It’s becoming a real pain to check each card one at a time on a windows box, just to figure out if a card is going to over heat it’s memory in the racks under linux. Our compute load is heavy on the memory and when we are adding 100 gpus a month to our compute/render farm it’s really becoming [is] a problem to not have memory junction/memory temps in linux. nvidia-smi needs this like yesterday, what is taking so long to get this done?