1) Questions and Answers : Windows : Is this normal? (Message 46642)
Posted 14 Apr 2022 by myjunkmalebox
Post:
Thanks for the info.
2) Questions and Answers : Windows : Is this normal? (Message 46640)
Posted 14 Apr 2022 by myjunkmalebox
Post:
When you let one Atlas-Task running with 10GByte RAM, it's ok.
Otherwhise, you have to find a way with app_config.xml, to optimize how many Atlas-Task with how many RAM and Cores.


This does not clearly answer my question. Does the windows task manage under report CPU usage due to the VM, similar to the RAM usage under report?
3) Questions and Answers : Windows : Is this normal? (Message 46638)
Posted 14 Apr 2022 by myjunkmalebox
Post:
The limiting factor usually is the amount of memory available to Boinc. An Atlas task using 8 cores needs 10200 MB of memory (3000 + ncores * 900 MB).


That'll do it, the server in question only has 16GB (I've got another 16 coming in the mail).

I'm not worried about seeing the actual ram usage, the behavior was just odd and was interested in an explanation.

I don't think any one addressed my follow up question, does this visibility limitation apply to CPU usage as well?
4) Questions and Answers : Windows : Is this normal? (Message 46628)
Posted 13 Apr 2022 by myjunkmalebox
Post:
Thank you for the feed back. Does the visibility issue apply to processor usage as well?

Is there something that prevents BOINC from running two jobs at the same time? It's got enough cores to run two at once, but only ever runs 1.
5) Questions and Answers : Windows : Is this normal? (Message 46622)
Posted 13 Apr 2022 by myjunkmalebox
Post:
I've got a server dedicated to BOINC, it's got several projects configured but LHC is the only one giving work at the moment. I see one job running using 8 cores (of 24). When I look in task manger my CPU usage is 0%. It's showing 10GB of ram used, but looking at the list of what's using ram the largest item is 400Mb and the 2nd largest is 62mb. All in I'd say 1gb of ram is accounted for in the list of applications. I also noticed when the job finished the used ram% plummets, even though the list of applications looks exactly the same. Most of the jobs this machine does validate successfully so I feel like the work is getting done.

Is this just an artifact of virtual box?



©2024 CERN