Message boards : Number crunching : Note to my team
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Pete49

Send message
Joined: 18 Sep 04
Posts: 35
Credit: 250,303
RAC: 0
Message 6248 - Posted: 2 Mar 2005, 5:26:03 UTC
Last modified: 2 Mar 2005, 5:26:49 UTC

I am curtailing my LHC resources due to the credit problems at that project. WU's continue to be reported with 0 cpu time / 0 credit regardless of the fact that hours and hours of work has been crunched.

I feel I have a fiduciary responsibility to the team to generate credits regardless how I personally feel about the project.

9 machines averaging 8 hours per day is yielding only a third of the RAC that other projects with equal share are generating on the same machines.

Resolving this problem does not appear high on their list of things to fix.

<img src="http://www.boincstats.com/stats/teambanner.php?teamname=GasBuddy"> <img src="http://www.boincstats.com/stats/banner.php?cpid=84c0cf7846cbf28338406e54b3eb8a83">
ID: 6248 · Report as offensive     Reply Quote
STE\/E

Send message
Joined: 2 Sep 04
Posts: 352
Credit: 1,393,150
RAC: 0
Message 6250 - Posted: 2 Mar 2005, 10:00:07 UTC
Last modified: 2 Mar 2005, 10:07:14 UTC

Pete, I have Hundreds if not approaching Thousands of WU's in my Account that show 0 Time, but many of them show Credit Granted also, so it's not all lost.

We are most definitely losing some credit but it may not be as bad as it seems ... And I'm sure it is a Very High Priority for the Dev's to get this problem fixed because I think they know in order for this Project to succeed it has to be fixed ASAP or more & more people are going to drop out and go to the other Projects ... :)
ID: 6250 · Report as offensive     Reply Quote
Profile adrianxw

Send message
Joined: 29 Sep 04
Posts: 187
Credit: 705,487
RAC: 0
Message 6260 - Posted: 2 Mar 2005, 11:40:11 UTC
Last modified: 2 Mar 2005, 11:48:09 UTC

Whilst I sympathise with the developers, it does not alter the fact that the hours being spent on units here are not being credited correctly, and at other projects they are. People working on several projects WILL look at this as a loss of potential credits elsewhere, and will be tempted to drop the project.

It is all very well the Known bugs page saying it affects everyone equally, but that is a very blinkered statement. It quite simply is untrue in the wider view.

Example.

It seems to be getting worse, not better. Look at the results prior to 1/3 and then after.

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
ID: 6260 · Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 2 Sep 04
Posts: 545
Credit: 148,912
RAC: 0
Message 6266 - Posted: 2 Mar 2005, 15:48:34 UTC

From my perspective it is worse than that ...

If the project cannot make sure that the program tracks time correctly, can they be trusted to make the hard parts are correct?
ID: 6266 · Report as offensive     Reply Quote
Profile littleBouncer
Avatar

Send message
Joined: 23 Oct 04
Posts: 358
Credit: 1,439,205
RAC: 0
Message 6273 - Posted: 2 Mar 2005, 17:48:09 UTC
Last modified: 2 Mar 2005, 17:50:50 UTC

[QUOTE: from Paul]
can they be trusted to make the hard parts are correct?
[/QUOTE]

exactly!

Thats's why I drop this project.

greetz from Switzerland
littleBouncer
(sorry I don't want disturb a teams-conference)


ID: 6273 · Report as offensive     Reply Quote
Profile Logan5@SETI.USA
Avatar

Send message
Joined: 30 Sep 04
Posts: 112
Credit: 104,059
RAC: 0
Message 6324 - Posted: 3 Mar 2005, 22:43:39 UTC
Last modified: 3 Mar 2005, 22:46:58 UTC

'Ya know, all this talk of dropping LHC@home seems to be a premature no? Sure there's problems, but if you look at the alternatives, most of them are far worse:
------------------------------------------------------------------------------
S@H "Classic" - Restricted to running 8 hours per day because of Building wide power problems....NO ETA

S@H BOINC - Down indefinately because of Building wide power problems...NO ETA

Pirates@home - Project cancelled within the next few months

CPDN - Core Client compatibility issues

LHC@Home - Stable and running well....
-------------------------------------------------------------------------------

Out of the alternatives, LHC is the "project to beat" even with the zero credit issue....

<B>Besides, it should NEVER be about credit or scores, but instead it should ALWAYS be about the science first, last and always. I crunch them as I get them, no matter what I get on the other end as a 'reward'.

Maybe more people should look at what their priorities are if they're getting upset over something that's not a big deal in the larger picture of the project and the science....dontcha think???




