On Thu, Jul 8, 2010 at 10:44 AM, Martin Langhoff
<[email protected]> wrote:
> I had to tweak Jamie's example a bit, and it doesn't scream
> hyper-performance, but pushing tons of pixels across USB v2 was never
> going to be fast anyway.

Xinerama works by asking each driver to do the same high-level drawing
operation. If they're all hardware-accelerated, then that's probably a
good thing. If they're both hitting software fallbacks, I suppose
you're getting half the performance you could reasonably expect as you
draw the same pixels twice. I think somebody's working on using one
GPU to accelerate drawing for another, which might lead naturally to
better performance in your particular use case.

>  - There is no mouse / pointer displayed on Screen2 -- this seems to
> be a Xinerama flaw. Is there a way to get the pointer to display in
> Xinerama?
>
> (I wonder if using "software-rendered" pointer would help, but I
> cannot find the swrender options anymore...)

Hmm, yes. I believe the cursor code currently assumes that once it's
found one screen that should contain the cursor, it's done. If that's
true I wouldn't expect software cursors to help. I think overlapping
Xinerama screens are very rarely tested. Perhaps Peter Hutterer can
comment on whether this is an expected behavior and how hard it is to
fix?

>  - The internal screen and desktop are 1200x900 and we cannot change
> that. If I use an external screen with 1280x960 it distorts the
> desktop, if I add Virtual, it ignores my 'modes' line and falls back
> to un-declared 1182x864 which clips the screen.
>
> How can I get 1280x960 with a 'virtual desktop size' of 1200x900?
> (With black/grey/white banding in the margins...)

The intel driver has a "PANEL_FITTING" output property which does
essentially what you want, but it looks like sisusb has nothing
similar. I don't think there's any generic server infrastructure that
will get you that result. You might be able to get your window manager
to keep windows from crossing into the border...

Jamey
_______________________________________________
[email protected]: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Reply via email to