I doubt it, they never have in the past & they say they have no way of doing it. I wondered why my Credit was down some there but never looked at my Wu's Until somebody sent me a PM at the Project notifying me of the Validation errors. Right now I haven't got any more since they supposedly fixed the problem ...
Proud member of SETI.USA since 28īth December 2005.
Joined old MB Dec. 28th 2005 - 5837 posts
I noticed something with Dirt.. if I used any GPUs that were ONLY SP Math, I ended up with a lot of invalid work (using 8800s and 250s from nvidia)... but if I use any of the 500 cards which do DP math, I rarely get any errors out of Dirt.
Yeah the 9xxx vs 8xxx cards are basically new biosesyou can turn pretty much any 8K card into a 9K if you know what you are doing, but there's no real gain for it.
I know there's a comparison someplace, but I think NV GPUs only went to DP math with the mid-high end 400s and 500s if I remember correctly. ahh here's the list...
2.0 GF100, GF110 GeForce (GF100) GTX 465, GTX 470, GTX 480, Tesla C2050, C2070, S/M2050/70, Quadro Plex 7000, GeForce (GF110) GTX570, GTX580, GTX590 2.1 GF104, GF114, GF116 GeForce GTS 450, GTX 460, GTX 550 Ti, GTX 560, GTX 560 Ti, 500M, Quadro 600, 2000, 4000, 5000, 6000
Damn, I thought spingadus's problem was an isolated incident. I see 77 errored wu's on my account. Not good.
I also found this on Dirt's website
Edit: Just found the RSS feed for this buried in my list of unread posts. Gotta catch up...status update on server errors
We have essentially the web site, scheduler, and validator running. The assimilator, perfecting, etc. are not running. The error rate has been 0 since we dropped the other daemons, watched the load dropped, and restarted the machine. However, this is not to say that the problem(s) are resolved as we are testing to see if our current theories have correct.
Martin, our host, and I have been exploring particularly thermal and power issues. (Martin found the upload handler was writing 16KiB chunks to disk but optimization to that helped load but didn't change uploaded files getting corrupted.) Also, the upload handler is essentially the least changed, untouched, and until now basically the upstream boinc code. At present the load is staying low and so are the CPU thermals.
Our I/O upgrade only changed the RAID controller and the drives but 3 of the drives were running hotter than they should have been. This extra heat of course impacts the heat in the system.
We have the disk space to keep running in a minimal mode like we are now. Our host already has a heavier power supply and high cfm fans ready. We also are planning some contingencies if we still have problems at that point.
25 Oct 2011 21:10:07 UTC
Last edited by Fire$torm; 10-25-11 at 11:23 PM.
FWIW, I haven't had any new errors for the last 8 hours. I've split my gpu between distrrtgen and PG now. It irks me to not get credit for the crunching that I've done. The project could just give everyone credit for the invalid wu's if they wanted. I may move away from chasing the high credit gpu projects and spread my cards over the other nvidia projects for my MM's.
I've had 3 that I can find since yesterday, looks like quel wrote a script to grant credit to the Validation Error Wu,s, so they could have all along done that but just didn't want to but the problem was so severe yesterday they almost had to ...