Hi All,

I'm running several CPU and GPU projects on a C2Q WinVista32 host with BOINC 6.10.18. I've noticed recently that BOINC sometimes uses fewer than 4 CPUs even though my preferences are set to use all processors, all the time. I've never seen this before and the only thing I can think of that has changed as far as my mix of projects is that I'm now running AQUA multi-theaded WUs, something I hadn't done in a long time.

AQUA's multi-threaded app usually uses 4 CPUs, but when one of those isn't running, I've seen everthing from 1-4 CPU projects running (in addition to a fraction for the GPU app). I haven't been able to determine how long this condition goes on between times that all 4 CPUs are being used. I've tried pausing/resuming the BOINC manager as well as stopping/starting the client and manager altogether. Not that I thought it would, really, but neither of those actions "corrected" things.

Thanks for any input.

Cheers,

MarkR

Hello Mark,

This is caused by the new type so called single core aqua's "FOKKER_PLANCK"
I replied in the news section.
I've noticed if you suspend the running aqua task all free cores starts work of other projects.
If the aqua task resumed and 1 task of another project finishes that core is not used until that aqua task finished too or suspends.
To prevent this behaviour you should deselect the FOKKER_PLANCK in your aqua preferences and only use it on real (or VM-ed) single cores, until the AQUA project fixed it.

Cheers,

CP

Thanks, CP. I should have checked the AQUA forum first since I suspected their app might have been an issue... Anyway, I found this post as well as another post that together suggest avoiding BOINC 6.10.14-6.10.18 and possibly upgrading to BOINC 6.10.19 as a solution. It would seem that the AQUA developers are aware of this.

Since BOINC is still off line and will be for another couple of hours, you can check for that version of BOINC on the WCG mirror.

Not meaning to hijack the thread, but...


I noticed that they've been down for some time. What news did I miss?

Big scheduled power outage on the Berkeley campus, affecting the hill where the SETI and BOINC labs are located. They are back up now, but another large outage is scheduled for next Sunday (3 January)

Thanks for the news update!


Thanks for the reference, Jord.