You are not logged in.
Pages: 1
So my 1080Ti
Linux Kernel:
Last edited by Histole (2026-08-25 05:33:09)
Offline
You should check the behavior w/ an uncomposited X11 WM but iirc any kind of multihead implies activation of the composition pipeline.
(You'd also get this on a single output when scaling or rotating it?)
Online
You should check the behavior w/ an uncomposited X11 WM but iirc any kind of multihead implies activation of the composition pipeline.
(You'd also get this on a single output when scaling or rotating it?)
So if I connect just the one display, it idles in P8. Though if I have two monitors connected it idles in P0. I had this issue on windows before I dove into Linux, and it was solved via nvidia inspector’s multi display power saver.
I could check on X11, but I think the benefits of Wayland outweigh the power savings on X11.
Last edited by Histole (2026-06-16 18:11:33)
Offline
The point is to confirm that this isn't just a glitch of your wayland compositor and the estimate actually that you'll not get any different behavior there ![]()
https://forums.developer.nvidia.com/t/n … /118113/20
I assume this hasn't been addressed, but users there report that lowering the refresh rate or disabling gsync allowed their GPUs to downclock (might completely depend on the generation)
Online
The point is to confirm that this isn't just a glitch of your wayland compositor and the estimate actually that you'll not get any different behavior there
So I haven’t gotten around to trying X11 yet, as I was working out the Ly issue. I suppose I should add that on Windows I also had a similar phenomenon, albeit it wasn’t in P0 all the time, it did occasionally drop to P8. What fixed it was using Nvidia inspector’s “multi display power saver” which forces the GPU to P8 unless an excluded application is ran. Is there anything similar on Linux?
Will report back once I figure out how to try X11.
Last edited by Histole (2025-08-09 00:25:57)
Offline
I should add that on Windows I also had a similar phenomenon
you've actually mentioned that
nvidia-smi allows you to set a power limit, but neither is that the performance state nor is capping this/freezing the performance necessarily great.
users there report that lowering the refresh rate or disabling gsync allowed their GPUs to downclock
Online
you've actually mentioned that
Oops yes you're right, wasn't paying attention.
users there report that lowering the refresh rate or disabling gsync allowed their GPUs to downclock
Last edited by Histole (2026-08-25 05:32:52)
Offline
This would be more about the crtc than the GPU (ie. the thing that needs to pump put the pixels, not the one that calculates their colors)
Online
Pages: 1