Task resets when BOINC is suspended

Message boards : Number crunching : Task resets when BOINC is suspended

To post messages, you must log in.

AuthorMessage
Profile ukishun

Send message
Joined: 28 Apr 11
Posts: 4
Credit: 18,756
RAC: 0
Message 70187 - Posted: 30 Apr 2011, 22:59:12 UTC
Last modified: 30 Apr 2011, 23:02:39 UTC

So I'm new here (just started a few days ago) and I seem to be having trouble with this task:

FOLD_N_DOCK_YgaP_D2symm_2_SAVE_ALL_OUT_IGNORE_THE_REST_25975_2913_0

(https://boinc.bakerlab.org/rosetta/result.php?resultid=419308408)

It seems that every time BOINC suspends calculations (set preferences at "while processor usage is less than 40 percent"), this task resets and goes back to about 5%. This includes the elapsed time which goes back to about 21 minutes.

Does anyone know why this is happening? Or is this normal? The rest of the tasks I've done don't seem to have this problem.
ID: 70187 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 70188 - Posted: 1 May 2011, 1:29:16 UTC

Any time a task is suspended and removed from memory, it will restart at its last checkpoint (or at the beginning if no checkpoints have been reached yet). So, this is normal if your tasks are being suspended (and it sounds like CPU usage over 40% running other tasks is causing that to happen), and the task is removed from memory (which would be dependent upon your preferences for memory).

If your machine is going to be suspending tasks frequently, you will want to be sure to check the box on the disk and memory usage tab of your preferences to leave applications in memory when suspended. Once the task is suspended, any real memory it is using will begin being swapped out as the other tasks require memory. So it's actually leaving the application in VIRTUAL memory, not RAM.

So yes, it sounds normal up until you confirm whether you have the leave in memory box checked on this machine.
Rosetta Moderator: Mod.Sense
ID: 70188 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile ukishun

Send message
Joined: 28 Apr 11
Posts: 4
Credit: 18,756
RAC: 0
Message 70189 - Posted: 1 May 2011, 1:41:07 UTC - in response to Message 70188.  

Any time a task is suspended and removed from memory, it will restart at its last checkpoint (or at the beginning if no checkpoints have been reached yet). So, this is normal if your tasks are being suspended (and it sounds like CPU usage over 40% running other tasks is causing that to happen), and the task is removed from memory (which would be dependent upon your preferences for memory).

If your machine is going to be suspending tasks frequently, you will want to be sure to check the box on the disk and memory usage tab of your preferences to leave applications in memory when suspended. Once the task is suspended, any real memory it is using will begin being swapped out as the other tasks require memory. So it's actually leaving the application in VIRTUAL memory, not RAM.

So yes, it sounds normal up until you confirm whether you have the leave in memory box checked on this machine.


This didn't happen with the other tasks that's why I thought it automatically does that. Thanks for the quick reply!
ID: 70189 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 70190 - Posted: 1 May 2011, 3:03:00 UTC

It happend. It is just a matter of whether the task had just checkpointed. Some R@h tasks checkpoint more frequently then others, so the lost time may not have been as pronounced on other tasks.

If your machine is running without incident, you may want to increase the CPU threshold for when BOINC suspends operations as well as checking the leave in memory box.
Rosetta Moderator: Mod.Sense
ID: 70190 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile ukishun

Send message
Joined: 28 Apr 11
Posts: 4
Credit: 18,756
RAC: 0
Message 70220 - Posted: 2 May 2011, 12:39:32 UTC - in response to Message 70190.  

It happend. It is just a matter of whether the task had just checkpointed. Some R@h tasks checkpoint more frequently then others, so the lost time may not have been as pronounced on other tasks.

If your machine is running without incident, you may want to increase the CPU threshold for when BOINC suspends operations as well as checking the leave in memory box.


Okay I'm still getting the problem at this same task. When I suspend the task, then turn off BOINC (so I can shut down the computer), the task goes back to the 5% (and 21 minutes elapsed). I've already checked the "Leave applications in memory while suspended". I've also set processor usage to 0% (meaning it's always on).

For now I'll probably just try to have this task run continuously when the computer will be on all day.
ID: 70220 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 70222 - Posted: 2 May 2011, 16:53:15 UTC

Right, leave in memory only works while the machine is still running. It shouldn't need to run all day, but it needs to run longer then it has been to reach the next checkpoint. If a task like this restarts 5 times from the same point and has made no progress (i.e. not reach the next checkpoint), R@h will mark it as completed and return the work you've done so far. So, either way, that task will not be a problem for much longer.
Rosetta Moderator: Mod.Sense
ID: 70222 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Task resets when BOINC is suspended



©2025 University of Washington
https://www.bakerlab.org