Also I would like to know if I start 8 instances of command windows to run Phoenix miner on 8 gpus of the same rig, is there any performance penalty on that?
This is what I did for claymore right now. each miner takes -di # (# is a single digit between 0 to 7 representing 1 gpu)
8 instances of claymores, each mine to a unique eth addr.
If one gpu crashes the miner but didn't cause system hang, the other 7 conitnue working w/o restart. the crashed gpu would just restart automatically 30 second ~ 20 minutes late depending on how bad the situation is. If I start all 8 in one instance, it would restart all 8 gpus as many times as it could until the single crash one became stable.