Rybka is lonely on top of the FRC Rating followed bt Hiarcs

[pichy]

http://www.computerchess.org.uk/ccrl/404FRC/

[Spock]

Indeed. Actually I think it under-performed, it should have done better. But I'm certainly not complaining

I have now started to upgrade the list to 64-bit. Currently running is Glaurung 2 e/5 x64 (I hope it plays OK, I don't think Tord tested it under FRC as I recall) and I'll then run Naum 2.2 x64, followed by Rybka 2.3.2 x64. Then in future, 64-bit versions will automatically be played where they are available instead of 32-bit.

[Uri Blass]

I wonder if it is really rybka2.3.2 because based on my knowledge vasik implemented knowledge in rybka recently and not only implemented the rules of Fischer random chess.

Does it show the same analysis as rybka2.3.2a in normal chess positions?

Uri

[Spock]

It is Rybka 2.3.2 series but not 2.3.2a. It's a special exe that only plays FRC and not normal chess.

I'm going to cancel testing of Glaurung e/5 x64, I'm getting too many time losees.

[Uri Blass]

I do not understand how a version can play only FRC because normal chess is a private case of FRC

If that rybka2.3.2 can play FRC then she can also play normal chess in 1/960 of the games.

Uri

[Spock]

I do not understand how a version can play only FRC because normal chess is a private case of FRC

If that rybka2.3.2 can play FRC then she can also play normal chess in 1/960 of the games.

Uri

I'm only repeating what Vas told me. You'd have to ask him for clarification

[Spock]

Well the latest news here is that Volker Annuss has discovered an illegal move by Rybka in one of it's games, and it went on to win There were also some time losses by Rybka, so clearly all is not well with this version, it has some bugs. I'm considering what action to take, which could include removing it from the list completely becuase I don't know how many more there might be. Certainly I need to make one change. Although we do test private versions as we did here, the policy is that they cannot appear on the "pure list". Rybka will still appear at the top of the "best versions" list but will be excluded from the pure list database. The publicly available Hiarcs 11.1 will be at the top of that list. And in this case that also preserves the integrity of that database. If anyone knows of an automated way of searching through Rybka's 1000 games for illegal moves, I'd be delighted to hear it.

[Uri Blass]

Well the latest news here is that Volker Annuss has discovered an illegal move by Rybka in one of it's games, and it went on to win There were also some time losses by Rybka, so clearly all is not well with this version, it has some bugs. I'm considering what action to take, which could include removing it from the list completely becuase I don't know how many more there might be. Certainly I need to make one change. Although we do test private versions as we did here, the policy is that they cannot appear on the "pure list". Rybka will still appear at the top of the "best versions" list but will be excluded from the pure list database. The publicly available Hiarcs 11.1 will be at the top of that list. And in this case that also preserves the integrity of that database. If anyone knows of an automated way of searching through Rybka's 1000 games for illegal moves, I'd be delighted to hear it.

unfortunately movei's tool to detect illegal move does not work for FRC games.

It translates every castle to e1g1 or e1c1 or e8g8 or e8c8 and it is clear that it is not correct in FRC.

fixing it by using the square of the king does not help because f1g1 can be king move and not castling move.

Uri

[Spock]

Good news - Volker has explained here how to use Hermann to check for illegal moves
http://kirr.homeunix.org/chess/discussi ... 556#p23556

And when I ran it, there was only one game. So I feel more comfortable now with Rybka remaining on the list. Although as explained before, as a private engine it will not appear on the pure list in future, only on the best versions list and the complete list.

[Uri Blass]

Good news - Volker has explained here how to use Hermann to check for illegal moves
http://kirr.homeunix.org/chess/discussi ... 556#p23556

And when I ran it, there was only one game. So I feel more comfortable now with Rybka remaining on the list. Although as explained before, as a private engine it will not appear on the pure list in future, only on the best versions list and the complete list.

After correcting movei I can confirm that movei found also only one illegal move against Naum.

Uri