here is the best solution, if you have a gpu that is having issues that means you've got an underlying problem somewhere which should be checked out, if you rig CANNOT RUN for 30+ days without any problems there's an issue you need to fix and its as simple as that, stop trying to use some script or little program to reboot the rig when clearly you've got a problem somewhere. the one thing everyone posted on this thread has FAILED to even mention.
a good running rig can run for months with no problems what so ever, the fact you are having issues shows there's a problem somewhere
Very true statement.
Sometimes, though, things are outside the average miner's control: There could be an infrequent bug in the hashing algorithm; switching between algo's can make GPU's twitchy; dicey networks; etc.
All of those do require analysis to determine root cause and permanent corrective action (as you mention). But until a person has enough time to dive into it, it's nice that the rig reboots itself to keep the mining process moving. Now if the rig is rebooting frequently then jump on it immediately. But if it's once a week it's not as urgent of an issue (as compared to other urgent issues... for me at least).
