You are not logged in.
Whenever I check the current state (idle) of my NVIDIA GPU with tools such as nvidia-smi, nvidia-settings, nvtop etc. my GPU always seems to be running at the maximum speed (2250MHz clockspeed, 9000MHz memory rate) and drawing approximately 30W. After a few seconds of monitoring it, it suddenly returns to its normal state (210 MHz clockspeed, 405MHz memory rate) and drawing as little as 1W. I don't know if this is a "feature" or not, but it really bugs me.
Here's some information about my system:
Laptop: Asus ROG Strix 18 (G814JZ_G814JZ)
CPU: Intel Gen 13th Core i9-13980HX
GPU (dedicated): NVIDIA RTX 4080-Mobile | Using nvidia-open 560.35.03
GPU (integrated): Raptor Lake-S UHD Graphics
WM: i3-wm
Offline
Does it really or do those diagnostic tools fire up the card for a moment and then return to normal?
Offline
It is normal, likely inevitable. The driver wakes the GPU to query it.
=> https://wiki.archlinux.org/title/PRIME# … Management
tail /sys/class/drm/card*/device/power_stateto check the power status.
Online
Does it really or do those diagnostic tools fire up the card for a moment and then return to normal?
The latter might be the case, actually. The terminal does stutter for a second or so before querying the results (when i do it for the first time in a while). As long as it's not actually happening while idling, that's fine.
Offline
It is normal, likely inevitable. The driver wakes the GPU to query it.
=> https://wiki.archlinux.org/title/PRIME# … Managementtail /sys/class/drm/card*/device/power_stateto check the power status.
I see, thanks for the information.
Offline