I revisit the C vs Swift post to see how things have improved.
Baseline
I’m running on newer hardware, so a new baseline is required. Like the original,
I’m using -Ofast, and this time, the run completed in about 17.45 seconds.
time orbitals
Computing row 2880 of 2880
Done! Return code = 0
orbitals 17.45s user 0.04s system 99% cpu 17.498 total
Swift
After reviewing the original code, I noticed that I had refactored it to use arguments to specify the number of pixels and iterations. These are not constants, as they are in the C version, which can permit the compiler to perform additional optimizations. I decided to test that theory. Running the current version in Swift 5.3 yielded a resunt in about 20 seconds. Changing it to use constants put it ahead of the C version.
time orbitals-swift
orbitals-swift 16.23s user 0.07s system 99% cpu 16.321 total