Perhaps Rybka 4 IS proving itself.

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.
Steve B
Posts: 3697
Joined: Tue Jul 31, 2007 2:26 pm
Contact:

Re: Perhaps Rybka 4 IS proving itself.

Post by Steve B » Mon Oct 11, 2010 7:18 pm

Graham Banks wrote:
Steve B wrote:History is repeating itself because you had ONE post removed where you accused a member of lying and called his character into question and offered nothing else in that post??
actually people left here due to posts like yours
your post was a violation of our charter
i removed it not once ..but twice
and i asked you to kindly not post it again via PM
why dont you insult members of the other forum like you so freely do here?

Steve
Even better, if it's such a paradise, why doesn't he stay over there!
at issue here of course is not what side of the Ippo debate one is on ..
but whether or not we can have a civil discussion and not personally attack those that disagree with our positions
De Vos's post was a personal attack pure and simple
for the most part the debate has been civil but
it seems there is a small group here who believe they can ignore the charter simply because they are on one side of the debate
well they cant
Steve

Milos
Posts: 3387
Joined: Wed Nov 25, 2009 12:47 am

Re: Perhaps Rybka 4 IS proving itself.

Post by Milos » Mon Oct 11, 2010 8:58 pm

Steve B wrote:anyone whose post has been removed is welcome to repost anything they please regarding Rybka's strengths/weaknesses... however.. please do so Sans the personal comments/attacks
There is no point of reposting, since some posts are obviously not removed in order to protect the charter. If certain moderators were not biased they would also try to protect other points of the charter except point 3.
For example point 4 which says:
"Are not flagrant commercial exhortations"

Posts of M Asnari, who is part of the Vasik Rajlich team and is constantly and flagrantly falsifying and distorting facts and results in order to advertise the product whose team he is part of - Rybka - are nothing but against point 4 of the charter.

tomgdrums
Posts: 736
Joined: Wed Dec 02, 2009 8:48 am

Re: Perhaps Rybka 4 IS proving itself.

Post by tomgdrums » Mon Oct 11, 2010 9:06 pm

Milos wrote:
Steve B wrote:anyone whose post has been removed is welcome to repost anything they please regarding Rybka's strengths/weaknesses... however.. please do so Sans the personal comments/attacks
There is no point of reposting, since some posts are obviously not removed in order to protect the charter. If certain moderators were not biased they would also try to protect other points of the charter except point 3.
For example point 4 which says:
"Are not flagrant commercial exhortations"

Posts of M Asnari, who is part of the Vasik Rajlich team and is constantly and flagrantly falsifying and distorting facts and results in order to advertise the product whose team he is part of - Rybka - are nothing but against point 4 of the charter.
Why do you hate rybka so much?

Milos
Posts: 3387
Joined: Wed Nov 25, 2009 12:47 am

Re: Perhaps Rybka 4 IS proving itself.

Post by Milos » Mon Oct 11, 2010 9:12 pm

tomgdrums wrote:Why do you hate rybka so much?
If I were like some moderators I could conclude that you are actually calling me a Rybka hater which could be interpreted as a personal attack which is against point 3 of the charter which some moderators are so keen to enforce.
Oh, wait, this is only valid for member who are against Rybka, for others it's not so certain, so I would say you are on the safe side... ;)

User avatar
M ANSARI
Posts: 3408
Joined: Thu Mar 16, 2006 6:10 pm

Re: Perhaps Rybka 4 IS proving itself.

Post by M ANSARI » Mon Oct 11, 2010 9:12 pm

Actually I had wanted to share some of the information we had during beta testing and after beta testing, where IMHO I really thought that R4 had serious TC issues. Unfortunately this quickly had to change to silly claims and personal attacks that have nothing to do with the original thread ... which wanted to know why R4 results in a publicized Live match were so much better than was expected. The result was no surprise to me as this result was in a match at long time controls and with ponder ON. I had very similar results early on and I did post on the R4 forum that I thought "R4 was the most underrated engine out there". I said that simply because in my testing R4 had serious issues with time management and it was very easy to change the performance of R4 by changing the time parameters. Unfortunately it was not possible to come up with a "perfect setting" and we had to compromise and choose a setting that would do OK at all time controls. We did manage to talk Vas into leaving some TC parameters on, and so depending on what TC setup you want to play, you most certainly can improve on the default settings.

