1st CEGT Quad Marathon Championship 40/400 repeated

[Heinz Van Kempen]

Hi all ,

the text below was already posted in Rybka forum and there is nothing new for those who are reading there. Additionally I answered to a lot of questions and proposals there, but will answer also here of course if other questions and proposals will be made.

Hi all ,

Conditions:

time control:

40/400 repeated on Core 2 Quad E6600 (5 men) - 4 GB RAM
One single game lasts averagely 24 hours because this is real time control and not adapted to slower hardware like CEGT and CCRL are usually doing to adapt the very different hardware of testers to one level

The engines are using 1024 MB of RAM each

books:

The generic books by Harry Schnapp already well tested in CEGT 40/120. The special HS book for Rybka will not be used in order to give equal conditons

EGTB:

5 men

Like in CEGT 40/120 additionally all games are replayed and checked. Uri Blass and I noticed that especially Zappa sometimes plays 0 second moves. This are not always so-called easy moves (forced ones or obvious recaptures) and I do not know if this is part of the strange time management for Zappa or because of the ChessBase GUI. Anyway I changed to Shredder Classic GUI after game 24. As far as I could see those moves did not harm Zappa´s play causing it to lose one game or to deteriate it´s position, but maybe Zappa would play even better without them? So far I could not detect this 0 second moves in Shredder Classic GUI, but I have only one game up to now for comparison.

Interim results and comparison:

CEGT Extreme 40/400 repeated 2007

1 Rybka 2.3.2a mp 4CPU ½½½½½1½½01½001½1½½½½½1½½ 13.0/24
2 Zappa Mexico X64 v.2 4CPU ½½½½½0½½10½110½0½½½½½0½½ 11.0/24

For comparison the result of Rybka 2.3.2a mp 4CPU against Zappa Mexico 4 CPU in CEGT 40/120 (but please note that this was the first Zappa Mexico version before the update in December):

CEGT Quad tournament time control 2007

1 Rybka 2.3.2a 64 4CPU ½101½10½1½½101½1½½½½½½1½½½½½½1½½10½½1½0½111½1½½11½ 31.0/50
2 Zappa Mexico X64 4CPU ½010½01½0½½010½0½½½½½½0½½½½½½0½½01½½0½1½000½0½½00½ 19.0/50

Procedure and planning:

Planned is to run this marathon matches using one machine constantly all over the time (in case ot interest).

Only the top two engines will play this match. So if Zappa will not be developped further, we might soon see a new Naum, Hiarcs, Deep Fritz 11 or another engine facing Rybka in the CEGT Marathon Championships. If a new Rybka comes out this will replace the old one, but the current match should include at least 50 games.

Availability, download, replay and updates:

You can replay the games with or without comments here:

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

and additionally download them using the following links at the bottom of this download page:

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

Updates will be available each weekend at the same time when the CEGT rating lists are given. I will post more rarely in the fora, but will report in German language in CSS and Schachwerkstatt on Monday.

Early conclusions and impressions:

I am especially keen to see this match, because a lot of people claimed that Zappa would come close to Rybka giving a lot of time or using even faster hardware, what can be simulated this way. I have no idea how this time control on Quad can be compared to which faster time controls on 8 CPU´s or more.

I do not like to draw early conclusions but this up to now is by far the closest match including Rybka 2.3.2a I ever saw and the match is not decided yet.

Striking is the high percentage of draws. Two thirds of the games ended in draws, only 8 out of 24 had a winner. I hope this is mere accident and will change soon and not an indication about what we will see in computer chess in the future. In the CEGT 40/120 quad match between Rybka and Zappa we had 28 draws out of 50 games, also many but not that extreme.

[Chris Tatham]

Hi Heinz

Thanks for the update. This looks like a mammoth but also very interesting undertaking. I hope you don't have any PC/engine crashes mid-game!

Keep up the good work. Hope you are keeping well!

Best wishes


Chris

[Heinz Van Kempen]

Hi Chris ,

thanks, I am fine and hope you as well.

No crashes with this engines luckily. Would really be a pity with those time controls.

[Yarget]

Hello Heinz!

This is simply (as always when you are involved) firstclass work! I follow your 40/120 Ratinglist very closely and your latest idea (the 40/400 match between the current 2 topengines) is very interesting.

Keep on the good work

Best regards
Per

[Heinz Van Kempen]

Hi Per ,

thanks. Being a good tester yourself you will have noticed that it is always the best to run what gives you personally the most fun and this is what I am having with this project.

As practically all other things are already done by the testing groups and those who run their own tournaments and matches for years already - including endgame and positional testing - I felt obliged to do something with extreme time controls additionally, because I spent half of my life with correspondence chess.

Anyway only one machine is used for this giving around seven games each week, because the quad list with tournament time control should be expanded.

[Tony Thomas]

Wow, if it was possible to get 1000's of game at this time control, I am sure that lot of engines will come out with a better engine book. Great work by the way. Yes, Zappa is a very close opponent. However, I doubt that Zappa will win as they are both using general books.

[Graham Banks]

it is always the best to run what gives you personally the most fun

Wise words.