I have developed a render engine and the test is organized like this:
Pentium IV 1.6G
GeForce 2 32MB
256MB DDRRAM
Windows XP Pro
Window size 640 * 480
22,000 triangles
512 by 512 BMP textures * 6 with LINEAR
light effect
all normal vectors,
depth buffer
fog effect
And the refresh rate is 100fps.
Is the speed acceptable?
It is completely acceptible with me for you to have that speed. Carry on!
good call, Coriolis 
22000*100 = 2.2 million triangles per second at 640x480 isn’t remarkable for a geforce2 - it’s capable of far more.
Many questions:
Are your 22,000 triangles big in the viewport? If they are then you may be experiencing the fill rate limit.
What do you mean by ‘lighting effect’? Is this standard OpenGL lighting? How many lights are you using?
Are you blending any of the triangles in the framebuffer (glEnable(GL_BLEND)) ?
How many of these 512x512 textures (32bit?) are you using per frame? (That’s how many “individual” textures, not the same one repeated)
Fog does impact performance a significant amount, so maybe this explains your low trispersecond.
Insert glCullFace(GL_FRONT_AND_BACK) in your program to measure whether you’re fill bound or geometry bound. Dollars to donuts you’re fill bound.
>>22000*100 = 2.2 million triangles per second at 640x480 isn’t remarkable for a geforce2 - it’s capable of far more.<<
i doubt doom3 will reach 2.2million tris/sec even on a radeon9700
Originally posted by zed:
i doubt doom3 will reach 2.2million tris/sec even on a radeon9700
Yes but does the subject of this post have AI? Collision detection? Lighting/shadow volume calcs? Network code? Multiple objects/render states? etc.
Originally posted by zed:
[b]>>22000*100 = 2.2 million triangles per second at 640x480 isn’t remarkable for a geforce2 - it’s capable of far more.<<
i doubt doom3 will reach 2.2million tris/sec even on a radeon9700[/b]
If you count all the triangles it draws rather than just the triangles you see, I bet it will.