Hi Joe,

OK this screenshot makes more sense... and is indeed how one would
typically drive multiple monitors with the OSG.

One optimization that is worth doing with this type of setup is use
one graphics context per graphics card, and split the each window
across two displays.  This typically provides the best performance.
But... other posts on similar about Visita/XP suggest that XP is
capable of this, but Visita has lots this capability, which sucks.
Linux is capable of doing the one context driving two screens, so if
can port then you have a solution there.

The only other thing you can try on the OSG side is to change the draw
dispatch serialization flag to off, as right now its on by default.
Use the env var OSG_SERIALIZE_DRAW_DISPATCH set to OFF, then run your
app.   In theory one should always have OSG_SERIALIZE_DRAW_DISPATCH
set to OFF, but in my own testing have found PC OGL drivers/hardware
perform worse when you allow multiple threads driving multiple cards
to all do draw dispatch in a parallel, perhaps later drivers might
fix/hardware this lack of scalability.   I'd recommend that you do a
search on this env on the osg-users archives to see the points I've
raised in the past on this topic.

In your case I don't actually expect the draw serialization to make a
difference to you only getting fps, as the swap buffers aren't
serialized - they all still run in parallel, and your not breaking
frame due to draw dispatch anyway.  What may help is the swap groups
extension - it shouldn't be necessary to get 60fps, but perhaps the
driver is just a bit crappy and needs the extra push to do the right
thing.

Robert.

On Tue, Oct 28, 2008 at 5:40 PM, Joe Lyga <[EMAIL PROTECTED]> wrote:
> Hi Robert,
>
> I guess I forgot to mention what the original problem was.  Normally, the
> program is run across all four monitors.  However, even with practically
> nothing loaded in view, I'm only getting around 30 fps.  I was running it
> independently on two monitors at a time in order to isolate the problem, and
> there's definitely a difference between the first two and the second two.
> Here is the stats of it running on four monitors, but I don't think it shows
> as much as the stats with it on two each at a time.
>
>
>
> _______________________________________________
> osg-users mailing list
> [email protected]
> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
>
>
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org

Reply via email to