Hi all!
I’ve recently come across an issue with launching games from steam. For now I’ve seen it happen on two games (or rather demos): Crow Country and Sophonce. They launch fine but have some flickering black blocks on the screen that make them unplayable. Here are some images:
https://pasteboard.co/qXH4H5gm7sIS.jpg
https://pasteboard.co/7k9XM394Zlsi.jpg
https://pasteboard.co/6vPC8GYxhHu8.jpg
This only happens when they go full-screen and on wayland, on X11 works fine. I’ve tried many versions of proton including eggroll variants.
The games work fine when launched with wine without steam in the middle if i do wine <game.exe>
. If i do it this way the games detect that I’m running on an ultra-wide screen and play as such but through steam they stay on 16:9 instead.
My system:
Os: Nobara 40
Kernel: 6.8.12-200.fsync.fc40.x86_64
DE: Gnome 46.2
CPU: AMD Ryzen 7 7800X3D (16) @ 5,05 GHz
GPU 1: AMD Radeon RX 7900 XTX
RAM: 32GB
Any help is appreciated,
Thanks!
I know this issue. It’s reproducible when
The problem has long been reported in Mesa project, but nothing was done to help. My bet is that the bug sits in amdgpu kernel driver and not user space.
https://gitlab.freedesktop.org/mesa/mesa/-/issues/8705
EDIT: maybe it’s worth to report in kernel bugzilla or wherever amdgpu kernel driver bugs would go. I don’t reproduce this on my end anymore, because I changed my screen and it uses DP
EDIT2: I could only reproduce it on RDNA2 and yours is RDNA3. I had Polaris (RX 570) for quite some time and it was running on Wayland. Maybe it only happens on newer cards, maybe it’s regression added along the way
Bingo, yes this is the same issue for what they say. I am connected through DisplayPort though on all my screens. Shame that it is not being worked on but from my testing I assume that when steam moves proton to support wayland natively it will be fixed. Thanks for the info, very interesting!!
The HDMI/DP was my wild guess. For now the problem should also go away if you just change in-game resolution for whatever is your screen. Maybe the problem isn’t being addressed because it was reported to Mesa user-space driver, but maybe the problem is in kernel module. It’s also not 100% always reproducible.