Message boards : LHC@home Science : Why only one result
Message board moderation

To post messages, you must log in.

AuthorMessage
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18822 - Posted: 9 Jan 2008, 22:17:06 UTC

I was very excited to get some tasks from LHC@home. I am running the project on two superb Core 2 Duo machines, and I saw tasks on both.

Yet, I only have one result.

I do not understand why other tasks failed.

>>RSM
Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18822 · Report as offensive     Reply Quote
otij

Send message
Joined: 15 Mar 07
Posts: 11
Credit: 1,038
RAC: 0
Message 18824 - Posted: 10 Jan 2008, 6:02:32 UTC

They not reward for result that "crash into the walls" ;
also you probably checked late
you had more then one result but all "failed".
you put more then 1.5 seconds cpu time on your side for sure
but you not going to be rewarded for it.
Even if it is not
looking fair , live with it like me for now until
a project admin hear about it :(

check other thread about overcloking also
ID: 18824 · Report as offensive     Reply Quote
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18825 - Posted: 10 Jan 2008, 11:18:14 UTC - in response to Message 18824.  

They not reward for result that "crash into the walls" ;
also you probably checked late
you had more then one result but all "failed".
you put more then 1.5 seconds cpu time on your side for sure
but you not going to be rewarded for it.
Even if it is not
looking fair , live with it like me for now until
a project admin hear about it :(

check other thread about overcloking also


QUESTIONS
What means "crash into walls"?

I saw all of the failures in results, that is why I posted my question.

What means " you put more then 1.5 seconds cpu time on your side for sure
but you not going to be rewarded for it."

I do not have any overclocking on my machines.

>>RSM

Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18825 · Report as offensive     Reply Quote
Chaz

Send message
Joined: 18 Feb 06
Posts: 8
Credit: 13,541
RAC: 0
Message 18833 - Posted: 14 Jan 2008, 0:00:08 UTC

Crash into the walls refers to those runs on sixtrack that finish extremely quickly because the simulated magnet configuration leads to a particle stream that is unstable or misdirected and the particles crash into the walls of the loop rather than spinning around in a coherent stream. I am a lay person so there maybe someone who can give a more technical explanation but this is the gist of it.

I crunch on a dual core with no problems but these very short "crash and burn" runs sometimes don't crunch for long enough to gain credits. However, its v. useful for the scientists to know that certain set ups bring disasterous results.
ID: 18833 · Report as offensive     Reply Quote
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18834 - Posted: 14 Jan 2008, 1:18:05 UTC

O.K., I understand now, that the crashing has not to do with my crunching, but the project design of the task.

So, I am sticking with it. I am interested in C.E.R.N. since 1991, when I watched a film, "Creation of the Universe", narrated by Timothy Ferris. It visited C.E.R.N. and FERMI labs and I was hooked.

So, lets get somehthing going.

>>RSM
Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18834 · Report as offensive     Reply Quote
otij

Send message
Joined: 15 Mar 07
Posts: 11
Credit: 1,038
RAC: 0
Message 18836 - Posted: 14 Jan 2008, 12:45:41 UTC - in response to Message 18825.  

because your computer will put for exemple
one hour to compute the unit , but
when you check on the website
it says 0.0 seonds to a few seconds,
with zero reward.


QUESTIONS
What means "crash into walls"?

I saw all of the failures in results, that is why I posted my question.

What means " you put more then 1.5 seconds cpu time on your side for sure
but you not going to be rewarded for it."

I do not have any overclocking on my machines.

>>RSM
[/quote]
ID: 18836 · Report as offensive     Reply Quote
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18849 - Posted: 18 Jan 2008, 22:03:37 UTC - in response to Message 18836.  

I think I learned about "crashing into walls".

In the LHC forum at allprojectstats.com, John Hunt gave a link to a bunch of YouTube videos about the LHC at CERN.

http://www.youtube.com/watch?v=fJPMfnz2E

In one video, produced at Google, a woman from Berkeley is giving a lecture about LHC. She says that in a New Yorker article about LHC looking for Higgs, it will be like "finding a needle in a needle factory."

I searched on the phrase with only the domain newyorker.com and came up with it.

http://www.newyorker.com/reporting/2007/05/14/070514fa_fact_kolbert?currentPage=all

The author says that it takes 1800 times the energy to push as proton as it does an electron or positron, and that all particles want to go straight, so I think that very simply tells us why we are not getting a good result when the task experiences this result of "crash into walls."

I recommend the videos and the article.

>>RSM

Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18849 · Report as offensive     Reply Quote
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18850 - Posted: 18 Jan 2008, 22:09:12 UTC - in response to Message 18849.  

I think I learned about "crashing into walls".

