I tried to use 550 driver, But in 550 driver it doesn’t detect external display. On 545 it detects external display.
May be anybody can give some script to repair display when it freezes?
Hi.
I have a Dell G16 7630 and i have the same issue
nvidia-bug-report.zip (495.3 KB)
.
For a few days, when I am using external monitors randomly they freeze. Only laptop monitor seems to work, but with some delay.
After restarting, both monitors restart to work… and then they freeze again…
I can not be able to know what is the reason…
I try to update driver version to 550, but with this version both monitor appear not to be connected… so I reverted on 535.
This is my bug report
1 Like
Hi.
Just to chime in, I’ve having this problem with my Lenovo Legion 7:
Arch kernel 6.8.4-arch1-1
RTX 4070 running 550.67
Screen freezes randomly but not constantly. I’ve had days of continuous work without the freeze, but some days it freezes like twice within a few minutes.
Sometimes it recovers when I logout (xfce4/sddm), sometimes I need to reboot.
glxgears runs perfect on all 3 monitors causing no issues.
Noticed when trying to use other tools that require GPU (in this case, LM Studio, ollama) triggers the condition randomly, but not instantly upon execution.
Got a Samsung G7 32" 240hz (down to 144hz) on HDMI and a small 10" on a USB-C to HDMI dongle at 75hz.
Hope this helps.
Nothing has changed after installing latest driver version 550.76, released yesterday.
@amrits, is any progress in solving external monitor freezes? How could we help you to solve it faster?
3 Likes
Good day. Thanks to everyone who’s contributed to this massive but informative thread. I’ve got some data to contribute. I’m temporarily booted into Windows for unrelated reasons and can’t get my exact Linux driver versions from here, but I didn’t want to lose track of this thread. So I’m going to post from memory now and will edit the details after I boot back into Linux.
I’m able to reproduce the external monitor freezing on the following environment:
- System: Dell Precision 7740 mobile workstation, 8 core Intel CPU, 64 GB system RAM
- GPU: NVIDIA Quadro RTX 4000 Mobile, 8 GB VRAM
- Operating System 1: Linux Mint 21.3 (based on Ubuntu 22.04) – Reproduces problem
- Operating System 2: Windows 10 – Does not reproduce problem
- NVIDIA Driver (Linux): 550.54.15 installed from NVIDIA APT repository
- NVIDIA Driver (Windows): 535.99 installed from NVIDIA website download
- Laptop panel is 4K (3840x2160) and is always used as primary
- External monitors are each 4K LG connected via HDMI, in “Join” mode, one at a time (different offices)
- Have both Wayland and Xorg packages installed, but AFAICT what’s actually running is Xorg. (See my note at end of post.)
Details of Problem
- Reproducible if GPU switching is enabled in BIOS (i.e., Optimus/PRIME active)
- Occurs when primary monitor is running a dedicated GPU app such as Unreal Engine Editor
- So far has been observed only when at least one app is on secondary monitor
- So far always had one or more non-dGPU apps on second monitor
- Can occur even when there is also a dedicated GPU app (again, UE Editor) on second monitor
- Definitely NOT reproducible if integrated Intel GPU is disabled in BIOS
- Definitely DOES reproduce with prime-select set to “on-demand”
- (Updated) Problem does NOT reproduce with prime-select set to “nvidia” (based on several hours of runtime)
- Unplugging and replugging the secondary monitor reliably clears the problem (but it will recur later).
TL;DR: The problem is reproducible on driver 550.54.15 and on Quadro GPU, and only (for me) if Intel GPU is also enabled.
Anecdotally, I have noted that some app startups and window refresh for non-dGPU apps are distinctly sluggish, and when I read someone’s comment in this thread about slow data copying between iGPU and dGPU, my background (long ago) in hardware design led my “engineer’s spidey-sense” to note that the specific type of lag I’m seeing (which looks like BitBlt functioning at almost-normal speed but lagging by a few frames) may be consistent with that data copy issue.
