Nvidia ssh -X

Hi all,
I have two computers. One with an external graphics card Nvidia RTX 5000 (PC1) and the other only with an integrated graphics card from intel (PC2). I would like to log in to PC1 from PC2 using ssh -X and run programs that would use the RTX 5000 and not the integrated graphics card. Unfortunately, I can’t do it.
Any advice?

Thank you

Define “use RTX 5000”. What exactly final result should be? What do you want to run and what do you expect to happen?

PC1 uses Nvidia RTX 5000 for KDE and other programs. I want that when I log in via ssh -X from PC2, I use all programs that I run via ssh -X to render Nvidia RTX 5000.

When you use ssh -X from PC1 to PC2, all graphics you see on PC2 are generated on PC2 (triggered by the high level intsructions send over the connection). Thus the hardware (and the drivers that belong to them) on PC2 is used.

Programs that you run in “ssh -X” session talk to X server on client side. You may be able to use nVidia via CUDA aware programs, but this certainly not “everything”.

Look at running headless X server on PC1 and using something like VNC to connect to it from PC2. This way programs will run and render locally and content of display will be forwarded to remote system. See x11vnc, xpra or similar.

I have found that a “Quick and Dirty” way to setup a machine for “ssh -X” is to simply setup VNC using YaST

https://doc.opensuse.org/documentation/leap/reference/html/book-opensuse-reference/cha-vnc.html

After you set that up on the server side,
It will also support “ssh -X” if you have the port open and the ssh daemon running.

TSU

Hello all,
I solved this problem with help virtualgl.
Thank you for your help. :wink:

Your original request was “everything must be rendered locally” and virtualgl only redirects OpenGL, so it is not “everything”. But whatever works for you …