Near constant disk activity
13 Oct 2010 11:52:55 UTC
Topic 195380
(moderation:
I'm running S5CG1 right now, with BOINC 6.10.58 on an XP machine. Whenever the Einstein@home task is running, it is constantly hitting the disk. This is driving me crazy, and can't be good for my laptop. I have BOINC set to run all the time, but at times I can't do anything because the disk activity is so intense.
Do I have something configured incorrectly? How do I fix this?
Near constant disk activity
)
Change the task check pointing time to a longer interval. I have mine set to 900 seconds, 15 minutes, and it seems to be fine. Also when you stop Boinc does the disk activity stop? If not you might want to check for some malware with this software http://malwarebytes.org/. Download the free version, install it, update it and perform a quick scan. If all is okay then you might have other problems but malware bytes is decent at finding the bad guys.
I just looked and are you running MW at the same time on the same laptop? That could be your problem too, too much going on due to multiple projects. You may need to think about getting another pc and becoming a farmer or even a rancher some day! The more pc's the higher the tag, with your 2nd pc you are a gardener, then after 5 you are farmer than after 15 you become a rancher! I am currently a farmer but could be a rancher if I could figure out how to pay the electric bill! I have 10 pc's running right now but have 6 more sitting on the floor just needing to have their hard drives installed and they too could be crunching.
I suspect that when running
)
I suspect that when running BOINC, your disk is not only accessed for checkpointing. BOINC itself will update the client_state.xml file rather frequently, and the debug output of the apps is written to a file called stderr.txt in the slots sub-directory. I don't think any of the preferences settings will have an influence on those files??
CU
HB
RE: I suspect that when
)
Actually the BOINC Client should honor the "write to disk at most" setting as well, I remember some change to the Client to write the client_state.xml oly if something has really changed. Stderr of the App should be buffered, too, and also not be written more often that configured, unless the buffer fills up in the time configured.
jmontana, how many S5GC1 tasks do you have running at the same time? If you have configured to "write to disk not more than every 60 seconds" and you have six Apps running, this would still mean a checkpoint every 10 seconds on average.
BM
BM
Another possibility is
)
Another possibility is virtual memory thrashing. You'd need to examine your disk(s) free space(s), and check which disk is assigned to the swap file.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal