Hello,

I have some trouble figuring out the correct Xorg configuration. I have a laptop with a GTX1650 and an interal GPU of the AMD Ryzen 4600H. The Laptop monitor is connected to the internal GPU, while the external monitor is connected to the nvidia card via HDMI. I managed to get it running through the following Xorg.conf:

The XServer expands the Laptop's monitor to the right-hand side. Mouse and window moving is possible, but the problem is that the external monitor extremely suffers from Tearing and the nvidia GPU is always used for rendering. I want the Tearing to disappear and the internal AMD GPU to do Rendering by default, the nvidia gpu only on demand.

There are many things I don't understand at all. Why does the Xserver automatically place the external monitor at the right and the internal one at the left? Why is the internal monitor active at all? In the ServerLayout section only the external one is selected.

I hope you can help me. Tell me, if you need more information.

Thank's in advance.

Max

The problem I see is: AMD & Nvidia.

If you want to use Nvidia proprietary drivers, they will overwrite the Mesa libGL* files in /usr/lib(64) in patent protected ways.
If you install the AMD proprietary drivers, they will overwrite the Mesa libGL* files in /usr/lib(64) in alternative patent protected ways.

To make the drivers fast, you're going to need proprietary drivers. To make them compatible, you're going to have to use slower OSS drivers.

You can see what's going on by posting the version numbers of your amd & nvidia drivers, and the output of 'ls -l /usr/lib/libGL*'
Examine what the symlink libGL.so points at. On Mesa, Mine points at libGL.so.1.7.0. The proprietary drivers will overwrite that symlink to libGL.so.<Nvidia-version> or libGL.so.<AMD_version>.


The AMD GPU is in an APU, right? Sell the Nvidia card, and convert to AMD GPUs is the simplest way out. You may not need the second card. It depends how many monitors you have.

It seems that the proprietary NVIDIA driver does not wipe MESA. Only using the AMD APU is not possible because the HDMI connector is physically connected to the NVIDIA chip.

I've used the nvidia - not today or yesterday, but I've used it.

There's a kernel module, libGL* files and an X driver. I'm not an authority here any more, and you will have to get some others to comment. Some look like nvidia drivers [e.g. libGLESv1_CM_nvidia.so.450.80.02] but they have apparently renamed them in their code to allow Mesa and nvidia drivers to exist. I'm surprised, but pleased.

Don't know what AMD are up to there, so I'm not sure I'm much use to you at all.