Komodo 2.03 release is imminent

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
lkaufman
Posts: 3835
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Komodo 2.03 release is imminent

Post by lkaufman » Wed Jun 15, 2011 10:25 pm

Don is correct that the true time loss is less that 20%. However at the speeds you are testing I think that 20% is probably close to 40 elo. We have the same problem as you in our testing, namely the tradeoff between systematic error and sample-size error. In my opinion the way you test is ok if you are comparing two versions of the same engine where nothing has changed that would have a favorable or unfavorable effect on one at hyperspeed. In the present case this means that comparing Komodo 2.03 to 2.01 should be fine this way, but not compared to 1.3 due to the 20 ms lag option. But for comparing unrelated engines it is just way too fast to have much predictive value for real-time play. All of the Ippo-derived engines look ridiculously strong at hyperblitz levels like yours, but other than Houdini they do not excel relative to SF2.1.1 and Komodo 2.03 at time limits like the IPON 5' +3".

Frank Quisinsky
Posts: 4852
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: Komodo 2.03 release is imminent

Post by Frank Quisinsky » Wed Jun 15, 2011 11:01 pm

Hi Don,

Komodo / Doch versions are in SWCR the most passiv engines in the group of stronger engines. More as 10.000 40-minutes games played with your versions and the statistics are allways the same. Easy to see because in SWCR I play without resign mode up to mate.

Your engine made the most points vs. the other one in the early endgame. I think more passiv is better for blitz games. More spekulative better for longer time controls. This is allways the same, not important the engine you can find out it.

In this case it would be great if you can add 1-2 UCI parameters for more daynamic and speculative aggressive chess. This is enough, not to many parameters, thinking on your file you programmed for Doch.

Congratulation for the big jumpings from Komodo 1.2 to 1.3 and now it seems to the very big jumping from 1.3 to 2.01 / 2.03. I will start my SWCR test on Saturday in the evening.

Thanks to Larry too!
Larry, what is the reason for this passiv chess in older Komodo versions from your point of view? Only two engines in SWCR have weaker results in short mate games as Komodo / Doch. The early endgame is fantastic, perhaps Komodo is at the moment the strongest engine we have in this game part.

Best
Frank
I like computer chess!

Dave Acevedo

Re: Komodo 2.03 release is imminent

Post by Dave Acevedo » Wed Jun 15, 2011 11:20 pm

Komodo 2.03 How exciting!

Thank you very much Don and Larry and the Komodo Team.

Since Fruit I have noticed that the engines can not anymore be classified as aggressive or tactical etc., but it finds the move that makes the most sense among those 64 squares. To watch chess engines do this is a great wonder.

lkaufman
Posts: 3835
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Komodo 2.03 release is imminent

Post by lkaufman » Thu Jun 16, 2011 1:21 am

Frank Quisinsky wrote:Hi Don,

Komodo / Doch versions are in SWCR the most passiv engines in the group of stronger engines. More as 10.000 40-minutes games played with your versions and the statistics are allways the same. Easy to see because in SWCR I play without resign mode up to mate.

Your engine made the most points vs. the other one in the early endgame. I think more passiv is better for blitz games. More spekulative better for longer time controls. This is allways the same, not important the engine you can find out it.

In this case it would be great if you can add 1-2 UCI parameters for more daynamic and speculative aggressive chess. This is enough, not to many parameters, thinking on your file you programmed for Doch.

Congratulation for the big jumpings from Komodo 1.2 to 1.3 and now it seems to the very big jumping from 1.3 to 2.01 / 2.03. I will start my SWCR test on Saturday in the evening.

Thanks to Larry too!
Larry, what is the reason for this passiv chess in older Komodo versions from your point of view? Only two engines in SWCR have weaker results in short mate games as Komodo / Doch. The early endgame is fantastic, perhaps Komodo is at the moment the strongest engine we have in this game part.

Best
Frank
I'm not too sure what you mean by "passiv(e)" chess. Are you talking about a subjective opinion about the types of moves it plays, or do you perhaps mean that it is a bit weaker than other top engines at finding surprising middlegame tactics? If you mean the later, I can believe that this is true, because unlike many other engine teams we do not devote any serious effort to maximizing performance on tactical suites, we just try to maximize the rating consistent with keeping realistic evaluations. I am personally more interested in having a program play well positionally than I am in making it outperform others in tactics, since all the programs play way above 3000 level tactically but probably below 2500 positionally if such a thing could be measured.
As for UCI parameters, we already provided one with the new engine for how aggressive you want to make king safety. If you want to make the engine more inclined to gambit a pawn to expose the enemy king, raise that value. In fact almost the last change we made to the program before release was to raise the default from 5 to 6, making it significantly more aggressive/speculative. But we are pretty sure that the (new)default will give the best rating, or at least that no change will help more than a tiny bit.
If I have not answered your question adequately, please explain more clearly what you mean by "passiv(e)" play and what is the basis for the statement.
I look forward to your testing.

Best regards,
Larry

Cubeman
Posts: 644
Joined: Fri Feb 02, 2007 2:11 am
Location: New Zealand

Re: Komodo 2.03 release is imminent

Post by Cubeman » Thu Jun 16, 2011 2:52 am

Don wrote:The Komodo 2.03 release is imminent. Release 2.01 was short lived due to problems.

Version 2.03 is to be viewed as a separate version, not just a bug fix release as it has some improvements, perhaps 5-10 additional ELO.

But it also has a major bug fix that has caused problems in testing. This bug has been in Komodo since the beginning of time ...

Get it here: http://komodochess.com

Komodo 2.03 is significantly stronger than Komodo 1.3.

