I don’t appear to have the x11 one, but I do have egl-gbm:
paru -Qs egl ✔ 01:46:04
local/egl-gbm 1.1.2-1
The GBM EGL external platform library
local/eglexternalplatform 1.2-2
EGL External Platform interface
local/freeglut 3.6.0-1
Free OpenGL Utility Toolkit
local/lib32-libglvnd 1.7.0-1
The GL Vendor-Neutral Dispatch library
local/lib32-wayland 1.23.1-1
A computer display server protocol
local/libglvnd 1.7.0-1
The GL Vendor-Neutral Dispatch library
local/nvidia-egl-wayland-tkg 565.57.01-260
NVIDIA EGL Wayland library (libnvidia-egl-wayland.so.1.1.17) for 'nvidia-utils-tkg'
local/wayland 1.23.1-1
A computer display server protocol
As for the egl-wayland, no, I was using the non git version from the official arch repo, which is still p bleeding edge tbf.
But when it broke, I switched to the one provided by nvidia-all, nvidia-egl-wayland-tkg 565.57.01-260.
I should mention that I was on Hyprland, both when it broke and got fixed, that might play a role in it since hyprland has it’s own protocols afaik, aquamarine.
Since driver 555 I’ve been experiencing wine-vulkan errors after random intervals under Wayland and I still am as of 565.57.01.
Edit: I can reproduce this by playing KINGDOM HEARTS II FINAL MIX for a while and eventually it will happen. Sometimes though the game will instead just deadlock the rendering thread without the error popup.
Edit 2: Here’s the nvidia-bug-report.log.gz (324.1 KB) taken after this error occurred but before closing the application.
That’s what I thought, too. I wonder if anyone else with a Nvidia dGPU is facing the same problem (external monitor plugged to HDMI is not recognized). If 6.12 goes stable like this, it will be a no-go for me, as I depend heavily on an external monitor for daily work ={
The weird thing is that inxi -Gxx recognizes both the GTX 1650 dGPU and the external display, but GNOME display selection only shows the built-in display:
I’m using GSK_RENDERER=vulkan right now and it works for me after recent gnome packages update (previously i had the same problem as you with gnome apps not opening). But maybe it’s because i have newer packages from arch.
Cool! But is suspend/resume working just fine for you? Would you mind sharing your setup (desktop? notebook? iGPU? dGPU? external monitor?) so that I could try to compare with mine?
BTW I just saw there is a new 6.12rc7 on Rawhide, I’ll try it later on to see if I can finally use an external monitor again.
EDIT: just tested with kernel 6.12.0-0.rc7.59.fc42 (latest from Rawhide), external display is not available.
❯ uname -r
6.12.0-0.rc7.59.fc42.x86_64
❯ modinfo -F version nvidia
565.57.01
I’m seeing a lot worse FPS, and higher frametimes in Vulkan applications under Linux, than under Windows. This is not exclusive to programs running under Proton, but it’s especially prevalant there. It’s also in native applications that use Vulkan.
I’ve run the same programs under Windows 11 (LTSC 24h2) and under CachyOS with the bore scheduler, and with the eevdf (standard) scheduler. And it’s consistently better under Windows. I even installed Windows in a Qemu virtual machine and passed the nvidia GPU through to the virtual machine, so that the virtual machine adheres to the Linux kernel and scheduler, just to rule that out. And the Windows virtual machine was still consistently faster in Vulkan applications than those same applications underneath the host Linux OS.
I’ve been seeing a lot of flip event timeout on head 0 errors lately on Nobara 40. Usually happens when I have a YouTube video running. The screen will lock up, but I can still hear the sound. I can also hit the space bar to pause. The only workaround I have found so far is to switch to TTY3 and then after that loads (without putting my login info in), I switch back. 9 times out of 10, the desktop environment is responsive again. The 1 other time out of those 10 times, it won’t be fixed and will require a hard reboot.
This usually happens several times a day, a lot of times happening within 10s of minutes of each other.
I’d love to go to the bottom of the sleep/suspend issues… X11 works fine, but Wayland just cannot recover on my side, no matter what I try. The best I could get is a GDM screen with missing visual assets. Far better than the unresponsive black screen I was getting before, but still broken.
6.6.59-200.fc41.x86_64 with 565.57.01
Desktop PC with GDM/Gnome Wayland on a RTX 4800
I feel your pain… same here :-( It’s pretty bad as it is, and to make matters worse, kernel 6.12 which was supposed to fix these issues brings another problem for me by losing the ability to detect the external monitor. I’ll have to choose the lesser of two evils, which will mean I’ll be stuck with kernel 6.11 and broken suspend/resume.
A couple of questions regarding your configuration:
is the “mode” by itself on the first line correct?
until now, I wasn’t using nvidia-persistenced and was getting expected suspend/resume; also, it is not mentioned on the HOWTO; are you sure it is needed?
My GPU (GTX 1650) doesn’t support S0ix power management, so I’m out of luck here. As for the PreserveVideoMemoryAllocations I was not using it either, do you know if this became mandatory with the 565 driver?
From reading the documentation, my understanding is that we do not need the persistenced service running.
For PreserveVideoMemoryAllocations, I do think we need it. I’ve seen mentions of it while reading some discussions in Arch forums. I’d love to get a proper explanation of what we need and why it fail…
Thanks for the info. As for the PreserveVideoMemoryAllocations, my understanding from the docs is that it allows proper/full suspend/resume at the cost of more disk space:
The GPU state saved by the NVIDIA kernel drivers includes allocations made in video memory. However, these allocations are collectively large, and typically cannot be evicted. Since the amount of system memory available to drivers at suspend time is often insufficient to accommodate large portions of video memory, the NVIDIA kernel drivers are designed to act conservatively, and normally only save essential video memory allocations.
The resulting loss of video memory contents is partially compensated for by the user-space NVIDIA drivers, and by some applications, but can lead to failures such as rendering corruption and application crashes upon exit from power management cycles.
However, so far It didn’t seem to be necessary for may day-to-day use. YMMV
I’ve tried giving as much information as possible to the nvidia team in my comments on this post. Including bug reports and full repro steps for quite a few bugs. However after weeks on weeks (almost 6 months on some), I’ve seen no reply or acknowledgement from the nvidia team for many of my issues. To make it easier for you, I will link my current issues below for you to look through and create internal bug reports:
Still exceedingly high vram usage in games (spider-man remastered, doom eternal, assetto corsa) and wayland desktops (Hyprland). Supposedly fixed, but isn’t:
Nvidia drivers not assigning correct monitors, leading to opposite display for bios and for tty. without fbdev both bios and tty display on my 1440p monitor, with fbdev, bios is on main monitor, tty is 1080p and very small on 1440p monitor:
Exceedingly high power draw / high clocks when cuda apps are running. Issue is getting out of hand now, needs a fix soon. My fans on my gpu spin up just when watching a youtube video. This should not be happening:
hdmi doesn’t work on 6.12 on asus rog zeph g16 2024 (4070)
I use external monitor >85% of time, so had to revert to 6.11, but at least suspend works nicely on 6.12 now
