i am trying to implement a raytracer with CUDA. But i have some questions:
I’m reading some obj-files, where the vertices for the raytracing are stored. When I upload this data to the GPU with cudaMalloc, it spends too much time and the kernel loading failed. Is it better to upload this data in a texture or in constant memory? It should be thousands of triangles…or is there another solution, what I forgot?
When i am rendering an image with two or three triangles, my frame rate is only 10 fps. What could be the problem? I would like to have some ray packets, but i don’t know how to do this?
At the rendering process i could see some noise, as it would be the rays… I think, when i could raise the frame rate, this problem could be solved.
Greetings thopil <img src=‘http://hqnveipbwb20/public/style_emoticons/<#EMO_DIR#>/crying.gif’ class=‘bbc_emoticon’ alt=’:’(’ />