Two notes here in one picture:

First is, when my CUDA cards fail under Linux, I don't get any information about which card has failed, and the error messages are extremely vague.  Overclocking controls are pretty limited in Linux for Nvidia cards so I can only assume that it's the memory throwing a fault and causing the card to crash).  Is there any way to indicate which card failed (and once one fails, the other one goes next, so I always get two of the same failure messages)

Second -- Check out that 3090 hashrate for almost 6 hours of mining!