Komodo 1.2 update available

[Dr.Wael Deeb]

Thanks Deeb, that worked!

Happy you managed

[Arturo Ochoa]

No I don't think the interface does. When i load up Komodo 1.2 within Fritz there are very few UCI options, one however is "use own book" and then you can name the book you want to use but without a file reference this is meaningless - i.e there's no way to select the book.

The only book formats I've seen work (easily) with Fritz are ctg. files and in-built UCI ones as Hiarcs does.

I don't suppose you'll produce a book in either of these formats?

Is there any way to convert from polyglot book format to chessbase?

The answer is no,but it would be great if you develop a feature in Komodo to create an opening book from a pgn file....then I'll personally support you with a killer opening book,say up to 40 halfmoves....
Dr.D

There are already tools to create a polyglot book from pgn files. So is there something specific you have in mind?

I know Don,but it's rather a little bit coplicated to create a polyglot opening book,at least for me....is there any automated tools out there I can use

What I have in mind is a simple command line like in Bright chess engine:

builfbook bright.pgn bright.ini

Now the ini file contains the parameters which will determine the moves that will be choosen from the pgn file....
Here's a segment from Bright's ini file which deals with this:

#
# book building options
#
# file : output file, if empty: derived from pgn file name: x.pgn -> x.bbk
# minelo : drops games by players with lower elo than minelo
# maxelodelta : drops games between players with higher elo difference than maxelodelta
# minply : drops games with less plies than minply
# maxply : drops moves beyond this game ply
# mingames : drops moves occuring in less games than mingames
# minscorew : [0..100] drops white moves scoring less than minscorew
# minscoreb : [0..100] drops black moves scoring less than minscoreb
# keeppct : [0..100] for each position: drops moves that cover less than keeppct
# of the games that the position occured in
# (e.g. keeppct=80 will usually drop e3, f3 etc from the initial position)
#
book.build.file=drdeeb2010.bbk
book.build.minelo=
book.build.maxelodelta=
book.build.minply=
book.build.maxply=40
book.build.mingames=2
book.build.minscorew=0
book.build.minscoreb=0
book.build.keeppct=98

My thoughts tells me that this is a rather trivial task for you....
Flattery in action regards ,
Dr.D

Basically, what you are talking about already exists in most of the Polyglot book creation options. It is not complicated and it is quite a joke to create a book there.

I have a private polyglot book used by The Baron in different Internet Tournaments and now I am tuning it with the Polyglot features of Scid. The only thing, which I find nasty, is you must save the book in the Scid Gui every time you add a new move or modify a score.

Personally, in my opinion, a killer book created from a simple pgn file is a fantasy. The manual tuning is always there, no matters how much you can hate doing it… Maybe killer for certain type of opponent, but not in general.

[swami]

Komodo 1.2 is now #3 from 63 Engines Tested!

Komodo 1.1 had 741/1000.

http://sites.google.com/site/strategict ... st-results

[jpqy]

Thanks swami for this great result from Komodo 1.2 ..3th place,and this with a single core engine!

[swami]

Thanks swami for this great result from Komodo 1.2 ..3th place,and this with a single core engine!

You're welcome, Vael.

All engines were tested with 1 CPU. Yes, It's a pretty good result from Komodo. I believe it's atleast 25 elo better than version 1.1.

[Don]

I am curious about how much this test is affected by the hardware it is run on. For instance will running it on a computer 10x faster make it perform like a program that that can give you 10-1 time odds?

The reason I ask is that we are interested in somehow "measuring" the knowledge contained in the program. Unfortunately I think this is a mostly impossible task. If you make the program faster, or improve the search, it also "appears" to be positionally stronger too.

It's rather ironic that I am more interested in the knowledge part and Larry is more interested in the search part. The irony is that Larry is better suited to help with the knowledge and I am better suited for working on the search! We both push each other to work on the other thing.

I am not sure what this test measures, but I am pleased that it seems to indicate that Komodo is reasonably strong in knowledge.

Komodo 1.2 is now #3 from 63 Engines Tested!

Komodo 1.1 had 741/1000.

http://sites.google.com/site/strategict ... st-results

[Don]

This table is a bit confusing. Why are there 2 grades and what does a score of S mean?

Komodo 1.2 is now #3 from 63 Engines Tested!

Komodo 1.1 had 741/1000.

http://sites.google.com/site/strategict ... st-results

[jpqy]

S = Super Strong!!!

You have just to make STS2 , STS7 , STS8 & STS9 better !

Don, you can download the program here: http://sites.google.com/site/strategict ... e/sts-stat

[swami]

This table is a bit confusing. Why are there 2 grades and what does a score of S mean?

Komodo 1.2 is now #3 from 63 Engines Tested!

Komodo 1.1 had 741/1000.

http://sites.google.com/site/strategict ... st-results

S grade means "Superior" or "Outstanding". It's the grade awarded in Indian Institute of Technology.

1st score out of 100 means number of positions that the engine found the solution.

2nd score for 1000 is a partial credit scoring (underlined below)

1kr5/3n2p1/q2rpn1p/p7/Ppp2P2/5BP1/1PQ2B1P/2RR2K1 w - - bm Rxd6; id "STS(v10.0) Simplification.008"; c0 "Rxd6=10, Kh1=4, Qe2=7, Rd4=4";

... from all the 100 positions for each epd.

[Don]

Thanks you. That makes it much clearer to me now.

This table is a bit confusing. Why are there 2 grades and what does a score of S mean?

Komodo 1.2 is now #3 from 63 Engines Tested!

Komodo 1.1 had 741/1000.

http://sites.google.com/site/strategict ... st-results

S grade means "Superior" or "Outstanding". It's the grade awarded in Indian Institute of Technology.

1st score out of 100 means number of positions that the engine found the solution.

2nd score for 1000 is a partial credit scoring (underlined below)

1kr5/3n2p1/q2rpn1p/p7/Ppp2P2/5BP1/1PQ2B1P/2RR2K1 w - - bm Rxd6; id "STS(v10.0) Simplification.008"; c0 "Rxd6=10, Kh1=4, Qe2=7, Rd4=4";

... from all the 100 positions for each epd.