Hey,

I'm running into a problem when loading my game levels in a separate thread. The problem is, part of the level loading generates cubemaps (requiring the graphics device), but my gamescreen rendering code also requires the graphics device. This causes concurrency issues (for example, if the cubemap code happens at the same time as the rendering code, an exception will be thrown). This isn't safe, it sounds like per the DirectX guidelines that all rendering code should be in a single thread. Is there a good workaround for this, or a better way to handle this?

Thanks for your help!

-Bryan

Hey Nick,

Thanks for the reply! I was thinking along these lines but was concerned with the overhead of locking. As I understand there is overhead when obtaining a mutex (even if there is nothing blocking), and I didn't like the idea of that overhead in my render code. I guess I could give it a shot though and see if affects performance. Maybe the overhead is very small (or i could be totally wrong about their being overhead) compared to the work done by the rendering-