I try to get the max TPS value available on my Ti4600. I render 10 triangle strips 125000 triangles each and get about 46.7 FPS
that is about 58.4 MTPS.
All vertices in the strip are unique (I try to use just 20 unique vertices to force them all to be in the cache and this to my surprise doesn’t add anything to FPS)
Texturing, lighting, blending, Z, alpha, stecil tests are off. Vertex format is just coordinate 3f. VAR allocated from video memory. So is it the maximum or I forgot something? I’ve read here that somebody get about 300 MTPS on 9700. Is it the actual difference between Ti4600 and 9700?
I’ve been able to get about 60 MTris/second on GF4_4600 in a simple test app. In a real world scenario (i’m working on a terrain engine) I can get to about 35
Did you turn off rasterization to make sure you’re not fill bound?
glEnable( GL_CULL_FACE );
glCullFace( GL_FRONT_AND_BACK );
In a real world example, we are doing 18’2 MTris with a Ti4200 (per pixel lighting + cubemap), but with our scenario (because of the large polygons) we are very far from this!
I don’t know how the geforce pipeline works, but is possible to obtain transformation and rendering time disabling rasterization, as you said?
Ummmmm…very good trick…I didn’t know how to do that…