Timeline for How do I properly use multithreading with Nvidia PhysX?
Current License: CC BY-SA 3.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 10, 2012 at 17:27 | comment | added | Roy T. | @xcrypt the physics call should be somewhere in the update thread so that you can simultaneously prepare for drawing. This way you dont lag behind so much, it also teaches about good points to synchronize so that your physics aren't 10 frames behind your drawing, etc... | |
| Apr 10, 2012 at 17:18 | comment | added | xcrypt | @RoyT. I got to read that article now. It's pretty good for rendering, but I can't see how it would help me with physics. Nvidia PhysX requires that you do one huge Simulate() call, you cannot call it separately per object. (And I don't think that would be a good idea for physics). Am I missing something? | |
| Apr 10, 2012 at 1:37 | comment | added | xcrypt | @RoyT. I tried about everything, different browsers, I even asked some friends to open it for me, nothing works. altdevblogaday.com works, but it won't let me open that specific page. | |
| Apr 9, 2012 at 21:16 | comment | added | Roy T. | @xcrypt, how weird, does this tinyurl link work: tinyurl.com/cybzwb5 ? Else try to find the article "Threading and Your Game Loop" on altdevblogaday.com | |
| Apr 9, 2012 at 14:03 | comment | added | xcrypt | @RoyT. I can't seem to open the page you linked to :( | |
| Apr 8, 2012 at 3:17 | history | edited | House | CC BY-SA 3.0 | deleted 46 characters in body; edited title |
| Apr 7, 2012 at 16:42 | comment | added | Roy T. | To alleviate that you should create several sync bounds. If you have multiple threads they should all finish at the end of each frame for example. For a really good example see: altdevblogaday.com/2011/07/03/threading-and-your-game-loop | |
| Apr 7, 2012 at 14:47 | comment | added | xcrypt | The problem escalates with the number of threads you have that are dependent on the physics, you might end up with one thread(for example AI) to use the physics from 7 frames ago in it's calculations, and that is a serious problem. The other option you propose is impossible since inbetween a simulate and fetchresults call you can't compute new physics | |
| Apr 7, 2012 at 14:21 | comment | added | Daniel Carlsson | As long as your having 60fps the delay is only 17ms, human reaction time is longer than this so its unlikely to cause any great problems. The other option would be to do the simulation in the logic part and stall waiting for the results before rendering. You would lower your framerates but it would give you something to compare the results with, see if the problem really is with the delay or not. | |
| Apr 7, 2012 at 13:56 | history | asked | xcrypt | CC BY-SA 3.0 |