Hello dstm,
Thank you for your great work. I have been using your software to mine for a while.
I would like to ask, if it possible to make your software "restart" after a cuda memory fail on a gpu.
I have several rigs running and I am not always checking on them, it`s not nice to see that I had one rig not working for a entire night or day because there was a error.
This happens more in rigs where I have 6 different gpu`s from nvidia, it`s not easy to find the best OC. So if, a error happens, it should be logged and the mining should restart. That way I would check for the logs and know that I should reduce a OC on a gpu but i wouldn`t lost several hours of mining.
Thank you very much and keep the good work.
Sorry about my poor english.
I've already done some tests with respect to this. This can't be done reliable from the context of zm. If an unstable/overclocked GPU crashes it stops responding to driver calls most of the time such that it's impossible to restart the GPUs from zm's context. You need something external - like a simple script - which restarts zm on GPU crashes. However this won't work reliable too - sometimes you have to restart your whole system after a GPU crash. If possible zm reports the affected GPU but even this is not always possible on unstable hardware.