ID: 6324 · Report as offensive     Reply Quote
Profile JigPu

Send message
Joined: 1 Sep 04
Posts: 26
Credit: 600,998
RAC: 0
Message 6325 - Posted: 4 Mar 2005, 0:17:45 UTC

I wouldn't exactly put LHC as "Stable and running well", but it's not like this is some kind of catastrophe. Berekeley is definatly having serious issues at the moment, and the other projects I don't really have much of an interest in. I'd rather donate un-credited CPU time to a project I really enjoy helping than get credit with a project I really don't enjoy. No, I'm not really helping my team blast through the stats this way, but we're doing well enough anyway to overtake a few more spots in the SETI team (our main one), and any contribution that does manage to get credit over here is just gravy for our baby LHC team.

(Note: the views presented in this post may be skewed since I've not had the displeasure of crunching very many zero time WUs [..since I haven't crunched through many LHC WUs period ;)])

Puffy



ID: 6325 · Report as offensive     Reply Quote
STE\/E

Send message
Joined: 2 Sep 04
Posts: 352
Credit: 1,393,150
RAC: 0
Message 6328 - Posted: 4 Mar 2005, 2:28:38 UTC
Last modified: 4 Mar 2005, 2:28:55 UTC

'Ya know, all this talk of dropping LHC@home seems to be a premature no? Sure there's problems, but if you look at the alternatives, most of them are far worse:
------------------------------------------------------------------------------
S@H "Classic" - Restricted to running 8 hours per day because of Building wide power problems....NO ETA

S@H BOINC - Down indefinately because of Building wide power problems...NO ETA

Pirates@home - Project cancelled within the next few months

CPDN - Core Client compatibility issues

LHC@Home - Stable and running well....
-------------------------------------------------------------------------------

What happened to Predictor@Home & Eistein@Home ... ???

ID: 6328 · Report as offensive     Reply Quote
Profile adrianxw

Send message
Joined: 29 Sep 04
Posts: 187
Credit: 705,487
RAC: 0
Message 6334 - Posted: 4 Mar 2005, 11:29:48 UTC

Logan_5, people wouldn't be talking about dropping the project if it was running well. In my opinion and experience, Predictor is probably the project that gives me the least trouble.

What you have to remember is that the average cruncher of wu's is not a scientist, and would not know what a force carrier was unless it was something he could shoot at in his next online game session. They are competitive, not altruistic.

The problem here at LHC is pretty insidious. When SETI goes down, as it often does, you don't get fresh units. You simply crunch those you have, and they'll upload at some future time. Credit will come. Work done = credit given. Here, the equation is work done != credit given. See the difference? People competing for credit across multiple projects are seeing their CPU cycles going to waste, /as far as they and their team are concerned/!!!

You only need look at the web pages run by many of the larger teams to see that altruism is a very long way down the list of attractions, although if questioned of course, they would say the opposite! It's psychology, not plasma that will solve this.

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
ID: 6334 · Report as offensive     Reply Quote
Profile Logan5@SETI.USA
Avatar

Send message
Joined: 30 Sep 04
Posts: 112
Credit: 104,059
RAC: 0
Message 6349 - Posted: 4 Mar 2005, 15:34:23 UTC

Said on the front page of the LHC@Home Website:

4.3.2005 09:20 UTC
We are very concerned and apologize to users for the 'Zero CPU' problem. We suspect a problem with our use of the Boinc API (which has changed recently) but we are short of effort to fix it quickly. <B><I>We considered halting the project but as the results are in fact very valuable we would like to continue, with the support of those users who can accept possible credit errors.</B></I> Thank you all again for your crunching for LHC!


Looks like the project admins agree with me....no?

ID: 6349 · Report as offensive     Reply Quote
Profile adrianxw

Send message
Joined: 29 Sep 04
Posts: 187
Credit: 705,487
RAC: 0
Message 6354 - Posted: 4 Mar 2005, 16:38:04 UTC - in response to Message 6349.  

We are very concerned and apologize to users for the 'Zero CPU' problem. We suspect a problem with our use of the Boinc API (which has changed recently) but we are short of effort to fix it quickly. We considered halting the project but as the results are in fact very valuable we would like to continue, with the support of those users who can accept possible credit
errors. Thank you all again for your crunching for LHC!

>
> Looks like the project admins agree with me....no?
>

No, if they did, why bother acknowledging the problem, apologising for it, and endeavouring to fix it. I would say that they are agreeing with me, they recognise the competitive element for what it is, (a substantial part of their resource), and are actively courting it.

