Rybka 4 vs Houdini 1.02 - Long TC Match - LIVE BROADCAST

[Albert Silver]

I think the end games are played very stupidly. Rybka had an easy win, with rook exchanges, but chose a longer path.

When did Rybka have an easy win? I'm not saying it couldn't have won the endgame, as there were several interesting moments, but I find it far from easy.

[Martin Thoresen]

The match is finished, Rybka wins 26.5 to 21.5 for Houdini.

I want to thank all of you who were following the match, hope you got some interesting games in your collections. The website had 12.535 pageloads during this match, with visitors from 64 countries.

The full PGN can be downloaded from the bottom of the website.

The next match is Rybka 4 vs Stockfish 1.8 which will start this weekend. I will use the same 24 openings, so it will also be 48 games.

See you all then!

Best Regards,
Martin

[Albert Silver]

The match is finished, Rybka wins 26.5 to 21.5 for Houdini.

I want to thank all of you who were following the match, hope you got some interesting games in your collections. The website had 12.535 pageloads during this match, with visitors from 64 countries.

Bear in mind that includes many repeats, since to refresh I had to refresh form the browser, and everytime I did that, the counter increased.

The full PGN can be downloaded from the bottom of the website.

Where?

Edit: Thanks for fixing it.

And thanks for the fascinating match!

[Martin Thoresen]

Bear in mind that includes many repeats, since to refresh I had to refresh form the browser, and everytime I did that, the counter increased.

That's how pageloads are counted. Other stats was 3907 unique visitors.

The full PGN can be downloaded from the bottom of the website.


Where?

There there, I need at least a few minutes to use Notepad and add the link.

It's up now.

[pacifist]

Thanks for this great match! Maybe with endgame tablebases of 6 men (not all but some critical) Rybka could have performed better and of course Houdini with tablebases has the potential to be a good candidate for 1rst place. For now Rybka seems to me the best engine at long games and for analyse purposes. I wait for next match impatiently since Stockfish plays very elegantly and dymanic .

[S.Taylor]

OK then, I got it now. Rybka knew that Houdini, (white), would and could have avoided the rook exchange, therefore it didn't bother choosing that line.

[Martin Thoresen]

Thanks for this great match! Maybe with endgame tablebases of 6 men (not all but some critical) Rybka could have performed better and of course Houdini with tablebases has the potential to be a good candidate for 1rst place. For now Rybka seems to me the best engine at long games and for analyse purposes. I wait for next match impatiently since Stockfish plays very elegantly and dymanic .

Thanks for your interest!

Which 6-men are the most critical?

I am using a 2 Terabyte RAID setup for my tablebases and it seems that they are pretty fast so I hardly doubt 6-men should pose any problems. It might be interesting to download a few of the most critical ones.

I agree about Houdini, it should get some sort of endgame help, for example Gaviota tablebases.

I am also awaiting the next match, I've always had a soft spot for Glaurung/Stockfish.

[S.Taylor]

I think the end games are played very stupidly. Rybka had an easy win, with rook exchanges, but chose a longer path.

When did Rybka have an easy win? I'm not saying it couldn't have won the endgame, as there were several interesting moments, but I find it far from easy.

I explained below that i didn't recheck it enough before posting. I made a mistake. Houdini could and would have avoided exchanging rooks, which it had to keep on at all costs.

[LaurenceChen]

Thanks for this great match! Maybe with endgame tablebases of 6 men (not all but some critical) Rybka could have performed better and of course Houdini with tablebases has the potential to be a good candidate for 1rst place. For now Rybka seems to me the best engine at long games and for analyse purposes. I wait for next match impatiently since Stockfish plays very elegantly and dymanic .

Thanks for your interest!

Which 6-men are the most critical?

I am using a 2 Terabyte RAID setup for my tablebases and it seems that they are pretty fast so I hardly doubt 6-men should pose any problems. It might be interesting to download a few of the most critical ones.

I agree about Houdini, it should get some sort of endgame help, for example Gaviota tablebases.

I am also awaiting the next match, I've always had a soft spot for Glaurung/Stockfish.

