1) Questions and Answers : Windows : There are no tasks to do! (Message 22802)
Posted 13 Jun 2011 by Tony DeBari
Post:
The original purpose of this project was to run particle beam simulations to aid in the design and tuning of the LHC. Now that the LHC is up and running, there isn\'t much left for us to do. We will not be crunching the data from actual particle collision experiments here for reasons that have been stated elsewhere in these fora. We may be given additional beam simulations to crunch for future upgrades to the LHC, but that is by no means certain.

Bottom line - the unfortunate reality is that there is simply nothing for this project to do at the moment. Anyone unhappy with their CPU(s) idling while attached to LHC@Home are strongly encouraged to find other projects to contribute to, as LHC@Home tasks are likely to be very few and very far between for the foreseeable future.


-- Tony D.
2) Message boards : Number crunching : We will NOT be crunching the results of the LHC collisions (Message 22801)
Posted 11 Jun 2011 by Tony DeBari
Post:
If the LHC+@Home/Text4Theory@Home team succeeds in providing proof-of-concept for the scientists and can create a viable, well-behaved project for the BOINC user base, there is every expectation that said project will be used to crunch the data from actual LHC collision experiments.

The expectation you speak of is your expectation and yours alone. I can\\\\\\\\\\\\\\\'t find anything on the Test4Theory@home site that indicates anyone associated with CERN in an official capacity shares your expectation. Did I miss something?



From the LHC++@Home home page:

LHC++@Home is based at CERN and uses the CERN-developed CernVM and BoincVM systems to harness volunteer cloud computing power for the CERN Large Hadron Collider (LHC) physics experiments. We will host several sub-projects as we work with different physics groups.

Seems pretty clear to me.


-- Tony D.
3) Message boards : LHC@home Science : Will this task ever get more work units? (Message 22798)
Posted 11 Jun 2011 by Tony DeBari
Post:
I think they\\\'re doing all the good stuff and crunching onsite for more immediate LHC data results, and they just occasionaly throw us a loaf of bread to entertain themselves with digital mayhem, or to see if BOINC is still working.

Maybe when they\\\'re ready to start releasing some serious WU output from the LHC 24/7/365 I\\\'ll be ready with an ATI Firestream or Nvidia Tesla card. So I hope LHC will support both ATI OpenCL and Nvidia CUDA, as well as the main CPU.

Just got my first ever WU today, accidentally killed it with the screen saver, lost it for good, oh well it was just a bread crumb, it bounced off my beak.


Data from the LHC particle collision experiments will probably never make it here. There are too many technical hurdles to overcome in adapting the analysis tools to BOINC, and CERN already has their own grid specifically designed for that kind of work. What we may see is further rounds of SixTrack beam simulations to assist in performing future upgrades to the LHC, much in the same way that the beam simulations we ran previously helped to design and build it. Even so, I wouldn\'t count on GPU support, as the project to date hasn\'t even been able to produce a Linux version of the current SixTrack app - much to the chagrin of many hopeful volunteers.

However, there is a project in alpha right now that hopes to provide a way for the LHC data to be crunched by BOINC hosts through the use of a virtual machine running the same OS and analysis tools as the CERN grid. Check out the LHC++@Home Web site for more info.


-- Tony D.
4) Message boards : Number crunching : We will NOT be crunching the results of the LHC collisions (Message 22797)
Posted 11 Jun 2011 by Tony DeBari
Post:
My computer is number 12 on that project and has been working since November 28 2010.
Tullio


Nobody cares what your computer number is or how long it\'s been working there.


And that was necessary why, exactly?

The data being crunched as part of the Test4Theory@Home alpha test may not be from the LHC proper, but it is live data from actual studies being done by CERN scientists. At present, there are at least three different types of jobs being fed to the volunteer testers, with more to be added as progress on both the virtual machine and the BOINC interface to the hypervisor continue.

The ultimate goal is to demonstrate that the virtual machine running on an arbitrary BOINC host appears sufficiently indistinguishable from CERN\'s internal grid running Scientific Linux as to able to natively run the tools already in place to analyze the LHC collision data. If the LHC+@Home/Text4Theory@Home team succeeds in providing proof-of-concept for the scientists and can create a viable, well-behaved project for the BOINC user base, there is every expectation that said project will be used to crunch the data from actual LHC collision experiments.


-- Tony D.
5) Message boards : Number crunching : Clean up DB please -- old Pending WUs (Message 22703)
Posted 4 Mar 2011 by Tony DeBari
Post:
This WU has been Pending since August of 2010. Please apply credit and/or flush from database.

WU 3927892


I'll see your August 2010 and raise you 3 1/2 years. This is my oldest "pending" WU, from February 2007.

WU 1644322

Unfortunately, cleaning up the database is another item on the staff's PPPPP(*) list...


-- Tony D.

