Red Hat Enterprise Linux 8 and Nvidia Optimus


Yesterday I decided to go ahead and install RHEL8 on my laptop and use it as my everyday workstation as well as a learning platform to move forward on my Red Hat certification path.

In case you're interested, you can actually use RHEL for free with a No-Cost RHEL Developer Subscription as long as you don't use the machine "in production".

After a pretty straightforward installation of RHEL 8 I was immediately confronted with a first issue. My external HDMI monitor wasn't detected.

My laptop uses the Nvidia Optimus technology to seamlessly switch between the integrated Intel GPU and the Nvidia discrete graphics. I couldn't get it to work with the default opensource nouveau drivers, so I went ahead and installed the non-free Nvidia proprietary drivers.

By default, RHEL installs the (newer) Wayland display server, but the Nvidia drivers don't seem to work with those so I had to switch to Xorg. Either way, I didn't mind since the Nvidia drivers offer better performance.

Installing the non-free Nvidia drivers on RHEL 8

The nouveau kernel module should be blacklisted first, to prevent it from loading during the next boot:

# echo 'blacklist nouveau' >> /etc/modprobe.d/blacklist.conf

The next step will be to install the Xorg display server and the kernel development tools:

# dnf groupinstall "base-x" "Legacy X Window System Compatibility" "Development Tools"
# dnf install elfutils-libelf-devel "kernel-devel-uname-r == $(uname -r)"

Backup and rebuild the initramfs:

# mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r)-nouveau.img
# dracut -f

Make sure you have Dynamic Kernel Module Support installed. This will automatically rebuild the Nvidia kernel modules when you update the kernel, so you won't have to reinstall them afterward.

# dnf install dkms

Download the Nvidia drivers from https://www.nvidia.com/

Change the default runlevel:

# systemctl set-default multi-user.target

Reboot your system and install the driver:

# chmod +x NVIDIA-$version.run
# ./NVIDIA-$version.run

Test the new driver by switching to the graphical.target runlevel:

# systemctl isolate graphical.target

Correct the default runlevel:

# systemctl set-default graphical.target

Configuring Xorg

First, search for your hardware and take note of the PCI bus it's operating on. The PCI bus for my Intel GPU is 00:02.0 and for my Nvidia GPU 01:00.0.

# lspci | grep -EA1 'VGA|3D'
00:02.0 VGA compatible controller: Intel Corporation UHD Graphics 630 (Mobile)
01:00.0 3D controller: NVIDIA Corporation GP107M [GeForce GTX 1050 Ti Mobile] (rev a1)

Next, if you have an existing Xorg configuration (/etc/X11/xorg.conf), make a backup, then replace the existing config with the below. Make sure you modify the BusID's.

Section "ServerLayout"
    Identifier "layout"
    Screen 0 "nvidia"
    Inactive "intel"
EndSection

Section "Device"
    Identifier "nvidia"
    Driver "nvidia"
    BusID "PCI:01:00:0"
EndSection

Section "Screen"
    Identifier "nvidia"
    Device "nvidia"
    Option "AllowEmptyInitialConfiguration"
EndSection

Section "Device"
    Identifier "intel"
    Driver "modesetting"
    BusID "PCI:00:2:0"
EndSection

Section "Screen"
    Identifier "intel"
    Device "intel"
EndSection

The last steps consist of autostarting the dual-display configuration using xrandr. Execute nano /etc/xdg/autostart/nvidia-optimus.desktop and add the following lines :

[Desktop Entry]
Type=Application
Name=NVIDIA Optimus
Exec=sh -c "xrandr --setprovideroutputsource modesetting NVIDIA-0; xrandr --auto"
NoDisplay=true
X-GNOME-Autostart-Phase=DisplayServer

Copy the same file to /usr/share/gdm/greeter/autostart/nvidia-optimus.desktop.

You should now have a working dual display after logging out and logging back in or rebooting!