So the question is how much too late he will be ;)
At the computer tournament Alexander Naumov was already there and after installing Windows XP 64 bit Arena he noticed that his engine was slower under Arena and he tried to fix it by installing the ATI drivers for the graphics card - tomorrow we will see if that worked :)
Ask Vas how much time he has put into the new chess960 engine (testing and so on).
> he didn't know much about the "Deutsche Bahn"
Are trains in Germany late too!???
Greece is a great place to live and the worst place to work.
>At the computer tournament Alexander Naumov was already there and after installing Windows XP 64 bit Arena he noticed that his engine was >slower under Arena and he tried to fix it by installing the ATI drivers for the graphics card - tomorrow we will see if that worked :-)
Was slower under Arena relative to what GUI?
Also why he believes that the problem will be fixed if he installs newer graphic card drivers?
Because the drivers weren't found during installation and he thought that could be a solution :)
Don't forget to take some new pictures again in Mainz. It'll be very interesting to have again a new report here in the page.
Is it a computer-computer tournament the one that Rybka will participate?
Are the opponents Shredder, Naum and Spike?
If this is true then this tournament is a dubious one, since Hiarcs and Deep Sjeng at least, are very good FRC players also and they will not play.
It's a fisher random chess computer tournament.
>The opponents are Naum, Sjeng and Shredder, so the strongest ones.
How do you know that they are the strongest ones?
CCRL has Hiarcs 12 as 5th but the error bars are such that it may be even 3rd.
And also the confidence level, since it's not 100% obviously, it always leaves room for Hiarcs for example to be even the strongest FRC engine.
Also the time control of CCRL is 40/4 so may be that with the time control that will be used on Mainz, Hiarcs or Spike for example will be those that would be in the top 4.
Also even if these 4 are the strongest ones, why choose the top 4 and no the top 5 or the top 10 or even better allow a free participation?
The format has the advantage that you have only top programs playing on a small stage. before it was not really attractive, many unknown programs and confusing for the spectators.
Now i don't understand this you've said:
>twice as much games per day- so 6 games tomorrow
Rybka's opponents are 3 so a Round-Robin tournament with alternating colors for each side means 6 games.
But the way you use the phrase "so 6 games tomorrow", makes me believe that there will be games in the day after tomorrow also.
So is this a double RR tournament or what?
And actually what is exactly the program if you know?
The program? Which one? Rybka is Rybka 3 :)
>The program? Which one? Rybka is Rybka 3 :-)
:-P No i meant the program=schedule of the tournament.... :-)
>Rybka plays against every opponent 4 times (with black and white twice) afaik.
Oh so all programs will have 12 games each.
So it will be a 4 days tournament right?
It seems the tournament has some serious misinformation written to its official page. Very misleading.
I see there:
>"Category 29: The strongest chess tournament…ever!
>Livingston Chess960 computer chess world championship"
I don't know if it's the strongest Chess tournament ever, but i don't understand on what base they say this. Even that way, perhaps CEGT 120/40 QUAD list is the strongest Chess tournament ever, or anyway even if it's not, i don't see why this "Livingston Chess960 computer chess world championship" must be the strongest one.
Also this tournament is not Chess as they say in the title. It's a variation of Chess: FRC-Chess.
Also about this also very misleading statements:
>"The engines that will battle it out in Mainz are the numbers 1-4 on the Chess960 computer rating list. Arbiter Hans Secelle calculated that the >average rating of Rybka, Shredder, Naum and Deep Sjeng adds up to (3052 + 2970 + 2956 + 2947) = 2981. This means that we'll have a >Fide-category 29 (!) tournament or, in other words, the strongest tournament in the history of chess.... (Eric van Reem)"
I explained earlier that CCRL 40/4 FRC list is not appropriate to make the conclusions that they make. Error bars, time controls, and not use of latest versions(perhaps Hiarcs beta is stronger than the Naum that participates in the tournament).
Also the above article has another serious mistake:
They assume the ELO of the programs that have in CCRL 40/4 FRC list corresponds exactly to FIDE ELO.
Obviously this is unjustified and wrong.
For example if CCRL have chosen to use a base rating for Glaurung 2.0.1 64bit, 2500 instead of 2800, then this Mainz tournament average ELO would be 2681 instead of 2981.
Machines have AMD Phenom 9600 4x2,30 GHz with 2048 MB RAM und Windows XP 32-bit.
PS: I think, Vas will install xp 64-bit, if it´s allowed. I hope, it´s not allowed! :-)
The conditions could be more unfavourable for rybka.
For example single processor machines.
I can't seem to find them....
Also it's obvious(just like Jeroen said) that Rybka wants to castle as quickly as possible and this obviously bad for several reasons.
BUT FRC-Chess is not Chess. So where is your objection?
I guess you believe FRC-Chess is more or less like Chess and that Rybka that is extremely good on Chess should normally be also extremely good at FRC-Chess.
Because FRC-Chess after a point, where castle rights are forbidden or already castle has occurred, the big majority(not all) of the FRC-Chess positions are equivalent to Chess positions so Rybka should play them extremely well also.
But it seems it doesn't. And this is a logical question.
So i have to ask Vasik or Larry why?
Why Rybka seem not to have this invincible aura in FRC-Chess also?
Doesn't Rybka 3 play well the openings? But Rybka 3 is supposed to know much more about openings than any other program. Or this is only about Chess openings?
Does have a bug in FRC-Chess implementation?
Did the version of the tournament had a bug?
Does Rybka 32-bit is so much worse than 64-bit and if 64-bit was used then the situation will be completely different?
Or the results are normal?
Do you have an evidence that rybka3 has lower branching factor than rybka2.3.2a?
If rybka's low branching factor was so bad for the opening how rybka could do so well in cegt and ccrl when programs play with no opening book and have to find theory moves by themselves?
Edit:Note that CCRL includes also chess960
Rybka's branching factor in 2.3.2a was claimed to be around 2. To make up for a slower evaluation, a more selective search is used in R3. I assume this means a lower branching factor. Maybe you can provide some reasonable alternatives?
Of course CCRL did _not_ use opening books for their FRC games.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill