I’ve been writing Wired.jl, which is a Julia library for calculating static magnetic fields in 3D space using the Biot-Savart law. As I was writing the library, I was surprised by how slow my code was running, despite my best efforts to keep it fast and efficient, so got the idea to benchmark the core algorithm by writing it in other languages.
At the time, I thought that perhaps Julia wasn’t as fast of a language as I expected, and that by writing it in Julia, I was “leaving some speed on the table”. I also wanted to try my hand at learning another language like Fortran, or brushing up on my C skills, so I thought it would be a fun little project.
As it turns out, I wasn’t writing the Julia version very well. Now, the Julia implementation is the fastest one. I’m sure that others could write Fortran or C code that’s far faster than mine, but to my surprise, my best efforts resulted in slower implementations in those two languages.
I published this section of the website to:
- Publish some of my notes and lessons learned, in case someone else could learn from them
- Document the optimizations I performed to make the
Wired.jllibrary as fast and efficient as possible
All source code for this series of benchmarks is published here:
https://github.com/freestatelabs/wired-benchmarks/
Prior Art
Note that I am by far not the first person to think of writing this sort of code; for instance, see:
- https://github.com/vuthalab/biot-savart
- Biot Savart Law integrator BioSaw (I couldn’t find the source code anywhere)
- Computation of the Biot-Savart line integral with higher-order convergence using straight segments, code located here