Houdini 3

[bhlangonijr]

Hello Robert.

Just out of curiosity: Do you own the domain www.houdinichess.com?
It redirects to Houdini's old website with only the free version (1.5).

[Houdini]

Hi Robert,

Don't forget to dedicate some resources for testing at long time controls too as some people here are implying Houdini 1.5 is stronger than version 2 at the regular OTB time control (40 moves in 120 minutes).

Best regards

Mel, some of the most vocal of the "some people here" have never actually tested Houdini 2 . If you have any doubts about Houdini 2 at long TC, check the neighboring thread, or the very interesting matches of various engines against Deep Junior 13. Houdini 2 scores about 10% better than its closest competitors.

As for Houdini 3, it will do quite well at long TC - some search algorithm improvements will have an increasing effect with higher search depths.

Robert

[FriedmannC]

Great news, Robert! Do you think HOUDINI 3 at its maximum hardware could defeat the RYBKA CLUSTER in a match consisting in 24 games played at classical time controls ( 2h30'/40 moves+ 1h15'/20 moves + 1h for the rest of the game with 15'' increment starting from the 1st move)? The book would be the same for both HOUDINI 3 and the RYBKA CLUSTER up to 10th move.
Best regards!

[Houdini]

Hello Robert.

Just out of curiosity: Do you own the domain www.houdinichess.com?
It redirects to Houdini's old website with only the free version (1.5).

Ben-Hur, yes I own that website too, it's currently just a backup web site with no real use.

Robert

[Houdini]

Great news, Robert! Do you think HOUDINI 3 at its maximum hardware could defeat the RYBKA CLUSTER in a match consisting in 24 games played at classical times controls ( 2h30'/40 moves+ 1h15'/20 moves + 1h for the rest of the game)? The book would be the same for both HOUDINI 3 and the RYBKA CLUSTER up to 10th move.
Best regards!

Teodoriu, I don't speculate about chess engines, especially not about Houdini's direct competitors.

Robert

[Nelson Hernandez]

You describe match conditions, as would be the case in a bona fide world championship match. In that case it isn't really a battle of engines as much as it is a battle of programmers, and in that event there ought not be any limitation on opening books. That's a major part of the battle of wits.

A problem is the asymmetry in hardware, though. Last I heard the Cluster was 296 cores, and last I heard Houdini topped out at 32 cores and has no clustering software. We don't know how strong the Rybka edition running on the Cluster is, but it is surely stronger than Rybka 4.1. So the bottom line is that Rybka would enter the contest with an edition that might be roughly equal to Houdini 3 (we don't know) but had nearly 10x the hardware, scaling not 1:1 but at some respectable level. If I were a betting man I'd go with the Cluster, and especially if opening books were not prescribed by the rules, as the Rybka team has a lot of resources there.

[Jimmy Huggins]

As for Houdini 3, it will do quite well at long TC - some search algorithm improvements will have an increasing effect with higher search depths.

Any chance you can expand on this thought with us?

[FriedmannC]

Thank you for your reply and I apologise if this topic is too science-fiction. Best regards!

[Houdini]

As for Houdini 3, it will do quite well at long TC - some search algorithm improvements will have an increasing effect with higher search depths.

Any chance you can expand on this thought with us?

Jimmy, not really, that's way beyond the scope of this simple "progress report".

Robert

[Nelson Hernandez]

I have reason to think it is far from science fiction. But it is true that the right conditions and incentives are not in place for such a contest to take place. For now all you need is money and an energetic match impresario; after that it is a matter of selling the concept of a well-publicized match as an adjunct to an existing human tournament, negotiating a deal between the principals, assembling a team to facilitate and promote the match, and seeing things through to completion. Unfortunately the first two conditions are not in place.