Start Your Engines!

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

Moderator: Ras

User avatar
GenoM
Posts: 914
Joined: Wed Mar 08, 2006 9:46 pm
Location: Plovdiv, Bulgaria
Full name: Evgenii Manev

Re: A subthread removed

Post by GenoM »

Graham Banks wrote:When the moderation team moved some recent posts from this thread with profanities in them, we neglected to look at the post headers of some remaining posts (the downside of using flat view as opposed to flat view).
Those posts have been now pointed out and have been moved also.
Apologies for missing this originally and apologies to those of you who made good posts, but didn't change the header.

Regards, Graham (moderation)

PS - Albert has since pointed out to me that those profanities were actual engine names which I hadn't realised.
Graham,
my regards to you and to the whole moderation team but to remove posts because of their Re:- headers is, softly said, half-baked.
take it easy :)
User avatar
Graham Banks
Posts: 44898
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Content of all removed posts in here.

Post by Graham Banks »

Christopher Conkie:

It's a fruit, glaurung, crafty, rybka, shredder thing.

No download yet.....

But be advised there will be one soon.

after that we will release Shafta 32.0 (Black hole version)
Evgenii Manev:

Belka 1.8.7 is a beast
try it

http://www.zshare.net/download/513431570bf44a/
David Dahlem:

Many thanks for this link. Here is a logo for this engine ...

http://www.geocities.com/davedahlem/Belka187.jpg

Regards
Dave
Eelco de Groot:

Thanks Evgenii! Thanks Dave for your logos!

I am running Belka 1.8.11 in a test against Glaurung 2.01 and Glaurung in the Mjolnir settings, I am using 1.8.11 because I did not manage to download the 1.8.7 version at the time before your first links disappeared Evgenii Smile . Glaurung in the default settings on my single core Athlon is having not a very good match sofar and if 1.8.7 is even stronger as you said...


I am hoping that Tord is maybe willing to test some other settings because I really believe the Mjolnir settings could be better, at least against Belka. And if they are better against Belka maybe they could work better against Rybka too, although I do not think the two have really that much in common, as far as I can see, just by watching some games.

So far in these games Belka seems to be a more aggressive attacker than Rybka. But the way it uses its Queen did remind me of Rybka and it can make some very impressive moves sometimes. Even in limited timecontrols, Belka is not afraid to take some time to calculate something through for up to a minute for one move if it thinks the move is really strong, even in my 8 minutes + 2 seconds testing. Much like a strong GM would know when a move is really critical and can decide the game, and have the patience and coolbloodedness to use the available time to the max! Very good!

Regards, Eelco
Eelco de Groot:

Christopher you have received the sources I believe for both Strelka 1.0 and 1.8 Your analysis after several months I believe did not provide any results. I have not read all the threads but I'm sure that it would have made some headlines, if you had. The only one who could give a definitive answer I believe is Vasik Rajlich. Personally I think Yuri Osipov would not even object to Vasik inspecting his sources as well, as he invited all programmers to come and exchange information.

Why do you not ask Mr. Osipov if you can send your sources now to team Rybka instead of posting any more nonsense about your army of clones?

Please stop generalizing about all Russian chess programmers being Rasputins here!

Eelco
Uri Blass:

Vasik said the following:
"Strelka is obviously a Rybka 1.0 clone"

see

http://rybkaforum.net/cgi-bin/rybkaforu ... 1#pid33005
Andrej Sidorov:

Its interesting that Rajlich doesn't have and doesn't want to read sources of Strelka that Osipov offered to send him.
Tony Thomas:

He said that without even looking at the sources. What a pity.
gbanksnz at gmail.com
Terry McCracken
Posts: 16465
Joined: Wed Aug 01, 2007 4:16 am
Location: Canada

Re: Content of all removed posts in here.

Post by Terry McCracken »

Graham Banks wrote:
Christopher Conkie:

It's a fruit, glaurung, crafty, rybka, shredder thing.

No download yet.....

But be advised there will be one soon.

after that we will release Shafta 32.0 (Black hole version)
Evgenii Manev:

Belka 1.8.7 is a beast
try it

http://www.zshare.net/download/513431570bf44a/
David Dahlem:

Many thanks for this link. Here is a logo for this engine ...

http://www.geocities.com/davedahlem/Belka187.jpg

Regards
Dave
Eelco de Groot:

Thanks Evgenii! Thanks Dave for your logos!

I am running Belka 1.8.11 in a test against Glaurung 2.01 and Glaurung in the Mjolnir settings, I am using 1.8.11 because I did not manage to download the 1.8.7 version at the time before your first links disappeared Evgenii Smile . Glaurung in the default settings on my single core Athlon is having not a very good match sofar and if 1.8.7 is even stronger as you said...


I am hoping that Tord is maybe willing to test some other settings because I really believe the Mjolnir settings could be better, at least against Belka. And if they are better against Belka maybe they could work better against Rybka too, although I do not think the two have really that much in common, as far as I can see, just by watching some games.

So far in these games Belka seems to be a more aggressive attacker than Rybka. But the way it uses its Queen did remind me of Rybka and it can make some very impressive moves sometimes. Even in limited timecontrols, Belka is not afraid to take some time to calculate something through for up to a minute for one move if it thinks the move is really strong, even in my 8 minutes + 2 seconds testing. Much like a strong GM would know when a move is really critical and can decide the game, and have the patience and coolbloodedness to use the available time to the max! Very good!

