Work queue in optix

Does anyone know if optix is allocating work per thread or per warp? Is per thread to much overhead? Could a per warp prefix scan + a per block allocation be a good mix?
I would be interrested in any details I could get about this topic.

Hi David,

Thank you for your interest. Please repost your message to the OptiX forum and we’ll do our best to answer it there:
http://developer.nvidia.com/forums/index.php?showforum=43

Yours will be the first technical question up there because we haven’t released OptiX yet, but then some one needs to be first .

  • Phil

I don’t have premissions to start a new thread there.

Tried again, logged in but still can’t create a new thread. zelcious is my user name if you have forum privileges.

You may need to open another account there in order to post because it’s a different “zone”.
Sorry for the trouble.

  • Phil

I have another account! I can log in but I can’t start a new thread… or reply on an existing.

Hi , I have download the package wich contains the optix dll ,but I dont know where to install it