CEGT - rating lists March 06th 2011

[Werner]

Hi all,

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

As the host of our coordination forum foren-city has no more support and we had a lot of problems to log in we now use an alternate forum here:
http://cegt.siteboard.eu/index.php

40 / 20:
New games: 2723; 73 different engines
Total: 492.677

NEW Engines
106 Gull 1.2 x64 nLP: 2975 - 595 games (same rating as version 1.1)
287 Equinox 0.95 x64 1CPU: 2853 - 800 games (6 points ahead of version 0.91 - so only a small difference)
673 Francesca MAD 0.16: 2547 - 850 games (-36 to version 0.15)
So we have no good news for all these engines

UPDATES

1 Houdini 1.5a x64 6CPU: 3295 - 600 games (+4)
b] 1 Houdini 1.5a x64 2CPU:[/b] 3248 - 634 games (-2)

9 Stockfish 2.0.1 x64 4CPU: 3186 - 2176 games (-3)


40 / 4:
No update this week!

40/120
See here our new single-list:
http://cegt.siteboard.eu/viewtopic.php?f=4&t=18
A big „Thank you“ to all testers as usual!!

Links

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
Elo-comparison: http://www.husvankempen.de/nunn/Replay/ ... arison.htm
Games of the week: http://www.husvankempen.de/nunn/40_40%2 ... on/gow.jpg

Werner Schuele
CEGT-Team

[Vinvin]

40/120
See here our new single-list:
http://cegt.siteboard.eu/viewtopic.php?f=4&t=18
A big „Thank you“ to all testers as usual!!

Hello Werner, why don't you reuse old match from the old 40/120 list (single cpu vs single cpu)?

[Werner]

Hi Vincent,
as Wolfgang is now online again - he will answer. He manages the new list.
As I noted before - I have a lot of problems to log in into the forum at foren-city.de

[silentshark]

Hi all,

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

As the host of our coordination forum foren-city has no more support and we had a lot of problems to log in we now use an alternate forum here:
http://cegt.siteboard.eu/index.php

40 / 20:
New games: 2723; 73 different engines
Total: 492.677

NEW Engines
106 Gull 1.2 x64 nLP: 2975 - 595 games (same rating as version 1.1)
287 Equinox 0.95 x64 1CPU: 2853 - 800 games (6 points ahead of version 0.91 - so only a small difference)
673 Francesca MAD 0.16: 2547 - 850 games (-36 to version 0.15)
So we have no good news for all these engines

UPDATES

1 Houdini 1.5a x64 6CPU: 3295 - 600 games (+4)
b] 1 Houdini 1.5a x64 2CPU:[/b] 3248 - 634 games (-2)

9 Stockfish 2.0.1 x64 4CPU: 3186 - 2176 games (-3)


40 / 4:
No update this week!

40/120
See here our new single-list:
http://cegt.siteboard.eu/viewtopic.php?f=4&t=18
A big „Thank you“ to all testers as usual!!

Links

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
Elo-comparison: http://www.husvankempen.de/nunn/Replay/ ... arison.htm
Games of the week: http://www.husvankempen.de/nunn/40_40%2 ... on/gow.jpg

Werner Schuele
CEGT-Team

Thanks, Werner, it's disappointing to lose 36 ELO between versions! In my own tests, I have 0.16 33 ELO points above 0.15! Perhaps the differences will even up after more games?

[Werner]

Thanks, Werner, it's disappointing to lose 36 ELO between versions! In my own tests, I have 0.16 33 ELO points above 0.15! Perhaps the differences will even up after more games?

Hi Tom,
normally 850 games are enough to see a +30 improvement. The only thing is: version 0.15 does not have so many games. But I never saw a 60 points jump after I have made more than 500 games. So I am not sure what happened.
Of course you can download the games and have a look at them.

best wishes
Werner

[Werner]

This week I made 100 games more for each version - sorry but the results are not better

Code: Select all

Francesca MAD 0.15    2200 - Bobcat 20110220 x64 1CPU     2200   24.0 - 26.0    +17/-19/=14    48.00%
Francesca MAD 0.15    2200 - GNU Chess 5.07.170.5b x64    2200   23.5 - 26.5    +15/-18/=17    47.00%

Code: Select all

Francesca MAD 0.16    2200 - Bobcat 20110220 x64 1CPU     2200   19.5 - 30.5    +9/-20/=21    39.00%
Francesca MAD 0.16    2200 - GNU Chess 5.07.170.5b x64    2200   18.5 - 31.5    +13/-26/=11    37.00%

[Wolfgang]

40/120
See here our new single-list:
http://cegt.siteboard.eu/viewtopic.php?f=4&t=18
A big „Thank you“ to all testers as usual!!

