GPGPU Programming example: spiral animation
After having come close to General Purpose GPU programming and having studied “CUDA by example”, I decided to take the time to write a simple CUDA example.
Many of the small projects I had in my mind were not suitable for CUDA, since they were using Tensorflow that lie above the cuDNN. Therefore, I decided to create a useless graphic animation to prove some techniques learnt in the book.
The problem exposed in this post is an animation based on a spiral. The spiral computed in this project is the Archimedean Spiral:


The figure is built starting from multiple spiral spinning at the same distance. Every pixel in the image has a tone in the grey scale that depends on the distance to the nearest spiral.
After the paralleled algorithm was written, I passed to CUDA implementation of a 512×512 animated image. The video here below shows well the dynamic behaviour.