CEGT 40/20 update (May 27th 2007)

[Werner]

Hi all ,

Our newest rating lists are now online. All links at the end of the posting as usual.

40/20
This week we added more than 1600 games to our database. New in our list is Hermann 2.0 (single version)

Chiron 0.87
Chiron 0.87 could hold its good rating after now 579 games. Its 50 Elo better than previous version but did not reach Delfi 5.1

Colossus 2007a
Colossus 2007a has now a more stable rating after 399 games. The versions have a continous improvement and reached 2600 elo mark.

Glaurung 2-Epsilon/2
The new Glaurung 2-Epsilon/2 engine could hold its high ranking from last week. The rating of the 64-bit version must be looked with care. Maybe its a llittle bit too high after only 60 games.

Alaric 7.04
Alaric 7.04 did not hold its high rating from last week. The engine is clear better than previous version but could not reach 2700 elos.

Deep Sjeng 2.5 2CPU reached now 2738 Elos. The difference to the single version is now 26 points.

40/120
Our 40/120 list was also updated with more game for List MP, Naum2.1 MP and Zap!Chess Zanzibar MP x64

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.html

Werner Schüle
CEGT-Team

[Uri Blass]

Hi all ,

Our newest rating lists are now online. All links at the end of the posting as usual.

Glaurung 2-Epsilon/2
The new Glaurung 2-Epsilon/2 engine could hold its high ranking from last week. The rating of the 64-bit version must be looked with care. Maybe its a llittle bit too high after only 60 games.
A big thanks to all testers as usual.

Werner Schüle
CEGT-Team

My guess is that it may be better to decide about constant difference between 32 version and 64 bit version in your lists based on speed improvement.

If you decide that the difference between 32 bit and 64 bit is 20 elo then instead of having:

49 Glaurung 2-Epsilon/2 x64 2813 78 75 60 73.3 % 2638 33.3 %
88 Glaurung 2-Epsilon/2 2729 20 20 765 51.6 % 2718 34.0 %

You can decide to solve the equation:
2813*60+2729*765=(x+20)*60+x*765

After getting a solution x you can get the estimate that the rating of the 32 bit version is x when the rating of the 64 bit version is x+20.

Uri

[Werner]

Hi Uri,
I have some troubles to understand what you mean (my English could be better ) - and google translation is very bad.

There are always different ways to get elo of the "same" but faster engine:
- a) make games and calculate the results
- b) if its the same engine you can measure the speed difference (here it is 28,5%) and calculate the elo difference. I do not know the exact formula, but the result will be around 20 elos in that case if I say 100 % are about 60 elos.

[Uri Blass]

Hi Uri,
I have some troubles to understand what you mean (my English could be better ) - and google translation is very bad.

There are always different ways to get elo of the "same" but faster engine:
- a) make games and calculate the results
- b) if its the same engine you can measure the speed difference (here it is 28,5%) and calculate the elo difference. I do not know the exact formula, but the result will be around 20 elos in that case if I say 100 % are about 60 elos.

The idea is to solve some mathematical equation to get better estimate for the rating of glaurung based on expected difference in rating.

I did not solve the equation that is easy to do but I guess that we may get some estimate that is close to

Glaurung2-epsilon/2 x64 2755
Glaurung 2-Epsilon/2 2735

Uri

[Dave McClain]

Hi Uri,
I have some troubles to understand what you mean (my English could be better ) - and google translation is very bad.

There are always different ways to get elo of the "same" but faster engine:
- a) make games and calculate the results
- b) if its the same engine you can measure the speed difference (here it is 28,5%) and calculate the elo difference. I do not know the exact formula, but the result will be around 20 elos in that case if I say 100 % are about 60 elos.

The idea is to solve some mathematical equation to get better estimate for the rating of glaurung based on expected difference in rating.

I did not solve the equation that is easy to do but I guess that we may get some estimate that is close to

Glaurung2-epsilon/2 x64 2755
Glaurung 2-Epsilon/2 2735

Uri

Uri,

Has it been firmly established by enough games with various programs and time controls that 64 bit offers a small advantage over 32 bit ELO ratings?

[Dirt]

Hi all :D,

Our newest rating lists are now online. All links at the end of the posting as usual.

Glaurung 2-Epsilon/2
The new Glaurung 2-Epsilon/2 engine could hold its high ranking from last week. The rating of the 64-bit version must be looked with care. Maybe its a llittle bit too high after only 60 games.
A big thanks to all testers as usual.

Werner Schüle
CEGT-Team

My guess is that it may be better to decide about constant difference between 32 version and 64 bit version in your lists based on speed improvement.

If you decide that the difference between 32 bit and 64 bit is 20 elo then instead of having:

49 Glaurung 2-Epsilon/2 x64 2813 78 75 60 73.3 % 2638 33.3 %
88 Glaurung 2-Epsilon/2 2729 20 20 765 51.6 % 2718 34.0 %

You can decide to solve the equation:
2813*60+2729*765=(x+20)*60+x*765

After getting a solution x you can get the estimate that the rating of the 32 bit version is x when the rating of the 64 bit version is x+20.

Uri

I think you are assuming that the only difference in play between the 32-bit version and the 64-bit version is speed. This is probably what is intended by the author, but I don't think testers should rely on it.