Reports | October 06, 2008 18:08

Rybka clinches 2nd Computer World Championship title

Computer World Championship 2008Rybka, universally considered the best computer engine, comfirmed her status once again at the Computer World Championship held in Beijing last week. In a ten-player round robin she finished on 8/9, one point clear of Hiarcs.

The 16th Computer World Chess Championship was held September 28-October 5 in the Beijing Golden Century Golf Club in Fangshan, Beijing. It was part of the Computer Games Championship which also included tournaments for draughts, checkers, backgammon, Go and others.

Originally twelve chess programs would play, but later this number was brought down to ten. Unfortunately among them were not Fritz and not Zappa. For some reason Fritz stopped playing in computer tournaments, and the programmer of Zappa, the engine that beat Rybka in a match last year, is not active anymore.

Luckily there were enough strong programs to expect an exciting world championship. Besides Rybka, the big favorite for the title, the participants list included Hiarcs, Junior, Cluster Toga, Shredder and Sjeng.

Both Shredder and Junior have won several computer world championships. During the last two years Hiarcs has been playing very well in international tournaments and after its great second place behind Rybka in May this year, Sjeng also belonged to the list of dangerous outsiders.

Venue of the Computer World Championship 2008

The venue of the 13th Computer Games World Ch and the 16h Computer Chess World Ch

A World Championship is always "unlimited" which means that every participant may use any hardware he or she likes. On a personal note, I still have difficulty "accepting" this concept and would prefer to see a world championship in which opponents play on the same hardware, but alas this was not the case.

This situation was not a bad one for Vasik Rajlich - the Rybka programmer - because shortly before the tournament he had managed to get Rybka running on a cluster of 5 PCs and so he could use 5 x 8 = 40 processors. This hardware configuration makes Rybka about 150 ELO points stronger than on an ordinary Quad.

The programs Cluster Toga and Jonny were running on similar clusters, Junior tried its luck on a brand new 12-core PC while Hiarcs and Shredder were using 8-core Skulltrail machines. It should be mentioned that the only Chinese participant, Mobile Chess, was playing on a Nokia mobile phone :-).

Venue of the Computer World Championship 2008

This is what a computer chess playing hall looks like.
Usually there's a friendly atmosphere where chatting is allowed!

The tournament went great for Rybka. After seven rounds, there was only one rival left: Hiarcs had 6 points (Rybka 6.5 uit 7) while the rest was too far behind. In the penultimate round, Rybka played Hiarcs and reached a winning position within twenty moves, which meant the tournament was decided and Rybka had won her second world title.

Rybka only drew with Cluster Toga and Junior and won her other seven games. Hiarcs was the deserved runner-up; it only lost to Rybka and defeated Shredder and Junior among others.

[TABLE=415]

  • SOS: Sum of Opponent Scores
  • SoDOS: Sum of Defeated Opponent Scores

[TABLE=416]

Many thanks to Jeroen Noomen for providing the photos and a large part of the above text

Links:

Peter Doggers's picture
Author: Peter Doggers

Founder and editor-in-chief of ChessVibes.com, Peter is responsible for most of the chess news and tournament reports. Often visiting top events, he also provides photos and videos for the site. He's a 1.e4 player himself, likes Thai food and the Stones.

Chess.com

Comments

wright's picture

"It should be mentioned that the only Chinese participant, Mobile Chess, was playing on a Nokia mobile phone :-)."

LMAO!!!

thorex's picture

A tournament on differnt hardware isn't even worth mentioning. Rybka might be the strongest program anyway, but the hardware range is too big to confirm this seriously.

Look at Falcon for example. It managed to win with a moderate configuration (my 400?¢‚Äö¬¨ home pc is far stronger!) against a 24-cluster Toga. I think this is quite impressive.

Chessmaster's picture

"A tournament on differnt hardware isn?¢‚Ǩ‚Ñ¢t even worth mentioning. Rybka might be the strongest program anyway, but the hardware range is too big to confirm this seriously."

First of all, it's difficult or impossible to use unified hardware in practice in these tournaments, since the programs might be written for entirely different system architectures. You can't force the programmers to code their chess programs for a specific platform.

Second, the World Computer Championship has always allowed programs on different hardware to compete against each other (partly because of the reason mentioned above). Faster hardware doesn't necessarily mean you'll win: Fritz did for example defeat Deep Blue's predecessor in the WCC many years ago.

Third, the latest version of Rybka is so superior compared to other programs (by 150-200 points to its nearest rivals), that it would most likely have won even when running on much slower hardware.

Usana's picture

The problem with having all programs use a standard hardware is that some hardware configurations may be optimal for certain programs. By limiting the hardware, you're limiting the program strength. Better to let the programmer who knows their software best to choose whats best.

semipatz's picture

"Better to let the programmer who knows their software best to choose whats best."

