Jump to content
  • Advertisement


This topic is now archived and is closed to further replies.


OGL Engine Performance Problems

This topic is 5772 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hi there! Me and my friend are working on an OGL engine for an RTS game. We have learnt everything from the nehe tutorials, but now we have doubts about that, our engine has a good performance. On my machine (1900+ XP, Radeon8500LELE 64MB) it displays about 1,9 million triangles per second. The scene is: A 20x20x20 array of boxes made of triangles (not using quads), with an 512x512 texture on each side (all the same), using mipmapping. Is this performace all right, or it can be improved somehow? Please help!

Share this post

Link to post
Share on other sites
1.9 is ok is your only using the likes of glBegin/glEnd.

It depends a bit on fillrate too.

general ways to improve performance, in order of improvement:

vertex array object (or vertex array range for nvidia) extension.
hence using vertex arrays.
texture binds! (not a problem for you I guess)
keeping the cpu and gpu in sync. This means not doing things like calling glReadPixels.
Keeping your vertex data as static as possible. Copying to AGP memory in blocks of 32 or 64 bytes. Keeping vertex strides to 32bytes if possible (this doesn't effect radeons much at all, since they are well designed, but gives nvidia cards a pretty good boost).
optimizing for the vertex cache. (ditto above)
using mipmapping. Reduces texel stride for distant pixels. Hence improves memory access.
reducing front to back rendering if overdraw is a problem.
using triangle strips instead of lists (once again, doesn't effect radeons any where near as much as geforces)
reducing usage of lighting, etc, when not needed.

do all that and it should be possible on your system to be getting around the 15-20 million tris/sec mark. I can on my radeon 9000 pro, which isn't as fast as a 8500. (well, depends I suppose)

| - Project-X - my mega project.. big things comming soon - | - adDeath - an ad blocker I made - | - email me - |

[edited by - RipTorn on January 23, 2003 4:26:16 PM]

Share this post

Link to post
Share on other sites
State changes. They make on hell of a difference. Make sure you have a good state tree, or something similar. I personally use an ordered trie...

Death of one is a tragedy, death of a million is just a statistic.

Share this post

Link to post
Share on other sites
thank you for your answers! But i have a few questions (accordin to my bad english i think)
what do these mean:
"Keeping vertex strides to 32bytes if possible"
(an example would be nice)
"vertex data as static as possible"
what do you mean under static?

and heres my current structure (just an essence), what do you think?

Share this post

Link to post
Share on other sites

  • Advertisement

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!