Anyway with R4 it is also not as simple as not having a good time management setup, and I will share some opinions I have. Of course R4 will do much better if Vas would spend some time on a sophisticated TC module. But it is complicated because it has a lot to do with R4 (and R3) using processes instead of threads. With processes there seems to be a latency every time the engine is initialized. While this is minute it is much more than what an engine would go through in a threaded engine. It was even brought up to Vas to see if he would possibly create a threaded R4, but he was against it. Although I am not certain, I think that Vas sees R4 using processes as being better for the future since it seems that with processes you can overcome the slight latencies of engine initialization and get better scalability as more and more cores are added. Again in testing this seems to be plausible because on big hardware R4 performed dramatically better than it should against other engines as did R3. Also in Ponder ON games R4 performance was better than in Ponder OFF games (no engine initialization in P ON games).

You could possibly say that there is a threaded version of R3 out there, and by that I mean the R3 clones. If you play one of the strongest clones against R3 on say a dual core and watch the games while looking at the clock, you will quickly see how R3 falls way behind on time, and will try to blitz out its moves after it is way behind on the clock. When you add to this the fact that there is more lag in a process engine than a threaded engine, the final seconds remaining will be to the advantage of a threaded engine. Now when you move to bigger hardware (8 core and more), the better scalability seems to more than compensate for the process initialization lag. You have to remember that R4 actually came out of Rybka cluster, thus its TM setup is something that would be more tuned to that environment.

Anyway that is how I explained to myself why in my testing R4 results are so much better with Ponder ON and Long Time Controls. If someone has another theory I would be very interested in hearing it and having a dialog in a civil way as this issue IMHO is very interesting.

Milos
Posts: 3387
Joined: Wed Nov 25, 2009 12:47 am

Re: Perhaps Rybka 4 IS proving itself.

Post by Milos » Mon Oct 11, 2010 9:26 pm

M ANSARI wrote:Anyway that is how I explained to myself why in my testing R4 results are so much better with Ponder ON and Long Time Controls. If someone has another theory I would be very interested in hearing it and having a dialog in a civil way as this issue IMHO is very interesting.
I have a different explanation that your test results are either product of completely distorted and faulty test methodology or simple falsifications.

P.S. Dear (Rybka responsible) moderator, is this allowed to be said on this site or, in order to be in line with your understanding of the charter, we have to blindly believe anything that any member of Rybka team says, no matter how ridiculous that is???

Wayne Lowrance
Posts: 60
Joined: Thu Feb 18, 2010 2:35 am

Re: Perhaps Rybka 4 IS proving itself.

Post by Wayne Lowrance » Mon Oct 11, 2010 9:32 pm

Dad Blame it, here we go again, attacking Vas, I am sick of this
Wayne

Steve B
Posts: 3697
Joined: Tue Jul 31, 2007 2:26 pm
Contact:

Re: Perhaps Rybka 4 IS proving itself.

Post by Steve B » Mon Oct 11, 2010 9:35 pm

Milos wrote:
Steve B wrote:anyone whose post has been removed is welcome to repost anything they please regarding Rybka's strengths/weaknesses... however.. please do so Sans the personal comments/attacks
There is no point of reposting, since some posts are obviously not removed in order to protect the charter. If certain moderators were not biased they would also try to protect other points of the charter except point 3.
For example point 4 which says:
"Are not flagrant commercial exhortations"

Posts of M Asnari, who is part of the Vasik Rajlich team and is constantly and flagrantly falsifying and distorting facts and results in order to advertise the product whose team he is part of - Rybka - are nothing but against point 4 of the charter.
Thats right
if we moderate a post when someone attacks an Ippolit supporter(which we have) we are biased and corrupt
if we eliminate a post where some one supporting Rybka is attacked (which we have)we are biased and corrupt
your particular post called into question a members Morality
the point is and will always be that personal attacks are not permitted
PERIOD
Steve
Last edited by Steve B on Mon Oct 11, 2010 9:39 pm, edited 1 time in total.

User avatar
michiguel
Posts: 6388
Joined: Thu Mar 09, 2006 7:30 pm
Location: Chicago, Illinois, USA
Contact:

Re: Perhaps Rybka 4 IS proving itself.

