I tried several methods to control the fan speed even if i succeed it reverts back after boot or evetually . When the GPU runs at 100% , power is at 170w max the SW cap still kicks in throttling back the power causing oscillating in the training. Also Gpu gets above 80 deg with the fan only going at 50% , when i managed to increase the fan to 80% the gpu was running at much lower temp 65% for same model and workload . I there a definitive way to control both Fan and Power throttling manully . I tried both ubuntu 22.02 and 24.02 same result , also different drivers …frustrated :(
I have resolved this by enabling coolbits from Xorg.conf prior to log in otherwise it wont work .Hope this heps someone ,not straight forward and not much help from Nvidia support on this . I dont understand why the FAN lets the GPU run hot above 85 deg while the fan sits aat 40-50% spin on normal run without manual control