(*) Perpetually Postponed Pending Priority Projects
6) Message boards : Number crunching : SixTrack is not available for your type of computer (Message 22698)
Posted 3 Mar 2011 by Tony DeBari
Post:
Gee. Why not?

(Opteron185, SuSE 11)


At the moment, the only platforms supported by SixTrack are Windows-x86 and Windows-x86_64. There have been repeated requests for a Linux app, but circumstances have thus far not allowed the project to produce one.


-- Tony D.
7) Message boards : Number crunching : LHC+@Home (Message 22687)
Posted 23 Feb 2011 by Tony DeBari
Post:
So how does one go about getting an invite to join LHC++@Home (aka Test4Theory@Home)? I joined LHC+@Home just before it went off the air, and although I never got VirtualBox to work properly on my computer (a Dell laptop running 32-bit Vista), I\'d still like to try to get it running and help the cause if I can.


-- Tony D.
8) Message boards : Number crunching : WU : seconds or minutes (Message 22650)
Posted 20 Dec 2010 by Tony DeBari
Post:
You are right, that could be a reason.
But the server says the crunch is a success.


Beams crashing into the collider\'s wall is a \"successful\" result as far as the project is concerned. They tend to learn as much (if not more) from failed simulations as from successful ones. The only results that will earn no credit are ones in which the crunching process itself fails for some reason; i.e. results marked as something other than \"SUCCESS\".

Anyway, I am happy to crunch a WU after wainting so long. Even for 0,05 credit ; it is better than 0 credit.


So say we all...


-- Tony D.
9) Message boards : Number crunching : Waiting on Linux App (Message 22643)
Posted 15 Dec 2010 by Tony DeBari
Post:
I think the reason that not many people get WUs is that as soon as they are pushed to the server there are all these systems setting around just waiting to be fed and the resulting feeding frenzy is like piranha attacking a cow that was dumb enough to step into the river.


Well, the Linux people don\'t get any WUs because the project doesn\'t have a Linux app with which to crunch them. Hence the reason for this thread. As for the rest of us, it\'s a matter of hitting the server at the precise moment that work is available. In other words, pure luck of the draw.


-- Tony D.
10) Message boards : Number crunching : 38.73 / 91.77 pending credits on a work unit - why just me? (Message 22607)
Posted 15 Nov 2010 by Tony DeBari
Post:
I\'m not sure anything can be done about old tasks with the project in its current state. I have hundreds of tasks dating back to July 2005 spread across 6 computers that for one reason or another are stuck in a completed or pending state. As we have been told, updating the server-side code and cleaning out the database are on the \"To Do\" list, but they\'re not very high priority at the moment.


-- Tony D.
11) Message boards : Number crunching : Computations without end (BUG?) (Message 22512)
Posted 1 Sep 2010 by Tony DeBari
Post:
Thanks for the reply Tony. Since I have upgraded to latest version and this time installed using the special rights option for UAC hoping the no-end problem would be due to unability to write a state file or something similar. As it turns out I can now receive and correctly return results! I do not know if anything else has changed, but Vista/Win7 users might wan\\\'t to reinstall BOINC granting it full rights for lhc@home to work.


You\'re welcome! I\'m glad to hear you got it straightened out. Good point about the User Account Control (UAC) feature in Vista/Win7. I never though to play around with the UAC settings on my Vista laptop. There\'s no UAC in WinXP, so I\'m still at a loss to explain why my other PC is having this issue. Still, as neither of those computers are CUDA-capable, I can live with BOINC 5.10.45 (replaced 5.8.16) for a little while longer.

I should also point out that the BOINC data directory I gave in my other post applies to WinXP only. On Vista/Win7 machines, the corresponding folder is C:\\\\ProgramData\\\\BOINC. Apologies for the omission.


-- Tony D.
12) Message boards : Number crunching : Confused on WU error (Message 22483)
Posted 17 Aug 2010 by Tony DeBari
Post:
I have several these waiting to crunch on a couple of my machines as well. All have been returned by my wingmen with the same error and continue to be re-issued, probably until the max error count of 10 is reached. Looks like it\'s a totally chowdered batch of work units.

That ::thud:: you hear is my daily WU quota dropping like a stone... :-)


-- Tony D.
13) Message boards : Cafe LHC : There is definitely new work at lhc@home (Message 22463)
Posted 10 Aug 2010 by Tony DeBari
Post:
All that is well and good, but the reality is that at the current time, work unit demand still greatly exceeds supply. Put another way, the lack of work is not due to the small user base; rather, the small user base is due to the lack of work. Some work is now flowing, yes, but still not nearly enough to keep the existing crunchers running continuously for any significant length of time. The project even caps the daily download quota at 10 wu/CPU to try to achieve a more equitable distribution of work among the relatively few users that are still here. Adding more users through some sort of awareness-raising campaign won\\\'t appreciably increase the amount of work done and will very likely result in many more complaints of \\\"Why don\\\'t I ever get any work?\\\"

Until LHC@Home can supply the 24/7/365 workflow of the other projects you listed, keeping the user base relatively small does more good than harm, IMHO.