Depends on the way you add the spin.

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
ID: 6354 · Report as offensive     Reply Quote
Toby

Send message
Joined: 1 Sep 04
Posts: 137
Credit: 1,691,526
RAC: 383
Message 6355 - Posted: 4 Mar 2005, 16:45:33 UTC

Logan, as much as I would like to believe most people just "do it for the science", I know this is not the case and I do get tired of people saying that. Credit is a part of the BOINC system. There are a lot of people who are genuinely interested in the science but also enjoy the stats and given a choice of projects, will chose to run the one that has a working credit system. There is nothing wrong with that. They are all doing worthy science. A lot of design effort has gone into the credit system as well as a lot of effort from people such as myself who provide stats websites so the projects don't have to mess with it beyond displaying basic top 1,000 lists. It SHOULD work as advertised or the system is broken.

I myself couldn't care less. I haven't even looked at my results page in several days and am going strong on LHC - I just took the #1 spot on my team! :) LHC makes up 23% of my RAC and is still on the rise. I say let 'em go. That way there is less competition for work units and those who stay advance in the rankings faster :)
- A member of The Knights Who Say NI!
My BOINC stats site
ID: 6355 · Report as offensive     Reply Quote
Profile Logan5@SETI.USA
Avatar

Send message
Joined: 30 Sep 04
Posts: 112
Credit: 104,059
RAC: 0
Message 6374 - Posted: 4 Mar 2005, 22:44:11 UTC
Last modified: 4 Mar 2005, 22:54:52 UTC

Ya, know, it really doesnt matter why ANYONE is involved with Distributed Computing, and when it comes down to it, there are 2 schools of thought on the subject.

There are the people who don't do it for the Science or anything remotely like that. They do it for the stats, the numbers, the glory, seeing themselves ranked "X" in the world....blah...blah...blah. Anything to do with something like scientific research is secondary to their primary goals.

Then there are ALSO people like me who have been doing DC since 1999 <B><I>for the science</B></I> because they truely beleive that their contribution has some merit and meaning toward someones scientific research.

People can beleive me when I say I do this for the science or they can ridicule me for my convictions and call me a geek, propellerhead, nerd...etc..etc I could honestly care less, because the bottom line for me is that *I* think the Science should be the most important aspect of why we do what we do.

IMHO: The science benefits from every completed WU returned, but the peple who actually like to crunch or who actually beleive in the project's goals that they are running seem to make more of a contribution the science then those who do it for other reasons....

And for the record, the project admins DO think that the scientific value of ALL results returned, even the "Zero Credit" results are valuable which is why they decided to keep the project running instead of shutting it down for who knows how long to try and fix it....

So, yea, I think that they DO agree with me by continuing the project for the science, not the glory.

ID: 6374 · Report as offensive     Reply Quote
Profile Logan5@SETI.USA
Avatar

Send message
Joined: 30 Sep 04
Posts: 112
Credit: 104,059
RAC: 0
Message 6376 - Posted: 4 Mar 2005, 23:04:33 UTC - in response to Message 6354.  
Last modified: 4 Mar 2005, 23:05:08 UTC

> No, if they did, why bother acknowledging the problem, apologising for it, and
> endeavouring to fix it. I would say that they are agreeing with me, they
> recognise the competitive element for what it is, (a substantial part of their
> resource), and are actively courting it.
>
> Depends on the way you add the spin.

adrianxw & Toby:

I guess this sums up the differences between our perceptions of this.

Where you see "spin" and competitive elements being emphasized, I see things far differently.

When the Admins have said <B><I>"We considered halting the project but as the results are in fact very valuable (to us) we would like to continue, with the support of those users who can accept possible credit errors."</B></I>, this tells me that they want people dedicated to the Science to continue, and help them by returning results which are only flawed by the credit reporting problems, not the value of the research performed... What does this mean to the both of you??

I hope that this has clarified my position and helps the both of you to understand what it is better then 'yall already do.


ID: 6376 · Report as offensive     Reply Quote
Profile adrianxw

Send message
Joined: 29 Sep 04
Posts: 187
Credit: 705,487
RAC: 0
Message 6377 - Posted: 4 Mar 2005, 23:34:19 UTC

I don't particulaly want to be rude Logan, but you are still not getting the bigger picture.

The sentence you highlight tells me exactly the same as it does you. The time flawed units give the researchers the same info as the non-time flawed ones. They want to get their work done. Of course they do, they wouldn't have invested the time necessary to get the BOINC client built if they did not.

