Hi,

Over the last 5 days the WU's being processed on my 2 pcs don't appear to be getting credit correctly.

They have both been running 100% 24/7 for maybe a month and there were no issues prior to 5 days ago.

A couple of days ago I aborted any unprocessed WU's cause I was thinking that maybe the reporting times had slipped and it was processing overdue WU's, but that would be unexpected as they are new GPU's. But this didn't change anything.

I don't know if its a factor, but in the boinc manager on the task tab the CPU time only seems to count up very slowly, nowhere near real time. After a WU has been crunching for a few hours CPU time can be displaying only 10 to 20 seconds.

Any ideas?
Thanks, Far

Apols for that MrS, they should be visible now.

A Q6600 and a Q9450, both with one GTX260+ each (the 216 shader versions).

The CPU's are lightly (and stably) overclocked, the GPU's are stock.

I updated Boinc on one of them to version 6.4.7 in case that helped but no luck.

Can you see anything amiss?
Thanks,
Far

When looking at the 'Redundant Results',

http://www.gpugrid.net/workunit.php?wuid=262335

It looks like someone else completed this one after you ran it for 4 days...

Are you sure the Q6600 is running 24/7..? A GTX260 should not take that long to finish a wu.


____________


Consciousness: That annoying time between naps......

Experience is a wonderful thing: it enables you to recognize a mistake every time you repeat it.

OK, cool I understand the CPU time counting slowly then, that makes sense.

Yeh, the machines sure are running 24/7 (the heat and noise is difficult to miss, not to mention they are also used for backups & I'd hear pretty fast if they were failing:-)

Host 25189 appears to be running OK again as of the last 24 hours - but then I upgraded it from 6.4.5 to 6.4.7 approx 24 hours ago.

Host 24362 is unhappy. Its been running its WU for 4 days on a GTX260+, and I have checked that it has been running several times over this period. I will also upgrade it to 6.4.7 now as well.


Should we be using the developmental versions of win Boinc manager for more gpu stability?


Thanks,
Far

Upgrading to 6.4.7 fixed it for the other one also. Strange when 6.4.5 had been working AOK for the maybe 3 weeks that they have been doing GPU calcs. Nothing was changed at this end. There was one power failure but how that could affect them in a way that boinc needed a reinstall I don't know. Altho one of the machines would only boot in safe mode and needed its graphics drivers cleaned out and reloaded to boot without a BSOD... Which is obviously GPU related..

Wow... The timeframe matches perfectly for when the power went out, and could also explain why both machines boinc results went whacko at the same time.

OK, so there is the possibility that a power failures when running GPU boinc can put boinc in a state that it needs to be reinstalled. It will appear to be running without issues and loading up the machine, but will be achieving nothing.

As i can understand elapsed CPU time will be exchanged with elapsed wall clock time in upcoming Boinc manager version