I bet...

[Adam Hair]

Of the top engines, only Komodo, Hiarcs, Protector, and Stockfish are able to apply checkmate before the 50 move limit, when using 5 sec + 0.1 sec time control.

I have tried 2 seconds (total thinking time), but Gaviota beta and the above engines all forfeit on time. That may be due to GUI lag.

[Thomas Mayer]

Thomas,

I just tried Quark 2.35. It allowed the white king to go to h1 and did nothing to force him out.

yep, there were some versions where it was broken...

but I will try myself... funny, I even have no chess program at all installed on my new laptop...

Greets, Thomas

[Adam Hair]

Thomas,

I just tried Quark 2.35. It allowed the white king to go to h1 and did nothing to force him out.

yep, there were some versions where it was broken...

but I will try myself... funny, I even have no chess program at all installed on my new laptop...

Greets, Thomas

It sounds like you are trying to live by the motto found in your avatar picture

[Adam Hair]

Of the top engines, only Komodo, Hiarcs, Protector, and Stockfish are able to apply checkmate before the 50 move limit, when using 5 sec + 0.1 sec time control.

I have tried 2 seconds (total thinking time), but Gaviota beta and the above engines all forfeit on time. That may be due to GUI lag.

I turned off "animate" in Winboard, and now I am able to confirm that the engines mentioned above can checkmate with only 2 seconds of total thinking time. Though, Hiarcs 12 and Stockfish 2.2.2 64-bit also drew once a piece.

[Rebel]

I don't have TB's installed but self-ply at depth=6 does the trick for mine.

[Robert Flesher]

You engine cannot mate with 5 seconds on the clock (probably even with 0.1 s increment), as black, w/o using tablebases, against an engine with TBs.

[D]n3kb2/8/8/8/4K3/8/8/8 w - - 0 1

Miguel

This amuses me as Hiarcs 6.0 could do this without any troubles. These days the elo monsters are having troubles? I guess this shows the trend of engines improving in the area of search and removing knowledge. It would be interesting to ask Mark Uniake how he taught/coded this mate into Hiarcs 6.0. Did he program in the W pattern? Or was it just based on a few specific rules. If I remember correctly I believe Hiarcs 6.0 executed the mate with the w-pattern. The same pattern that any human would use after studying this mate.

[geots]

You engine cannot mate with 5 seconds on the clock (probably even with 0.1 s increment), as black, w/o using tablebases, against an engine with TBs.

[D]n3kb2/8/8/8/4K3/8/8/8 w - - 0 1

Miguel

Miguel, as good a time as any. Is there any way you could see that there is a separate download for the gaviota bases that ONLY include 3 and 4 man. As a matter of principle, I would never set aside double figure MBs that take up hard drive space. The rewards to me are not worth the space I am giving them.

Best,

George

[Trahald]

I find this to be a very interesting topic, and it could be fun to test Gaviota with other pawnless endgames such as KQ v KR or KRB v KR as in the Carlsen-Svidler video.

[D]3r4/8/8/8/8/2B2k2/R1K5/8 b - - 0 1

Carlsen has just played Rxa2 to reach the diagram and Svidler was unable to hold. Of course it was only a blitz game, but I was surprised the first time I saw this video because I would have assumed the drawing method to be common knowledge, and rather trivial to a player of Svidler's quality. It is also remarkable to watch Carlsen's handling of this, but that is almost an everyday occurrence!

I wonder if any engine (without TB of course) could hold this against an opponent with TB's with a minute or two each on the clock.

While on the subject of tablebases I wonder if Miguel could give a word about how his GTB is progressing?I look forward to the day when 6 man and perhaps even 7 man? GTB's could become a reality.

Finally for those who are interested in the KBN v K mate, I wonder how many of you are familiar with this alternative method of delivering the mate. It seems there are many chessplayers who find this easier than the 'W' maneuver, perhaps it could be effective for engines as well?

http://www.youtube.com/watch?v=PWZ7h2yrJME

[jdart]

Arasan has had special code for this for a very long time.

But it is mate in 31 from your start position, so of course it won't see the mate.

--Jon

[lucasart]

You engine cannot mate with 5 seconds on the clock (probably even with 0.1 s increment), as black, w/o using tablebases, against an engine with TBs.

[D]n3kb2/8/8/8/4K3/8/8/8 w - - 0 1

Miguel

No need for TB here. It's very easy: just have a piece on square table for the lone king to incentivise to push it towards any corner of the color of the bishop. The search will do the rest and see the forceful mate once it's close enough.

try stockfish for example. i bet you anything it mates w/o problems here, even with 5"+0.1"