BAD Work Unit

Hi,

Windows 10 Pro recently decided to upgrade to latest patches and it's knocked out both of my folding rigs.

When windows updates, F@H client complains of a missing DLL which is resolved by reinstalling the NVIDIA drivers HOWEVER, when it does run, every work unit downloaded processes for a couple of seconds before bombing out and reporting BAD_Unit.

Gutted as I've just bought a 1080 GTX and was going to relocate my 980ti hydro copper into my server to join in folding - no point until this "bug" is sorted though - neither of my rigs will fold and both did reliably until the latest software update.

I know that linux would probably be better, but both rigs need to be on Win10 so moving away is not an option for me due top my MS developer needs....

Anyone any ideas or come across this?

 

Try older drivers ?

 

Which nVidia drivers are you using?

 

This. As Doc said to me not too long ago, use something around the 372.xx mark.

 

I just "ugraded" to the current ones when windows upgraded.
I'll download 372.something later and install...

Is it a known issue?

 

Yes - it's a known issue - see noizdaemon666's Core21 thread

 

That's great Doc/NoizDaemon666 - thanks both for the heads up

Will restore older drivers tomorrow when I can get time back infront of my server.

+rep guys

 

switched out the driver for version:

373.06-desktop-win10-64bit-international-whql

installed the driver and physix only, selecting clean install and.... so far so good.

 

Hi.
I thought I would try the nVidia hotfix 376.48 driver. All seemed well for a while, but on a few occasions it seems the cpu would continue folding, but new jobs would not download for my two 1070's. I've now rolled back to the 372.70 driver and all seems to be well again. Any thoughts?

 

Just downloading the hotfix:

http://nvidia.custhelp.com/app/answers/detail/a_id/4288

myself

 

....Well.... it's folding now, which is progress.

Let's see if it continues to fold or it does what Clanseven's does....

 

I've had clients sit at the waiting stage for no apparent reason. In these cases, I've found that if I delete the particular Slot in the client, save, and then re-add it (which usually results in a new GPU Folding Slot ID), it picks up a new WU OK. I don't think that this issue is anything to do with the graphics driver.

 

I know I'm tempting fate, but I'm 30% the way through the first workload on the hotfix driver and all is well... fingers crossed.

 

100% completed and uploaded. Nice to be able to periodically fold once again

Might turn the heating off in my office and see if I can keep it warm more productively

 

That's my issue in a nutshell Doc, but it didn't happen before I switched to the hotfix driver and it hasn't happened since I rolled back, but that could just be a coincidence. Let's see what happens. Thanks for the replies guys.

 

I've done a few work units on and off over yesterday and all seems well for me with this new hotfix - no issues so far with it pulling down new work units

 

That's good news for you and the team.

 

Was worried why my new heating rig wasn;t affecting my production rate, then I saw my mistake

http://folding.extremeoverclocking.com/user_summary.php?s=&u=723782

Ahem - it says Tattysnux not Tattysnuc.

Make sure you type in your NAME as it was registered AS WELL AS THE team no and passkey.

DOH DOH Frickin Doh!!!!

 

This is a widely documentedissue, 373.06 is the last stable driver for F@H that Nvidia released. Anything later will cause Core 21 units to crash. I've now disabled automatic driver updates on my Windows 10 machines so they don't grab the latest driver from Windows Update any time I add or swap around cards.