Naum4 - Rybka 3 - 100 games 60 min. test. After 16 games...

Discussion of computer chess matches and engine tournaments.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 7:09 pm

Naum4 - Rybka 3 - 100 games 60 min. test. After 16 games...

Post by Tomcass » Sun Dec 21, 2008 12:08 pm

Quad 6700
Gui Fritz11
Ponder: Off
Books: Perfect15 for Naum4 and RybkaII for Rybka3.
Game in 60 minutes.


Tomcass, Blitz:60' 0


1 Naum 4(x4) ½½110½0½½½11½½½½ 9.0/16
2 Rybka 3 32-bit ½½001½1½½½00½½½½ 7.0/16

Comments:

Strong start of Naum4.
Number of draws 67% (Clearly below my forecast 85%)
In the mid game: Naum4 around 4.000 KNPS, Rybka3 around 130 KNPS

Best regards from Barcelona.

Tom.

Uri Blass
Posts: 8948
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Naum4 - Rybka 3 - 100 games 60 min. test. After 16 games

Post by Uri Blass » Sun Dec 21, 2008 12:49 pm

Tomcass wrote:Quad 6700
Gui Fritz11
Ponder: Off
Books: Perfect15 for Naum4 and RybkaII for Rybka3.
Game in 60 minutes.


Tomcass, Blitz:60' 0


1 Naum 4(x4) ½½110½0½½½11½½½½ 9.0/16
2 Rybka 3 32-bit ½½001½1½½½00½½½½ 7.0/16

Comments:

Strong start of Naum4.
Number of draws 67% (Clearly below my forecast 85%)
In the mid game: Naum4 around 4.000 KNPS, Rybka3 around 130 KNPS

Best regards from Barcelona.

Tom.
Long time control helps the program with better opening book
and I read that perfect15 is the best opening book.

Uri

User avatar
M ANSARI
Posts: 3500
Joined: Thu Mar 16, 2006 6:10 pm

Re: Naum4 - Rybka 3 - 100 games 60 min. test. After 16 games

Post by M ANSARI » Sun Dec 21, 2008 1:54 pm

P15 book is heavily tuned against Rybka II book. Better would be to test with same book.

Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 7:09 pm

Re: Naum4 - Rybka 3 - 100 games 60 min. test. After 16 games

Post by Tomcass » Sun Dec 21, 2008 4:41 pm

After this test I will repeat it with Perfect 15 for both engines. By now... I am enjoying this match very much. :wink:

Tom.

Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 7:09 pm

Naum4 - Rybka 3 - 100 games 60 min. test. After 34 games

Post by Tomcass » Mon Dec 22, 2008 6:21 pm

Tomcass, Blitz:60' 0


1 Naum 4(x4) +10/-7/=17 54.41% 18.5/34
2 Rybka 3 32-bit +7/-10/=17 45.59% 15.5/34

Regards from Barcelona.

Tom.

Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 7:09 pm

Naum4 - Rybka 3 - 100 games. Standings after 60 games.

Post by Tomcass » Wed Dec 24, 2008 12:22 pm

Tomcass, Blitz:60' 0

1 Naum 4(x4) +20/-12/=28 56.67% 34.0/60
2 Rybka 3 32-bit +12/-20/=28 43.33% 26.0/60

Naum4 is taking a clear, although non decisive advantage, against Rybka3. I start to feel that my double forecast:

50% for each engine and
a huge amount of draws

will be wrong in both cases. :-)

Regards and Merry X'mas from Barcelona,

Tom.

Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 7:09 pm

Naum4 - Rybka 3 - 100 games. FINAL STANDINGS.

Post by Tomcass » Sat Dec 27, 2008 12:38 pm

Quad 6700
Gui Fritz11
Ponder: Off
Books: Perfect15 for Naum4 and RybkaII for Rybka3.
Game in 60 minutes.


Tomcass, Blitz:60' 0

1 Naum 4(x4) +35/-16/=49 59.50% 59.5/100
2 Rybka 3 32-bit +16/-35/=49 40.50% 40.5/100

A clear, impressive victory of Naum4 over Rybka3. My 50% each forecast has proven clearly wrong. In my Quad 6700 hardware Naum4 is much stronger than I expected.

The books selection has had some importance in the final result. But also here the results are different that what I expected. Here you have the 'average opinion of both engines' after finishing their book moves:

When Naum plays white, his evaluation was + 0,22 on average.
When Rybka plays black, his evaluation was + 0,02 on average.

