Komodo CCT release

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

Moderators: hgm, Rebel, chrisw

User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 4:27 pm

Komodo CCT release

Post by Don »

In private correspondence a lot of people have asked me about the version of Komodo that won the CCT 15 tournament by 1.5 points and competed so well against 16 cores programs in the TCEC, many requesting an opportunity to purchase this program.

The program that competed was our latest development version of Komodo and is approximately 30 ELO stronger. Since I am discontinuing development of Komodo from this code base, we decided that it would be a shame to just orphan such a strong program, so we are offering it as a one-off purchase on the Komodo web site. There are no user discounts for existing customers or anything like that and it's not even part of our normal version scheme, thus we call it Komodo CCT.

Some of you may be wondering why you should purchase this when you have already been promised Komodo 5 MP and in fact many of you have pre-ordered it and have been waiting for some time. The answer for you is that you probably should just wait for Komodo 5.1 MP which will be free for anyone who has already pre-ordered it. We have set the release date to May 28, 2013 which is not a target but a firm date. In fact the program is ready now and working well and we will use the time to make some usability and strength enhancements but we are maintaining a release-ready version at all times so the date will not slip. In the future we plan to do all releases that way with firm dates and we will ship the best of whatever we have on the release date.

As you may have heard, the new MP version is a rewrite, basically a port of Komodo to C++11 and a major code cleanup and it is still a work in progress. As such it is still in a bit more primitive state than Komodo CCT. There is some minor missing knowledge and a lot of cruft stripped out of it. But there is also some things done much better, so we expect this to be a better base to proceed from.

Unfortunately, the Komodo 5.1 MP release is currently about 30 ELO weaker than the new Komodo CCT version when running on 1 core, but of course on 2 cores or more it will play much stronger. It will also be at least as strong as Komodo 5. Even though the strength matches Komodo 5 it is not the same program and internally is closer to Komodo CCT. If you want the strongest possible Komodo running on multi-core machines then Komodo 5.1 MP is what you want. If you are fan of single processor programs then Komodo CCT will be stronger.

After we release Komodo 5.1 MP we will be working to bring it up to, and beyond, the strength level of Komodo CCT. So in November we plan to release the upgrade to Komodo 5.1 MP which will be our strongest program yet running on 1 core or many cores.

If you find any of that confusing, here is a table to help make it clearer:

Code: Select all

Code Base     Language  Program(s)
------------  --------  ---------------------
Doch          C         Komodo 3, 4, 5,  CCT
Beekay        C++11     Komodo 5.1 MP

Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
User avatar
geots
Posts: 4790
Joined: Sat Mar 11, 2006 12:42 am

Re: Komodo CCT release

Post by geots »

Extremely well explained. Good job!


gts
Daniel Shawul
Posts: 4185
Joined: Tue Mar 14, 2006 11:34 am
Location: Ethiopia

Re: Komodo CCT release

Post by Daniel Shawul »

Doesn't this fall under flagarant commercial exhtortions? I have seen posts removed for much less.
User avatar
pohl4711
Posts: 2433
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

Re: Komodo CCT release

Post by pohl4711 »

Don wrote: The program that competed was our latest development version of Komodo and is approximately 30 ELO stronger.
Nice. LS-ratinglist testrun will start after the Stockfish 3.0 JA testrun is finished (May 8 ?!?).

Stefan
Tom Likens
Posts: 303
Joined: Sat Apr 28, 2012 6:18 pm
Location: Austin, TX

Re: Komodo CCT release

Post by Tom Likens »

Don,

What has been your experience with the new C++11 standard? Are you still compiling all this with MingW?
And finally, which version of gcc are you using, have you made the switch to 4.8 yet (it seems to be much
more C++11 compliant)?

regards,
--tom
User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 4:27 pm

Re: Komodo CCT release

Post by Don »

Tom Likens wrote:Don,

What has been your experience with the new C++11 standard? Are you still compiling all this with MingW?
And finally, which version of gcc are you using, have you made the switch to 4.8 yet (it seems to be much
more C++11 compliant)?

regards,
--tom
I'm using mingw 4.8 on windows, and I found a distribution fixed up to support the C++11 way of doing threads - that seems to be the real sticking point and it was a lot of work making this work. The good news is that everything is much more cross platform. I don't think there is a single #ifdef in the program to deal with cross platform idiosyncrasies. That was exactly what I had hoped for.

So I actually made the binary on windows, where normally I use the mingw cross-compiler and compile Komodo on Linux even for windows.

Not every distribution of mingw works out of the box for multi-threaded programs though (or else I couldn't figure out how.) I had downloaded 3 or 4 before stumbling on to this one. I'm sorry to say I don't remember where I got it but I will look for you.

I also had some trouble making a static compile, so my testers were having to place a couple of DLL's in the directory with the binary. However, Dann Corbit helped me figure out how to do this and it was simpler than I thought.
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
User avatar
Graham Banks
Posts: 41415
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Komodo CCT release

Post by Graham Banks »

Thanks Don.
Looking forward to trying this out. :P
gbanksnz at gmail.com
Werewolf
Posts: 1795
Joined: Thu Sep 18, 2008 10:24 pm

Re: Komodo CCT release

Post by Werewolf »

Don Dailey you are a blessing. Thanks so much for sticking with this and seeing it through.

Can I ask: how many cores does Komodo 5.1 MP support, and how many do you aim to support in Nov?


Regards,

Carl
Rein Halbersma
Posts: 741
Joined: Tue May 22, 2007 11:13 am

Re: Komodo CCT release

Post by Rein Halbersma »

Don wrote: I'm using mingw 4.8 on windows, and I found a distribution fixed up to support the C++11 way of doing threads - that seems to be the real sticking point and it was a lot of work making this work. The good news is that everything is much more cross platform. I don't think there is a single #ifdef in the program to deal with cross platform idiosyncrasies. That was exactly what I had hoped for.

So I actually made the binary on windows, where normally I use the mingw cross-compiler and compile Komodo on Linux even for windows.

Not every distribution of mingw works out of the box for multi-threaded programs though (or else I couldn't figure out how.) I had downloaded 3 or 4 before stumbling on to this one. I'm sorry to say I don't remember where I got it but I will look for you.

I also had some trouble making a static compile, so my testers were having to place a couple of DLL's in the directory with the binary. However, Dann Corbit helped me figure out how to do this and it was simpler than I thought.
Hi Don,

Which C++11 features do you actively use and find most useful? Which ones do you avoid?

Rein
Jesse Gersenson
Posts: 593
Joined: Sat Aug 20, 2011 9:43 am

Re: Komodo CCT release

Post by Jesse Gersenson »

Werewolf wrote:Can I ask: how many cores does Komodo 5.1 MP support, and how many do you aim to support in Nov?
Currently supports 64 cores.

http://komodochess.com/store/pages.php?cmsid=9