[vtk-developers] Rendering speed with large numbers of vtkActors (long)

[John Shalf]

john Tourtellott wrote:
> 
> 1. Thanks for sharing your info. I'll definitely try integrating some of
> your things into our code.

Way cool.  It'll be interesting to see how this works for
the Sun's GFX hardware.  I haven't done any
cross-architecture testing.
 
> 2. I also wanted to ask you about your item #5: "Lighting needs to be
> disabled when making glMaterial
> state calls.  (at least this is true on SGI's)." Do you mean that
> lighting has to be disabled in order for the glMaterial call to "work"?
> Or, in other words, does OpenGL ignore glMaterial calls if GL_LIGHTING
> is enabled? Perhaps I should make a little test program to check this on
> our Suns.

Neither... Apparently every change in glMaterial properties
forces a costly lighting model recalculation on the SGI
hardware if the GL_LIGHTING is on.  If you have the
GL_LIGHTING off when you change material properties, the
recalculation is deferred until you turn the lighting back
on again. This just reduces some redundant calculations that
otherwise create bubbles in the GL pipeline.  I'd never
heard of this before the performance engineering people at
SGI told me.

-john

> John Shalf wrote:
> >
> > Very interesting!
> > I've been working with Roger Brown at SGI to figure out why
> > performance on the SGI IR2 was so terrible with just a
> > single actor.  We started on this about a year-and-a-half
> > ago, but didn't know the appropriate venue to discuss
> > results.
> >
> > We were able to get a 3x performance improvement on the
> > "sphere" benchmark using the following modifications (some
> > of which are very similar to yours).  Some of these
> > improvements come directly from the OpenGL performance
> > tuning guide and others from profiling the GL pipeline
> > directly.
> >
> > ---------- snip ----------