So the AMDGPU driver has a problem: it sets clock speeds and power limits above what the GPUs actually support, causing unnecesary overclocking and system instability (https://gitlab.freedesktop.org/drm/amd/-/issues/3131).
So if the solution to a driver problem is to manually check a database and set the corresponding values based on a vendor ID and device ID, couldn’t this be automated somehow? I mean, wouldn’t it be better to set these values in the driver itself?
Not looking for a solution or anything really, just wanted to talk about it (hence why I posted in the water cooler and not te issue itself).
I would guess that an option would be for te solution to be automated, but that would be a patch on a patch.
Another solution would be for the driver to set these values from a database they handled. But I don’t know if this would be maintainable in the future.
I really wish this would be automated / fixed because it negatively impacts the experience using Linux with an AMD GPU.