As I remember it is x1800 ? If so - perfomance should be way higher. Try to look at acceleration slider inside windoze display->advanced->troubleshot tab.
I found the fire GL cards terrible in shaded view, but fast as anything in wireframe, and nvidia cards faster in shaded and slower in wireframe.
Depending on the textures shown and the resolution, ive had speeds like that before. And its not so much the number of polys but the number of near-intersections it has to work out that slows it down - one mesh subdivided will be faster than an array of tall, close boxes with the same poly count.
well I’ve checked the slider and it’s set to full acceleration.
the things I usually have on my viewport are buildings surely and I do display a lot of lines.
so I have always thought those spline rendering stuff was slowing me down but today I was dealing with a simple building model and it the slowness was driving me mad.
directX should be faster with static high poly counts, where OpenGL will be faster with wireframe views/lots of splines/deforming meshes. The DirectX feature that makes it so fast with static meshes is called “Use cached D3DX meshes” in the Customize>Preferences>Viewports Tab>Confiugure Driver dialog, so if you do decide to use DirectX make sure that is turned on. I’ve only ever used nVidia cards, so not sure what results you may get outside of that area.