Deep Fritz vs. Deep Junior -- Game 4: 0-1 (Junior again!)

[DomLeste]

Book moves up to 20 odd moves zzzzzzz

Clap these engines to 10 moves book thanks!

Rewrite history dont repeat it

Ok perhaps 15 moves book if someones going to cry LOL

Whats the future 30-40-60 move books come on!

[Uri Blass]

Book moves up to 20 odd moves zzzzzzz

Clap these engines to 10 moves book thanks!

Rewrite history dont repeat it

Ok perhaps 15 moves book if someones going to cry LOL

Whats the future 30-40-60 move books come on!

No need to do it.
It is not an advantage to have a bigger book.
I do not know about the book in game 4 but
Fritz lost in game 3 thanks to the fact that the bigger book that it used leaded it to lost position.

Uri

[Erik Roggenburg]

I was looking at CEGT's 40/4 list: http://www.husvankempen.de/nunn/40_4_Ra ... liste.html

and the 40/40 list of CEGT:
http://www.husvankempen.de/nunn/40_40%2 ... liste.html

Also, a while back I ran a few thousand games, and DF10 was quite a bit ahead of DJ10.

Code: Select all

4 Deep Fritz 10                  : 2721   19  19   960    60.2 %   2649   29.2 %
 
7 Deep Junior 10.1               : 2636   18  18   960    47.3 %   2654   31.1 %

These are a few of the reasons why I am surprised that DJ has fared so well. I am happy for the DJ team. They work hard on their engine, and it is an engine I have no problems purchasing with each new release.

[Leto]

I was looking at CEGT's 40/4 list: http://www.husvankempen.de/nunn/40_4_Ra ... liste.html

and the 40/40 list of CEGT:
http://www.husvankempen.de/nunn/40_40%2 ... liste.html

Also, a while back I ran a few thousand games, and DF10 was quite a bit ahead of DJ10.

Code: Select all

4 Deep Fritz 10                  : 2721   19  19   960    60.2 %   2649   29.2 %
 
7 Deep Junior 10.1               : 2636   18  18   960    47.3 %   2654   31.1 %

These are a few of the reasons why I am surprised that DJ has fared so well. I am happy for the DJ team. They work hard on their engine, and it is an engine I have no problems purchasing with each new release.

They're using Deep Junior 11, which is probably stronger than DF10.

[Enrico]

Hello.

Actually, it does not say Deep Junior 11. You are seeing the ".1" at the end. The number that they use is insignificant as it represents an old Chessbase version of the .ENG file. It's just used to make hassle-free, unprotected versions for testing.

Regards,

-elc.

Alexander Kure, Fritz' opening book author, does not look happy at all.

http://chessbase.de/nachrichten.asp?newsid=6690

It is interesting to note that in one of the images you can clearly see that they are using Deep Junior 11.

[Leto]

Hello.

Actually, it does not say Deep Junior 11. You are seeing the ".1" at the end. The number that they use is insignificant as it represents an old Chessbase version of the .ENG file. It's just used to make hassle-free, unprotected versions for testing.

Regards,

-elc.

It is interesting to note that in one of the images you can clearly see that they are using Deep Junior 11.

Looks like you're right. The new English chessbase article shows that they are using DJ 9.1:

[Harvey Williamson]

I was looking at CEGT's 40/4 list: http://www.husvankempen.de/nunn/40_4_Ra ... liste.html

and the 40/40 list of CEGT:
http://www.husvankempen.de/nunn/40_40%2 ... liste.html

Also, a while back I ran a few thousand games, and DF10 was quite a bit ahead of DJ10.

Code: Select all

4 Deep Fritz 10                  : 2721   19  19   960    60.2 %   2649   29.2 %
 
7 Deep Junior 10.1               : 2636   18  18   960    47.3 %   2654   31.1 %

These are a few of the reasons why I am surprised that DJ has fared so well. I am happy for the DJ team. They work hard on their engine, and it is an engine I have no problems purchasing with each new release.

