How do I properly use multithreading with Nvidia PhysX?

Posted by xcrypt on Game Development See other posts from Game Development or by xcrypt
Published on 2012-04-07T13:56:32Z Indexed on 2012/04/08 5:48 UTC
Read the original article Hit count: 275

Filed under:
|

I'm having a multithreading problem with Nvidia PhysX. the SDK requires that you call Simulate() (starts computing new physics positions within a new thread) and FetchResults() (waits 'till the physics computations are done). Inbetween Simulate() and FetchResults() you may not "compute new physics".

It is proposed (in a sample) that we create a game loop as such:

  1. Logic (you may calculate physics here and other stuff)
  2. Render + Simulate() at start of Render call and FetchResults at end of Render() call

However, this has given me various little errors that stack up: since you actually render the scene that was computed in the previous iteration in the game loop.

Does anyone have a solution to this?

© Game Development or respective owner

Related posts about physics

Related posts about multithreading