Regards, Eelco
Eelco de Groot:

Christopher you have received the sources I believe for both Strelka 1.0 and 1.8 Your analysis after several months I believe did not provide any results. I have not read all the threads but I'm sure that it would have made some headlines, if you had. The only one who could give a definitive answer I believe is Vasik Rajlich. Personally I think Yuri Osipov would not even object to Vasik inspecting his sources as well, as he invited all programmers to come and exchange information.

Why do you not ask Mr. Osipov if you can send your sources now to team Rybka instead of posting any more nonsense about your army of clones?

Please stop generalizing about all Russian chess programmers being Rasputins here!

Eelco
Uri Blass:

Vasik said the following:
"Strelka is obviously a Rybka 1.0 clone"

see

http://rybkaforum.net/cgi-bin/rybkaforu ... 1#pid33005
Andrej Sidorov:

Its interesting that Rajlich doesn't have and doesn't want to read sources of Strelka that Osipov offered to send him.
Tony Thomas:

He said that without even looking at the sources. What a pity.
I knew you would pull my post...proving my point...
User avatar
Eelco de Groot
Posts: 4681
Joined: Sun Mar 12, 2006 2:40 am
Full name:   Eelco de Groot

Re: Content of all removed posts in here.

Post by Eelco de Groot »

Graham Banks wrote:
Andrej Sidorov:

Its interesting that Rajlich doesn't have and doesn't want to read sources of Strelka that Osipov offered to send him.
Tony Thomas:

He said that without even looking at the sources. What a pity.
Thanks Graham, that works.

To Tony and Andrej:

You have a point there, but I think a chessprogram author like Vas, who is working on his program practically every night and day, probably knows it better than he knows his own mind. If Strelka is really a completely re-engineered Rybka, the 'new' sources would probably not make much sense to even Vas, let alone if some of its inner working are deliberately obfuscated to look like tables of some sort, but really aren't.

It is also possible that taken from Rybka are only some parts that were really new or different, real but maybe small breakthroughs, and that this was combined with other parts, from other programs. Hiding the crucial parts would possibly not be very difficult if it was some small piece of code.

That other programs were combined I believe seems obvious enough if you look at some of the original output of Strelka 1.0 and 1.8, but is difficult to prove and not really relevant to the cloning of Rybka code. Some of this was evident from the sources I suppose and from what Dann Corbit told us, parts of Beowulf for instance are in the Winboard interface.

But in other words: Vasik would probably be able to say much more by watching how Strelka actually plays, he is a strong chessplayer himself and he would certainly know what to look for. You can "obfuscate" all you want but the final program still has to play chess.

The second point of attack; Vasik could have learnt more if he took the trouble of looking at Strelkas machinecode, as this would necessarily not be encrypted anymore once running and some keyparts would certainly be recognizable for him. But I don't even think that Vasik did this, the way the program plays and the output that in some instances was still very much like Rybka, we all know that, would have been enough I think. For Vasik it would be.

My best,
Eelco
Last edited by Eelco de Groot on Wed Nov 28, 2007 5:31 am, edited 1 time in total.
User avatar
Eelco de Groot
Posts: 4681
Joined: Sun Mar 12, 2006 2:40 am
Full name:   Eelco de Groot

Re: Content of all removed posts in here.

Post by Eelco de Groot »

Terry, no offence but you were not really a participant in the discussion so your remark was a bit out of the blue so to speak and not very nice either. Graham was just trying to be fair to everybody.

Regards, Eelco
Terry McCracken
Posts: 16465
Joined: Wed Aug 01, 2007 4:16 am
Location: Canada

Re: Content of all removed posts in here.

Post by Terry McCracken »

Eelco de Groot wrote:Terry, no offence but you were not really a participant in the discussion so your remark was a bit out of the blue so to speak and not very nice either. Graham was just trying to be fair to everybody.

Regards, Eelco
My point is to do with moderation practice...it was relavant.

Terry
User avatar
GenoM
Posts: 914
Joined: Wed Mar 08, 2006 9:46 pm
Location: Plovdiv, Bulgaria
Full name: Evgenii Manev

Re: Content of all removed posts in here.

Post by GenoM »

Eelco de Groot wrote:The second point of attack; Vasik could have learnt more if he took the trouble of looking at Strelkas machinecode, as this would necessarily not be encrypted anymore once running and some keyparts would certainly be recognizable for him. But I don't even think that Vasik did this, the way the program plays and the output that in some instances was still very much like Rybka, we all know that, would have been enough I think. For Vasik it would be.

My best,
Eelco
According to Osipov, he had offered three times to Vasik to send him the sources of Strelka and Raijlich refused three times.
See (translation of Zruty from RybkaChess forum):
Osipov Yuri wrote:Strange man, Vasik.
Three times I offered him to look at Strelka sources - in July, September and October. In July he wasn't interested, just asked me how difficult it was to study Rybka under disassembler and whether it was more difficult with newer Rybka versions. I described everything in detail. In September he was too busy preparing for Zappa match. After my October letter he thought about something for three weeks and answered too dimly.
And now his recent claim. {Raijlich: "Strelka is obviously a Rybka 1.0 clone. "} I don't understand - he doesn't want to read Strelka's sources but has a strong opinion about it.
(for those who know russian -- http://kasparovchess.crestbook.com/view ... 18#p124718)

Weird story...
take it easy :)