Posts by Alun

1) Message boards : Number crunching : Multiple Computation Errors (Message 75453)
Posted 24 Apr 2013 by Alun
Post:
Ho hum - rosetta broken again?

Cryo WU's: failing for an out of memory error, on an 8gb system which has >6gb free for BOINC tasks. Latest release version of BOINC (7.0.64) if that's of any relevance.

At this point there's been less than a month since Christmas where Rosetta itself hasn't been broken at the project end. Does anyone do any testing on these work units before throwing them at us to waste time and resources (and ultimately money) on? Project suspended for now.
2) Message boards : Number crunching : Client errors (Message 75188)
Posted 2 Mar 2013 by Alun
Post:
BOINC 7.0.52 (x64 version) with the 314.07 Geforce drivers seems to work fine for Rosetta. Four work units completed overnight with no issues. Will run some more over the next days to confirm it's actually sorted out.

From looking at the successful workunits it seems a lot of people still aren't aware that there's a major problem with rosetta using earlier versions of BOINC and GeForce gpu drivers. I'm seeing people with 8-core units that've only returned errors for their past months of work.

Definitely worth putting out a BOINC notice informing people there's an issue and that there're a couple of fixes available for it. If people find out they're wasting hundreds of hours of processing time because someone couldn't be bothered to put out a BOINC notice warning them of issues they're likely to be a little... irritable.
3) Message boards : Number crunching : Client errors (Message 75068)
Posted 9 Feb 2013 by Alun
Post:
Is anyone actually actively investigating the issues with nvidia drivers & cards at the moment, or (as it seems from the forums) is it falling to the community to find the problem in the UoW's Rosetta applications?

Question / point: If it was purely a driver issue wouldn't we be seeing errors on other GPU projects running on the same box? GPUGrid, Milkyway, Einstein and SETI are all fine - only Rosetta gets borked by updated drivers...
4) Message boards : Number crunching : 100% error rate on work units since Christmas (Message 74887)
Posted 13 Jan 2013 by Alun
Post:
Sadly if the only solutions are rollbacks to 306 series GPU drivers (not recommended for my card) or an 18 month old version of BOINC that predates the recommended release by a year, I'll have to disable Rosetta until such time as it catches up with the current releases of drivers and BOINC.

Thanks for the help :)
5) Message boards : Number crunching : 100% error rate on work units since Christmas (Message 74885)
Posted 13 Jan 2013 by Alun
Post:
Hey there, I wonder if someone'd be able to give me a bit of advice on how to track down what's causing the processing errors I've been experiencing since christmas (maybe related to the replacement GPU fitted over the holidays?)

Every Rosetta work unit that I've processed since is showing an outcome of "client error" on the Rosetta website, though they all appear to be completing ok at this end as far as I can see. I've tried resetting the project to ensure there're no issues with previously downloaded items (to no effect) but have absolutely no idea where to look further to find the root of the problem. If I can't find a solution it's probably best if I suspend Rosetta processing until the exact issue is a little clearer. None of my other projects appear to be affected by the issue so it does appear to be Rosetta specific.

Any advice gratefully received, thanks :D

Alun J






©2024 University of Washington
https://www.bakerlab.org