World Computer Chess Championship 2015

Post by **Harvey Williamson** » Thu Jul 02, 2015 1:55 pm

Hiarcs has the best of it v Komodo

[Event "WCCC 2015"]
[Site "Leiden, The Netherlands"]
[Date "2015.07.02"]
[Round "7.3"]
[White "Hiarcs"]
[Black "Komodo"]
[Result "1/2-1/2"]
[ECO "D16"]
[LiveChessVersion "1.4.6"]

1.d4 { [%clk 1:45:28] } 1...c6 { [%clk 1:45:27] } 2.c4 { [%clk 1:45:38] } 2...Nf6 { [%clk 1:45:39] } 3.Nc3 { [%clk 1:45:48] } 3...d5 { [%clk 1:45:52] } 4.Nf3 { [%clk 1:45:58] } 4...dxc4 { [%clk 1:46:04] } 5.a4 { [%clk 1:46:11] } 5...e6 { [%clk 1:46:15] } 6.e4 { [%clk 1:46:21] } 6...Bb4 { [%clk 1:46:26] } 7.e5 { [%clk 1:46:32] } 7...Nd5 { [%clk 1:46:37] } 8.Bd2 { [%clk 1:46:43] } 8...b5 { [%clk 1:46:49] } 9.axb5 { [%clk 1:46:53] } 9...Bxc3 { [%clk 1:47:01] } 10.bxc3 { [%clk 1:47:03] } 10...cxb5 { [%clk 1:46:16] } 11.Ng5 { [%clk 1:47:13] } 11...Nc6 { [%clk 1:46:27] } 12.Qh5 { [%clk 1:47:21] } 12...Qe7 { [%clk 1:46:39] } 13.h4 { [%clk 1:47:29] } 13...Bd7 { [%clk 1:46:50] } 14.Rh3 { [%clk 1:40:00] } 14...b4 { [%clk 1:46:02] } 15.Ne4 { [%clk 1:33:01] } 15...O-O { [%clk 1:45:12] } 16.Bg5 { [%clk 1:29:08] } 16...f6 { [%clk 1:44:20] } 17.exf6 { [%clk 1:26:09] } 17...gxf6 { [%clk 1:44:17] } 18.Bh6 { [%clk 1:23:19] } 18...Kh8 { [%clk 1:40:17] } 19.Bxf8 { [%clk 1:23:26] } 19...Qxf8 { [%clk 1:36:12] } 20.Bxc4 { [%clk 1:23:36] } 20...bxc3 { [%clk 1:34:02] } 21.Nc5 { [%clk 1:19:41] } 21...c2 { [%clk 1:32:02] } 22.Bxd5 { [%clk 1:17:15] } 22...exd5 { [%clk 1:29:40] } 23.Re3 { [%clk 1:17:25] } 23...Nxd4 { [%clk 1:25:59] } 24.Qxd5 { [%clk 1:10:41] } 24...Nf5 { [%clk 1:24:29] } 25.Re4 { [%clk 0:53:19] } 25...Rd8 { [%clk 1:24:40] } 26.Nxd7 { [%clk 0:51:04] } 26...Qa3 { [%clk 1:24:49] } 27.Rxa3 { [%clk 0:49:51] } 27...c1=Q+ { [%clk 1:22:08] } 28.Ke2 { [%clk 0:50:02] } 28...Qxa3 { [%clk 1:22:16] } 29.Qxf5 { [%clk 0:48:05] } 29...Qa6+ { [%clk 1:21:40] } 30.Kf3 { [%clk 0:46:47] } 30...Qd3+ { [%clk 1:20:02] } 31.Kf4 { [%clk 0:46:58] } 31...Qd6+ { [%clk 1:15:52] } 32.Ne5 { [%clk 0:47:05] } 32...a5 { [%clk 1:14:11] } 33.Kf3 { [%clk 0:43:31] } 33...fxe5 { [%clk 1:11:15] } 34.Qxe5+ { [%clk 0:41:37] } 34...Qxe5 35.Rxe5 { [%clk 0:41:38] } 35...Rf8+ { [%clk 1:08:57] } 36.Ke2 { [%clk 0:39:32] } 36...Ra8 { [%clk 1:03:24] } 37.Re3 { [%clk 0:39:41] } 37...a4 { [%clk 1:00:26] } 38.Ra3 { [%clk 0:38:13] } 38...Kg7 { [%clk 0:56:39] } 39.g4 { [%clk 0:36:12] } 39...Kf6 { [%clk 0:55:09] } 40.Kd3 { [%clk 0:36:21] } 40...Ke5 { [%clk 0:53:57] } 41.Ke3 { [%clk 0:36:29] } 41...h5 { [%clk 0:49:00] } 42.gxh5 { [%clk 0:36:38] } 42...Kf6 { [%clk 0:45:50] } 43.Kf4 { [%clk 0:36:49] } 43...Kg7 { [%clk 0:43:51] } 44.Kg5 { [%clk 0:35:11] } 44...Ra5+ { [%clk 0:43:26] } 45.Kg4 { [%clk 0:34:24] } 45...Ra8 { [%clk 0:42:27] } 46.f4 { [%clk 0:32:53] } 46...Kh6 { [%clk 0:41:30] } 47.f5 { [%clk 0:32:11] } 47...Rg8+ { [%clk 0:38:55] } 48.Kh3 { [%clk 0:32:18] } 48...Ra8 { [%clk 0:38:10] } 49.f6 { [%clk 0:32:27] } 49...Kxh5 { [%clk 0:37:27] } 50.f7 { [%clk 0:32:37] } 50...Kg6 { [%clk 0:36:42] } 51.Rxa4 { [%clk 0:32:46] } 1/2-1/2

