Suggestions for properly abstracting multiple screens.

JoJo jojo onetwojojo at gmail.com
Mon Feb 25 21:51:28 PST 2008


Hi Dave

so you are saying
Step 1: Multiseat on discrete GFX
Step 2: Multiseat on dual o/p GFX OR dual o/p IGPs
(and vgaArbitrator in currently at Step1.)

Multiseat is currently supported with multiple gfx,
(each gfx listening on its own PCI,
or dual o/p GFX listening on multiple PCI)

No one is talking about Multiseat with 1 dual o/p GFX,
(much less anyone with dual o/p IGP)
All the multiseat resources (teh intarwebs) talk about multiple GFX
No one is reportedly working on it.

...ergo it does not exist.

of course, I must concede, I cannot correctly predict the
vgaArbitrator relevence,
with all the changes in the X world & the GFX h/w world. I invite you,
to take a shot at that.

(I did watch your ppt on kernel modesetting, mutiseat was not discussed)

we are keenly following your team's 3D work on RS690 as well.

-JoJo

On Tue, Feb 26, 2008 at 10:36 AM, Dave Airlie <airlied at gmail.com> wrote:
> > dude if I could afford to donate 1 board to you,
>  >  I could probably also afford to put in another gfx card and run 2 x.org.
>  >
>  >  probably AMD could donate one of those (<$70) RS690 boards to your team,
>  >  (or SUSE could lend you one of their earlier experimental boards)
>  >
>  >  also, you keep writing code for discrete gfx cards, but the
>  >  technological leaders,
>  >  are going in for integrated GFX, does that bother you? irrelevance.
>  >
>  >  I can understand, its a very fun project for you guys ;-)
>  >  Looks like vgaArbitrator will only handle the discrete gfx card scenario.
>  >  It seems unstated policy for x.org, that multiseat will be supported
>  >  only on discrete gfx.
>
>  Where did you figure that out? I'm interested in the logical
>  progression you have made to arrive at this answer...
>
>  Part of the plan for kernel modesetting is to enable better single GPU
>  multi-seat as well as integrated with the VGA arbiter to enable
>  better multi-card multi-seat..
>
>  These are two mostly different problem spaces, even though to the
>  uneducated it might look like solving the same problem.
>
>  Dave.
>



More information about the xorg mailing list