When Rybka plays white, his evaluation was + 0,21 on average.
When Naum plays black, his evaluation was + 0,01 on average.

So that, both programmes evaluated in a very similar way their positions when finishing the book moves. Therefore there is not a very important incidence in this field.

However, imho the most relevant point is the additional time awarded to Naum4, because of the higher number of book moves played. In other words, Naum played on average 8,6 moves more than Rybka under the book umbrel.la and this fact gave it between 6 and 7 additional minutes per game to think. A very relevant advantage in a 60 minutes game.

I know that my English is not perfect :oops: but I hape you can understand what I mean.

Best regards from Barcelona.

Tom.

Tomcass
Posts: 786
Joined: Sun Apr 16, 2006 7:09 pm

Naum4 - Rybka 3 - 50 additional games. Now Rybka+Perfect15!

Post by Tomcass » Thu Jan 01, 2009 7:58 pm

Tomcass wrote:Quad 6700
Gui Fritz11
Ponder: Off
Books: Perfect15 for Naum4 and RybkaII for Rybka3.
Game in 60 minutes.


Tomcass, Blitz:60' 0

1 Naum 4(x4) +35/-16/=49 59.50% 59.5/100
2 Rybka 3 32-bit +16/-35/=49 40.50% 40.5/100

A clear, impressive victory of Naum4 over Rybka3. My 50% each forecast has proven clearly wrong. In my Quad 6700 hardware Naum4 is much stronger than I expected.

The books selection has had some importance in the final result. But also here the results are different that what I expected. Here you have the 'average opinion of both engines' after finishing their book moves:

When Naum plays white, his evaluation was + 0,22 on average.
When Rybka plays black, his evaluation was + 0,02 on average.

When Rybka plays white, his evaluation was + 0,21 on average.
When Naum plays black, his evaluation was + 0,01 on average.

So that, both programmes evaluated in a very similar way their positions when finishing the book moves. Therefore there is not a very important incidence in this field.

However, imho the most relevant point is the additional time awarded to Naum4, because of the higher number of book moves played. In other words, Naum played on average 8,6 moves more than Rybka under the book umbrel.la and this fact gave it between 6 and 7 additional minutes per game to think. A very relevant advantage in a 60 minutes game.

I know that my English is not perfect :oops: but I hope you can understand what I mean.

Best regards from Barcelona.

Tom.
After the first test I was interested to know what could happen if Rybka3 had the Perfect15 book.

Quad 6700
Gui Fritz11
Ponder: Off
Books: Perfect15 for Rybka3 and Storm_2.1 for Naum4.
Game in 60 minutes.

Tomcass, Blitz:60' 0


1 Naum 4(x4) +39/-20/=91 56.33% 84.5/150
2 Rybka 3 32-bit +20/-39/=91 43.67% 65.5/150

This is the CUMULATIVE result after 150 games.

The second part of this match (Games 101 to 150) has given the following results:

Tomcass, Blitz:60' 0


1 Naum 4(x4) +4/-4/=42 50,00% 25,0/50
2 Rybka 3 32-bit +4/-4/=42 50,00% 25,0/50

My feelings:

- Storm_2-1 is also an excellent opening book. You can find it in Sedat's web site.

- Playing at 60 min game in a Quad 6700 the strength of Rybka3 and Naum4 is very similar. No doubt they are the best 2 commercial engines currently available.

- Probably Rybka3 does not benefit from the extra speed of a Quad 6700.

- In my first forecast for this match I expected about 85% draws. The result in the first 100 games was anly 49% draws due mainly to the difference between Perfect 15 and RybkaII books. Now, with two excellent books (Perfect15 and Storm2.1) the number of draws reaches 84%, very close to my first estimate.

I will test 50 additional games under the same parameters.

My best wishes for a great 2009 for all of you from Barcelona.

Tom.

Uri Blass
Posts: 8948
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Naum4 - Rybka 3 - 50 additional games. Now Rybka+Perfect

Post by Uri Blass » Sat Jan 03, 2009 2:58 pm

Tomcass wrote:
Tomcass wrote:Quad 6700
Gui Fritz11
Ponder: Off
Books: Perfect15 for Naum4 and RybkaII for Rybka3.
Game in 60 minutes.


Tomcass, Blitz:60' 0

1 Naum 4(x4) +35/-16/=49 59.50% 59.5/100
2 Rybka 3 32-bit +16/-35/=49 40.50% 40.5/100

A clear, impressive victory of Naum4 over Rybka3. My 50% each forecast has proven clearly wrong. In my Quad 6700 hardware Naum4 is much stronger than I expected.

