Rybka Chess Community Forum
Congrats to Vas, Lukas, Jiri and Hans, well done!!!
First of all a hearty congratulations to Vas and Lukas, Jiri and Hans and Jeroen and Nick (am sure you guys were involved in there somewhere
). Ironically many of Rybka's wins came in endgame positions that we had heavily criticized as being in need of improvement. It will be very hard to criticize Rybka's endgame play after it seemed to have squeezed water out of baked rock. What is sure is that any endgame mistake by any engine will be severely punished.
One thing about this tourney though, the entire tournament was very poorly broadcast and advertised. I think all parties should push for active Live transmission of the games, and better advertisement that such a tourney was taking place. I thought that the entire WCCC was extremely disappointing to follow.
And yes, I also was impressed by Rybka's endgame capabilities. This version has a completely new cluster algorithm which can effectively use many computers.
Again it was hard work to get things running - and I was way behind my schedule. During the first game I had to write a tool which collects logfiles. This shouldn't take much time, but testing it I ran into some network problems which had to be solved first - while the game was running. I decided to try some kind of transmission, but log files now are so huge that you shouldn't try to get positions manually. So I had to write a tool for this first. I did that during the second game. So I'm sorry that I couldn't make a transmission of these 2 games. Although the games are over now, there's still much work to do. I'm still not satisfied with the tools we use for transmitting data. And there's a bunch of new and very cool parameters that have to be optimized.
Lukas, do not apologize! Thanks to you we have been able to follow Rybka's games live, in the absence of proper live coverage by the ICGA. A huge thanks to you for this service!
CG Rybka team.Thank you Lukas for broadcasting all the rybka games.
Yes, thank you yet again for the live games.
can these parameters be made public soon?
Well, some are cluster specific - they can never make it into a normal Rybka version. The other ones are secret. But maybe I can convince Vas to put the one I like most into the Rybka 4 bugfix version.
> But maybe I can convince Vas to put the one I like most into the Rybka 4 bugfix version.
Rybka 4+?? Wake up, Lukas, you are dreaming....
Lukas he is saving his bag of tricks for the rental version. You should know this!
How you guys were able to stretch R4 elos as far as you did for what you were given is quite amazing to me. It doesn't take a genius to figure out the rental version has the last 2 1/2 years work of src/eval improvements into it.
Lukas your efforts were appreciated by everyone, Thank you ever so much
PS hate to share this in a way, but I had some nice thoughts about Rondo.
Zap has always been one of my favorite and dependable engines.
Well, the games showed that Rybka has poor endgame knowledge and didn't know that all rook endgames are a draw. she won them instead.
I don't really believe in the drawishness of rook endgames. That's why I didn't set up a rook endgame penalty. I claim this was a good decision.
oh, well,... as a tournament player I can say that many rook endgames are drawish indeed :) but it's a different story if the pawns are not on the same wing, one side has a very active king or can get a fast passed pawn. Activity and passed pawns are the key. you need both to win :)
No you can win even without passed pawns by creating them later
In this position(against Junior) rybka had no passed pawns and the pawns were on the same wing and rybka won the game.
8/3p4/1k2p3/4P2R/1r6/8/2K5/8 b - - 0 68
well, in this case the pawn on e5 will fall :) if that wouldn't be the case, it would of course be a draw. So there's a "combination" that wins the pawn. note that the center is not a wing :D
What I would like to see, before the tournament is broadcast, is the rules requiring that the commercial versions of the engines be used.
It is somewhat dishonest to claim 'Rybka' (or whatever engine) won the computer chess championship and use it to market the product, when really it isn't Rybka, but some souped up version that is significantly different.
If we follow your suggestion, all programs will get disqualified. None of the commercial programs was used, but everyone used some enhanced version. The amateur programs will have to be disqualified too, as they don't have a commercial version.
So thank you very much for your thoughtful proposal.
I'm sure that he must have been joking or sarcastic--the stupidity of the comment is, I think, so obvious that there's no way that he wasn't being sarcastic.
It's not the competition I object to, people are free to hold any tournament with whatever rules they see fit, but it is the misuse of results in marketing the product I object to. This of course does not just apply to Rybka. To claim to someone in an advert Rybka 4, for example, that 'Rybka is the 2010 World Computer Chess Champion' would be misleading as the claim is actually for a different bit of software than the one being advertised (I haven't seen this with Rybka 4 adverts yet, but I have seen similar adverts in the past.)