Carl Bicknell · Post by **Carl Bicknell** » Thu Jul 02, 2015 6:38 pm

The Komodo team have really let themselves down with their openings.

Great engine, terrible opening book.

RadioSmall · Post by **RadioSmall** » Thu Jul 02, 2015 7:08 pm

Carl Bicknell wrote:The Komodo team have really let themselves down with their openings.

Great engine, terrible opening book.

Your are completely wrong on this issue ....I emphasize it has nothing to do with opening book.........All this shows it the huge hyperbole that surrounds programs like Stockfish and Komodo ..... In the very ignorant talk chess forum people blindly assume everything must lose to Komodo and SF ... Hopefully this REAL WorldChampionship will be a Huge wakeup call for them........

Carl Bicknell · Post by **Carl Bicknell** » Thu Jul 02, 2015 7:35 pm

RadioSmall wrote: Your are completely wrong on this issue ....I emphasize it has nothing to do with opening book.........All this shows it the huge hyperbole that surrounds programs like Stockfish and Komodo ..... In the very ignorant talk chess forum people blindly assume everything must lose to Komodo and SF ... Hopefully this REAL WorldChampionship will be a Huge wakeup call for them........

NO ONE is claiming "everything must lose to SF or Komodo". But we know its elo, and therefore we know its one of the favourites if all other factors below are equal.

There are a variety of things which affect the outcome of any tournament:
a) The number of games played (the lower the more variance)
b) The strength of the engine
c) the speed of the hardware
d) the opening book

To say the opening book has NOTHING to do with it is crazy - the HIARCS team THEMSELVES sell opening books precisely because they DO add elo and affect the outcome of games!!

IanO · Post by **IanO** » Thu Jul 02, 2015 7:39 pm

RadioSmall wrote:
Carl Bicknell wrote:The Komodo team have really let themselves down with their openings.

Great engine, terrible opening book.
Your are completely wrong on this issue ....I emphasize it has nothing to do with opening book.........All this shows it the huge hyperbole that surrounds programs like Stockfish and Komodo ..... In the very ignorant talk chess forum people blindly assume everything must lose to Komodo and SF ... Hopefully this REAL WorldChampionship will be a Huge wakeup call for them........

Agreed. The critical game was decided by an endgame error. We won't know the full situation until we see the PGN files annotated with engine evaluations and search depths. Say what you will about Rybka, I miss Rajlich's opening prep and commentary. This event was once used to advance opening theory.

Massive clusters should not be underestimated! Remember that in 2013 Jonny on the same cluster was getting 2Bnps and reaching middle-game depths of 40 ply!

Carl Bicknell · Post by **Carl Bicknell** » Thu Jul 02, 2015 7:46 pm

IanO wrote:
RadioSmall wrote:
Carl Bicknell wrote:The Komodo team have really let themselves down with their openings.

Great engine, terrible opening book.
Your are completely wrong on this issue ....I emphasize it has nothing to do with opening book.........All this shows it the huge hyperbole that surrounds programs like Stockfish and Komodo ..... In the very ignorant talk chess forum people blindly assume everything must lose to Komodo and SF ... Hopefully this REAL WorldChampionship will be a Huge wakeup call for them........
Agreed. The critical game was decided by an endgame error. We won't know the full situation until we see the PGN files annotated with engine evaluations and search depths. Say what you will about Rybka, I miss Rajlich's opening prep and commentary. This event was once used to advance opening theory.

Massive clusters should not be underestimated! Remember that in 2013 Jonny on the same cluster was getting 2Bnps and reaching middle-game depths of 40 ply!

I think there's some confusion here.

Komodo lost to Jonny because of an endgame mistake - granted. But that doesn't necessarily mean Komodo had a good book.

