RTX A3000 mobile, TGP stuck at 35w after suspend

Graphics / Linux Linux
1

I have HP Zbook Fury 15 G8, running Arch Linux.

A3000 in my laptop is set to 80w TGP by default (90w max according to nvidia-settings), and that works just fine, when i do nvidia-smi or nvidia-settings after fresh boot, i see 80w.

But when i suspend and then resume, TGP goes down to 35w and it is stuck there

I attached the report from nvidia-bug-report.sh
nvidia-bug-report.log.gz (581.3 KB)

I also tried starting nvidia-powerd service, but doesnt start, error:

terminate called after throwing an instance of 'std::runtime_error'
  what():  cpuid_error
Aborted
2

There seems to be something severely wrong with the nvidia gpu in that laptop. Runtime suspend doesn’t seem to work, so the nvidia gpu doesn’t shut off. I also suspect while in suspend, the nvidia gpu is on and not cooled anymore, it’s at 63°C in the logs after suspend/resume while nothing running on it and it should be sleeping.
/sys/bus/pci/devices/0000:01:00.0/power/control: on
Please check for a bios update first.

3

I just checked, i have latest available bios
I also noticed a really strange thing with nvidia-smi, it says 752W/80W sometimes, not sure if related

4

Please set kernel parameter
nvidia.NVreg_DynamicPowerManagement=0x02
and create a new nvidia-bug-report.log ofter reboot.

5

nvidia-bug-report.log.gz (446.2 KB)
Here it is

6

Everything seems to be fine and working correctly. Please suspend and resume, then post the output of
cat /sys/bus/pci/devices/0000:01:00.0/power/control

7

No, it’s not fine at all. The system is reporting the gpu is sleeping but it reports a temperature of 68°C when waked up. This is really dangerous, please report it to HP.

8

it is “on” after resume

Ok, thank you, but the last log i uploaded is just after reboot, i didn’t suspend/resume, should i upload another log after suspend/resume?

9

Not necessary, the first log you attached is sufficient.