In the LHC forum at allprojectstats.com, John Hunt gave a link to a bunch of YouTube videos about the LHC at CERN.

http://www.youtube.com/watch?v=fJPMfnz2E

In one video, produced at Google, a woman from Berkeley is giving a lecture about LHC. She says that in a New Yorker article about LHC looking for Higgs, it will be like "finding a needle in a needle factory."

I searched on the phrase with only the domain newyorker.com and came up with it.

http://www.newyorker.com/reporting/2007/05/14/070514fa_fact_kolbert?currentPage=all

The author says that it takes 1800 times the energy to push as proton as it does an electron or positron, and that all particles want to go straight, so I think that very simply tells us why we are not getting a good result when the task experiences this result of "crash into walls."

I recommend the videos and the article.

>>RSM


Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18850 · Report as offensive     Reply Quote
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18851 - Posted: 18 Jan 2008, 22:10:51 UTC - in response to Message 18850.  

I missed erred in the URL for the videos, left out a character.

Should be

http://www.youtube.com/watch?v=_fJPMfnz2E

Sorry.

>>RSM

Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18851 · Report as offensive     Reply Quote
Neal Watkins
Avatar

Send message
Joined: 14 Nov 05
Posts: 32
Credit: 712,132
RAC: 60
Message 18856 - Posted: 20 Jan 2008, 23:57:49 UTC

It is disheartening to wait for weeks, only to get 2 work units that immediately crash into the walls (processing time < 2 sec) and then
find that there are no more WU's to distribute. I guess I'll wait
another month to (maybe) get some work....
ID: 18856 · Report as offensive     Reply Quote
Neal Watkins
Avatar

Send message
Joined: 14 Nov 05
Posts: 32
Credit: 712,132
RAC: 60
Message 18858 - Posted: 21 Jan 2008, 0:43:30 UTC

Why not start the credit calculations at ...say...50, instead of zero. That way you get some credit for running one that crashes into the walls. Afterall, that's good to know....right?
It's not like we can do anything with the credits. I can't cash them in for Linden dollars in Second Life can I?
I would be nice to look at the stats tab and see at least a little bump every couple of weeks.
ID: 18858 · Report as offensive     Reply Quote
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18860 - Posted: 21 Jan 2008, 2:32:25 UTC - in response to Message 18858.  

Why not start the credit calculations at ...say...50, instead of zero. That way you get some credit for running one that crashes into the walls. Afterall, that's good to know....right?
It's not like we can do anything with the credits. I can't cash them in for Linden dollars in Second Life can I?
I would be nice to look at the stats tab and see at least a little bump every couple of weeks.


I understand the need to volunteer even if at first there are all of the failures. If you believe in the LHC as a scientific endeavor, then whatever happens with our tasks is a help.

I personally am fascinated by the work at C.E.R.N. since 1985. So, I am delighted to be on this project.

If you go to http://www.allprojectstats.com and visit the LHC forum, there are links to videos about C.E.R.N. and LHC, even Dr Higgs himself.

>>RSM

Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18860 · Report as offensive     Reply Quote
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18865 - Posted: 21 Jan 2008, 13:58:11 UTC

Hey-

Things are looking up.

I now have 4 successes out of seven tasks.

>>RSM
Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18865 · Report as offensive     Reply Quote
Home

Send message
Joined: 4 Feb 07
Posts: 1
Credit: 1,715
RAC: 0
Message 18875 - Posted: 24 Jan 2008, 17:10:37 UTC

Hey, I was wondering if it wouldn't be better if a computer first tries the WU's. I mean that it runs a WU for say 2-3 seconds and if it doesn't crash into the walls then it is send to the volunteers on the grid.
I think it would save a lot of bandwidth, the waiting for WU's to complete (because they are in a wait list on one of the volunteers' computers), and maybe even more trouble(?)...
If course that pc would put a delay on the final processing, one should test if it would make a profit in time/bandwidth
ID: 18875 · Report as offensive     Reply Quote
Richard Mitnick
Avatar

Send message
Joined: 20 Dec 07
Posts: 69
Credit: 599,151
RAC: 0
Message 18876 - Posted: 24 Jan 2008, 17:37:33 UTC

O.K., I now have had two successful tasks on each of my two computers.
I have three failures, possibly from the "crashing into walls" business.

What I would like to know is, when these simulations run as tasks via BOINC, and when they fail, are we testing the simulation process in BOINC, which is certainly fine with me, or, are we testing simulations of tasks that will actually go into the LHC?

>>RSM
Please check out my blog
http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings
ID: 18876 · Report as offensive     Reply Quote

Message boards : LHC@home Science : Why only one result


©2024 CERN