You can try these to start:

kbppkb.0.nbb.emd
kbppkb.0.nbw.emd
kbppkb.1.nbb.emd
kbppkb.1.nbw.emd
kbppkb.2.nbb.emd
kbppkb.2.nbw.emd
kbppkb.3.nbb.emd
kbppkn.0.nbb.emd
kbppkn.0.nbw.emd
kbppkn.1.nbb.emd
kbppkn.1.nbw.emd
kbppkn.2.nbb.emd
kbppkn.2.nbw.emd
kbppkn.3.nbb.emd
kbppkr.0.nbb.emd
kbppkr.0.nbw.emd
kbppkr.1.nbb.emd
kbppkr.1.nbw.emd
kbppkr.2.nbb.emd
kbppkr.2.nbw.emd
kbppkr.3.nbb.emd
kbppkr.3.nbw.emd
kbppkr.4.nbb.emd
kbppkr.4.nbw.emd
kbppkr.5.nbb.emd
kbppkr.5.nbw.emd
kbppkr.6.nbb.emd
knppkn.0.nbb.emd
knppkn.0.nbw.emd
knppkn.1.nbb.emd
knppkn.1.nbw.emd
knppkn.2.nbb.emd
knppkn.2.nbw.emd
knppkn.3.nbb.emd
knppkr.0.nbb.emd
knppkr.0.nbw.emd
knppkr.1.nbb.emd
knppkr.1.nbw.emd
knppkr.2.nbb.emd
knppkr.2.nbw.emd
knppkr.3.nbb.emd
kppkpp.nbw.emd
kpppkb.nbb.emd
kpppkb.nbw.emd
kpppkn.nbb.emd
kpppkn.nbw.emd
kpppkp.nbb.emd
kpppkp.nbw.emd
kpppkr.nbb.emd
kpppkr.nbw.emd
kqppkq.0.nbb.emd
kqppkq.0.nbw.emd
kqppkq.1.nbb.emd
kqppkq.1.nbw.emd
kqppkq.2.nbb.emd
kqppkq.2.nbw.emd
kqppkq.3.nbb.emd
kqppkq.3.nbw.emd
kqppkq.4.nbb.emd
kqppkq.4.nbw.emd
kqppkq.5.nbb.emd
kqppkq.5.nbw.emd
kqppkq.6.nbb.emd
krpknp.0.nbb.emd
krpknp.0.nbw.emd
krpknp.1.nbb.emd
krpknp.1.nbw.emd
krpknp.2.nbb.emd
krpknp.2.nbw.emd
krpknp.3.nbb.emd
krpknp.3.nbw.emd
krpknp.4.nbb.emd
krpknp.4.nbw.emd
krpknp.5.nbb.emd
krpknp.5.nbw.emd
krpknp.6.nbb.emd
krpknp.6.nbw.emd
krpknp.7.nbb.emd
krpknp.7.nbw.emd
krpknp.8.nbb.emd
krpknp.8.nbw.emd
krpknp.9.nbb.emd
krpknp.9.nbw.emd
krpknp.a.nbb.emd
krpknp.a.nbw.emd
krpknp.b.nbb.emd
krpknp.b.nbw.emd
krpknp.c.nbb.emd
krpknp.c.nbw.emd
krpkrp.0.nbw.emd
krpkrp.1.nbw.emd
krpkrp.2.nbw.emd
krpkrp.3.nbw.emd
krpkrp.4.nbw.emd
krpkrp.5.nbw.emd
krpkrp.6.nbw.emd
krppkr.0.nbb.emd
krppkr.0.nbw.emd
krppkr.1.nbb.emd
krppkr.1.nbw.emd
krppkr.2.nbb.emd
krppkr.2.nbw.emd
krppkr.3.nbb.emd
kbpkbp.0.nbw.emd
kbpkbp.1.nbw.emd
kbpkbp.2.nbw.emd
kbpkbp.3.nbw.emd
kbpkbp.4.nbw.emd
kbpkbp.5.nbw.emd
kbpkbp.6.nbw.emd
kbpknp.0.nbb.emd
kbpknp.0.nbw.emd
kbpknp.1.nbb.emd
kbpknp.1.nbw.emd
kbpknp.2.nbb.emd
kbpknp.2.nbw.emd
kbpknp.3.nbb.emd
kbpknp.3.nbw.emd
kbpknp.4.nbb.emd
kbpknp.4.nbw.emd
kbpknp.5.nbb.emd
kbpknp.5.nbw.emd
kbpknp.6.nbb.emd
kbpknp.6.nbw.emd
kbpknp.7.nbb.emd
kbpknp.7.nbw.emd
kbpknp.8.nbb.emd
kbpknp.8.nbw.emd
kbpknp.9.nbb.emd
kbpknp.9.nbw.emd
kbpknp.a.nbb.emd
