A total of three events was staged by the ICGA:
- The 17th World Computer Chess Championship – all computers restricted to eight cores at the most
- The 14th Computer Olympiad, Chess – no hardware limit
- The 17th World Computer Chess Championship (Blitz)
All three titles were won by Rybka, the first on an eight-core system (obviously), the second on a 52-core cluster and the Blitz again on the eight-core system.
Result of the 17th World Computer Chess Championship
|1||Rybka||USA||Intel Xeon W5580 @ 3.2GHz x 8||8.0|
|2||Deep Sjeng||BEL||AMD 3.2Ghz x 4, Intel X5560 @ 2.8Ghz x 8||6.5|
|2||Shredder||DEU||Intel Xeon W5580 @ 3.2GHz x 8||6.5|
|2||Junior||ISR||Intel Xeon W5580 @ 3.2GHz x 8||6.5|
|5||Hiarcs||GBR||Intel Xeon W5580 @ 3.2GHz x 8||6.0|
|7||The Baron||NLD||AMD Opteron 270 @ 2.0Ghz x 4||3.0|
|8||Equinox||ITA||8x Intel(R) Xeon(R) CPU X5355 2.66GHz||2.0|
|10||Joker||NLD||Core 2 Duo||0.5|
Here's a cross table of the event – the rankings in the above ICGA table are of course official.
Result of the 14th Computer Olympiad – no hardware limit
|3||Deep Sjeng||BEL||cluster, 56 cores||3.0|
Result of the 17th World Computer Chess Championship (Blitz)
Junior had to withdraw after three rounds because of hardware difficulties.
Resumé of the World Championship
By Rybka author Vasik Rajlich
The ICGA held the 17th rendition of their events this past week in Pamplona. We won the Open Hardware Computer Chess Olympiad ahead of Shredder and Deep Sjeng, the World (8-Core) Computer Chess Championship ahead of Junior, Shredder and Deep Sjeng, and the blitz tournament ahead of Shredder.
Rybka author and triple computer chess world champion Vasik Rajlich
I'd like to thank the ICGA for making the events happen, and to give a huge thanks to the guys on our team who worked hard to make Rybka's participation possible. Our team was:
Hardware and Testing
The credits start with Lukas Cimiotti, who over the past year has systematically built and tweaked a monstrous piece of hardware. His current cluster contains one w5580, two Skulltrails, another Harpertown oct, and five i7-920s, for a total of 52 cores.
A monstrous piece of hardware – the Rybka cluster with 52 cores
We had various problems getting everything to work, so in addition to his hardware and tuning work Lukas also contributed a lot of testing and debugging help. This clustering project would probably never have been started without Lukas, and it certainly wouldn't have come as far along as it has.
We finally got everything working this past Tuesday morning, and for now have only preliminary performance figures. The cluster won a match against a Nehalem Quad (i7-920) running the latest Rybka version with a score of +37 =25 -2 (+213 Elo). I suspect that with more games this Elo gap will be slightly lower.
Lukas is a regular in the Playchess engine room (handles: Rechenschieber, Victor_Kullberg) and will probably run his cluster there on occasion.
Nick Carlin handled our opening book, taking advantage of material previously published by Jeroen Noomen, and did really well. All of our book positions were equal or better, and all were complex and offered plenty of winning chances.
Nick is from the new breed of computer chess opening authors, who rely on systematic, automated methods and on statistical analysis. He uses all of the available resources, from Jeroen's work to Playchess games to Aquarium editing tools to Polyglot. For this event, he made algorithmic innovations in the area of sharpening the book exit points – his aim was to drop Rybka off into rich positions with plenty of winning chances, and to prevent opposing authors from doing the opposite. Judging by the games, these methods work quite well.
Unfortunately, Nick has decided to take an extended break from computer chess after these events, as the time required to stay on top of everything is just too high for him. Jeroen is also still taking a break after his last book release and has started to apply his skills to the stock market – hopefully he will soon be rich and will then return to what is best in life. All of this should underline just how much work is involved in the book preparation. The responsibility is high, as one mistake can spoil an entire event and wipe out the work of everyone on the team. Modern opening theory is simply a huge load and we will have to think about how to handle it.
Vas Rajlich at hom testing new versions of his program (yes, the computers are on a
bed frame with no mattress, and yes, those are normal household fans assisting in the cooling).
Hans van der Zijden was our on-site representative and operator and won the blitz tournament, where operator skill is important. His workload was heavier than usual, as there were three tournaments this time instead of the usual two. You can read his reports on our web site.
Victor Zakharov and his team from Convekta (ChessOK) let us use internal versions of several unreleased tools which they have developed. They were also extremely responsive to feature requests. The most important of these tools was a remote engine server & client networking tool, which we used for intra-cluster communication as well as to host our playing engines. This tool has a number of nice features, one being automatic smooth reconnection capability.