-- Tony D.
14) Message boards : Number crunching : Computations without end (BUG?) (Message 22442)
Posted 6 Aug 2010 by Tony DeBari
Post:
Update...

I did a little digging and found this thread in the Sixtrack topic of the Questions and Answers forum. No explanation is given as to why, but apparently some computers have a problem running the SixTrack app under BOINC 6.x.x. (The thread refers to SixTrack 4.67, but clearly the issue persists in the current version 4209.00.) The solution is to downgrade to a 5.x.x version of the BOINC client. The one I found on the download page is 5.8.16, which did in fact solve the problem on my Athlon MP computer. The HP is at my parents\' house; I\'ll try the downgrade on it next time I\'m over there.

To anyone else who wishes to try this: Remember that BOINC 6.x.x introduced a new directory structure under Windows (don\\\'t know about the other OSes). When you downgrade back to a 5.x.x, you will have to copy the contents of C:\\Documents and Settings\\All Users\\Application Data\\BOINC into C:\\Program Files\\BOINC so the older core client will be able to find all the necessary files. Oh, and you should back up both BOINC folders and uninstall your 6.x.x client before installing 5.6.18. I\'m not sure what would happen if one were to install a 5.x.x client over top of a 6.x.x client. You\'ll give up some features - most notably GPU support - but it\'s either that or detach from LHC@Home until this mysterious bug in Sixtrack is fixed.


-- Tony D.
15) Message boards : Number crunching : Computations without end (BUG?) (Message 22440)
Posted 5 Aug 2010 by Tony DeBari
Post:
I am seeing the same behavior on two of my machines. One is a relatively old dual Athlon MP box running Windows XP, and the other is a fairly new HP notebook with a Core2 Duo running Vista. They have literally nothing in common except that every LHC task they have run in the last year or so seems to complete (or terminate early in the normal manner) but never actually exit and upload. CPU use drops to 0% and the task remains in the \\\"Running\\\" state indefinitely until aborted. I have tried everything I can think of to correct this, but so far nothing has worked.

For comparison, this is an even older dual Pentium III box that runs without a hitch. Any thoughts or suggestions would be greatly appreciated.

-- Tony D.
16) Message boards : Number crunching : 10 second work unit (Strange Quirks - Not Quarks) (Message 22437)
Posted 4 Aug 2010 by Tony DeBari
Post:
After many weeks of not receiving any work units I was pleased to see a new work unit this morning. The work unit was listed as an 8 hour task... But after only 10 seconds the unit terminated and is listed as 100 percent complete.

(snip)


This will happen from time to time - there is nothing wrong with either your setup or the LHC science app. The way I understand it, this occurs when the virtual beams do something they aren\'t supposed to do - like hit the wall of the virtual collider - and the simulation ends early. The task should report normally and will be granted a tiny amount of credit if it validates successfully against that of your wingman.


-- Tony D.
17) Message boards : Number crunching : Lengthy WU’s (Message 22231)
Posted 15 Apr 2010 by Tony DeBari
Post:
I have had some WU’s with a very long running time. Now five in total, two done with results and two with compute error after a long run. Number five is crunching now and has done 31% in approx. 12 hours. Still 35 hours to go.
Is it wise to let this one run out or will it result in an error too?



If it looks like it will complete before the deadline, I would let it run. Work is still pretty scarce, and the worst thing that will happen is that it will error out and be reissued to another machine.


-- Tony D.

[Edit - fixed typos...]
18) Message boards : Number crunching : When can we expect the next batch of wu's? (Message 14503)
Posted 8 Aug 2006 by Tony DeBari
Post:
I must have blinked and missed it, but one of my computers got some work on the 5th. 20 work units with a total granted credit of 0.2 cobblestones. Yes, that's a total of 0.2 - each WU claimed 0.02 and was granted 0.01. Must have been those super extra short WUs...

-- Tony, thinking he should go out and buy a MegaMillions ticket tomorrow...
19) Message boards : Number crunching : How long will the next batch last? (Message 14453)
Posted 27 Jul 2006 by Tony DeBari
Post:
<deleted>
20) Message boards : Number crunching : How long will the next batch last? (Message 14452)
Posted 27 Jul 2006 by Tony DeBari
Post:
Is it possible to delete own posts?

You can't delete a post, but you can edit it to remove what you originally typed, thereby leaving an empty post in the thread. However, when you do this, it is customary to leave some indication that you in fact deleted your original post. See my next post for an example.

Regards,

-- Tony

[edit - increased size of quoted text to make it more legible.]


Next 20


©2022 CERN