The miner can detect various errors and reboot the PC if needed, but low hashrate is not an error obviously.
Well if it could shutdown right after such error it would be great, now it just restarts with low hashrate after every single error like this:
ERROR: Can't find nonce with device [ID=0, GPU #0], cuda exception in [StreamContext<struct search_results,struct Ethash::KernelLaunchTag>::synchronize, 51], hardware stack error