Of course there will be a percentage of crunchers that run the wu's for purely altruistic reasons. The point is, as I made it before, that does not include everyone, I would go as far as to say that such people are a minority. You wish to be included with that minority, fine by me, where are people calling you a geek or ridiculing you? I ask you to substantiate that claim.

By the way, I have been crunching wu's from S@H classic since 1999 as well, (look for cruncher adrianxw under Team Denmark if you'd care to verify that registered 14/7/1999) - being a long serving cruncher does not entitle special privilige.

The scientists that are getting their good science out of BOINC are doing so because they have people prepared to run their clients. If they have a lot of work to do, they will want as many crunchers as possible.

The LHC group are competing for crunchers not only with the other BOINC projects, but also the other distributed/grid systems "out there" in the wild, United Devices et al, some of whom charge for their services, but offer great deals to their crunchers.

For the free donation of CPU time to continue, those that have a competitive streak in them have to be kept on board. The LHC@H people realise this, that is why they are trying to fix it. The current application is, as I understand it, to tune up the superconducting magnets in the development stage of the LHC. I imagine once the thing starts producing results, searching for meaningful particle interactions may become the teams project. Scanning the VAST amounts of data that the detectors from such a machine generates requires colossal computational resources, (look at the output from the older machines at CERN or the Tevatron at FermiLab, or SLAC for examples). Right now, the LHC@H team want to be building a large loyal cruncher base if they intend to utilise this low cost resource to help the next phase of LHC work.

If you cannot see this, then fair enough. You try running a project, and maybe you'll learn.

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
ID: 6377 · Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 2 Sep 04
Posts: 545
Credit: 148,912
RAC: 0
Message 6382 - Posted: 5 Mar 2005, 0:03:30 UTC - in response to Message 6377.  

> Right now, the
> LHC@H team want to be building a large loyal cruncher base if they intend to
> utilise this low cost resource to help the next phase of LHC work.

For me, the key was the fact that as we began to identify possible issues to explore the problem we did get responses from the project team members, and now, the post on the new pane. So, instead of pulling off the project, those machines that had been off are now back on munching away.

I think that the subtle point is often missed is that there is never too much communication to the participant base. Bruce Allen regularly answers posts, not always to the satisfaction of the participants, but that is vital. Markku (I hope I spelled it right) when a Sysadmin here did the same.

The UCB team does try, but fails more than they succeed (In *MY* biased opinion, your mileage may vary) with instances like yesterday when the message on the news pane said they were up when they most certainly not ...

The fact that they are looking, and even an occasional, we looked at "x" and found nothing ... would be of interest to me. Heck, I would like to see some of the WU that failed with 0 time reissued to see if they always get zero time. Even of more interest to me are those where the time is 0 on some returns, but non-zero by other participants. Maybe it is just a random effect, but only an examination of the data in the database can tell us that ...
ID: 6382 · Report as offensive     Reply Quote
Jayargh

Send message
Joined: 24 Oct 04
Posts: 79
Credit: 257,762
RAC: 0
Message 6384 - Posted: 5 Mar 2005, 0:30:34 UTC

Yes Paul and how about adding my original thread to start this off,http://lhcathome.cern.ch/forum_thread.php?id=1226 I challenge all the "whiners" and non-whiners to post thier anomalous workunits instead of bashing admin and make it somewhat easier to perhaps isolate the problems. After all Isn't this what we are all after?
ID: 6384 · Report as offensive     Reply Quote
Profile Logan5@SETI.USA
Avatar

Send message
Joined: 30 Sep 04
Posts: 112
Credit: 104,059
RAC: 0
Message 6386 - Posted: 5 Mar 2005, 0:36:44 UTC - in response to Message 6377.  
Last modified: 5 Mar 2005, 0:54:43 UTC

adrianxw:

OF COURSE I agree with you when you say that running a project such as this from the ground up will be (and is) a daunting task, but the history lesson you provided really isin't applicable for this discussion.

I disagree that the LHC project is only as you say <B><I>"For the free donation of CPU time to continue, those that have a competitive streak in them have to be kept on board. The LHC@H people realise this, that is why they are trying to fix it".</B></I>

If everyone who wasn't in this for the science left, then the project would still benefit (even more so {imo}) by becoming more focused on the science then it can be when they have to pull the project in 2 opposing directions to try and please everyone. Life shows that "you can't please everyone all of the time" so there will always be a small minority of people who are never pleased with the way things are at any given moment in time...such is life and how this is dealt with by the LHC project admins will determine a lot toward the future direction this project takes, as it appears to be working toward a long-term presence.