Post by michiguel » Mon Oct 11, 2010 9:38 pm

M ANSARI wrote:Actually I had wanted to share some of the information we had during beta testing and after beta testing, where IMHO I really thought that R4 had serious TC issues. Unfortunately this quickly had to change to silly claims and personal attacks that have nothing to do with the original thread ... which wanted to know why R4 results in a publicized Live match were so much better than was expected. The result was no surprise to me as this result was in a match at long time controls and with ponder ON. I had very similar results early on and I did post on the R4 forum that I thought "R4 was the most underrated engine out there". I said that simply because in my testing R4 had serious issues with time management and it was very easy to change the performance of R4 by changing the time parameters. Unfortunately it was not possible to come up with a "perfect setting" and we had to compromise and choose a setting that would do OK at all time controls. We did manage to talk Vas into leaving some TC parameters on, and so depending on what TC setup you want to play, you most certainly can improve on the default settings.

Anyway with R4 it is also not as simple as not having a good time management setup, and I will share some opinions I have. Of course R4 will do much better if Vas would spend some time on a sophisticated TC module. But it is complicated because it has a lot to do with R4 (and R3) using processes instead of threads. With processes there seems to be a latency every time the engine is initialized. While this is minute it is much more than what an engine would go through in a threaded engine. It was even brought up to Vas to see if he would possibly create a threaded R4, but he was against it. Although I am not certain, I think that Vas sees R4 using processes as being better for the future since it seems that with processes you can overcome the slight latencies of engine initialization and get better scalability as more and more cores are added. Again in testing this seems to be plausible because on big hardware R4 performed dramatically better than it should against other engines as did R3. Also in Ponder ON games R4 performance was better than in Ponder OFF games (no engine initialization in P ON games).

You could possibly say that there is a threaded version of R3 out there, and by that I mean the R3 clones. If you play one of the strongest clones against R3 on say a dual core and watch the games while looking at the clock, you will quickly see how R3 falls way behind on time, and will try to blitz out its moves after it is way behind on the clock. When you add to this the fact that there is more lag in a process engine than a threaded engine, the final seconds remaining will be to the advantage of a threaded engine. Now when you move to bigger hardware (8 core and more), the better scalability seems to more than compensate for the process initialization lag. You have to remember that R4 actually came out of Rybka cluster, thus its TM setup is something that would be more tuned to that environment.

Anyway that is how I explained to myself why in my testing R4 results are so much better with Ponder ON and Long Time Controls. If someone has another theory I would be very interested in hearing it and having a dialog in a civil way as this issue IMHO is very interesting.
I am willing to discuss in a civil manner, but it looks like only insults are answered. For instance, I am still waiting you answer my previous civilized questions :-)

Anyway, I do not think your theory of processes and threads make any sense for various reasons. In fact, I cannot imagine why it would affect an engine when it is running with a single core. Even running multiple cores, the TC should have nothing to do with that.

Miguel

User avatar
M ANSARI
Posts: 3408
Joined: Thu Mar 16, 2006 6:10 pm

Re: Perhaps Rybka 4 IS proving itself.

Post by M ANSARI » Mon Oct 11, 2010 9:45 pm

Milos wrote:
Steve B wrote:anyone whose post has been removed is welcome to repost anything they please regarding Rybka's strengths/weaknesses... however.. please do so Sans the personal comments/attacks
There is no point of reposting, since some posts are obviously not removed in order to protect the charter. If certain moderators were not biased they would also try to protect other points of the charter except point 3.
For example point 4 which says:
"Are not flagrant commercial exhortations"

Posts of M Asnari, who is part of the Vasik Rajlich team and is constantly and flagrantly falsifying and distorting facts and results in order to advertise the product whose team he is part of - Rybka - are nothing but against point 4 of the charter.

Dude ... that is really funny . M Ansari is NOT part of the Vasik Rajlich team and advertising for Rybka. M Ansari is an Architectural and Civil Engineer whose main business is real estate development. The idea that I am working for Vas or that I have some financial gain from Rybka has to be the funniest thing I have yet to hear. Do you know how much Rybka brings in income? Trust me it would have to be dramatically more profitable (emphasis on dramatically) before it would pique my interest financially. I do like chess and consider it as a hobby, and chess engines bring chess to a new level. I met Vas on this very forum as well as many others that I now consider friends. I was probably one of the first to test Rybka when it first came out on this forum, and have had a dialog with Vas ever since. I am a member of the Rybka beta testing group ... but trust me Vas doesn't pay me for that :P I am quite sure that I am not doing anything against point 4 in the charter :D

Post Reply