It’s something that I have noticed on a LOT of C++ class/API/engines that encapsulate Opengl’s commands.
I think that generaly, the biggest slowdown on the majority of their C++ class/API/engines is in the “pre-culling” process (cf. for don’t draw the majority of polygons that aren’t on the view fustrum)
=> they want always make this “big culling” on software when hardware with T&L can make this very more speedly with the more simple and hardware accelerated fustrum culling …
i found the problem. I had to set the draw buffer to GL_BACK explicitely in SoQt’s class Inventor/Qt/viewers/SoQtViewer.cpp in the actualRedraw method at the end of the method just before the actualRedraw call.
Don’t know why this is necessary only on GeForce3.