And as far as people ridiculing me personally for what I beleive in as far as DC, Personally?? never has happened, but if anyone mentions the seti@home project to someone not familliar with the DC concept or folding@home, or einstein@home or pirates@home, you WILL get some odd looks and comments directed toward you. Don't beleive me? next time you're out with your "non distributed computing" friends, mention this or any other DC project and I guarantee you that at least one of them wil laugh at what you say.

Human nature tends to 'classify' everyone and everything, so yea, people hearing geeky sounding things coming from someone will automatically assume (for right or wrong) that they ARE a geek or whatever, and some people have louder mouths then others so words have been in the past and will continue to be in the future used to portray what we all do in a 'less then positive' light....

Oh, I pointed out my Distributed Computing credentials only to show that I'm not some n00b at this and know very well what is at stake for these projects should the glory become more important then the science.

BTW: Feel free to be rude if you think that you need to be to make your point, but it's I that think YOU are not understanding or appreciating my perspective on this.

Thanks
-L5
ID: 6386 · Report as offensive     Reply Quote
Toby

Send message
Joined: 1 Sep 04
Posts: 137
Credit: 1,691,526
RAC: 383
Message 6416 - Posted: 5 Mar 2005, 9:18:59 UTC

Logan: I think we are making more of an issue about this than it is really worth. We all agree that the science is the most important thing here. All I'm saying is that the credit issue is important to a lot of people and it shouldn't be ignored (which it isn't). All I hear you saying is "screw the credits!" over and over. From what I have seen here I think you would have me completely shut down my stats website and put up a "sorry, stats don't matter. If you aren't in it for the science then get out!" message. I suspect that this is a little more extreme than you really are but that is the impression I get. The fact is that I get several hundreds of thousands of hits on my web site every month and I am only one out of 3 or 4 major sites and many more smaller stats sites. Credits ARE important to the BOINC community and as such, any BOINC project that wishes to be a success must deal with them and hand them out on a consistent and fair basis.

People being "more focused on the science" won't make their CPUs crunch faster. The only way to do more science is with more users and getting more users depends on satisfying those who may be motivated more by credit than science. Hmm... isn't that essentially what Adrian said? :) As I said above, I don't really care if a few credits go missing but I also enjoy the competitive aspect that the credit system allows to exist. It drives some people to buy more computers or to recruit friends for their teams, thus accomplishing more science. The science these people do is no less valid than the science that you do.

At the very least I think we can agree that fixing the credit problem will certainly not hurt the project and MIGHT just help it out a little...
- A member of The Knights Who Say NI!
My BOINC stats site
ID: 6416 · Report as offensive     Reply Quote
Profile adrianxw

Send message
Joined: 29 Sep 04
Posts: 187
Credit: 705,487
RAC: 0
Message 6422 - Posted: 5 Mar 2005, 16:08:52 UTC
Last modified: 5 Mar 2005, 16:11:55 UTC

Logan, if everyone that did not have 100% altruistic motivation left the project, the project would continue, but with a lot less computing resources available to it then it currently has or at least has the potential to acheive.

Toby made the point more eloquently than I, but I will echo it. The motivation of the cruncher does not alter the computational power of his computer. My 2.533GHz machine would still be a 2.533GHz machine if you owned it.

To get more science done, you need more crunchers. To get the maximum number of crunchers on board, you need to pander to a broad spectrum of people. From the people such as yourself, through the scientifically educated and well informed, but slightly competitive, (me raises hand), through the less informed but interested, through to the totally ignorant, wildly competitive gamers. Hint, it is the last category that tend to keep the fastest machines....

----
should the glory become more important then the science.
----

That comment raised my eyebrows. I wonder how many scientists have cheated and fabricated results in the past to get their papers published, their ego's boosted, and the possibility of a prize? For every one of the well publicised exposé's that reach the headlines, there will be many that do not. I guarantee it is happening today, and it will still be happening tomorrow. Just being a scientist, does not make you a saint ;)

I do not believe anyone I socialise with would laugh at me if I told them about DC and the various projects. On the contrary, I suspect they would be interested, and may even want me to tell them more about it, or get involved themselves. Of course, if I harked on about it at every dinner party or social event, I would expect to be branded a bore, but that is not the same thing at all. If you feel your "friends" would ridicule you if you "came out" as a DC cruncher, then I would advocate a review of your list of friends.

At the end of the day, you have decided your position, and it seems will not budge from it, so simply repeating the arguments is not going to acheive anything. Do as you wish, that is your right, but allow others theirs as well huh?

*** Edited for an attrocious spelling error!





Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
ID: 6422 · Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : Note to my team


©2024 CERN