The goal of this project is to write a CUDA raytracer that can handle millions of spheres as fast as possible. All those spheres have the same radius and share the same material.
This is based on my optimization project Ray Tracing in One Weekend in CUDA
The current renderer can only handle scenes that fit completely in constant memory, which is 64KB in most graphic cards. The main goal of this project is to figure out how to render scenes that exceed this limit.