| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project - Phase 2 Forum
Thread: New Project Setting |
| No member browsing this thread |
|
Thread Status: Active Thread Type: Sticky Thread Total posts in this thread: 124
|
|
| Author |
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
This download bandwidth issue just happened to me, too...and the problem with devoting multiple boxes to WCG is you (or I, at least) don't necessarily notice when the cached tasks are drained dry...so the system sits there idle - and then when you do start back up after editing the client_state.xml, every CEP2 task's disk I/O is practically synchronous, so efficiency takes a hit on top of the time wasted idle. Arrrggggghhhhh....this time-waster, on top of one of my SVN repositories deciding there are two of me working on the same project (although that might just be electrons conforming to client desires for me to work twice as fast). We have to reimplemented the bandwidth override feature in the new server code. We are still fighting some fires related to the upgrade (fortunately they are getting smaller now) and this is on the short list to get reimplemented. Sorry for the delays. |
||
|
|
Richard Mitnick
Veteran Cruncher USA Joined: Feb 28, 2007 Post Count: 583 Status: Offline Project Badges: 
|
Re: the request at FB for reports, What I have is when I go to the result page, and filter for CEP2, I see very little except pending, a lot of results pending. One or two valid, one inconclusive.
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks for the update Kevin! Hope things get into calmer water for the WCG team soon. Take care, many thanks, and best wishes
Your Harvard CEP team |
||
|
|
sgoll
Advanced Cruncher Joined: Oct 24, 2006 Post Count: 87 Status: Offline Project Badges: 
|
Well, right now I see a lot of workunits will finish hart after 12 hours of computing time. I think there is a (good) reason fpr that setting ... but on a core i520, a c2d@1866 and a celeron E3400 (my boxes) this setting is short before the regular ending.
----------------------------------------I do not know what influence this hard limit has to the results, but it seems that that WU are in the range of 10 hours computing time. I'm curious: why not raising the limit or making smaller WUs? When (if?) the computing time raises with the remaining WUs this issue will hit more and more cruinchers. Can the project science advisor leave some words about this thoughts? Thanks, Stephan  |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7680 Status: Offline Project Badges: 
|
Well, right now I see a lot of workunits will finish hart after 12 hours of computing time. I think there is a (good) reason fpr that setting ... but on a core i520, a c2d@1866 and a celeron E3400 (my boxes) this setting is short before the regular ending. I do not know what influence this hard limit has to the results, but it seems that that WU are in the range of 10 hours computing time. I'm curious: why not raising the limit or making smaller WUs? When (if?) the computing time raises with the remaining WUs this issue will hit more and more cruinchers. Can the project science advisor leave some words about this thoughts? Thanks, Stephan I believe this has been answered elsewhere, but here goes. The 12 hour limit was put in basically for everyone's peace of mind about long running jobs, especially if you have a slow running older system. The scientists can use the results of the WU even if after 12 hours it has only completed some but not all of the subtasks. The size of the work unit is dependent on the size of the molecule being analyzed. Also from an answer from one of scientists a long time ago, the software - Q-Chem is complicated and they have taken pains to size the units so they can be done well enough by most of the available resources on the grid. When the project started there was quite a discussion about letting the WU run to completion if the person elected to do so, but so far the hard 12 hour limit has stayed. See:http://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,30547_offset,0 The answer will be be in Cleanenergy's post, #14 from the top. Cheers
Sgt. Joe
----------------------------------------*Minnesota Crunchers* [Edit 1 times, last edit by Sgt.Joe at May 18, 2012 1:36:26 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The hard limit was to be extended to 24 hours IIRC. Limitless would I think not go, as a 24 hour job, if stuck with a factor 10 leeway [standard] would run 240 hours before it goes south. Accordingly the output file would likely become monstrous. We'll see when the techs get to work on this, and Beta test.
--//-- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Yes, our friends at IBM had been working on this but I think they work got interrupted with new projects coming up.
Best wishes Your Harvard CEP team |
||
|
|
tekennelly
Cruncher Joined: Oct 10, 2005 Post Count: 45 Status: Offline Project Badges:
|
I am running 3 to 4 CEP2 process right now at 100% without much impact. Note that the Boinc data directory is on a 4 GB ram disk. Also, I am using Win 7 and an Intel I7 with 16 GB of memory (minus 4 for the ram disk).
Rather than reinstalling BOINC, so that I could place the BOINC folder on the ram disk, I just copied the entire BOINC folder to the ram disk and then used a symbolic link from the hard drive where the BOINC folder originally existed and pointed it to the ram disk. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi tekennelly,
Thanks for sharing your experience. Best wishes from Your Harvard CEP team |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I am noticing that some of my jobs are extending past the 12 hour line, some are as much as 18 to 20 hours to complete. Note this is NOT a problem, from what I understand, the longer it runs, the closer to 'completion' for that task and the better the data is. I do not mind at all letting them run longer, and am curious is this going to be kept this way with the longer potential run times?
THank you Aaron |
||
|
|
|