This is my 2nd project for CUDA computing class.
This project is my implementation of ray casting of spheres.
The rendered scene is processed on GPU using CUDA API and CUDA OpenGL interop.
The rendering was optimized using BVH algorithm which allows to quickly determine which parts of the scene should be processed by which threads.
The performance of my solution was measured on RTX 2060 and I've been able to average above 100fps for scene with 1000 spheres
For contrast I've also developed a similar CPU algorithm which wasn't able to get anywhere close to the GPU speeds
10 000 spheres rendered in real-time (at around 50fps) |