To Whom it May Concern,

Greetings. My name is Lars Ewell and I work here in Hampton.

I have been running GATE Monte Carlo simulation software on a cluster node here running Scientific Linux 6.6 (carbon). I recently noticed a significant increase in time when running a benchmark test. I did a 'top' and noticed that boinc was running in the background. I thought it was possible that this was responsible for the increase.

Is there a way that I can suspend or stop the boinc job running in the background so that I can test this hypothesis? Unfortunately, I do not root privilege on this cluster.

Thanks in advance.

regards,

Lars Ewell

Assuming you are sharing cpu/gpu/memory resources with everyone else then the only way to know is by the 5/10/15 minute load average on the host.

To change the schedulers priority your user or group must be at the same level or above and any security policies in effect would affect rights. Otherwise your sol unless the sysadmin can help you.

On top of the excellent information given by Andrew, I'd like to throw a couple facts that may help you better understand the BOINC framework:

1) BOINC should never be taking up a significant portion of CPU resources. BOINC is merely a management program that schedules when science tasks are to be run, and handles upload and downloads of results and tasks. When BOINC is not uploading, downloading, or launching a new science application, it should be sitting idle on the system.

2) When BOINC launches a science application, it by default launches it at the lowest CPU priority supported by the OS. In the case of Linux, that should be Nice 19. This is done to allow all user tasks, which should be running at a higher priority, to run with minimal impact or performance loss.

Given the above: A) are you certain it was BOINC consuming CPU cycles and not a project launched by BOINC and B) what priority level does your benchmark program run at?