VirtualGL config not working - cannont remove nvidia modules

[SMcDuffee]

Hi

I'm trying to use VirtualGl on a server which I connect to through VNC.  VGL doesn't seem to be configuring itself to use my nvidia GPU and it looks like it's because I need to rmmod nvidia out of the kernel but I can't get it to happen.  I've killed the display manager and also tried booting into a root recovery mode btu I can't get it to work.  nvidia-drm can be removed but uvm and modeset I can't remove.  Can anybody help?  I've tried looking around for info on this but nothing has worked.  When I VNC in and use vgl it runs very slow and looks like it's using a software implementation - not my card.  Thanks for any help.

Sean

[Jason Edgecombe]

Hi Sean,

First, some tips that might get your Nvidia setup to work:

Check to see if you're actually using the nvidia card by running "vglrun glxinfo | grep -i opengl"  in a terminal from within the VNC session. If that says Nvidia, then you're using the Nvidia card.

I had an issue with my Nvidia cards only an abysmal 1 frame per second. I found a fix by setting "HardDPMS" to false in the Device section of the Xorg.conf file.

Here is the bash snippet that I use to fix my servers, but you'll need to modify to your xorg.conf path to fir your situation.

grep -qs HardDPMS /etc/X11/xorg.1.conf || sed -i '/Identifier     "Device0"/i \ \ \  Option         "HardDPMS" "false"' /etc/X11/xorg.1.conf

As for removing the nvidia module, have you tried blacklisting the kernel modules and rebuilding your initrd?

Thanks,

Jason

---------------------------------------------------------------------------
Jason Edgecombe | Linux Administrator
UNC Charlotte | The William States Lee College of Engineering
9201 University City Blvd. | Charlotte, NC 28223-0001
Phone: 704-687-1943
jwed...@uncc.edu | http://engr.uncc.edu |  Facebook
---------------------------------------------------------------------------
If you are not the intended recipient of this transmission or a person responsible for delivering it to the intended recipient, any disclosure, copying, distribution, or other use of any of the information in this transmission is strictly prohibited. If you have received this transmission in error, please notify me immediately by reply e-mail or by telephone at
704-687-1943.  Thank you.

--
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/948d92b2-7ee9-4e22-b543-688a7819d87d%40googlegroups.com.

[DRC]

I need to understand what you mean by "it looks like it's because I need
to rmmod nvidia out of the kernel but I can't get it to happen."  Please
post the console output of what you're trying to do and the relevant
error messages.

Also, have you verified that the nVidia driver is loaded at all?  When
you run glxinfo on the main display (the "3D X server", usually :0)
without invoking VirtualGL, does glxinfo indicate that the X server is
using the nVidia driver?  If not, then you need to resolve that problem
prior to attempting to configure VirtualGL.  Basically,
hardware-accelerated 3D rendering needs to be working properly on the 3D
X server without the use of VirtualGL before VirtualGL can be expected
to work properly.

There is a known issue whereby, on some Linux distributions, the nouveau
(open source) driver interferes with the nVidia proprietary driver, but
I seem to recall that nVidia's installer automatically takes care of
that these days.

DRC

[SMcDuffee]

Hi Jason

Hmm yes you're right.  vglrun of glxinfo does show nvidia with the correct card and driver version.  gears runs crazy slow let alone anything else.  I'll look into that HardDPMS setting.  Thanks

Sean

To unsubscribe from this group and stop receiving emails from it, send an email to virtual...@googlegroups.com.

[DRC]

NOTE: I was unable to reproduce the problem whereby the nvidia module cannot be removed.  In my testing, that only occurred if the display manager was running.  If you have any further information regarding how to reproduce that (for instance, the specific Linux distribution used), then please let me know.

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/a821b3b1-b90e-456a-9305-0333cabeb32e%40googlegroups.com.