As for analysis and Dynamism, wouldn't it be good to have a value of its own as well as contempt has, some kind of "White Dynamism" too? And wouldn't that be so for King's Safety as welllkaufman wrote:My subjective feeling is that the evals using the default of 120 seem a bit too high for the side with the Dynamic plus, but that's what tests well. I think your 110 is a very reasonable choice for analysis.
Komodo 11.2
Moderators: hgm, Rebel, chrisw
-
- Posts: 3187
- Joined: Sat Feb 16, 2008 7:38 am
- Full name: Peter Martan
Re: Dynamism in Komodo 11.2
Hi Larry!
Peter.
-
- Posts: 2453
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: Komodo 11.2
Of course not. I just updated my website. Finally the calculation of my new gamebase for Stockfish-testruns was finished (one month of calculation...). Now, I am testing the latest Stockfish. When this testrun is finished, I will test Komodo 11.2.JJJ wrote:
And does it mean you will have to reset your test since a new Komodo is out ?
Stefan (SPCC)
-
- Posts: 12038
- Joined: Mon Jul 07, 2008 10:50 pm
Re: Komodo 11.2
congrats on the achievement.lkaufman wrote:We urge testing groups to test this version despite the short two month period since the last one, as the gains would usually take us four months or more.
-
- Posts: 5960
- Joined: Sun Jan 10, 2010 6:15 am
- Location: Maryland USA
Re: Dynamism in Komodo 11.2
It's possible, but we haven't really investigated such asymmetric evaluation and I don't know whether it would be very useful.peter wrote:Hi Larry!As for analysis and Dynamism, wouldn't it be good to have a value of its own as well as contempt has, some kind of "White Dynamism" too? And wouldn't that be so for King's Safety as welllkaufman wrote:My subjective feeling is that the evals using the default of 120 seem a bit too high for the side with the Dynamic plus, but that's what tests well. I think your 110 is a very reasonable choice for analysis.
Komodo rules!
-
- Posts: 8514
- Joined: Thu Mar 09, 2006 3:25 am
- Location: Jerusalem Israel
Re: Komodo 11.2
I'm sure this would be nice too.Nordlandia wrote:How long before Android version is ready?
I launch matches between SF Dev and latest Komodo on Galaxy S8+
But the main thing is Komodo being top again, in big PC computers.
-
- Posts: 8514
- Joined: Thu Mar 09, 2006 3:25 am
- Location: Jerusalem Israel
Re: Komodo 11.2
Wonderful!lkaufman wrote:Komodo 11.2 has been released at komodochess.com. We skipped over 11.1 as the elo gain from 11.01 is unusually large (estimated 24 elo) and also to avoid confusion between 11.01 and 11.1. Gains came from all areas, search, eval, and time management. As usual, it is free to subscribers (and to anyone who bought 11.01 in July) and discounted 20% to those who bought 10.00 or later version. It is the version that won the World Computer Championship this month with some speedups, a minor bugfix, and a changed UCI option default. We urge testing groups to test this version despite the short two month period since the last one, as the gains would usually take us four months or more.
Now it must CERTAINLY be time for next TCEC where komodo might beat all the others including SF.
Quick TCEC! Before SF comes out to overtake komodo again!
If komodo is top AND plays best positionally too, then this makes komodo much more top than just top!
-
- Posts: 3187
- Joined: Sat Feb 16, 2008 7:38 am
- Full name: Peter Martan
Re: Dynamism in Komodo 11.2
Well, Larry, I would expect it to be as useful as for Contempt, which makes any other evaluation asymmetric anyhow too, doesn't it?lkaufman wrote:It's possible, but we haven't really investigated such asymmetric evaluation and I don't know whether it would be very useful.
Of course one could be confident with that kind of asymmetric eval only, but wouldn't it change the symmetry of King's Safety and Dynamism in analysis mode with each ply moving forward or backward like that's the case with Contempt?
So wouldn't a "White Dynamism" and or a "White King's Safety" be as necessary as a White Contempt is for analysis mode compared to playing for one side only mode
Peter.
-
- Posts: 1494
- Joined: Thu Mar 30, 2006 2:08 pm
Re: Komodo 11.2
It has been made and is being tested now. So soon.Nordlandia wrote:How long before Android version is ready?
I launch matches between SF Dev and latest Komodo on Galaxy S8+
Mark
-
- Posts: 1494
- Joined: Thu Mar 30, 2006 2:08 pm
Re: Dynamism in Komodo 11.2
You really want to use "White Contempt" when analyzing back and forth. Otherwise, Contempt will be used for whatever side Komodo is calculating a move for. "White Contempt" locks down Contempt for a specific color's point of view. Contempt is definitely asymmetrical.peter wrote:Well, Larry, I would expect it to be as useful as for Contempt, which makes any other evaluation asymmetric anyhow too, doesn't it?lkaufman wrote:It's possible, but we haven't really investigated such asymmetric evaluation and I don't know whether it would be very useful.
Of course one could be confident with that kind of asymmetric eval only, but wouldn't it change the symmetry of King's Safety and Dynamism in analysis mode with each ply moving forward or backward like that's the case with Contempt?
So wouldn't a "White Dynamism" and or a "White King's Safety" be as necessary as a White Contempt is for analysis mode compared to playing for one side only mode
]The present Dynamism is not tied to a specific side. It inflates or deflates a number of dynamic parts of the eval, without changing the more static evaluation features. So it is symmetric. Different values for each color are possible, but I am unsure of what gain it might bring. If you have some sample positions that might show what is desired, I could see what we can do.
Mark
-
- Posts: 6
- Joined: Thu Jul 09, 2015 2:28 pm
- Location: São Paulo - Brazil
Re: Komodo 11.2
What a stupid and biased observation. Comments like this one took Thorensen out of the TCEC command. How much ignorance.
Commentary on this Taylor's opinion.
Commentary on this Taylor's opinion.