Quote from: Birman on May 16, 2014, 12:40:37 PM

Quote from: Brewins on May 16, 2014, 12:07:03 PM

Quote from: Stedsm on May 16, 2014, 11:45:42 AM

Hi,

I have one GPU with it's WU stuck @ 99.99% for several hours now. I paused the other GPU for 30min and no change. i seen this problem from another user but he didn't get an answer. Should I select that GPU and use the "Finish" option,reboot the PC? Project is 13001

pause the miner then completely close down the client. wait 10 seconds and open it again. That normally fixes it for me

I have the same problem
2 of cards folding 14h to 99.99% and they are idle 9h. and dont want to finish.
My PC have same problem - I paused and restart - after 2h everythig was OK byt only with 6000 credits (for R9 290x) ?!?!

kqpahv posted some information directly from the folding forums about this a few pages back.

The issue you are seeing with the WU that appear to be stuck at 99.99% is actually a sync issue in the client. What is actually happening is that your GPU has encountered an issue. You can confirm this by looking at the log tab inside the FAH Control client for the specific work unit and you will see it actually stopped updating long before 99.99%. I had no errors in my logs and didn't see any errors in Windows event viewer but the data being sent in the log stopped long before the 99.99%

The problem is almost guaranteed to be an overclocking problem on your graphics cards. I had to tone my R9 290X TRI-X cards down to standard 290X speeds. One computer still had hung work units so I slowed the core clock speed down to 990 MHz instead of 1000 and will keep dropping it until the GPU's can handle the work. Keep in mind these cards were factory over clocked but that made no difference with folding. The overclocking got in the way and actually slowed down my WU submission.

TLDR: Lower your GPU clock and memory speeds until the "hung WU" issue no longer happens. As a reference I'm using a core clock speed of 1000 MHz and memory clock of 1250 on my R9 290X TRI-X cards (much lower than the factory defaults). Even at these speeds I still had to reduce the core clock speed on two of these cards down more to 990 MHz and will continue to lower them 10 MHz at a time until the work units get submitted without incident.