[vtkusers] VTK vs. straight OpenGL performance (was Re: VTK wireframe performance)

[Sebastien BARRE]

At 2/1/2002 01:43 PM, Dave Reed wrote:
> > From: Sebastien BARRE <sebastien.barre at kitware.com>
> >
> > At 2/1/2002 04:27 PM, Christophe ATTANASIO wrote:
> >
> > >Why VTK have so poor performance (less or egal  than shading) in
> > >wireframe mode ??
> >
> > Nothing to do with VTK, but with your OpenGL card and drivers.
> >
> > Yes, it seems strange that what used to be a rock-solid assumption
> > "wireframe rendering is faster than surface rendering" is not true anymore
> > for today's graphics cards.
>
>And in my experience the exact opposite (i.e., polygons are faster
>than wireframe) is a good assumption for today's graphics cards.

Sorry, bad "French Translation on Friday", I did not meant that I doubt the 
assumption "wireframe faster than surface" is no more true, I meant the 
opposite (i.e. it is puzzling but this is exactly what is happening 
nowadays: this old rock-solid assumption we all knew is no more true in 
many cases).

>On a related note, it does seem that rendering the same number of
>polygons with VTK is signifcantly slower than with just plain
>OpenGL. For example just using GLUT, I can use glutSphere with a
>tesselation of over 1 million polygons and it runs quite nicely on my
>Radeon 7500 card, but a VTK example with around 300,000 triangles (in
>strips) is slower.

On the other hand it has been reported that displaying *many* actors 
(several thousands) would slow down rendering compared to a similar OpenGL 
solution (or a single actor with the same total geometry, even if you 
consider the actor overhead). There has been a whole discussion about that 
in the past (I guess it was related to the fact that OenGL is 
context/stack-based, and pushing/popping was kind of costly).

>You may remember last fall I started a thread about some "missing
>traingles" when using a triangle stripper. The original author of Mesa
>found the problem (the triangles were actually being rendered, but as
>backfaces in some situations and it's fixed - starting in Mesa 4.0.1 I
>think). While Brian was looking at the problem using VTK (since that's
>the only program anybody had found that showed the problem), he
>mentioned that there were some "optimizations" that could be made in
>VTK's OpenGL calls. Unfortunately, he didn't elaborate.
>
>Has anyone ever looked into trying to optimize the OpenGL calls in
>VTK?

Ken changed a couple of lines 2 or 3 months ago, changing an OpenGL call if 
I remember. The result was impressive, it was more than 40% faster in most 
cases.

>Or does anyone who knows about the VTK internals think there are
>some changes that could be made to use OpenGL more efficiently? Is
>there interest in this, etc.?
>
>Dave



--
Sebastien Barre