The books selection has had some importance in the final result. But also here the results are different that what I expected. Here you have the 'average opinion of both engines' after finishing their book moves:

When Naum plays white, his evaluation was + 0,22 on average.
When Rybka plays black, his evaluation was + 0,02 on average.

When Rybka plays white, his evaluation was + 0,21 on average.
When Naum plays black, his evaluation was + 0,01 on average.

So that, both programmes evaluated in a very similar way their positions when finishing the book moves. Therefore there is not a very important incidence in this field.

However, imho the most relevant point is the additional time awarded to Naum4, because of the higher number of book moves played. In other words, Naum played on average 8,6 moves more than Rybka under the book umbrel.la and this fact gave it between 6 and 7 additional minutes per game to think. A very relevant advantage in a 60 minutes game.

I know that my English is not perfect :oops: but I hope you can understand what I mean.

Best regards from Barcelona.

Tom.
After the first test I was interested to know what could happen if Rybka3 had the Perfect15 book.

Quad 6700
Gui Fritz11
Ponder: Off
Books: Perfect15 for Rybka3 and Storm_2.1 for Naum4.
Game in 60 minutes.

Tomcass, Blitz:60' 0


1 Naum 4(x4) +39/-20/=91 56.33% 84.5/150
2 Rybka 3 32-bit +20/-39/=91 43.67% 65.5/150

This is the CUMULATIVE result after 150 games.

The second part of this match (Games 101 to 150) has given the following results:

Tomcass, Blitz:60' 0


1 Naum 4(x4) +4/-4/=42 50,00% 25,0/50
2 Rybka 3 32-bit +4/-4/=42 50,00% 25,0/50

My feelings:

- Storm_2-1 is also an excellent opening book. You can find it in Sedat's web site.

- Playing at 60 min game in a Quad 6700 the strength of Rybka3 and Naum4 is very similar. No doubt they are the best 2 commercial engines currently available.

- Probably Rybka3 does not benefit from the extra speed of a Quad 6700.

- In my first forecast for this match I expected about 85% draws. The result in the first 100 games was anly 49% draws due mainly to the difference between Perfect 15 and RybkaII books. Now, with two excellent books (Perfect15 and Storm2.1) the number of draws reaches 84%, very close to my first estimate.

I will test 50 additional games under the same parameters.

My best wishes for a great 2009 for all of you from Barcelona.

Tom.
I can only guess that the meaning of your result is that storm2.1 is probably stronger opening book than perfect15 because it is known that rybka is stronger than naum.

Uri

Uri Blass
Posts: 8948
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Naum4 - Rybka 3 - 100 games. FINAL STANDINGS.

Post by Uri Blass » Sat Jan 03, 2009 3:02 pm

Tomcass wrote:Quad 6700
Gui Fritz11
Ponder: Off
Books: Perfect15 for Naum4 and RybkaII for Rybka3.
Game in 60 minutes.


Tomcass, Blitz:60' 0

1 Naum 4(x4) +35/-16/=49 59.50% 59.5/100
2 Rybka 3 32-bit +16/-35/=49 40.50% 40.5/100

A clear, impressive victory of Naum4 over Rybka3. My 50% each forecast has proven clearly wrong. In my Quad 6700 hardware Naum4 is much stronger than I expected.

The books selection has had some importance in the final result. But also here the results are different that what I expected. Here you have the 'average opinion of both engines' after finishing their book moves:

When Naum plays white, his evaluation was + 0,22 on average.
When Rybka plays black, his evaluation was + 0,02 on average.

When Rybka plays white, his evaluation was + 0,21 on average.
When Naum plays black, his evaluation was + 0,01 on average.

So that, both programmes evaluated in a very similar way their positions when finishing the book moves. Therefore there is not a very important incidence in this field.

However, imho the most relevant point is the additional time awarded to Naum4, because of the higher number of book moves played. In other words, Naum played on average 8,6 moves more than Rybka under the book umbrel.la and this fact gave it between 6 and 7 additional minutes per game to think. A very relevant advantage in a 60 minutes game.

I know that my English is not perfect :oops: but I hape you can understand what I mean.

Best regards from Barcelona.

Tom.
I disagree with your conclusions.
evaluation out of book proves nothing because programs do not know to evaluate positions out of book correctly.

It is possible that white has a clear advantage and both programs evaluate the position as equal.

I do not believe that the only advantage of naum was time advantage
because
Rybka with 50 minutes per game is supposed to be better than naum with 60 minutes per game

Uri

Post Reply