Re: glCopyTexSubImage blocks the CPU?

Andras Balogh wrote:

> Well, I guess I'll have to do it myself then. But then it would be 
> nice to  have a non blocking version of Map(), that would just "peek", 
> and return  null if not possible to map yet, instead of blocking. 
> Would something like  this be possible?

Actually, you don't even need to Map(), you can call TexSubImage using 
the same PBO once the ReadPixels has completed.  I believe NV_fence will 
allow you to test this without forcing a Finish(), allowing you to do 
other work in the CPU for maximum overlap.  Now when you call 
TexSubImage the driver will synchronously merge the subregion into the 
compressed texture.  Anyway I think this should work...

Its a cool idea, let us know how it works out for you.