[vtk-developers] Re:Modifications to vtkOpenGLPolyDataMapper.cxx vtkOpenGLProperty.cxx

[John Biddiscombe]

Before committing the changes to the PolyDataMapper and Property I ran a
Sphere-Bench test myself but found that wireframe was considerably slower
(I'm using a GEForce 2 MX) graphics card.

General explanation seems to be that GL_LINE_LOOP is faster than GL_POLYGON
since it does less bookkeeping and wotnot. Is this likely to be true on the
newer generation of cards too?

In the light of this.

Is a speed hit on wireframe worth the gain of backface culling ability. Or
in other words, do we leave things the same or commit the (improved from the
purists point of view) code, but possibly slower on some(many?) platforms.
I'm guessing that at some point in the past when the mapper was created a
similar decision was made...

Can we check speeds of tests via the dashboard (and see how many platforms
would be slower)?

JB