| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Uncovering Genome Mysteries
Thread: UGM - priority too low? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 9
|
|
| Author |
|
|
katoda
Senior Cruncher Poland Joined: Apr 28, 2007 Post Count: 170 Status: Offline Project Badges: 
|
Hello,
----------------------------------------recently I switched from UGM + CEP2 to OET1 + CEP2 processing. As expected, I started to receive "no tasks available" message from OET1 and then the system sends me workunits from other projects - but only from MCM and FAAH. I've read somewhere that UGM has lower priority than other projects, but is the priority so low that the system does not send UGM workunits at all to the users without UGM project ticked?  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
As I understand it, the system averages over some time so, as you have received UGM units in the recent past, I suspect that it thinks that you've had your quota for a while. I'm sure you'll start to get them again eventually.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
No, UGM1 is run consciously 'under-balance', at low feeder priority, and will be for the full duration of the project. If you want more, the only way to achieve is to make an exclusive selection in the device profile(s). See this forum for posts by uplinger for background [storage limitations]
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Rob, I'm confused. Are you saying No to me? Are you agreeing with katoda that UGM's priority setting is "so low that the system does not send UGM workunits AT ALL [my emphasis] to the users without UGM project ticked" when their selected project(s) have no WUs?
I understood that there was proportionality, not "pick from the top of the priority list". Please explain. Thanks! |
||
|
|
KWSN-A Shrubbery
Senior Cruncher Joined: Jan 8, 2006 Post Count: 476 Status: Offline Project Badges: 
|
If you download 150 tasks at the same time you may get 1 or 2 ugm units.
----------------------------------------That is what Rob is telling you. If you want more, you must select only ugm. The few tasks going out there are going to the machines that have only ugm selected. The grid does not care if a certain machine or user gets a proportional amount of work, the overall output will balance according to priorities set on the server.  |
||
|
|
katoda
Senior Cruncher Poland Joined: Apr 28, 2007 Post Count: 170 Status: Offline Project Badges:
|
Thanks for all the explanations :)
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
'Not the answer you were looking for', but efforts are made to increase the UGM1 feeder share. See http://www.worldcommunitygrid.org/forums/wcg/viewpostinthread?post=482995 . Did notice a slight increase in numbers for this project, but still small on a MCM1/UGM1/CEP2 mix machine.
Rather than using 'mix' profiles on the 2 octos, set one to do MCM1+CEP2 and the other UGM1+CEP2. This delivers artificially a 50/50 share between MCM1 and UGM1. Alternately, periodically flipping selections achieves the same or using app_config on one machine, but that is of the dirty trick department. This method still requiring periodic science selection flipping to backfill the UGM1 buffer share [for those insisting on equal share ;0]. (Sure an opened can to start more price shooting, but technical limits are technical limits ;o) |
||
|
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12398 Status: Offline Project Badges:
|
In order to get Ebola units whenever they come back on line, I have been changing my device profile every few days. I have OET1 only set for a few days while crunching my cache and it checks for OET1 whenever it submits a return.
When I am about to run out of work, I add MCM1 and UGM1 to the profile. About 180 units download of which usually about 3 are UGM1. Today I got none! I then change the profile back to OET1 only and so on....... If you really want to boost your UGM1 output, you could do the same but using UGM1 instead of OET1. But bear in mind that you would be cutting other people's share. Mike |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Few days ago uplinger noted he was tweaking the weight to the point where there was no backlog of uploads to scientists developing. Observed since about 1AM in the buffer that the balance of fetched work is in 'quantitative' terms rather equal. Received 8 MCM1 and 8 UGM1. Former run 8.3 hours, latter run 5.7 hours average on my host, so set app_config to do 3 for UGM, 5 for MCM and 1 for CEP, yes a small core overlap with a 1.3 day buffer depth. Hopefully don't have to do much MM to keep all cores busy and not deplete one or the other.
|
||
|
|
|