My external monitor keeps freezing whenever I run a D3D12 game (in that case FH4).
I’m using optimus-manager, and set it as hybrid mode. The game is running on the NVIDIA GPU.
I either have to restart my computer, or go to a TTY then go back to my desktop to have it working again.
RTX4070 + AMD Ryzen 7 8845HS on EndeavourOS.
Note that the integrated display of my laptop is disabled, I only use the external one.
edit: thx @jtellm, adding VKD3D_DISABLE_EXTENSIONS=VK_KHR_present_wait lets me run the game. External display will still randomly freeze, or whenever I alt-tab out of the game.
Yeah, they release a “BETA” driver, and ignore almost all feedback. But that’s nothing new unfortunately.
We are paying customers and have worse feedback from a trillion dollars corporation, than a random hobbyist project on Github. Sad.
I’ve been able to use the driver for a bit, and it worked perfectly in the beginning, but now that i’ve updated to the latest linux kernel, i’m only able to boot with the LTS kernel, and all wayland window managers are either completely broken (Hyprland doesn’t work), or mostly unusable (kwin, it does some weird stuff with my mouse cursor being duplicated, and the center of the mouse cursor being way below the actual mouse cursor. Also low fps on the desktop without anything else opened).
I have the problem consistently with Nioh 2 and only Nioh 2 (freeze after 1/4 h or when a special shader is used, I don’t know). A friend of mine I was co-oping could play without freezes, but all of a sudden it started happening to him too. Both of use just use Steam+Nioh 2, and we didn’t use any special settings that I’m aware of. I can screenshot and post my in-game settings if you want.
This may be unrelated to 565 drivers, but I’ve been doing some more testing for Nioh 2, and I could reproduce the 565 freeze, even on 550, but only when using gamescope (version 3.15.14). Maybe it’s a coincidence, but it’s the exact same behavior as playing with the newer drivers.
With these params (base gamescope without any upscaling), NO FREEZE, I played for about 45 minutes and it was very stable:
gamescope -- %command%
With these params (lower resolution with basic upscaler + sharpness), I freeze 1 to 5 minutes after starting a mission:
(for some reason it seems that --sharpness 0 is the highest sharpness and 20 is the lowest)
If I manually turn on FSR or NIS during a mission, mess around with the sharpness, it eventually triggers the freeze
gamescope -- %command%
...
[during a mission]
SUPER + U [Enable FSR]
SUPER + Y [Enable NIS]
SUPER + I, SUPER + O [-/+ sharpness]
Maybe this specific game really doesn’t like being injected with any kind of post-processing, I don’t know if the newer drivers do something like that by default, I did have some custom Profiles in nvidia-settings with some sharpness, but they are completely disabled (and also still there on 550, which didn’t cause an issue, I played for 6 hours yesterday without a single freeze)
The freezes don’t seem to happen if I just stay in the menus (or maybe they take a longer time to happen and it didn’t happen to me)
Just throwing everything at the wall seeing what sticks right now
Hi, I would like to provide new information about bug ID [4825128] (freezing in games), and also complement my old report:
In the last few weeks, I had the opportunity to test the nvidia 565 further and better understand whether the freezing would still persist or not.
In general, in my Arch Linux - XFCE4 - X11 environment, to play without any problems, I use the nvidia 555.58.02-16 drivers. I play on an external monitor, using optimus-manager in hybrid mode. All steam/proton games runs only in my dedicated nvidia GPU, and browser/discord/any other software runs mostly in my integrated intel GPU. Also, all the nvidia drivers that I have install and testing are from Arch Linux AUR (nvidia aur).
At the beginning of November 2024, I tested the nvidia 565.57.01-2 version and as I previously reported, it made the games freeze. And a quick way for me to check this behavior was through Euro Truck Simulator 2, where it would freeze moments before reaching the start menu.
Last week I was testing with nvidia 565.57.01-8 and to my surprise, the initial freezing that always happens in Euro Truck Simulator 2 no longer occurred. I tested this version more extensively by playing Metro 2033 Redux and Metro Last Light Redux and after hours, the games stills freeze (3 times in my tests) but only if I had something in the background using my GPUs, such as: streaming, video, discord call, etc. I highlight here that I played for long hours without any problems, and the three times it froze were when I was either in an audio call on discord or listening/watching a stream.
Furthermore, I didn’t have the chance to test in games in DX12 (vkd3d), I only tested in DX11 (dxvk) and all with Proton-Experimental [bleeding-edge].
Anyway, I don’t know if this information helps in any way, but I found the behavior of the driver versions in relation to this freezing problem curious. In short, this is what I have noticed happening in my environment:
nvidia 555.58.02-16: not freeze. Stable to play without any problems.
nvidia 565.57.01-2: freeze. Some games take longer to happen, but in Euro Truck Simulator 2 it is “simple” to reproduce because it happens right from the start.
nvidia 565.57.01-8 freeze. But only if you are using the GPUs in the background (watch streaming, discord call).
I’ve also started to experience gpu hangs and freezes. Not sure if it’s related to kernel version 6.12 or a recent update in pipewire. Usually there’s an audio glitch seconds before.
These hangs/freezes are really bad, most times I need to shutdown forcefully with the power button.
Dec 03 12:59:03 hostname1 pipewire[1182]: pw.node: (alsa_output.pci-0000_0a_00.4.analog-surround-21-93) graph xrun not-triggered (352 suppressed)
Dec 03 12:59:03 hostname1 pipewire[1182]: pw.node: (alsa_output.pci-0000_0a_00.4.analog-surround-21-93) xrun state:0x7a9e15a95008 pending:1/2 s:4109389497543 a:4109392185351 f:4109392197821 waiting:2687808 process:12470 status:triggered
Dec 03 12:59:03 hostname1 pipewire[1182]: pw.node: (Rocket League-56) xrun state:0x7a9e172ed008 pending:0/1 s:4109392185351 a:4109342500139 f:4109342502899 waiting:18446744073659866404 process:2760 status:triggered
Dec 03 12:59:03 hostname1 kernel: NVRM: Xid (PCI:0000:08:00): 32, pid=17600, name=RocketLeague.ex, Channel ID 00000048 intr 00008000
Dec 03 12:59:03 hostname1 kernel: NVRM: Xid (PCI:0000:08:00): 32, pid=17600, name=RocketLeague.ex, Channel ID 00000048 intr 00008000
Dec 03 12:59:05 hostname1 kernel: NVRM: Xid (PCI:0000:08:00): 56, pid='<unknown>', name=<unknown>, CMDre 00000007 0000041c ffffffff 00000004 00800000
Dec 03 12:59:10 hostname1 kernel: NVRM: Xid (PCI:0000:08:00): 79, pid='<unknown>', name=<unknown>, GPU has fallen off the bus.
Dec 03 12:59:10 hostname1 kernel: NVRM: GPU 0000:08:00.0: GPU has fallen off the bus.
Dec 03 12:59:10 hostname1 kernel: NVRM: kgspRcAndNotifyAllChannels_IMPL: RC all channels for critical error 79.
Dec 03 12:59:10 hostname1 kernel: NVRM: _threadNodeCheckTimeout: API_GPU_ATTACHED_SANITY_CHECK failed!