New eight-Core Rule
The ICGA added a new eight-core tournament this time around and gave it exclusive "World Championship" status, causing quite some controversy. While it's hard to go wrong with more events and more computer chess, this particular rule is badly formulated. The normal tradeoff with hardware limitations is that looser limitations encourage innovation while stricter limitations make the tournament less expensive and less of a hassle to participate in. An eight-core rule gives you the worst of both: it prevents innovation with clusters of off-the-shelf components while emphasizing the use of expensive workstations. I hope that this rule is retracted for future events.
Here's a game played by the 52-core cluster in the Olympiad:
Rybka - Shredder [C67]
Computer Olympiad Pamplona (5), 18.05.2009 [Felix Kling]
1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.0-0 Nxe4 5.Qe2. A rare move. 5.d4 is of course the main line. 5...Ng5 This is also a rare move. The opening looks like a clever choice by Jeroen, since engines seem to have problems to understand what is going on. [5...Nd6 6.Bxc6 dxc6 7.Qxe5+ Qe7 8.Qa5 Qd8 9.Qc3 is the most principle line. Black has some problems to develop it's pieces, since the black bishop has to defend the pawn on g7. White has some small advantage I'd think.] 6.Nxg5 Qxg5 7.d4 Qe7 [In the first round of the Olympiad, Deep Sjeng played the opening better and managed to equalize, but lost the game later on: 7...Qf5 8.c3 d5 9.dxe5 a6 10.Ba4 Be7 11.f4 Qg6 12.Be3 Bf5 13.Qf2 0-0 14.Nd2 Rad8 15.Nf3 Bd3 16.Rfe1 Na5 17.Nd4 Nc4 18.f5 Qg4 19.Bd1 Qh4 20.g3 Qh3 21.f6 gxf6 22.exf6 Bd6 23.Bf3 Be4 24.Bxe4 dxe4 25.Bf4 Rfe8 26.Bxd6 Nxd6 27.Rad1 Kh8 28.Nc2 Rg8 29.c4 Rg6 30.Ne3 Rg5 31.b4 Rdg8 32.c5 Nb5 33.Rd5 R5g6 34.Qd2 Qe6 35.Rd7 b6 36.Rf1 Rf8 37.Re7 Qh3 38.cxb6 cxb6 39.Nf5 Na3 40.Qd7 Nc4 41.Rxe4 Rg4 42.Rd1 h6 43.Rxg4 Qxg4 44.Rd4 Qg5 45.Rxc4 Kh7 46.Qe7 1-0 Rybka - Deep Sjeng, Computer Olympiad 2009, Pamplona] 8.dxe5 Nd4 [The pawn can't be taken yet: 8...Qxe5 9.Qxe5+ Nxe5 10.Re1 f6 11.f4] 9.Qd3 Qxe5 [9...Nxb5 10.Qxb5 c6 looks like a better choice after the game. Taking on e5 is risky.] 10.Nc3 Rybka's last book move. 10...Bc5 11.Qd1 Ne6 12.Re1 Qd4 13.Qf3 0-0 14.Re4!
White suddenly starts to go for black's king. At the moment it looks like white has not enough pieces on the kingside for a successful attack, but we will see. 14...Qd6 15.Rh4 Qe5?! Look how badly the queen is placed. White can bring all the pieces to the attack with tempo! [15...h6 looks better, since e.g. 16.Bxh6 gxh6 17.Rg4+ Ng5 18.h4 f5 19.Rg3 f4 20.Rg4 Qe5 21.hxg5 d5 is good for Black.] 16.Bd2 f5 17.Re1 Qf6 Again Black has to move the queen. 18.Qh3 Qg6
19.Nd5! Shredder didn't see the following continuation. However, White's attack looks very strong now. The knight is happy to enter the attack. 19...c6 20.Rxe6 Qxe6 21.Nf4 Qxa2 22.Rxh7 cxb5. Removing the defender and threatening mate. 22...Qa1+ 23.Bf1 Of course there's no back rank mate yet.
23.g3! Now the white king is safe. Black is a rook + pawn + bishop pair up, but in chess the king is more important than any other piece! 23...Rf6 24.Bc3. Every piece is attacking, Black has just the rook as defender on the kingside. 24...Kf7 25.Qh4 Qa1+ 26.Kg2 Qa6 27.Bxf6 Qxf6 28.Qh5+. Black resigned in view of 28.Qh5+ Ke7 29.Nd5+ winning the queen. A great symbol for a game that shows the drawback of a queen being in the middle of the board! 1-0. [Click to replay]