| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 14
|
|
| Author |
|
|
Rosemary7391
Cruncher Joined: Feb 3, 2024 Post Count: 8 Status: Offline Project Badges:
|
I've set my preferences to "Use... at most 25% of CPU time" but some - only some - MCM tasks are running at 100%. Other MCM tasks are running at 25% as they should.
----------------------------------------The ones I have noticed doing this all seem to be retries, although that might be a coincidence. Some examples of affected tasks : MCM1_0243564_9301_2: https://www.worldcommunitygrid.org/contribution/workunit/787393046 MCM1_0243564_1803_2: https://www.worldcommunitygrid.org/contribution/workunit/787396824 MCM1_0243608_8049_2: https://www.worldcommunitygrid.org/contribution/workunit/787831219 Thanks for any ideas :) [Edit 2 times, last edit by Rosemary7391 at Dec 14, 2025 5:01:22 PM] |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2361 Status: Offline Project Badges:
|
If you ask me, "Use... at most 25% of CPU time" does not mean that all your tasks should use at most 25% of the CPU time, it means that BOINC should use at most 25% of the CPU time. It is meant to be used to reduce CPU temperature and energy usage. Basically, in your case using 25% it means that computing should be active for 25% of the time and inactive 75% of the time, or, in other words, computing should be done for 1 second, then suspended for 3 seconds, then resumed for 1 second, etc.
Another example: 75% means compute for 3 seconds, wait for 1 second, and repeat. A warning is in place: using this feature might sound like a good idea™, but as you can read in post 708115, limiting the CPU-time is counter-productive, as explained by alanb1951. A better way would be to reduce the number of running tasks by limiting the CPUs in use to a certain percentage. So, instead of using 100% of the CPUs, you could use your mentioned 25%. Adri PS Mentioning the name(s) of your task(s) doesn't help. If you really want us to take a deeper look at the state of your tasks, use their workunit-IDs or their URLs. Where do you find a workunit-ID? Go to your My Contribution -> Results, search for a specific taskname there, then click on the 'Result name', e.g. MCM1_0243564_9301_2 like you said, then copy the URL or select the last part of the URL (i.e. the number in the address bar). |
||
|
|
Rosemary7391
Cruncher Joined: Feb 3, 2024 Post Count: 8 Status: Offline Project Badges:
|
Hi Adri
Thanks - am not well at the moment so I'm probably not being very clear! What you describe, ie running 1 second and suspended for 3, is what I expect and what I'm seeing with the majority of tasks. The problematic ones are just running all the time. I already restrict the number of cores and combined with the % CPU time setting that works well for me. I do need the % CPU time to control fan noise and energy use as otherwise even one task is annoying. Have added the links to the first post, thanks :) |
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 291 Status: Offline Project Badges:
|
Rosemary - another way to reduce heat and fan noise is to lower the maximum allowed CPU frequency. Sometimes a small reduction can made quite a difference. If you are running Windows cannot help more, sorry. From distant memory think it might be in the advanced power settings somewhere. Used to do that on a windows laptop before I retired 10 years ago. Then converted it to Linux. (Hope you get well soon).
----------------------------------------
Run Time Stats https://grassmere-productions.no-ip.biz/
----------------------------------------[Edit 1 times, last edit by TonyEllis at Dec 14, 2025 4:30:43 PM] |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2361 Status: Offline Project Badges:
|
That's making it more clear, thanks, Rosemary, hope you're doing better sometime soon; 2025 (the whole of it) hasn't been my favourite as well.
By the names of your tasks we can see that they're probably retries with a shorter deadline. That said, thanks to the links that you provided we can look at them somewhat closer and indeed, the deadlines of the 2 tasks with suffix _2 are reduced to three days, that's 50% of that of an initial task (with suffix _0 or _1). That means that they are taking a higher scheduled priority as far as BOINC is concerned. They need to be returned sooner than initial tasks, so now that we know their deadlines it might not be a spectacular surprise that they are dealt a bit more CPU-time than the other tasks. I would say that it is what can be expected, so there's no need to worry. Hope you can enjoy this last month of the year a little bit more. Adri |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2361 Status: Offline Project Badges:
|
Hey Tony, nice to see you here, 3 minutes before my former post.
Adri |
||
|
|
Rosemary7391
Cruncher Joined: Feb 3, 2024 Post Count: 8 Status: Offline Project Badges:
|
Thanks both, appreciate your help and good wishes :)
Tony - that's an interesting idea, but I assume that's system wide? This is my main laptop, so I need the speed for my own work sometimes. Adri - Thank you, and I hope 2026 goes better than 2025 for you. I just got another retry in, which has also started running at 100% time. Deadline is Wednesday, so you may be right there. Bit frustrating if that's the case, 4 hours vs 1 hour isn't going to make or break that deadline (and I run a small cache, so there's no chance it would miss the deadline for that reason). I'll keep updating my list in the first post and see if they are all retries: I think all the ones I've noticed before have been. I don't think I observed this behaviour prior to the migration, but I'm not sure I saw many retries then either. |
||
|
|
Bryn Mawr
Senior Cruncher Joined: Dec 26, 2018 Post Count: 386 Status: Offline Project Badges:
|
A guess here but it seems reasonable. The task is a retry and has a higher priority than your other tasks, the manager sends it to core 1 where it runs for 1 second and stops, core 1 is then rested for 3 seconds.
However, as the task is high priority it is then sent to core 2 for 1 second, then core 3 and then core 4. Each of the cores is obeying the 25% rule but the high priority task is getting the 100% runtime it needs to complete on time. |
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 291 Status: Offline Project Badges:
|
Thanks both, appreciate your help and good wishes :) Tony - that's an interesting idea, but I assume that's system wide? This is my main laptop, so I need the speed for my own work sometimes. Adri - Thank you, and I hope 2026 goes better than 2025 for you. I just got another retry in, which has also started running at 100% time. Deadline is Wednesday, so you may be right there. Bit frustrating if that's the case, 4 hours vs 1 hour isn't going to make or break that deadline (and I run a small cache, so there's no chance it would miss the deadline for that reason). I'll keep updating my list in the first post and see if they are all retries: I think all the ones I've noticed before have been. I don't think I observed this behaviour prior to the migration, but I'm not sure I saw many retries then either.
Run Time Stats https://grassmere-productions.no-ip.biz/
|
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 291 Status: Offline Project Badges:
|
Hi Adri, currently writing another mainframe CPU story for you...
----------------------------------------Rosemary I assume that's system wide Botched my previous attempt, lets, try again If your other work is important for you, then maybe reducing the number of CPUs running WUs is worth a try. Fewer CPUs running at 100% should give somewhat the same number of results. The OS constantly moves any work, including WUs, between CPUs. They therefore will all be about the same temperature. As the WUs move around then each individual CPU will only be working part-time. However, if your laptop don't have enough CPUs, then you may not achieve fine enough control. Anyway, easy enough to try. Reduce the number of CPUs allowed to run WUs, stop the throttling and see how you go. Change the number of CPUs for desired temperature/fan-speed. This will also ensure retries run at the same speed as normal WUs. Easy enough to get back to where you are now if not successful. Another little help is to use a suitable object underneath the back of the laptop to lift it a bit. Be careful not to block any air vents. On my laptop the bottom runs much cooler as more air can easily flow. As well the keyboard is more comfortable for typing. Of course, not so practical if the laptop is moved around frequently.
Run Time Stats https://grassmere-productions.no-ip.biz/
|
||
|
|
|