On Sat, Dec 28, 2024 at 7:39 AM Gert Vanhaerents <gert.vanhaerents@xxxxxxxxxxx> wrote:
In the meantime I have contacted everyone who could have something to do
with it:
Kernel groups
System D
Nvidia
And gues: Everyone says it's not their fault.
But we don't give up. Linux is such a beautiful and solid system. Why
would it work with Windows and not Linux?u
Our analysis has now discovered that the problem does indeed come from
the kernel. The kernel does not allow several users to access a graphics
card at the same time. Indeed, to use it, several users need access to
the graphics card at the same time.
Can this simultaneous access be allowed by the kernel after all?
Or can the kernel be adjusted so that this would be possible?
In the meantime I am also working with a multiseat company to search
together for a solution for multiseat under Linux. This company is
working on creating an easy and user-friendly program to work via a
multiseat with Linux, and they have already a high-selling multiseat
program for Windows. The intention would be to create a program that
works like some multiseat programs in Windows: install, drag mice and
keyboards to the right screen, restart and it works as multiseat.
But such a program is only interesting if it can work for different
users on 1 video card. For example, for a multiseat of 6 users you need
3 video cards with 2 outputs each.
For now the programmers has a good béta program, but with the
restriction only one workstation on one video card, but then it's not
interesting at all.
You don't mention which GPU you have. Pre-Kepler GPUs can only scan out 2 outputs at a time, so even if you have 3+ connectors on the board, you can still only have 2 of them active at a time. This is a hardware restriction. Kepler+ enables 4.
If you check "lspci -nn -d 10de:" it should give you all the NVIDIA hardware in your machine.
Cheers,
-ilia
P.S. Looks like I provided this very same advice to you without response on Jan 16, 2024. So it's an annual thing?