HIARCS came out of the opening better than Komodo. That's not a comment on Komodo the engine or, for that matter, HIARCS the engine.
All I'm simply saying is the Komodo book isn't as strong as one might expect from an engine which (in some cases) is hundreds of elo above its rivals.

Peter Grayson · Post by **Peter Grayson** » Thu Jul 02, 2015 9:09 pm

All I'm simply saying is the Komodo book isn't as strong as one might expect from an engine which (in some cases) is hundreds of elo above its rivals.

Wasn't the consensus of opinion that Komodo should try and get out of established book lines as soon as possible to enable its superior knowledge to come into effect? I find it strange then that there should be complaints directed at the Komodo book and its author when it appears that is what was attempted.

With respect to "Komodo's 100's of Elo above its rivals" that is derived from relatively short line openings and is not truly reflective of performance based on the lines that could be expected in top competition when they are more likely to be based on established openings knowledge especially if the choice is directed by the engine author and his team who better understand the engine's strengths and weaknesses? The probability is that Komodo's relative performance differential is significantly less in the latter circumstance when engine opening knowledge may have little impact on the outcome. Perhaps if Elo was based on engine matches using longer contemporary opening lines the expectancy would not have been so high.

Seems to me excuses now forthcoming for Komodo's failure to wipe the floor with the opposition.

Peter

Carl Bicknell · Post by **Carl Bicknell** » Thu Jul 02, 2015 9:43 pm

Seems to me someone can't quote accurately.

ricard60 · Post by **ricard60** » Thu Jul 02, 2015 11:37 pm

Jony has 2400 cores, i wonder what would be its performance with 28 cores?

Hardware and not software regards
Ricardo

RadioSmall · Post by **RadioSmall** » Fri Jul 03, 2015 1:53 am

Carl Bicknell wrote:
RadioSmall wrote: Your are completely wrong on this issue ....I emphasize it has nothing to do with opening book.........All this shows it the huge hyperbole that surrounds programs like Stockfish and Komodo ..... In the very ignorant talk chess forum people blindly assume everything must lose to Komodo and SF ... Hopefully this REAL WorldChampionship will be a Huge wakeup call for them........
NO ONE is claiming "everything must lose to SF or Komodo". But we know its elo, and therefore we know its one of the favourites if all other factors below are equal.

There are a variety of things which affect the outcome of any tournament:
a) The number of games played (the lower the more variance)
b) The strength of the engine
c) the speed of the hardware
d) the opening book

To say the opening book has NOTHING to do with it is crazy - the HIARCS team THEMSELVES sell opening books precisely because they DO add elo and affect the outcome of games!!

You misunderstand what computer chess is all about ...Nobody said that a very good opening book does not have any effect on Elo ....But most of the strength increases in recent years (or decades I should say now) is due to software improvement ...In particular the game Komodo lost to Jonny is not because of an opening book which is not optimal ...We see this overemphasis on opening books in all your posts.......

Carl Bicknell · Post by **Carl Bicknell** » Fri Jul 03, 2015 7:31 am

RadioSmall wrote: You misunderstand what computer chess is all about ...Nobody said that a very good opening book does not have any effect on Elo

Actually you did. Above you said "I emphasize it has nothing to do with opening book"

RadioSmall wrote: But most of the strength increases in recent years (or decades I should say now) is due to software improvement ...In particular the game Komodo lost to Jonny is not because of an opening book which is not optimal

I'm not disagreeing with this. But the opening book is A factor - it didn't cause the loss against Jonny, but it didn't get it into a good position either.

RadioSmall wrote: We see this overemphasis on opening books in all your posts.......

eh? I've only made a few posts on the subject of opening books and most of those were in response to you.
It's not an over-emphasis, it's just to say the opening book does count for SOMETHING. And, in my opinion, Komodo's could have been better.

Post by **Mark Uniacke** » Fri Jul 03, 2015 6:26 pm

Congratulations to Johannes Zwanzger on winning the World Computer Chess Championship with his program Jonny.

Jonny has been steadily improving at each WCCC and proved superior in Leiden this year.

Post by **Fernando** » Fri Jul 03, 2015 6:32 pm

Mark Uniacke wrote:Congratulations to Johannes Zwanzger on winning the World Computer Chess Championship with his program Jonny.

Jonny has been steadily improving at each WCCC and proved superior in Leiden this year.

So it is just a matter to purchase 2400 cores and you enjoy it...

Fern

mclane · Post by **mclane** » Fri Jul 03, 2015 9:35 pm

I have heard there is a uniform tournament too. There the software has not 2400 cores. How is this done ?

ricard60 · Post by **ricard60** » Sat Jul 04, 2015 1:04 am

On the Architecture it is written that Jonny runs on a AMD x86-64, are these virtual cores?

https://icga.leidenuniv.nl/?page_id=1260