Hi, I’m currently reading from a VBO to obtain the vertices that were allocated by OpenGL. I’m able to retrieve all the data and works fine for now, but I’m wondering, is this is a good idea? Should I try to use texture memory objects to store the triangles into the GPU directly from Cuda instead?
Thanks!
If I am understanding your statements correctly, that would be a typical use-case for CUDA-OpenGL interop. You may want to google that and just start reading. I find this somewhat dated presentation to be still useful:
Thanks for your answer! I’ve actually already done that. I got it working, but I want to know if that is the fastest way to read vertices from the GPU. I’m writing a path tracer and I am using OpenGL to load the obj into GPU memory (VBO) then reading that part of the memory with Cuda, but I want to know if that is a good idea or not. Should I use a “texture<>” to store the vertices into the GPU instead of using OpenGL?
Thanks in advance, I know it’s a weird question.