kbpknp.a.nbw.emd
kbpknp.b.nbb.emd
kbpknp.b.nbw.emd
kbpknp.c.nbb.emd
kbpknp.c.nbw.emd
kbpkpp.0.nbb.emd
kbpkpp.0.nbw.emd
kbpkpp.1.nbb.emd
kbpkpp.1.nbw.emd
kbpkpp.2.nbb.emd
kbpkpp.2.nbw.emd
kbpkpp.3.nbb.emd
kbpkpp.3.nbw.emd
kbpkpp.4.nbb.emd
kbpkpp.4.nbw.emd
knpknp.0.nbw.emd
knpknp.1.nbw.emd
knpknp.2.nbw.emd
knpknp.3.nbw.emd
knpknp.4.nbw.emd
knpknp.5.nbw.emd
knpknp.6.nbw.emd
knpkpp.0.nbb.emd
knpkpp.0.nbw.emd
knpkpp.1.nbb.emd
knpkpp.1.nbw.emd
knpkpp.2.nbb.emd
knpkpp.2.nbw.emd
knpkpp.3.nbb.emd
knpkpp.3.nbw.emd
knpkpp.4.nbb.emd
knpkpp.4.nbw.emd
knppkb.0.nbb.emd
knppkb.0.nbw.emd
knppkb.1.nbb.emd
knppkb.1.nbw.emd
knppkb.2.nbb.emd
knppkb.2.nbw.emd
knppkb.3.nbb.emd
kqpkqp.0.nbw.emd
kqpkqp.1.nbw.emd
kqpkqp.2.nbw.emd
kqpkqp.3.nbw.emd
kqpkqp.4.nbw.emd
kqpkqp.5.nbw.emd
kqpkqp.6.nbw.emd
kqpkqp.7.nbw.emd
kqpkqp.8.nbw.emd
kqpkqp.9.nbw.emd
kqpkqp.a.nbw.emd
kqpkqp.b.nbw.emd
kqpkrp.0.nbb.emd
kqpkrp.0.nbw.emd
kqpkrp.1.nbb.emd
kqpkrp.1.nbw.emd
kqpkrp.2.nbb.emd
kqpkrp.2.nbw.emd
kqpkrp.3.nbb.emd
kqpkrp.3.nbw.emd
kqpkrp.4.nbb.emd
kqpkrp.4.nbw.emd
kqpkrp.5.nbb.emd
kqpkrp.5.nbw.emd
kqpkrp.6.nbb.emd
krbkrp.0.nbb.emd
krbkrp.0.nbw.emd
krbkrp.1.nbb.emd
krbkrp.1.nbw.emd
krbkrp.2.nbb.emd
krbkrp.2.nbw.emd
krbkrp.3.nbb.emd
krbkrp.3.nbw.emd
krbkrp.4.nbb.emd
krbkrp.4.nbw.emd
krbkrp.5.nbb.emd
krbkrp.5.nbw.emd
krbkrp.6.nbb.emd
krbkrp.6.nbw.emd
krbkrp.7.nbb.emd
krbkrp.7.nbw.emd
krbkrp.8.nbw.emd
krpkbp.0.nbb.emd
krpkbp.0.nbw.emd
krpkbp.1.nbb.emd
krpkbp.1.nbw.emd
krpkbp.2.nbb.emd
krpkbp.2.nbw.emd
krpkbp.3.nbb.emd
krpkbp.3.nbw.emd
krpkbp.4.nbb.emd
krpkbp.4.nbw.emd
krpkbp.5.nbb.emd
krpkbp.5.nbw.emd
krpkbp.6.nbb.emd
krpkbp.6.nbw.emd
krpkbp.7.nbb.emd
krpkbp.7.nbw.emd
krpkbp.8.nbb.emd
krpkbp.8.nbw.emd
krpkbp.9.nbb.emd
krpkbp.9.nbw.emd
krpkbp.a.nbb.emd
krpkbp.a.nbw.emd
krpkbp.b.nbb.emd
krpkbp.b.nbw.emd
krpkbp.c.nbb.emd
krpkbp.c.nbw.emd

[M ANSARI]

I am not sure about other games, but most certainly in the last game R4 had an easy win. I don't think it has anything to do with bad rook endgame play, but rather a bug that seems to affect some moves by R4 in the endgame. It tends to stall search and thus a poor move is produced. It is possible this is not due to R4 but due to instability of Houdini 1.02, but of course it could simply be a hash bug by R4.