Questions and Answers :
Unix/Linux :
Excessive page faults??
Message board moderation
Author | Message |
---|---|
Commander Shrom Send message Joined: 8 Jun 99 Posts: 1 Credit: 280,569 RAC: 0 |
I have accounts with as many projects on BOINC as I can, and keep all four CPUs on this machine pegged at 100% usage. I am running Mint 16 on an Asus® A55A™ series laptop. I look at top frequently, and when a new BOINC process appears, renice it to -1. (In top, for those who don’t know, this is done with the r command, which then prompts for the pid to operate on, hit enter, new nice value, and enter again to change (you probably will need to be root to do that which is why I use sudo top).) All four BOINC processes will show as very close to 100% CPU usage. The boinc program itself also commands nice of -1, and comes up relatively infrequently. I log all output it produces to stdout with a redirect to a file whose name is the date and time of creation in the format yyyymmdd_HHmmss. I recently changed the displayed fields to include the nMaj and nMin (number of major/minor page faults) columns and removing the PR (priority) since since the NI one is sufficient for that. I then observed something interesting: the setiathome_7.01 processes generate an enormous number of minor page faults, over a million every two minutes of runtime! Once they get to 64 million, however, they seem not to make any more, but they finish soon after that usually, around four hours of CPU time. It seems that a program that creates such large number of page faults cannot be running at peak efficiency, since those should slow it down. Has anyone else noticed this? Do other BOINC projects do this? Is there any solution, or is it not an issue? Thanks! Live long and prosper! |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.