what CPU temp is acceptable for a N100 mini pc (cheapo - gmktec G3)

So i have had a N100 based minipc for 3 months now (GMKtec G3). Works great. Best bang for buck cheapo tech I have bought in years.

For all this time I've been thinking how does the CPU always stay at 27c even when the fan is running... Stupid me. It's not 27c at all.

Turned out that I had to follow a guide to get it working!

Actual temp is shown below. Didn't expect that. Did a windows update on a vm in this thing just now and temp went up to 98c. haha

Is this a problem? It's not even summer yet and not hot today at all.

The internal nvme drive is running hot too (been like this from day 1 though)

Thanks. I have just found this issue so I guess I need to monitor the temp for a bit.

Seems to be 80c all the time (at idle) and that is quite hot I think.

Maybe I get some thermal paste and see if that improves the situation.

nvme ssd running hot too (80c) but stays at around the same temp even under load. Should I try to deal with that too?

Yes. You’re right for sure. It is not in the best spot but I can’t see it and that’s how I like it.

I’ve ordered a heat sink for the ssd.

Had a look inside the case but the cpu is on the bottom side of the board and I can’t access it. Poor design but that’s what you get for a usd 150 mini pc.

Fan does not come on at idle so it thinks it’s ok. 😂

I’ll check bios next. Might be able to set the cpu fan better.

If that doesn’t work, I’ll move the minipc to improve airflow and ambient temp.

Thanks for the help as always.

so as an update. I lowed max power from 10w to 8w in the bios but it has made no difference. I think maybe this will lower temp under load as power will max out at 8w?

I have found something interesting (and good news) - looks like the cpu monitoring in OMV is over reporting temp but I am not sure how/why.

output from cli looks like this.

root@minipc:~# sensors
coretemp-isa-0000
Adapter: ISA adapter
Package id 0:  +65.0°C  (high = +105.0°C, crit = +105.0°C)
Core 0:        +63.0°C  (high = +105.0°C, crit = +105.0°C)
Core 1:        +63.0°C  (high = +105.0°C, crit = +105.0°C)
Core 2:        +63.0°C  (high = +105.0°C, crit = +105.0°C)
Core 3:        +63.0°C  (high = +105.0°C, crit = +105.0°C)

acpitz-acpi-0
Adapter: ACPI interface
temp1:        +27.8°C

nvme-pci-0200
Adapter: PCI adapter
Composite:    +78.8°C  (low  = -273.1°C, high = +82.8°C)
                       (crit = +84.8°C)

and i get a correct result when I use the cpu-temp config in the cli

sensors coretemp-isa-0000 | awk '$1 == "Core" && $2 == "0:" { print $3 }' | grep -o '[0-9.]\+' | sed 's/\.//'

returns 640 as expected....

but in the gui - the gauge is 75-80

Thanks. I have used the guide so I think it is configured properly. Just the data direct from the senor in the cli differs from the gauge in OMV by around 10-15c - although I know the gauge is running the same command as the cli. It's bonkers.

I have no idea why. Doing some work using telegraf and influxdb to monitor temps straight off the sensor.

Also added a heatsink to the nvme ssd.

CPU gauge = temp of cpu core 0

CPU2 guage = temp of nvme ssd

grafana - before and after heatsink on on the ssd so it is working ok and has reduced ssd temp by about 15c

cpu steady at 65c but OMV does not agree

Quote from ruudm

For what it is worth: I am also using a N100 for over a year now. Regular temperature is 47. But I am not trying to run Windows in a VM, that is too much in my opinion. Computer is just for OMV (OS on USB), 10 containers and 1 VM for Homeassistant.

Thanks. I think not all N100 minipcs are created equal

Mine is cheap and located in a place with limited airflow.

Same with the nvme ssd. Poor airflow and cheap. It's not used for important data (e.g. used for docker, app data, vm's) and is backed up so not that worried.

I do want to know what the actual temps are over time so monitoring the sensors directly using influxdb / grafana

Thanks chente. I have been playing with the grep command but i'm useless.

I managed to get it right for the ssd

sensors nvme-pci-0200 | awk '$1 == "Composite:" { print $2 }' | grep -o '[0-9.]\+' | sed 's/\.//'

and I have tried the exact same as you posted but it doesn't return anything - but still working on it

However - it will not make much difference as core 0 and package id 0 are about the same - maybe its less twitchy

I thought this would work but it doesn't either.

sensors coretemp-isa-0000 | awk '$1 == "Package id 0:" { print $2 }' | grep -o '[0-9.]\+' | sed 's/\.//'

I am coming to the conclusion that it is working fine in omv gui.

I think It’s higher in the gui (by 5-10c) because the cpu temp sensor on the n100 is so sensitive (twitchy) that when the dashboard updates the page it spikes the cpu temp and the gui is reporting the cpu temp spikes.

As a test I setup the nvme ssd as a cpu temp in the gui. This is working fine.

This cheap minipc just runs hot especially as it’s not located in a good place for airflow.

Thanks for all the help guys.

No heatsink on cpu. Can’t access it.

Put a heatsink on the nvme ssd and it is working. Temp down by 15c so reasonable improvement.

So I have fixed the ssd heat issue and the cpu is not running as hot as I thought as gui is reporting spikes.

Thanks again everyone