Find myself with a strange cache situation. I have 13 active projects in my list. 8 core processor. I have a restriction for LHC to not run more that one job at a time [resource hog if it starts a bunch of multi core sucking VM's]. Cache has only LHC CPU work units and GPU units for other projects. Update says that no work requested not highest priority for the other 12 projects. BUT I have 7 cores with nothing to do!

This is not an efficient use of the resources made available to BOINC.

I suspect the one job at a time restriction for LHC isn't being picked up correctly by the work fetch and allowing it to grab CPU work when there are available cores no matter the resource percents.

The only other thing I can think of is the one job at a time restriction wasn't picked up by the work fetch and now there is more CPU work for LHC than the one core can finish on time.

You don't mention which version of BOINC you're using. This sounds like a problem I first reported with BOINC v7.6 three years ago (although it probably dates back to the first code allowing the use of max_concurrent in app_config.xml). Keith Myers hit the same problem last year, and with help from David Anderson, we think that the problem will be solved in v7.16 (several of us have been testing the code for months).

After the conference call yesterday, I'm hopeful that the release of v7.16 will finally get underway in the next few days - although an email from David this morning, asking for help, suggests that he fell at the first hurdle.

Hope someone else is doing the documentation, as that's impossible from a phone...

You don't mention which version of BOINC you're using. This sounds like a problem I first reported with BOINC v7.6 three years ago (although it probably dates back to the first code allowing the use of max_concurrent in app_config.xml). Keith Myers hit the same problem last year, and with help from David Anderson, we think that the problem will be solved in v7.16 (several of us have been testing the code for months).

After the conference call yesterday, I'm hopeful that the release of v7.16 will finally get underway in the next few days - although an email from David this morning, asking for help, suggests that he fell at the first hurdle.

Okay, version in use is 7.14.2 windoze, latest offered as stable, so if it is fixed in 7.16 I'll await it being offered as stable.

Typical when you limit # of tasks via LHC project page or app_config. Set LHC to 1% resource share and the other projects much higher.

Typical when you limit # of tasks via LHC project page or app_config. Set LHC to 1% resource share and the other projects much higher.

tnx, but that doesn't work. the work fetch will still gorge and suddenly there will be 8 VM's each using 8 cores running and the system turns into mush.
The correct answer is for BOINC to stuff more than one job inside the same VM, but that doesn't seem possible.