Ions Simulation

Idea:

This project was designed to explore optimization of single threaded looped code. My goal to explore different types of computing optimizations. Modeling N-body interactions between particles' gave me ample opportunity for optimization. This project was quite a milestone in my programming ability, as it had my first ever serious memory leak (as in the app could only run for ~5 mins before eating up all 16 GB of RAM). This was also the first time I was seriously able to look at optimizing loops and see measurable improvements in performance, getting up to ~60 million calculations of gravity (G(m1*m2)/d^2) per second.

Now, I get around 55-60 million force calculations per second and am able to simulate things like this:

This video is sped up by ~20x to show the interactions better. Since there are 2500 particles, each simulation update requires 2500^2 (6,250,000) individual G(m1*m2)/d^2 calculations, and so it only gives 8-9 updates per second. The calculations are spread across 8 threads on the 4 cores of my machine. After doing each of the ~50 million calculations of gravity per second, the program then breaks down the force F into it's x-axis (F*dx/dist) and y-axis components (F*dx/dist), and renders the image in a separate thread.

Previous Work:

This idea of this project originated many years ago from the IOS Application, 'Ions' by Douglas Applewhite. This application allows the user to play with having a max of 20,000 particles, with up to 20 'Fields', each being either repulsive or attractive. This idea fascinated me for quite a long time. As I had been recently learning to program when I discovered the app, this was one of the first programs that I ever wrote. The performance was initially mediocre, getting around 900,000 individual force calculations per second:

Current State:

This project has fulfilled a long-time goal of mine. I have always wanted a tool to just test high quantities of particles under different forces. While this page only demonstrates the gravitational force, I have also implemented twirling, stretching, directional, and other random types of fun forces to play with - see below. This is such a fun tool that I find myself starting it up once every few months to test random force ideas and see how they would play out in a complex system.