Generally speaking, advancements are made between the competition and the next release. Whether these advancements are used is up to the programmer, but it is still Rybka. Rybka won the WCCC, and they are selling Rybka. That the product name basis can remain the same with a slightly different product has been legal for many centuries.
I understand completely. It's the way things are though. The way it is has always been done, even if it isn't completely accurate.
You could argue that the closest commercially available engine to whichever favorite you had at the wccc would be that engines own commercial version. So in a way it isn't completely false advertising. So far the winners have more or less coincided with how the engines stack up on the rating lists as well.
These kind of things generally get exacerbated over time though. Like NASCAR racing where the cars have absolutely nothing in common with the brands that sponsor them.
A number of chess engines have competed with cluster support:
- Deep Sjeng
- Grid Chess
None of these cluster versions are available to the public...
I dont think Zappa has.I think previously it was a different version just capable of running on a supercomp and wasn't a cluster.Cluster rondo hasnt yet played a tournament game.
As I recall, Zappa ran on a 512 processor machine a few years ago. I suspect that Anthony had to deal with the same issues (breaking up the problem and not having a quickly accessible global transposition table) that clusters have to deal with.
I am looking forward to seeing cluster Rondo in action, but I guess Zach has a lot of things to work on right now.
there is a cluster rondo version but strictly private so far.
...for example, that 'Rybka is the 2010 World Computer Chess Champion' would be misleading as the claim is actually for a different bit of software than the one being advertised ."
I don't see the point in going into something that is pure, negative, supposition.
I am pleased that Rybka 4 cluster-buster won the WCCC.
I am proud to be a long time standing member of this forum, who can in some small way share in this victory. It has become a tradition.
With that I hail R4's elite sister who dwells on Mt Cluster, Super- little fish- Goddess!
> If we follow your suggestion, all programs will get disqualified. None of the commercial programs was used, but everyone used some enhanced version. The amateur programs will have to be disqualified too, as they don't have a commercial version.
> So thank you very much for your thoughtful proposal.
Good heavens man. He was just making a suggestion of what he would find interesting. For someone who likes to talk about "dreaming" you are pretty closed minded.
> For someone who likes to talk about "dreaming" you are pretty closed minded.
You've got balls of steel hey?
At the risk of again being called a Rybka-lover elsewhere, I'd like to add my congratulations to the Rybka team too.
Yet another heroic pummeling of the opposition. Nicely done.
My congratulations too! Rybka was in the "cluster event" the best engine, had the best book, and also had the best team.
Honestly? I was hoping that Rondo could be the champion. I always liked
Zappa and I think that Zach Wegner is a good and intelligent person.
I think what I really want is that others engines got the level of Rybka,
would be good for computer chess.
But the reality seems not to be this, so I congratulate the Rybka team
for another win in WCCC.
A very strong and impressive WM-sequence! Sure!
I'm interested to read more about the used hardware
"200 Nehalem EP Westmere, 2.93-3.6 GHz" sounds very interesting.
Can I read somewhere more about this?
I am waiting for the report from Lukas' wife on how she feels about the second computer room!
She really loves it. It's in the basement. Now the laundry dries much quicker and the living room which is above that room is a bit warmer - it was too cold before.
She really loves it.
Keep telling yourself that!
Congrats for all this nice team, which is going to break a few records.
Are there any new features of the Rybka version used durin this WCCC, (fug fixes i suppose, some new features in endgame etc.)
you mean fug bixes i suppose
Yes, congratulations to Rybka team on winning the WCCC title!
Now, I may be the weird one that was cheering for Rondo, but still, congratulations on a well deserved victory!
It didn't do bad either. lets see who gives tough competition to Rybka except Rybka itself.
Well almost. While I am glad rybka won, I had a strong pull for Rondo.
great win.i have always been impressed by Rybka's late middlegame and endgame play.
Congrats! looking forward to the openings report (if there will be one).
Congrats to the Rybka team. It was a well deserved victory. The only game I was not happy about was draw against Rondo. It was also the only time I think Rybka was in trouble of losing the game.
Congrats to the Rybka team!
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill