Where does CUDA fit into the render pipeline
Author
I want to start playing around with CUDA and experment with a voxel based rendering system. I'm just trying to get my head around where CUDA fits in. If I build a CUDA program, am I limited to only using the GPU, or can the GPU and CPU to communicte with each other ? I'm not too keen on developing a full application in CUDA as it would be a horriable waste of CPU power...
cuda doesn't fit in the pipeline at all, you should use it instead of the CPU to do software rendering
CUDA doesn't fit in the render pipeline at all. It's a API for using the GPU to solve highly parallel numerical programming problems. You can think of it as a way to turn your GPU into a second CPU, albeit a somewhat specialized one.
Author
Cool, so I've got complete control over how everything renders. Can I still have my CPU's running the gameplay logic, with the GPU acting as a software renderer ?
Edit, found my answer in the nvidia docs. Yes I can use both the GPU and CPU in an application
Edit, found my answer in the nvidia docs. Yes I can use both the GPU and CPU in an application
If it can run aside of DirectX or OpenGL you can offload all the tasks where you want no rasterizer involved or a simplified rasterizer like shadow maps and PostFX. I believe this will work in DirectX 11 where you can use the tranditional rasterizer based interface aside with a cuda-like interface. So you can run a ray-tracer on the side :-)
Quote:Original post by MrMark
I want to start playing around with CUDA and experment with a voxel based rendering system...
You might want to check out Ken Silverman's forum, in particular this thread http://www.jonof.id.au/forum/index.php?topic=1433.0
That guy initially uses some other system, but switches to CUDA fairly soon.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
