Particle Simulation in CUDA

So this was a part of the course that I took during my Master’s studies. I am extremely interested in the idea of high-performance computing and parallel programming. Hence, it was a no-brainer that I took GPU programming. This course served as an excellent gateway into CUDA, teaching me the nuances of general-purpose GPU programming and the role of memory access within GPUs.

The course comprised two main projects. The first was the creation of an HDR pipeline for displaying HDR images. This project demonstrated the capabilities of current GPUs effectively. Even on my seven-year-old hardware, the tone-mapping operation was reduced from 2 minutes of compute time (when using the CPU) to just 2 seconds on the GPU!

The second project involved particle simulation using a straightforward algorithm that applied verlet integration. The main focus was on dividing 3D space into grids and conducting local operations, providing practical insight into spatial partitioning strategies in GPU programming.

Please note: There could potentially be some inaccuracies in the simulation. Currently, my machine isn’t compiling the CUDA code, and I’m hesitant to update my Linux due to the risk of breaking several packages. Rest assured, I’ll update with the correct output as soon as I can render it successfully!