[Bug] External monitor fails to wake up from powersave mode if refresh rate is higher than 30Hz

I have an external monitor connected to my Advanced Optimus laptop.
The cable connecting the monitor and laptop is confirmed to be HDMI 2.1
The dGPU on the laptop is an nVidia 4090 laptop edition.

When this monitor is configured for an update rate of 30Hz, it wakes up normally from powersave. When higher display update rates are selected, it looks like the monitor is waking up, but it claims that there is no signal present and goes back to sleep.
A second or so later, the monitor looks like it wants to wake up again, but then goes to sleep.

I filed this bug report at KDE:
https://bugs.kde.org/show_bug.cgi?id=496631

There is a short video of how this looks like on youtube:

On a side note, this monitor states that it is capable of 144Hz, but the best I have seen in Windows and Linux is 120Hz.

Here is the output of nvidia-bug-report.sh:
nvidia-bug-report.log.gz (1.6 MB)

Any help in troubleshooting this issue would be greatly appreciated.

Hi there!

This seems to be related to the version of the nvidia drivers installed on my system.
I see the issue with version 565.77, and when trying an older version of the driver, 550.147, the issue is gone.

1 Like

One more data point to this bug is that when I use a USB-C to DisplayPort cable, this bug is still present on the latest released drivers, but not on 550.147

Since the last NVIDIA update with the new option for GAMES or Graphic Design ect… Same thing here. Gotta roll back or totally uninstall NVIDA drivers for my screen to work properly. Been waiting over a month for it to be fixed but still, No New Updates out. This is bad on their part.

I am not alone!!!
Probably should not be happy about this, but at the very least it means that this bug is reproducable on other computers. I am now running a single screen, so this bug is not affecting me at the moment. In a couple of months when I return home, I will have to downgrade my nvidia drivers again.

1 Like

Oh, just read something else:

565.77 is considered testing branch, and 550.144 is considered the stable branch. So, on the stable branch, this functionality works.

This bug is present in 570.124.04, which is the latest production release from nvidia

I observe the same thing, plugging from a 5080 on a 4K Samsung TV via HDMI (60Hz). Have you tried however to use CTRL+ALT+Fn2/3/4 to switch to console?

I use SDDM and this used to wake the TV/monitor. However I recently made SDDM run over Wayland, and this fails now (I use CompositorCommand=kwin_wayland --drm --no-lockscreen --no-global-shortcuts --locale1, maybe this is what causes it). I will do further experiments to reproduce.

I can confirm that if I run SDDM on X11, then I can use CTRL+ALT+Fn to switch to console (then back to desktop, Plasma compositor in Wayland) and go back from powersave. If I run SDDM on Wayland then I cannot do this, even without specifying a CompositorCommand.

EDIT: Actually it finally wakes up, switching to a TTY, but it takes many minutes.

The SDDM issue above may be related to some other bug in kwin_wayland or something else. That said I cannot wake up the screen with the mouse, which I could before upgrading to a 5080. Need to use SDDM on X11 and switch to a console before switching back to Plasma.

The easiest workaround I found is to just leave the external display at 30Hz. If I do forget to switch it, then I have to open display properties on the panel, set it to 30Hz to get the display to switch on, and then I can use the higher frequencies again.

Hi @evorster,

Apologies for the delay. We have filed an internal bug for tracking - #5178214.

  1. Do you mind verifying that this remains an issue on our latest driver 570.133.07 available here - Driver Details | NVIDIA
  2. If it continues to be an issue on 570.133.07, please capture a new NVIDIA bug report after you reproduce the issue. I have added the older bug report and your journal logs to the bug for Engineering review.
  3. Can you please provide the EDID for the external monitor. The steps to capture it are available here - Managing a Display EDID on Linux | NVIDIA. Please attach the EDID saved in binary format.

Thank you

Hi @ericgirard73 , Sorry for missing this report. Can you please capture a NVIDIA bug report so that we can confirm its the same issue as evortster’s report. Are you using the same BenQ EX3210U display?

Hi @zebcom,
The description of your issue sounds similar to Bug: 570.124.04 - freeze on monitor wakeup - "Flip event timeout"
Please help verify if this is resolved on 570.133.07 whenever possible

Thanks!

I can confirm that this issue is still present on the latest beta drivers.
The output of nvidia-bug-report.sh and the edid for the monitor is attached.

nvidia-bug-report.log.gz (1.4 MB)
BENQ.edid.zip (348 Bytes)

I will many thanks. Thing is I do not see these kwin_wayland_drm errors in journalctl, does it require more verbosity? Will try with the new driver on Arch Linux and report back.

Hi @abchauhan Thanks for checking it out. I think there is some progress, I am now able to switch to TTY console, but it may take a few minutes before it shows on the screen. I am unsure if pressing CTRL+ALT+Fn 3 works better when screen is on or off though. I will investigate more and report.

@abchauhan I describe the bug in NVIDIA Driver 570.86.15 Breaks Wayland Display Restoration on RTX 4070 SUPER with Samsung G95SC - Fixed by Downgrading to 550 - Graphics / Linux / Linux - NVIDIA Developer Forums. This is very weird. After screen is off, the only way to wake it up later is to switch to TTY3 while it is still off. If trying to switch while turned on, nothing wakes it (“no signal”). If I sitch to TTY3 while it is off, then I can turn the screen on and see the terminal (and can switch baclk to TTY1).

Hi
I’ve been trying for 30 min to find that bug report and I can’t and when I run the script it says that it need to be unblocked from my nvidia settings and I can’t find at all where to go. Using web directions is not even helping. So, I’m stock, sorry!

Hi there!
I have an open incident: 250307-000791
Would that help you to track it down?

Hi there!
I managed to sort this issue on my setup by upgrading the firmware of my external monitor.
It is a Benq EX3210U, and it’s firmware went from V3 to V17. Along with fixing this issue, it also enabled 144Hz through the normal HDMI connector, which is nice.
For me, this issue is now completely fixed.