Various platform specific versions will be posted as they become available and there may be multiple versions for your platform. If you don't see anything, check back frequently. At the time of this post only 2 version are ready, the Android version and the Mac version. But more will soon follow. Some versions may run better on your machine than others so you may wish to use the version that runs fastest on your hardware.

You can always get the latest release versions of Komodo here:

http://komodochess.com

Again, many thanks to Jim Ablett and Dann Corbit and Aart Bilk for helping me debug and compile. Clemens Keck was very helpful also in helping me test.

Feedback is always welcome.

Don and Larry
Various platform specific versions will be posted as they become available and there may be multiple versions for your platform. If you don't see anything, check back frequently. At the time of this post only 2 version are ready, the Android version and the Mac version.
Does that mean there might be other platforms like iOS (iphone/ipad) or even PocketPC, I for one would love to have an additional engine for PocketFritz and will be prepared to pay about $30 US for the download, my last purchase was a Naum 4.2 engine so I hope there will be a stronger engine available soon.

User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 2:27 pm

Re: Komodo 2.03 release is imminent

Post by Don » Thu Jun 16, 2011 3:34 am

Cubeman wrote:
Don wrote:The Komodo 2.03 release is imminent. Release 2.01 was short lived due to problems.

Version 2.03 is to be viewed as a separate version, not just a bug fix release as it has some improvements, perhaps 5-10 additional ELO.

But it also has a major bug fix that has caused problems in testing. This bug has been in Komodo since the beginning of time ...

Get it here: http://komodochess.com

Komodo 2.03 is significantly stronger than Komodo 1.3.

Various platform specific versions will be posted as they become available and there may be multiple versions for your platform. If you don't see anything, check back frequently. At the time of this post only 2 version are ready, the Android version and the Mac version. But more will soon follow. Some versions may run better on your machine than others so you may wish to use the version that runs fastest on your hardware.

You can always get the latest release versions of Komodo here:

http://komodochess.com

Again, many thanks to Jim Ablett and Dann Corbit and Aart Bilk for helping me debug and compile. Clemens Keck was very helpful also in helping me test.

Feedback is always welcome.

Don and Larry
Various platform specific versions will be posted as they become available and there may be multiple versions for your platform. If you don't see anything, check back frequently. At the time of this post only 2 version are ready, the Android version and the Mac version.
Does that mean there might be other platforms like iOS (iphone/ipad) or even PocketPC, I for one would love to have an additional engine for PocketFritz and will be prepared to pay about $30 US for the download, my last purchase was a Naum 4.2 engine so I hope there will be a stronger engine available soon.
I would like to see it run on as many platforms as possible, but I cannot spread myself any thinner than I already have. The Android version was made possible due the help of Aart Bik and probably also the fact that Android is Linux and Komodo was developed in Linux - so no real port was necessary. For other platforms it may not be nearly so easy. The spirit is willing but the flesh is weak - in other words I would love to provide binaries for other platforms but I'm not willing to put the enormous effort into this that would be required to see it through.

menniepals
Posts: 225
Joined: Wed Mar 08, 2006 7:31 pm
Location: Houston, Texas

Re: Komodo 2.03 release is imminent

Post by menniepals » Thu Jun 16, 2011 6:22 am

I hope the multi processor version will be a little cheaper than Rybka or Naum. :)

Sarciness
Posts: 43
Joined: Tue Nov 23, 2010 3:22 pm

Re: Komodo 2.03 release is imminent

Post by Sarciness » Thu Jun 16, 2011 2:35 pm

menniepals wrote:I hope the multi processor version will be a little cheaper than Rybka or Naum. :)
I hope it will be a lot cheaper!

Robert Flesher
Posts: 1232
Joined: Tue Aug 18, 2009 1:06 am

Re: Komodo 2.03 release is imminent

Post by Robert Flesher » Thu Jun 16, 2011 7:42 pm

Don wrote:The Komodo 2.03 release is imminent. Release 2.01 was short lived due to problems.

Version 2.03 is to be viewed as a separate version, not just a bug fix release as it has some improvements, perhaps 5-10 additional ELO.

But it also has a major bug fix that has caused problems in testing. This bug has been in Komodo since the beginning of time ...

Get it here: http://komodochess.com

Komodo 2.03 is significantly stronger than Komodo 1.3.

Various platform specific versions will be posted as they become available and there may be multiple versions for your platform. If you don't see anything, check back frequently. At the time of this post only 2 version are ready, the Android version and the Mac version. But more will soon follow. Some versions may run better on your machine than others so you may wish to use the version that runs fastest on your hardware.

You can always get the latest release versions of Komodo here:

http://komodochess.com

Again, many thanks to Jim Ablett and Dann Corbit and Aart Bilk for helping me debug and compile. Clemens Keck was very helpful also in helping me test.

Feedback is always welcome.

Don and Larry

Thanks Don and Larry, this new version is a real monster! SMP, and it's a sure purchase for me.

User avatar
Thomas Lagershausen
Posts: 238
Joined: Mon Jun 11, 2007 4:59 pm

Re: Komodo 2.03 release is imminent

Post by Thomas Lagershausen » Thu Jun 16, 2011 7:54 pm

I would say Komodo didn´t like to go any risk in his positions.

A so called passive chessplayer castle every time short.

A passive chessplayer didn´t go for a kingsattack with castling long.

A passive player looks always for a positional way to play the position.

It hates to offer a pawn for development.

A passive player didn´t like to go for fire on the board.

Komodo is playing chess like Karpov.

Very clean and cleare as cristal water, but absolutly unspectacular.

But this chessplayer is not the hero of the world.

Greetings
TL

Post Reply