Woes with onboard VGA

[Torkil Svensgaard]

Hi 

We have using VirtualGL on thin client servers for a while on our HPC installation  and it works well. However, when the OS (RHEL 9.3) / xorg starts up it disables onboard VGA which also makes IPMI Remote Console unavailable. 

I can get VGA back by doing these changes:

"
# cat /etc/X11/xorg.conf.d/10-nvidia.conf
Section "OutputClass"
    Identifier "nvidia"
    MatchDriver "nvidia-drm"
    Driver "nvidia"
    Option "AllowEmptyInitialConfiguration"
    Option "PrimaryGPU" "no" <--- changed from yes
    Option "SLI" "Auto"
    Option "BaseMosaic" "on"
EndSection

Section "OutputClass"
    Identifier "intel"
    MatchDriver "i915"
    Driver "modesetting"
    Option "PrimaryGPU" "yes" <---- added this line
EndSection
"

But then VirtualGL stops working. Any suggestions on how to fix this? 

Reinstalled made no difference.

Mvh.

Torkil

[DRC]

You probably need to configure the X server such that the Intel GPU is on Screen 0 and the nVidia GPU is on Screen 1. (You can probably make the nVidia GPU headless, so only the Intel GPU is actually driving a display.) Then you can set VGL_DISPLAY=:0.1 to use VGL with the nVidia GPU. However, please note that you cannot log into the local X server (the “3D X server”) while users are using VirtualGL, as doing so would cause all of VirtualGL’s GLX back end connections to the 3D X server to hang. VirtualGL’s EGL back end addresses that problem by avoiding the use of a 3D X server altogether, but the EGL back end still has some compatibility issues that are being ironed out. I generally recommend using the EGL back end only if you have verified that it works properly for the entire suite of 3D applications that your organization intends to use.

Unfortunately, I cannot provide more specific advice without diagnosing the specific system, which is a paid service.

On Dec 20, 2023, at 6:35 AM, 'Torkil Svensgaard' via VirtualGL User Discussion/Support <virtual...@googlegroups.com> wrote:

Hi 

--
You received this message because you are subscribed to the Google Groups "VirtualGL User Discussion/Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to virtualgl-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/virtualgl-users/4277950b-4fa9-43a9-bffd-26ae40e9f2ffn%40googlegroups.com.

[Torkil Svensgaard]

Hi

Thank you for the pointers. I got it working with this:

"

cat /etc/X11/xorg.conf.d/10-nvidia.conf
Section "Device"
    Identifier             "Screen0"
    Driver                 "modesetting"
    BusID                  "PCI:194:0:0"
    Option "PrimaryGPU" "yes
EndSection

Section "Device"
    Identifier             "Screen1"
    Driver                 "nvidia"
    BusID                  "PCI:129:0:0"
EndSection

Section "Screen"
        Identifier "intel_screen"
        Device "Screen0"
        SubSection "Display"
        EndSubSection
EndSection

Section "Screen"
        Identifier "nvidia_screen"
        Device "Screen1"
        SubSection "Display"
        EndSubSection
EndSection

Section "ServerLayout"
        Identifier "default"
        Screen "intel_screen" 0 0
        Screen "nvidia_screen" Below "intel_screen"
EndSection

"

and 

"

export VGL_DISPLAY=:0.1

"

I couldn't figure out how to get it to work with "OutputClass" instead of "Device" sections but alas, at least it works.

Mvh.

Torkil