Jump to content
  • Advertisement
Sign in to follow this  
ctian

Performance when using procedural texture

This topic is 4890 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, In my application, I use a texture procedurally computed by the cpu and transferred to the gpu for use in the pixel shader. The performance is very very poor. I used NVPerfHUD to test my program and found that the gpu spend about 50ms idling while the cpu spend about 75ms waiting for the gpu per frame. I think it is a terrible waste of cpu/gpu time. I am guessing it is due to the improper synchronization betweent the cpu and the gpu, i.e., when the pixel shader is reading the texture, the cpu has to wait untill it can lock the texture, and when the texture is locked by the cpu, the pixel shader has to wait untill the texture is unlocked. Is my guess right? Does anybody has the experience of handling procedural texture while keeping the performance from decreasing greatly? Thanks! b.t.w., My card is nVidia GeForce FX 5700LE. ctian

Share this post


Link to post
Share on other sites
Advertisement
Coder, Thank you very much! The performace is improved a lot when I use dynamic textures. The cpu seldom waits for the gpu now. But the gpu still spend some time idling (less than before). I think I should assign more jobs to gpu, right?

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

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

Participate in the game development conversation and more when you create an account on GameDev.net!

Sign me up!