Hello Werner, why don't you reuse old match from the old 40/120 list (single cpu vs single cpu)?

Hi Vincent,

sorry for my late answer but I was ill for about 5 weeks due to a pneumonia combined with an influenza ("flu") and problems with my pharmaceuticals. But fortunately this is over now.

When I restarted the 40/120-list I took 960 games (single vs. single) from our old list single/dual engines to have a basis for the new single list. This was mentioned in our old forum in this post:
http://cegt.foren-city.de/topic,463,-ne ... -list.html
(I hope you can read it, because the old forum is down normally, but at the moment it can be accessed.) There were more single-single games but I deleted an older Toga (1.2.1) and two older versions of Fritz (9 and 10) because four Fritz (9,10,11,12) were too much in my opinion.

On this basis I added various engines with at least 300 games. The actual list with 14 engines can be found here: http://cegt.siteboard.eu/f4t18-new-40-1 ... -list.html

Next entry will be Stockfish 2.0 x64 and then Rybka 4.0 x64.

Planned are: Spike 1.4, Houdini 1.5, Deep Junior 12 and more engines also from the middle class. But this will take much time...

[silentshark]

This week I made 100 games more for each version - sorry but the results are not better

Code: Select all

Francesca MAD 0.15    2200 - Bobcat 20110220 x64 1CPU     2200   24.0 - 26.0    +17/-19/=14    48.00%
Francesca MAD 0.15    2200 - GNU Chess 5.07.170.5b x64    2200   23.5 - 26.5    +15/-18/=17    47.00%

Code: Select all

Francesca MAD 0.16    2200 - Bobcat 20110220 x64 1CPU     2200   19.5 - 30.5    +9/-20/=21    39.00%
Francesca MAD 0.16    2200 - GNU Chess 5.07.170.5b x64    2200   18.5 - 31.5    +13/-26/=11    37.00%

Many thanks for doing the testing, it's really appreciated. What a pity the results aren't going in the right direction!

After thousands of games, I have 0.16 at 39 ELO better than 0.15. Maybe my testing is too narrow? Quite a lot of play between versions, with some battles vs things like Ruffian and Colossus thrown in.

Any how, my latest WIP version is currently showing 63 ELO better than 0.15. Maybe I need to release this one..

[Graham Banks]

Any how, my latest WIP version is currently showing 63 ELO better than 0.15. Maybe I need to release this one..

Please, please, please make it possible for your engine to play using a generic book as an option.

Cheers,
Graham.

[Werner]

Perhaps there is something wrong with the evaluation - see here:

[d]8/2r3p1/7p/3q3P/5P2/3k2P1/1Q3B2/6K1 w - - 11 63
(why should black repeat such a position?)
FrancescaMad 016:
10 00:01 448.765 1.447.629 +0,35 b2a1 d5h5 a1d4 d3c2 d4e4 c2c1 f2e3 c1b2 e4d4 c7c3 d4d2 c3c2
11 00:01 818.121 1.341.181 +0,29 b2b1
12 00:02 1.399.518 1.216.972 0,00 b2b1 c7c2 b1a1 d5f3 a1a3 d3e2 a3e7 e2d2 e7b4 c2c3 b4d4 d2c1 d4g7 f3d1
13 00:02 1.834.536 1.239.551 0,00 b2b1 c7c2 b1a1 d5f3 a1a3 d3e2 a3e7 e2d2 e7b4 c2c3 b4d4 d2c1 d4g7 f3d1 g1g2
14 00:05 5.960.716 1.273.657 0,00 b2b1 c7c2 b1a1 d5f3 a1a3 d3e2 a3e7 e2d2 e7b4 c2c3 b4d4 d2c1 d4g7 f3d1 g1g2 d1f3
15 00:07 8.857.448 1.278.131 0,00 b2b1 c7c2 b1a1 d5f3 a1d4 d3e2 d4e5 e2d3 e5d4
16 00:22 25.607.695 1.209.621 0,00 f4f5 d5f5 b2b1 c7c2 b1b3 c2c3 b3d1 d3c4 d1a4 c4d3
17 00:24 28.925.332 1.209.253 0,00 f4f5 d5f5 b2b1 c7c2 b1b3 c2c3
18 00:29 34.287.616 1.217.167 0,00 f4f5 d5f5 b2b1 c7c2 b1b3 c2c3
19 00:38 44.962.227 1.209.637 0,00 f4f5 d5f5 b2b1 c7c2 b1b3 c2c3
20 00:50 60.010.952 1.208.193 0,00 f4f5 d5f5 b2b1 c7c2 b1b3 c2c3 b3b1 c3c2

some remarks: not easy to test, as analyzing does not work - and under Arena with inf time control the engine does not accept stop! Only stop engine process helps after a few seconds!