Hi Michael,

On 10/10/2011 10:02 AM, Michael Raab wrote:
> Hi again,
>
> we have some new NVIDIA Geforce graphic cards (gtx580, gtx590) which do not
> accelerate all of our applications. Some application are more than 10 times
> faster than before but other apps are much slower. What I've investigated so
> far is that the apps that run slower have very much geometry distributed over
> a few nodes in the scenegraph. E.g.>  10 mill. tris with 20-30 nodes...
>
> Running gDebugger I found out that approx. 98% of the OpenGL calls that are
> generated by OpenSG1.8 are deprecated with OpenGL 3.1. May deprecated
> function calls cause these performance problems with new graphic boards? Does
> someone else had similar problems?

The main reason you see that is that OpenSG 1.8 doesn't use VBOs yet. It uses 
VertexArrays, if the data is single-indexed, or immediate mode otherwise. Both 
of those paths are not supported in OpenGL 3 any more, so you see the 
deprecated 
warnings.

However, by default OpenSG puts the geometrey into a display list, which at the 
last tests I did (admittedly some time ago) were still the fastest way of 
pushing geometry through the system.

So if you turned that off I could imagine seeing problems, or if the internal 
display list handler in the driver does not handle very large dlists well any 
more. ~500,000 triangles per node doesn't sound excessive though, do I'm not 
sure why that would be a problem.

Are you using display lists?

Hope it helps

        Dirk

------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2dcopy1
_______________________________________________
Opensg-users mailing list
Opensg-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensg-users

Reply via email to