II X64 4CPU  0½½1½½½½½½½00½½½0½0½  8.0/20

Against Zappa Mexico I upd. Rybka could win 24,5:17,5 (two more games added) and we have the same tendency once again with another win for Rybka inminent in a game in progress.


For a verdict we should of course wait for more games like always , but so far this long time games with non-adapted books gave a completely different impression than the short match in Mexico. In fact it is Rybka losing only 4 games out of totally 62 against the currently second best engine available and Zappa could surely avoid one loss or another by better access to the tablebases. When it comes to tablebase access the evaluation of Rybka is sometimes incorrect despite of shown search depths of more than 30 and 40 moves. See for example game 15 in the current match (between moves 72 and 76).


Harry Schnapp delivered a Shredder Classic book with more than 12,5 MB and apart from many other main lines like diverse Sicilian, Ruy Lopez, Petroff, French Defence, Queen´s Gambit, Slav, Nimzo-, Queen-, King´s Indian, English, we can see now for example also Ruy Lopez Marshall Gambit and Najdorf Poisoned Pawn variations, from time to time en vogue in GM practice, too. The match is played the way that both engines have to play the same line with reversed colors, too.


Games are available for Replay here:

http://www.husvankempen.de/nunn/Replay/cegtextreme1.htm

and also added to the downloads:

http://www.husvankempen.de/nunn/Replay/ ... d40120.htm

[Yarget]

Hello Heinz!

Continue these great high-end tests. This is simply as good as it gets. Zappa better enjoy these matches because Naum 3 will move very close to Rybka and certainly enter the second place in your 40/120 Quad-list.

Best regards
Per

[Heinz Van Kempen]

Hi Per ,

glad that you (and others) like those tests.

Like you I have a strong feeling that Naum 3 will soon be the next opponent instead of Zappa in this marathon matches. May well be that is is stronger than Zappa.

And hopefully with a new Rybka later we will also see less draws. Draw percentage currently is around 70%.

[Werner]

Hi Ernest, hi Per,
thank you for the report!
with next update we will have much more games with that engine - then I think the ratings will be more clear.

[ThatsIt]

Hi !

Only 320 games are irrelevant, especially under blitz-conditions.
Be sure that we will have much more games within the next update.

Best,
G.S.