They're using Deep Junior 11, which is probably stronger than DF10.

If Deep Junior had opened up a 2 game lead after only 4 games on any of the rating lists you would all be saying it is to few games.

Because its in a match and Junior has faster hardware than Fritz and perhaps a better opening book now its clear Deep Junior is better?!

[Uri Blass]

I was looking at CEGT's 40/4 list: http://www.husvankempen.de/nunn/40_4_Ra ... liste.html

and the 40/40 list of CEGT:
http://www.husvankempen.de/nunn/40_40%2 ... liste.html

Also, a while back I ran a few thousand games, and DF10 was quite a bit ahead of DJ10.

Code: Select all

4 Deep Fritz 10                  : 2721   19  19   960    60.2 %   2649   29.2 %
 
7 Deep Junior 10.1               : 2636   18  18   960    47.3 %   2654   31.1 %

These are a few of the reasons why I am surprised that DJ has fared so well. I am happy for the DJ team. They work hard on their engine, and it is an engine I have no problems purchasing with each new release.

They're using Deep Junior 11, which is probably stronger than DF10.

If Deep Junior had opened up a 2 game lead after only 4 games on any of the rating lists you would all be saying it is to few games.

Because its in a match and Junior has faster hardware than Fritz and perhaps a better opening book now its clear Deep Junior is better?!

Of course you are right.
I did not claim that Junior is better but only that there is no evidence that the commercial version of Fritz is better than the commercial version of Junior under the relevant conditions.

For some reason people use CEGT that is clearly different conditions than this tournament to get conclusions which version is better.

CEGT is ponder of fast time control(except 120/40) less processors not using own books.

SSDF is at least ponder on and own books.and Junior10 does well in that list.

CCRL 40/40 is at least 4 processors and not so fast time control and Junior10 has almost the same performance as Fritz.

I can add that Junior in the past was leading 5-0 against Fritz only to lose a match of 24 games.

Uri

[Leto]

I was looking at CEGT's 40/4 list: http://www.husvankempen.de/nunn/40_4_Ra ... liste.html

and the 40/40 list of CEGT:
http://www.husvankempen.de/nunn/40_40%2 ... liste.html

Also, a while back I ran a few thousand games, and DF10 was quite a bit ahead of DJ10.

Code: Select all

4 Deep Fritz 10                  : 2721   19  19   960    60.2 %   2649   29.2 %
 
7 Deep Junior 10.1               : 2636   18  18   960    47.3 %   2654   31.1 %

These are a few of the reasons why I am surprised that DJ has fared so well. I am happy for the DJ team. They work hard on their engine, and it is an engine I have no problems purchasing with each new release.

They're using Deep Junior 11, which is probably stronger than DF10.

If Deep Junior had opened up a 2 game lead after only 4 games on any of the rating lists you would all be saying it is to few games.

Because its in a match and Junior has faster hardware than Fritz and perhaps a better opening book now its clear Deep Junior is better?!

I didn't say DJ is better, I said that the version they are using is probably stronger than DF10. I of course would never say X is stronger than Y from just 6 games.

[geots]

In the ratings lists I see, Deep Fritz 10 rates higher than Deep Junior 10. What could be leading to Junior's solid performance in this match?

Opening preparation?
Better hardware?
Good luck?
Something else?

Perhaps it's Deep Junior 11?

The main reasons these results will vary from some or all rating lists is due to 3 reasons, and 3 only: 1. It is just a few games, and anything can happen in a short match. Not like they are playing best of 50. 2. The hardware each is using could make a big difference. 3. And this may be the most important- when testing groups are testing the engines, some are using generic books for each, and some are letting them use their own books. But either way it boils down to the same thing- the prorammers are not there in group testing to study previous games and force its program to avoid opening lines that are bad for it. That is a key element in most matches of this type- the preparation and study done by the programmer. Its really not that complicated to figure out. BTW, i dont know anything about multiple processors, etc.- but in a short match between the latest Fritz and Junior using 32 bit 1cpu- you could flip a coin.

Regards