Has Fruit finaly surpassed Rybka!?

[Tony Thomas]

http://wbec-ridderkerk.nl/html/premier.htm

No. This data is insufficient to conclude anything about the relative strengths between these two engines.

Yes Terry,but if this was the situation in a round robin tournament in a championship,a difference of 5 points is impressive....

You're making too much out of it. After 100 tournaments, well that would be a different case.

Terry

How often do you see a tournament were Rybka is behind after 28 rounds? Me and Wael runs computer tournaments on a regular basis, so you have to give us the benefit of the doubt that we might know something you dont.

[Tony Thomas]

http://wbec-ridderkerk.nl/html/premier.htm

No. This data is insufficient to conclude anything about the relative strengths between these two engines.

Yes Terry,but if this was the situation in a round robin tournament in a championship,a difference of 5 points is impressive....

Especially since none of the engines even managed a win against Rybka in the last edition.

Yes Tony,that and the fact that in this very particular tournament Fruit was never ahead of Rybka with 5 points....

You did notice that Rybka hasnt played the 7th round right? So technically it isnt 5 points behind, more like 1 or 2.

[Dr.Wael Deeb]

http://wbec-ridderkerk.nl/html/premier.htm

No. This data is insufficient to conclude anything about the relative strengths between these two engines.

Yes Terry,but if this was the situation in a round robin tournament in a championship,a difference of 5 points is impressive....

Especially since none of the engines even managed a win against Rybka in the last edition.

Yes Tony,that and the fact that in this very particular tournament Fruit was never ahead of Rybka with 5 points....

You did notice that Rybka hasnt played the 7th round right? So technically it isnt 5 points behind, more like 1 or 2.

Yes Tony,I noticed that,but I am still impressed

[Nid Hogge]

http://www.husvankempen.de/nunn/40_40%2 ... liste.html

#18 with only 71 games. Need to wait and see.

It has cameback from worse though. must be cautious when burying it too quickly. Iv'e seen plenty of these "Rybka in danger!?" threads from different sources and after all the hype and talking it has always remained on top. I suspect it won't be different this time but I hope i'll be proved wrong.. Can't wait for some real competition at the top.

[El Gringo]

http://wbec-ridderkerk.nl/html/premier.htm

And look at this :

Code: Select all

Fruit 2.4 Test Beta A x64 2cpu   - Rybka 1.2f  x64               18.0 - 22.0    +10/-14/=16    45.00%

Don't worry, be happy.....still a (long) way to go Mr Deeb


Best
Johan

[Terry McCracken]

http://wbec-ridderkerk.nl/html/premier.htm

No. This data is insufficient to conclude anything about the relative strengths between these two engines.

Yes Terry,but if this was the situation in a round robin tournament in a championship,a difference of 5 points is impressive....

You're making too much out of it. After 100 tournaments, well that would be a different case.

Terry

How often do you see a tournament were Rybka is behind after 28 rounds? Me and Wael runs computer tournaments on a regular basis, so you have to give us the benefit of the doubt that we might know something you dont.

Probably not often...maybe never, still that has no bearing on the relative strength of Rybka vs Fruit...it is indictive however other programs are climbing up the ladder fast and closing the gap.

Hiarcs which is more than 100 pts. weaker than Rybka beat Rybka and Rybka lost two tournaments to Hiarcs if I'm not mistaken, but I'm not going around and saying look at Hiarcs, it must be much better...sigh...

[Dr.Wael Deeb]

http://wbec-ridderkerk.nl/html/premier.htm

And look at this :

Code: Select all

Fruit 2.4 Test Beta A x64 2cpu   - Rybka 1.2f  x64               18.0 - 22.0    +10/-14/=16    45.00%

Don't worry, be happy.....still a (long) way to go Mr Deeb


Best
Johan

Still,I am impressed....

[Ryan Benitez]

Fruit 2.3.4n is the version that played in CCT10 with the bug fix added after the Nuam game. Fruit 2.4 Beta A is with a completely different weak pawn function. In my tests the one in 2.4 A should be better than 2.3.4n but not all tests agree. 2.4 Beta B is changes in tempo understanding from 2.4 A. This is an attempt at something that is not as simple as a generic side to move bonus for tempo. With no tuning to the ideas in the B version I would be surprised if it was an improvement on the first try. I do understand that tempo is more commonly solved as a search issue (no progress pruning) but I would like to at least try to solve it as an eval issue. Also about 64bit speed, in the current versions the GCC compiles on linux the 64bit versions are about 30-35% faster than the 32bit versions. On windows the 64bit compiles are only about 20% faster than the 32bit versions. A lot can be done to make the 64bit speed faster but I lack the motivation to do so at this time. Dedicating a weekend to optimizing could bring about 30 elo on 64bit systems, nothing ground breaking. Other than improved SMP the only thing that will continue to change in Fruit is tuning. The eval has little in common today with the eval of Fruit 2.2 and most of the values at this point are just best guess values. I have no guess as to how much change can be had from tuning. Some of my first gets values can not possibly be very good but I don’t know how much that matters overall. Also the way I would like to handle open files is not active because it did not optimize well. I know how to do it right now so fixing that may be the only code I write in Fruit in the near future. It may seem silly but my main focus right now is writing a computer Stratego program.

Thanks,
Ryan

[George Tsavdaris]

So when you will release current Fruit 2.3.4n or whatever newer in the public? You've said after CCT10 but you didn't

It may seem silly but my main focus right now is writing a computer Stratego program.

Why silly? Not silly at all.
Do you know of any other such engine that would play Stratego? It seems difficult to write a decent one engine. And to outplay humans seems very difficult also.....

[Beast]

I bet the HIARCS 12 isn't 100 weaker than Rybka.

Todd