Ways to cool Nvidia H100 below 85 C in ASUS ESC4000A-E122k?

[Martin5319]

Dear GPU and ASUS chassis experts,


I have mounted one Nvidia H100 with GPU air ducts in an ASUS ESC4000A-E122k chassis.

When running gpu-burn, nvidia-smi -q -a reports the GPU temperatures below:

            GPU Current Temp                               : 88 C                                                       
            GPU T.Limit Temp                               : -4 C

If I understand it correctly, the GPU throttles its performance due to thermal issues.

Activating full speed fan mode had no effect on the GPU temperatures.



Does anyone else has experience such issues? If yes, did you manage to fix the thermal issues?



When the GPU idles, nvidia-smi reports these numbers:

           GPU Current Temp                               : 47 C
           GPU T.Limit Temp                               : 42 C

Are these idle numbers OK for ASUS ESC4000A-E12?

 

[CyklonDX]

are you running this box in datacenter/air conditioned environment? i.e. you pull in 10-15'C ambient temp.
Not an asus chassis expert (no clue if anyone actually uses asus, at least i haven't met one yet.)
check if you have gpu-kit fans installed, the 22k rpm ones...

depending on your use case, if you can't lower ambient - i would recommend setting a temp limit, or limit wattage using nvidia-smi tool.
if you want more airflow, you can mount external fans (as such MCP-320-74702-0N-KIT).