| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Community
Forum: Chat Room
Thread: why does it run at high priority? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 20
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
cannot find preferences as on other projects
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It doesn't. World Community Grid software runs at the lowest possible priority.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
He probably means in Boinc under Tasks status that he has WU's showing 'Running - High Priority' rather than the software I suspect
---------------------------------------- [Edit 1 times, last edit by Former Member at Mar 4, 2008 6:55:09 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
...most of my DDDT's on one particular machine do that.......
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
@Scribe,
----------------------------------------Means you probably have a discrepancy between the the buffer size, actual crunch time and the real duration between what the client thought it would take and the actual durations. Or, you get allot of repair work with just about 2 days deadline. If it does not bother you, let it run. ciao @jwmark And for the priority in TaskManager, it's pointless to meddle with that as the code internally ensures it will continue to run at lowest priority. Net gain per job, 1 second (best you can hope for). There's no option in BOINC but to set the resource share or project weight (different name here) to allocate more/less time to WCG. Setting in the My Grid > Device Manager > Device profiles ttyl
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges: 
|
...most of my DDDT's on one particular machine do that....... Would it be a multicore machine, by chance? If I raise the extra buffer to 4 days of work (that was my most recent test experiment) my Pentium is OK but my quad "falls" in high-priority mode like you. As Sekerob says there is a discrepancy but I would place it inside Boinc itself. From long watching times it appears that Boinc takes the 4 cores in account when it requests work (I see a large number of seconds in its request) but it forgets about the four cores when it estimates if it will be able to process the queue in time. When Boinc starts to "panic" and stops renewing work requests (it is right doing so if it thinks it has too much, no concern with that) I have about 1.5 day of work for the four cores, i.e. 6 days of work for a single core for a 7-day deadline, hence the emergency mode. And if a WU finishes with a run time higher than the average which was in use while issuing the requests for work, then the emergency mode is in effect for a while. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
My quad goes indeed in/out HP when set on a larger buffer. It's driven also in part because it recomputes the amount of work in queue after each completed job. Have a longer than predicted task and the total time estimate inflates rapidly. Opposed, short jobs it only comes down slowly. All part of the protective mechanism, thus postpones fetching for a while.
----------------------------------------Noticed a few days ago that the maximum WUs per fetch has reduced from 10 to 6. After that the client scheduler recomputes the buffer size and sees if enough is there or more is needed to back-fill. With the changes of work size, here come the longer DDDTs, we'll be experiencing another round of calibration and battle between the servers and clients that are fetching on basis of what's just done and still in queue. Got 35 of version 5.15 sitting there at 55 minute estimate, but know they will take 4-5 hours.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
....single core only....
 |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges:
|
....single core only.... Well, next try... What is your work buffer length (network connection + extra work)? Have you begun to crunch the new larger DDDT WUs? What I have in mind is something like a job queue of 1.5 day or more and suddenly the first large DDDT WU finishes and, since the techs seem to have left them with the same duration estimate, your job queue suddenly happens to be 5 times longer than expected. Or if you had the problem long before the new larger WUs, that could happen if a very long WU follows a series of very short ones, just after your buffer was refilled. As far as I am concerned this experiment is finished: I have set my preferences back to their usual 0 + 0.03 settings. I just have to wait for the queues to clear, slowly on the Pentium, faster on the quad. Cheers. Jean. |
||
|
|
kyoq
Cruncher Joined: Apr 27, 2007 Post Count: 9 Status: Offline Project Badges: 
|
Hello!
I've a similiar problem, I havn't found anything like an answer on the forum. I have AMD X2 machine. Normally it would crunch 2 tasks in the same time. And it did. But since last, well i don't know, lets say - thursday, only one task is running. I've completed few task in this time. Always BOINC said "Running, High priority", the work unit was valid for minimum 10 days, completion time was always wrong (eg. 08:-8:-8, or 124:53:23) and almost completely frozen. I really don't care about the completion time but one of the reasons I've bought a dual core machine was to run WCG faster. Two times faster. Best regards Adam ps: sorry for my terrible english  ps2: i don't exactly know the correct names of columns in boinc client because i run polish version. |
||
|
|
|