Yeah, except inevitably some programmers manage to get a hold of hardware that is simply BETTER than others. As a result, some softwares play with a handicap. Most chess fans would really rather see a pure test of the engines themselves, I think.

arne's picture

I think Usana has a good point. Still, I'm sure all programmers will benefit from hardware being as fast and modern as possible. Surely no engine is running on a Commodore 64 :-) '
Personally, I think these matches are only interesting from an artistic point of view: sometimes a curious or good game will inevitably be played, with some memorable tactics. And just as often, a nice and memorable blunder will be made.
As for the competitive element: I don't see it as long as we can't quantify what it is we're measuring exactly. And I don't think it matters.

Arjon's picture

Actually, the real problem with using identical hardware for everyone is that one architecture would get supported above other architectures. Certainly, programs have been optimized for a certain number of cores, but the essential difference in the programs is still purely algorithmical and in evaluations.

The problem is that nobody can tell everyone to use 1 architecture, as this would be like free sponsoring.

Felix's picture

"...and would prefer to see a world championship in which opponents play on the same hardware"
Ok, what do you want? The strongest chess engine should be the winnder? Then you should use the rating lists and award the engine which is number one with the title.Those world championships are just show, if you let them play on the same hardware, this makes things not better.

Chessmaster's picture

"The problem with having all programs use a standard hardware is that some hardware configurations may be optimal for certain programs."

Actually, it's the other way around: The programmers take advantage of certain hardware. It's not that some configurations just happens to suit a particular program well: The programmers deserves credit for tweaking and optimizing their programs for a specific platform. Optimization is part of the game and it's been like that for many decades. Rybka, for example, is optimized for 64 bit architectures (it plays significantly better on 64 bit x86 CPUs).

"The strongest chess engine should be the winnder? Then you should use the rating lists and award the engine which is number one with the title."

Yes, the rating lists are much more reliable when it comes to assessing a program's strength compared to other programs, since these ratings are based on hundreds or thousands of games.

The fact that only 10 rounds or so are played in the WCCC brings a high dose of luck into the equation, although the strongest programs do tend to win the championship.

"Those world championships are just show, if you let them play on the same hardware, this makes things not better."

Besides, if they had to use the same hardware, then some programs wouldn't be able to participate simply because they're based on another system architecture. And that's not fair at all if you ask me.

Chessmaster's picture

"Personally, I think these matches are only interesting from an artistic point of view: sometimes a curious or good game will inevitably be played, with some memorable tactics. And just as often, a nice and memorable blunder will be made."

For many chess computer enthusiasts -- not to mention the programmers themselves -- winning the WCCC title is also a matter of prestige and fame. Such a title can mean a lot to a commercial chess programmer.

You don't get many headlines for topping the SSDF/CEGT/CCRL rating lists, but you do get media coverage and attention when you win a world championship.

test's picture

We already know which program is the best on identical hardware: just look at the computer rating lists.
Letting the programmers choose the hardware might be interesting, it does however un-level the playing field imo and diminishes my interest, if there was any to begin with.
And what do I care how some program performs on say IBM's Roadrunner supercomputer (1.026 petaflop/s), it's not like anybody would be able to use that at home

TrapArecev's picture

@test Hush! You are revealing Fritz' secret comeback scheme!

HCL's picture

Assuming one wanted to place artificial constraints, the variable to control for isn't hardware, it's budget. Given the same budget (say, $100,000), let the programmer choose the optimal hardware/software combo. There's no separating the two entirely.

However, even that is inferior to the tournament ideal: maximize chess strength, using whatever means you have, as determined by a "mano-a-mano" test of strength.

This free-for-all's probably the best format by far. (Who the heck wants to disqualify a program based on budget or hardware, if that particular combo makes the strongest in the world?)

HCL's picture

Btw, several of the Rybka games are absolutely amazing. It's blowing away other grandmaster-strength programs the way Fritz blows wimpy little me away. It's one heavyweight combination after another, until the opponent collapses.

capman's picture

absolutely greatest chess thinker.

Did24's picture

These games are anguishing.... I don t understand half the moves which have been played!! I feel that computers are not playing the game "chess" that i know :(

HCL's picture

These games are anguishing?¢‚Ǩ¬¶. I don t understand half the moves which have been played!! I feel that computers are not playing the game ?¢‚Ǩ?ìchess?¢‚Ǩ? that i know

Rybka can play these wild-looking sacs that are probably sound because it sees 50 moves ahead in every sub-tree, or something insane like that. Thus it dares to leave multiple pieces "hanging" for five, ten moves in a row. The tactics are truly un-human.

It's got an amazing move horizon, the ones grandmasters used to have nightmare premonitions of in the 1980s and early 1990s. It's showing off.

Latest articles