Thought you wanted to see this.

Topalov training with super computer Blue Gene P

Topalov confirmed that he has been using the Blue Gene /P supercomputer, with 8792 processors and working on 1 PFLOPS (petaFLOPS). Blue Gene is a computer architecture project designed to reach operating speeds in the PFLOPS (petaFLOPS) range, and currently reaching sustained speeds of nearly 500 TFLOPS (teraFLOPS). It is at least seven times more energy efficient than any other supercomputer, accomplished by using many small, low-power chips connected through five specialized networks. Blue Gene/P can be scaled to an 884,736-processor, 216-rack cluster to achieve 3-PFLOPS performance. A standard Blue Gene/P configuration will house 4,096 processors per rack.

Blue Gene/P has been installed on September 9, 2008 in Sofia, the capital of Bulgaria, and is operated by the Bulgarian Academy of Sciences and the Sofia University. Topalov has used exactly this modification of Blue Gene /P during his match with Viswanathan Anand for the World Chess Championship. For local media Topalov comments, "During my preparation for the match, with Silvio Danailov we have decided to look for additional resources for my preparation against Anand. We decided to use the processing power of Blue Gene /P with 8192 processors."

The only problem for Topalov was that there is no software that runs on such hardware. For Dnevnik Topalov revealed that he has collected a team of computer professionals to modify a program to work on the Blue Gene. Topalov will use the computer for his future preparation for the Candidate matches 2010 / 2011. Until then it will help young local chess players, who will be playing games and analysing with the supercomputer. On June 10th Topalov will join them to help in a series of 10 blitz games against the machine.

(Chessdom)

• Oh, we are narrowing down. Isn't it so that different correspondence chess organizations take different views on cheating, so that, generally, European organizations have a more liberal view on computer assistance than American?

• We all know that top players, starting foremost with Kasparov, use computer power to boost their performance. We also know that Fischer, crazy as he may have been in other ways, had a clear view on present day chess. He belived strongly that most top players only reproduced lines memorized, lines generated partly by computer aid.

• Kasparov also started the tournament in Leon, Spain, in Advanced Chess, won most times by Anand.

• I mean, it's hard to say what is truly a human move nowadays. So keeping chess "pure" is becoming tougher and maybe impossible. Fischer random is one way and otb of course (with the given exceptions above).

BalticKnight: Of course I'm well aware of the 2010 WCC between Anand and Topalov, but you mentioned "computer generated analysis at elite level" of their games. I couldn't find anything under the link you posted...

But a top 3 match analysis of their games clearly showed, that with a high probability none of them has cheated (which by the way no one would have ever supposed). http://www.chess.com/download/view/wcc-2010-anand-topalov-analysed-with-chessanalyse

When I said "cheating detection", I meant primarily chess correspondence (cc) games and not over the board (otb) games.

Wow! Didn't you follow the WC this year? For a person that has an interest in computers and chess it strikes me as somewhat amazing.

http://www.chessbase.com/newsdetail.asp?newsid=6340

Would you say, by using Rybka in this way, that Topalov was cheating?

BalticKnight: Where did you see "a lot of, ..., computer generated analysis at elite level. Topalov-Anand" ?? 

Well, in order to make sure of that there must be a way to seperate a human move from a move purely computer generated - is there? Or is it just an interference based on largly enough collection of moves that has a high degree of parallelity? We see a lot of, I guess, computer generated analysis at elite level. Topalov-Anand is the latest I can think of - would you say that Topalov was cheating?

BalticKnight: The top 3 match analysis method has been primarily used for cheating detection, i. e., detect moves calculated by engines (computer moves) and not by human beings.

What is a computer move? The simple answer of a move done by a computer seems a very lose definition. People who are into advanced chess might have some ideas about that.

By the way: Finding "good moves" is not the aim of top 3 match analysis. However, finding computer moves is pretty much satisfactory.

I know there is this kind of testing going on but I'm not into it. An author that springs to mind is Livshitz, like in Test Your Endgame Ability (with Speelman) or Test Your Chess IQ, GM Challenge. John Nunn also has written books that would fit. Any real well commented position would do I guess.

BalticKnight: Well, that's an idea. Let's fight the engines against each other not from the starting position but from 20 well chosen positions from GM practice. Each engine should play both black and white in each position. Do you have any such positions in mind?

Ok, so R4 is better in a small match than H. It still does not say anything about good moves. Feed like 20 positions to the both programs from GM practice, choose some critical positions where there is consenus about the merits of the "good" moves. Then see what happens.