Message boards : Questions and problems : BOINC not Suspending - Tasks Run Constantly
Message board moderation
Author | Message |
---|---|
Send message Joined: 14 Feb 25 Posts: 2 |
Up until today my client has abided by my preferences, suspends when in use and only compute between 1800 and 0600 (as when all the fans are running it makes it hard to work in the same room and do calls). Today it is running the Einstein@Home Gamma Ray pulsar search #5 full blast and doesn't want to stop. Any advice on calming it down and getting it to heed its configuration? It is also ignoring the maximum CPU time, set at 60% and currently using 65-80% based on number of threads. I am running Windows 11 Pro 24H2 Ryzen 9 5900X 64GB RAM BOINC v 8.0.2 Update: on reboot it seemed to sort itself out, but after maybe 20 minutes it is ramping up again Thanks |
![]() Send message Joined: 28 Jun 10 Posts: 2809 ![]() |
Is this for when computer is in use or for when idle? I have never used limiting the %of cpu time, preferring to limit the number of cores in use. |
![]() Send message Joined: 10 May 07 Posts: 1499 ![]() |
Do you have it set to run BASED ON PREFERENCES versus Run always and the correct venue (home work, school, blank/default) if preference set on website? |
Send message Joined: 7 Dec 24 Posts: 38 |
In reply to admiralblair's message of 14 Feb 2025: It is also ignoring the maximum CPU time, set at 60% and currently using 65-80% based on number of threads.Keep in mind that "Use at most x% of CPUs" & "Use at most x% of CPU time" are two completely different things. The "Use at most x% of CPUs" value has no bearing at all on the number of cores/threads in use (and vise versa). And the "Use at most x% of CPU time" is very stressful for a system, as it starts crunching, then stops, then starts, then stops, then starts, then stops etc, causing severe thermal cycling and stress. A few points to consider- if you have set computing to only occur between certain hours due to the noise, then there is no need at all for a limit on the number of core/threads, and there is no need for anything less than "Use at most x% of CPU time" to be set to 100% The other big problem with "Use at most x% of CPU time" (other than thermal stress) is that it is a duty cycle setting- the smaller the value, the longer it takes to actually process a Task. eg If it takes 2 hours of CPU time to process a Task, then it will take 4 hours in real time to process that Task if "Use at most x% of CPU time" is set to 50%. It will take even longer if you have computing set to "Suspend when non-BOINC CPU usage is above xx%" at anything less than 100% (ie disabled). And that will lead to deadline issues, especially if you run more than one project and run with a cache and limit the hours BOINC can process work- more than one project, or one project with reliable work- the best cache setting is 0.05 days and 0.01 additional days. Which brings us to your present issue- have you looked at the BOINC Manager to see if the Tasks that are running are at High priority? With a limit on CPU time, a limit on when the system can compute, with a limit on doing computing when other CPU usage is occurring, and a cache larger than 0, you are likely running into deadline issues and the Manager is trying to complete the work in time to not miss the deadline and to get Credit for it. Check your Tasks at Einstein and see how many are being returned in time, and how many have errored out because they missed the deadline. Grant Darwin NT. |
Send message Joined: 7 Dec 24 Posts: 38 |
Dr Who Fan's question is very relevant. Do you have it set to run BASED ON PREFERENCES versus Run always and the correct venue (home work, school, blank/default) if preference set on website? I only just noticed- in the Web based preferences, the Computing usage limits "Use at most xx% of the CPUs" and "Use at most xx% of CPU time" apply when the computer is and isn't in use. If you use the BOINC Manager local settings, there are two different settings for these values for those options- one for when in use, one for when not in use. If this is what is occurring, if you look in your Event log you should see some "Number of usable CPUs has changed from xx to xx." messages. eg 15/02/2025 11:02:54 | | Number of usable CPUs has changed from 12 to 2. 15/02/2025 11:07:25 | | Number of usable CPUs has changed from 2 to 12. 15/02/2025 11:15:06 | | Number of usable CPUs has changed from 12 to 2. 15/02/2025 11:18:10 | | Number of usable CPUs has changed from 2 to 12. 15/02/2025 11:18:34 | | Number of usable CPUs has changed from 12 to 2. Set the percentage of "Use at most xx% of CPUs" to be the same for both when in use and not in use. Grant Darwin NT. |
Send message Joined: 14 Feb 25 Posts: 2 |
I believe you are correct in that it is a deadline issue as once it cleared those tasks it is back to behaving normally. I do not recall setting the 60% CPU usage - which isn't to say I didn't do it. I have changed it to 100% to allow it free reign at night. Thanks for all the suggestions, I have taken the opportunity to look more deeply at the settings and fine tune. I had not noticed the advanced settings previously. |
Copyright © 2025 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.