| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: Mapping Cancer Markers Forum
Thread: Much Longer WUs |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 61
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Yep -- I was being overly optimistic when I said it was just the one batch. After the first task of the next batch went through only 50% over, it has gone back to 20-30x estimate. I may have to abort tasks and shorten the cache again. Sounds like you are running too long of a cache. Outages are rare and when they do occur they are not typically all that long. I run a very short cache and have over 38 years on MCM1 alone, WU's completed is over 145,000. I have very rarely had isle systems and the swing of WU run times do not impact me at all. |
||
|
|
Dayle Diamond
Senior Cruncher Joined: Jan 31, 2013 Post Count: 452 Status: Offline Project Badges: 
|
Looks like the tasks are going really long again. I've got several around the ten hour mark on my fastest rig.
|
||
|
|
KLiK
Master Cruncher Croatia Joined: Nov 13, 2006 Post Count: 3108 Status: Offline Project Badges: 
|
OK...just let them crunch...click & forget! ;)
---------------------------------------- |
||
|
|
Barnsley_Tatts
Senior Cruncher Joined: Nov 3, 2005 Post Count: 281 Status: Offline Project Badges: 
|
Some more monster WUs about.
----------------------------------------MCM1_0010378_2585 completed in 7.43 hours! Thats on an i7 rig.  |
||
|
|
vlado101
Senior Cruncher Joined: Jul 23, 2013 Post Count: 226 Status: Offline Project Badges: 
|
Some more monster WUs about. MCM1_0010378_2585 completed in 7.43 hours! Thats on an i7 rig. I have actually started to get some that are close to 9 to 14 hours. This one below for example had been running for 10 hours. MCM1_0010371_5065  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Great!
|
||
|
|
rbotterb
Senior Cruncher United States Joined: Jul 21, 2005 Post Count: 401 Status: Offline Project Badges:
|
Yep, just got some on my i7 that are now running 19+ hours. I'm really piling up the run time on these puppies. At this pace I'll hit sapphire sooner in 1Q2015 than I thought.....
|
||
|
|
terianne929
Cruncher USA Joined: Sep 3, 2011 Post Count: 23 Status: Offline Project Badges: 
|
The work units I downloaded had an estimated run time of 2:24:40, all 30 of them. BUT the first 4 that ran took over 5 hours (one took almost 6). To me, it seemed that if the runtime was underestimated by at least half what it would take to actually run them, 10 of them would be past the due date, waiting to run. Not a big deal when only one computer is assigned a WU, but when someone has to wait for credit because of a past due 'co-cruncher', I can see where that would be annoying.. My computer is NOT an old machine but does have a limitation of only 2 processors. I let the ones running finish and aborted the rest, hoping someone with a faster computer could run them. When I looked at the status of a few I aborted, I noticed at least one 'co-cruncher's' WU took over SEVEN hours - 3 times longer than the estimated run time of 2:24:40...
This didn't seem to be a problem with the first 1100+ WUs I ran for MCM - they completed as estimated or sooner. My BOINC manager is set to download up to 10 days work, but what I got in that 'batch' would have taken longer to run than the due date permitted. If this is a glitch, let us know when it is rectified. Until then I'll run the other projects.   |
||
|
|
deltavee
Ace Cruncher Texas Hill Country Joined: Nov 17, 2004 Post Count: 4891 Status: Offline Project Badges: 
|
My BOINC manager is set to download up to 10 days work, but what I got in that 'batch' would have taken longer to run than the due date permitted. If this is a glitch, let us know when it is rectified. Until then I'll run the other projects. terianne929, Thank you for being concerned about the impact to your fellow crunchers. Most who run with ten days of downloads are oblivious to what they are doing. Most crunchers rectify this by modifying their settings to download something less than ten days. Try 2 days and see if it fixes this. Almost everyone running with a ten day setting is going to have this problem. Also, it doesn't really matter in the long run. All workunits will get crunched and that is all that really matters. After a while we get used to the "No Reply" workunits and realize it's just part of the deal. [Edit 2 times, last edit by deltavee at Mar 31, 2015 10:54:17 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
My 'technical" recommendation is to never set a buffer greater than half of the shortest standard deadline, which happens to be MCM with 7 days. This means that any setting above 3.5 days can induce major client scheduling panic. Particular the high variability of runtime [all projects have non-deterministic algorithms], is the root cause of the frequent high priority / earliest deadline first task switching with any larger buffer setting. With the very high uptime of WCG, anything over 1 day is not needed.
----------------------------------------FYI: Well over 90 percent of work is returned within 4 days, speak gets validated if a wingman is needed. [Edit 1 times, last edit by Former Member at Mar 31, 2015 10:59:04 AM] |
||
|
|
|