Hi Harold,

I belive you haven't got any replies because your post was too
confusing to understand exactly what your seeing at your end.  For
instance you provide numbers but don't say what they are.  If you want
help you'll need to provide more specific information about the
results your are getting along with the units, and also provide info
on what OS, hardware and the nature of your models.   Also use the OSG
stats handler to get more information about the make up of your frame
- it'll tell you and others alot about what is the bottleneck.

Robert.

On Mon, Jul 20, 2009 at 8:33 AM, Harold Comere<[email protected]> wrote:
> Hi all,
>
> I have done the small polygon benchmark that i have worked on this two last
> weeks and it has confirmed a osg behavior wich i have noticed before without
> be able to explain it myself...
>
> The principle of my benchmark is to add to scene same model few times ( all
> with multiparent to save memory, each model has its own
> positionAttitudeTransform ) then render some seconds and compute average
> FPS. If FPS down under 15 FPS ( limit for interactive render ) benchmark
> stop and compute total rendered triangles, if not it redo it.
> So there is only one model per geode and one transform per model.
>
> The fact is if i use a model with :
> 5 000 triangles, i was able to display about 14 000 000
> 11 000 -> 20 000 000
> 34 000 -> 7 000 000
>
> So, two first results could be logical, because in first case there is more
> geodes and so more calls to opengl functions and graph is longer to
> traverse, so in second case perf are improved.
> But how the third case get so bad result ?
>
> All models are visible at same time, so i guess there is not any culling
> done.
> Does anyone has any explanation ?
>
> Thanks for your attention,
> Regards,
> Harold
>
> _______________________________________________
> 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