Finally, I’ll ask that folks be gentle with my naivete on Linux graphics issues. I’ve spent a couple decades creating infrastructure tooling on headless Linux servers, so I know Linux but don’t know Linux graphics stacks. My most recent Linux graphics knowledge came from building X11 and KDE from source to run on kernel 2.x, because that’s how you installed things back then, when FORTRAN was only THREETRAN! :)
(Updated 2024-05-09 to add exact Linux driver versions and latest test results)
4 Likes
Hi @scott179 , thanks for the detailed info and thorough testing. Regarding the copy performance issue between iGPU and dGPU for the external monitor, the code that mitigates the problem on Wayland (but doesn’t completely solve it, from what I understood) has been merged on GNOME 46, and the corresponding GNOME mutter issue has been closed. It is still unclear if this GNOME shell extension (hack) is also necessary (on my personal experience, it seems it is).
1 Like
Thanks for the additional info. It’s good that the problem’s been addressed upstream, though I’m running mutter 42.9 so won’t see the change for a while. I’m glad you pinged me, because I’m booted in Linux today and need to edit my post with the version info as promised. :) [Done]
1 Like
Just tried new driver version 550.90.07 and it still get freezes while running glxgears and vkcube with resizing last one. PRIME offload still unusable due to this freezes. Moreover, it gets freezes while resizing only running glxgears!
@amrits, is there any understanding when this bug will be fixed?
Since my last post, I happen to have upgraded to a new Dell Precision 7780 with the maxed-out Quadro RTX 5000 Ada, and I am reproducing this problem frequently on driver version 550.90.07, with Arch Linux (new install four days ago). I was running Linux Mint when I previously encountered this bug.
Using glxgears behind vkcube reproduces the problem instantly for me, with the proviso that it has to be done on the external monitor and not on the laptop screen.
I have reproduced this on two different HDMI monitors with different cables, and now on two different laptops.
[Deleted the paragraph that was here I discovered it was incorrect. See note at end of post.]
I would post my logs, but there is nothing in any log files on my system since the last occurrence of the error that is related at all to graphics – just some routine stuff about the wifi adapter rescanning for nearby SSIDs, and DHCP client refreshing my IPv4 address on the wired Ethernet connection.
UPDATE: The iGPU and dGPU were both enabled when I posted this message earlier today (I apologize for that error). I’ve now disabled the iGPU and Optimus and am running only on the dGPU. The problem seems better so far. With Optimus enabled, I could reliably reproduce the bug with glxgears and vkcube, and apps were causing it randomly every few minutes. With Optimus disabled, I’m now using both monitors and – so far! – no freezes. I repeated the glxgears+vkcube test and did not observe the problem with the iGPU disabled in BIOS.
@amrits I saw that you said you could reproduce the issue, but just in case the method you have right now is slow, or you just want an alternative, this is what I’ve found.
I’m on ubuntu 24.04 and the GNOME night-light causes this external monitor freezing as well.
First make sure that the current time of day is in the night-light time-window. These commands set night light to be on all day.
gsettings set org.gnome.settings-daemon.plugins.color night-light-schedule-from 6.0
gsettings set org.gnome.settings-daemon.plugins.color night-light-schedule-to 6.0
Next, run this command to repeatedly turn night-light on and off. The sleep seems to make it freeze faster. My external display is usually frozen within a couple seconds.
for i in {1..100}; do gsettings set org.gnome.settings-daemon.plugins.color night-light-enabled true; sleep 0.3; gsettings set org.gnome.settings-daemon.plugins.color night-light-enabled false; sleep 0.3; done
Can anyone else using GNOME confirm that this also freezes their external monitor?
Of course you can just manually turn it on and off, but maybe it takes a little longer for your screen to freeze than mine.
1 Like
I randomly decided to try upgrading to 550.90.07 (open version) and it is stable for me. Thinkpad P1, 3070, Ubuntu 22.04 on kernel 5.15.0-112-generic. Earnest thanks to everyone who has been working on this issue, hopefully it is not a fluke!
