| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Suggestions / Feedback
Thread: How are work units prioritised? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 6
|
|
| Author |
|
|
platinumjsi
Cruncher Joined: Aug 12, 2010 Post Count: 37 Status: Offline Project Badges: 
|
Sorry if this is in the wrong forum, couldn't see anywhere suitable as its a general question rather than relating to a specific project.
----------------------------------------So I have all the projects selected for my machines but the distribution of work I get is like this : about 50% of all units are mapping cancer markers About 30% are fight aids at home The other 20 % is a mix of Genome and Clean energy and never received a ebola unit Now I understand why clean energy is low in the unit counts due to the requirements but I find it odd that I rarely get Genome units?, as a test I set one of my machines to only get Genome and Ebola and its been happily crunching and downloading more Genome units all day so no shortage?, Ebola wise I understand there are no units at the moment but still odd I have never had a unit? Anyhow like I say I was just wondering how they were prioritised, I would have thought if the units were there I would get a fair spread?  |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
The work units have a few things that cause priorities to be set at various levels. For one, Ebola (OET1) does not have work units available at the moment while we fix up checkpointing. Uncovering Genome Mysteries (UGM) is running slow because after results are completed we need to send them back to the researchers. At the moment we are unable to send results fast enough back to them to clear them off our storage on our servers. As for FightAIDS@Home (FAAH) and Mapping Cancer Markers (MCM) these are running at normal priority and get leveled off based off total runtime for the projects. I have not reset these values in a bit and may need me to change them to make them as level as possible.
Thanks, -Uplinger |
||
|
|
platinumjsi
Cruncher Joined: Aug 12, 2010 Post Count: 37 Status: Offline Project Badges:
|
Ahh right thanks for clearing that up, ok in that case the machine that I have set to just do Genome and Ebola I will put back on everything, no sense in adding to the backlog if I can get work done that is useful right now :)
---------------------------------------- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The work units have a few things that cause priorities to be set at various levels. For one, Ebola (OET1) does not have work units available at the moment while we fix up checkpointing. Uncovering Genome Mysteries (UGM) is running slow because after results are completed we need to send them back to the researchers. At the moment we are unable to send results fast enough back to them to clear them off our storage on our servers. As for FightAIDS@Home (FAAH) and Mapping Cancer Markers (MCM) these are running at normal priority and get leveled off based off total runtime for the projects. I have not reset these values in a bit and may need me to change them to make them as level as possible. Thanks, -Uplinger Looking forward to the leveling, as since a very long time FA@H and MCM are running kind of in this share 2 parts MCM for 1 part FA@H. Mind you, Dr. Santiago seems to have his plate more than full with the Autodock analyses, i.e. there's no push to finish off the project prematurely, but that's my reasoning what 'level' could really be in the WCG planning (Don't feel being put on the spot  ) . |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges:
|
Sek,
With the stoppage for Autodock, there will some turbulance, so I will be trying my best to get it equal, but also need the grid running smooth too. So, it might take a few weeks to get it level. Plus there might be some more beta on the way soon, which takes extra attention. Thanks, -Uplinger |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Turbulence you call it
, the ratio improved to 3 parts MCM to 1 part FAHV  Be happy to do a cheat sheet for you to restore balance. There's a direct relationship between the substantial variability of MCM WU runtime and how much FA@H-Years is done, so much is obvious. Every time MCM goes long, FA@H is torqued off since the client buffers are filled with extended MCM work. This is probably worsened by using the <dont_use_dcf/> tag, the clients never learn how much to buffer. You've already got the 25 In Progress limit in place [overdue for review], so hardly anyone could get over-cached.  |
||
|
|
|