glutWireTeapot very slow???
Yes, but then there sure are a lot of calculations involved with that (doesn't sound like it though...), cause the fps wouldn't drop like that, if there wasn't some sort of bottleneck... I just assume it is the driver, but of course I could be wrong ;)
Author
i tested some of my 3ds models for line and solid modes, and i just realised that; if the object is lowpoly then fps is better @wireframe mode.. if not, then solid mode gives better results.. is it a normal response? did anyone tried that?
Quote:Original post by stroma
i tested some of my 3ds models for line and solid modes, and i just realised that; if the object is lowpoly then fps is better @wireframe mode.. if not, then solid mode gives better results.. is it a normal response? did anyone tried that?
Well, there shouldn't be a noticable difference at all with wireframe vs. solid in a low-poly model. As for the second part, yes, that's normal. A high-poly model in wireframe would need to render every triangle on the model, since you don't get any benefits of culling.
>>Well, there shouldn't be a noticable difference at all with wireframe vs. solid in a low-poly model.<<
yes there should the solid draws more pixels on the screen
this should make it clear(ignoring culling)
bollux numbers but u should get the idea
------------100 tris
wireframe 3*3*100 verts 1000pixels on screen
solid 3*1*100 verts 100000pixels on screen
------------10000 tris
wireframe 3*3*10000 verts 20000pixels on screen
solid 3*1*10000 verts 100000pixels on screen
yes there should the solid draws more pixels on the screen
this should make it clear(ignoring culling)
bollux numbers but u should get the idea
------------100 tris
wireframe 3*3*100 verts 1000pixels on screen
solid 3*1*100 verts 100000pixels on screen
------------10000 tris
wireframe 3*3*10000 verts 20000pixels on screen
solid 3*1*10000 verts 100000pixels on screen
Quote:Original post by zedzeek
>>Well, there shouldn't be a noticable difference at all with wireframe vs. solid in a low-poly model.<<
yes there should the solid draws more pixels on the screen
this should make it clear(ignoring culling)
bollux numbers but u should get the idea
------------100 tris
wireframe 3*3*100 verts 1000pixels on screen
solid 3*1*100 verts 100000pixels on screen
------------10000 tris
wireframe 3*3*10000 verts 20000pixels on screen
solid 3*1*10000 verts 100000pixels on screen
I said a low-poly model. So < 100 triangles wireframe vs. solid will not make a significant difference in the overall FPS, from my experience.
Quote:
I said a low-poly model. So < 100 triangles wireframe vs. solid will not make a significant difference in the overall FPS, from my experience.
hmm maybe i wasnt clear
------------10 tris
wireframe 3*3*10 verts 1000pixels on screen 100fps
solid 3*1*10 verts 100000pixels on screen 90fps
------------10000 tris
wireframe 3*3*10000 verts 20000pixels on screen 30fps
solid 3*1*10000 verts 100000pixels on screen 60fps
Quote:Original post by zedzeek
hmm maybe i wasnt clear
------------10 tris
wireframe 3*3*10 verts 1000pixels on screen 100fps
solid 3*1*10 verts 100000pixels on screen 90fps
I guess I wasn't clear either. But I'm not going to be ignorant and repost my original response.
For one thing, you provided no context in which those numbers came from. For all I know, you pulled them out of your imagination. In the tests I ran on my computer, I'm seeing no noticable (1-3?) difference in the wireframe vs. solid in low-poly models.
Quote:Original post by stroma
i tested some of my 3ds models for line and solid modes, and i just realised that; if the object is lowpoly then fps is better @wireframe mode.. if not, then solid mode gives better results.. is it a normal response? did anyone tried that?
I hate to guess at performance bottlenecks based on so little information, but it sounds like you are pushing the bottleneck around when you change the number of polys. It could be that you are bandwidth limited transferring the data of the high poly models, so you see no savings from drawing fewer pixels. In fact you could have some slowdown due to the overhead of wire frame drawing. If the card is optimized to draw filled polys I could imagine the wire frame overhead being significant. With the low poly models you may be fill rate limited, so drawing fewer pixels gives you savings there.
But this is all just speculation without the app to play with. You might try tracking down your bottleneck in each situation (try resizing the window and see if that affects frame rate, etc.) There are some good articles on that out there.
It's hard to get really useful timings with such high frame rates, small changes in state can have big affects at those speeds. Also FPS is not a linear scale with respect to speed. A drop from 300 to 100 is only a 6 or 7 ms difference, which is still significant, but not as bad as it sounds. That's the same as a drop from 45 to 35 fps for example.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