I have the same problem with a GTX 1660 ti on Debian 12 and Ubuntu 24.04. It seems that when only the external monitor is selected, that is, using only the dGPU, the system works without freezing. However, the bug appears when using Intel and Nvidia graphics simultaneously, no matter the driver version.
Did you try to run glxgears/vkcube test with external monitor connected to your laptop?
For what it’s worth, this bug only occurs using X11. Switching to Wayland and using PRIME in the same way with glxgears/vkcube, the issue does not occur.
I can confirm that the issue still remains on 555.58 on EndeavorOS. Switching to Wayland is unfortunately not an option for me due to various issues regarding NVIDIA driver performance on KDE (with a second monitor connected), as well as libinput never supporting any form of universal trackpad kinetic scrolling.
So I would, like many others, like to see this issue fixed as it hampers every day usability for any use case that needs to use the NVIDIA GPU (e.g. video editing). I’ve kept this thread open for half a year now hoping to see some developments on this issue, to no avail.
Did you try to run glxgears/vkcube test with external monitor connected to your laptop?
I tried it today with glxgears, no issues, though I use the trackpoint so I cannot resize so quickly ;)
I run a 4k monitor over HDMI, 30 Hz, 2x scaling.
I’ve tried to run latest version ( 550.100) of open NVidia kernel modules without success. This kernel modules with latest driver version produces exactly same freezes in glxgears/vkcube resizing test. And after some vkcube window resizing attempts I’ve got total freeze of external monitor, connected through HDMI connector. So for me these drivers and kernel modules don’t change anything.
I have the same problem.
I have an MSI Katana with RTX 4070, no problems on windows, I’m using Zorin OS (that is basically ubuntu, everything that works in ubuntu works in zorin) with Xorg X11.
The external screen randomly (more or less) freezes, especially when using an application such as Minecraft or Blender that runs on the Nvidia GPU. This is very annoying, I’ve tried lots of things and none work. The GPU works perfectly, I mean, i can render scenes, play videogames, it’s a breeze, but it freezes the secondary monitor (that I use a lot) and I can’t manage to fix it. Please help, fix it. If you need logs just tell me.
Edit: I switched to “Performance Mode”, using driver version 535 and it seems to be working. Though obviously it consumes more power.
Has anyone tested the latest open source 560 beta driver?
Maybe this time the bug is fixed…
I’ve just installed latest driver version:
Linux x64 (AMD64/EM64T) Display Driver
Version: 550.107.02
Release Date: 2024.7.29
Operating System: Linux 64-bit
Language: English (US)
File Size: 293.02 MB
And I’m very surprised with it’s testing results: previous driver versions (including 550.100) making freezes with vkcube resizing over running glxgears window, but 550.107.02 gets freezes while resizing only the glxgears window with Prime offload enabled (no vkcube is running). It seems, that we have negative progress…
@amrits, should 550.107.02 fix the glxgears/vkcube freezes while resizing their windows? Or it should fix only total discrete GPU display freezes? What is the expected behavior after installing 550.107.02 on laptop with NVidia dGPU running Prime offload?
PS: I’ve got again a total NVidia dGPU output freeze after intensive resizing running vkcube window. So this new driver version doesn’t change anything for me.
1 Like
Running Gentoo with kernel 6.10.2 and nvidia-drivers 560.28.03.
The problem is still there. I can’t run any graphics application or game on the dGPU without my screen freezing.
This happens on an Optimus laptop only. And if I activate the MUX switch and force the use of the dGPU, the problem isn’t there, but other problems occur :D
This is on an RTX 4070 and AMD Ryzen 9 8945HS.
Update:
Running the laptop on the dGPU with amdgpu installed and screens configured doesn’t freeze X at all. This seems to be the solution for me at the moment as I’m always docked and connected to the power supply when I use external monitors.
I would love for Nvidia to fix this though. I’ve had this issue happen on a laptop with Intel i7-12700H and RTX 3050Ti.