So it has been a while since I posted about computing the Mandelbrot set. This has partially been due to some observations I made looking at the run times of the Mandelbrot computations on different size images. I noticed that the time scaling of the computation and the rendering were almost exactly matched, so at no point would the raw computational power be the limiting factor.

Since my original reason for setting up this computation was to try out programming with CUDA, it was a bit disappointing to see that I wouldn’t really be able to leverage that parallelism in the computation. I suppose setting the GPU to the rendering task could give a pretty good speedup by itself, but that doesn’t sound as exciting to me.

Perhaps I’ll return to this sometime in the future, but other projects, my Librarian software in particular, have been occupying my time as of late. I think the next time I try to set up a simulation with a rendering task associated, I’ll try and use it as an opportunity to learn how to use the Unity engine.