Kateiacy,

I will review our page that has the Linux installation. It is a very good idea, so I will try to see what else we can add.

Thanks,
-Uplinger

codes,
It was only
- for CEP2 WUs
- for your machines
and we don't know yet if the problem is with the hardware, the OS or with the way CPU times are handled in the CEP2 WUs.
And nobody else has reported a similar problem to date.

So don't expect to see a statement as absolute as yours on the download page.

Thank you for reporting, pirogue.
Maybe that will help the techs to find what is going wrong with these WUs in these OS.

I think there is a real problem inside the CEP2 WUs with regards to handling CPU times at job transitions. Something like saving the CPU time too early at the end of a job and restoring it too late at the beginning of the next one.

Ubuntu is only better than the other distros but it has the problem too. Since the beginning of CEP2 it has been reported several times that the CPU time is more below the elapsed time than for other applications (for me, by about half an hour for a 5-hour WU), even in a dedicated machine doing nothing else, and we have assumed that handling the big files was the explanation to these low CPU activity stages reported by BoincTasks in Ubuntu as well.

Until now I was simply looking at the disk LED when I happened to notice a red line in BoincTasks. With no luck: nothing exceptional regarding disk activity.
After reading your post I have specifically watched a CEP2 process with Ubuntu's System Monitor while BoincTasks was showing no activity at all and in fact the CEP2 process has been running full speed during all that time.

At some point I wondered if there could be some kind of not reported ghosts jobs between the 16 official ones, but nope, no stealth process ID in the picture.
But when BoincTasks finally switched from 0 % CPU activity to 9x % the new job had already counted about 4 minutes CPU in System Monitor...

uplinger, any comment?