I was curious about ARM NEON and learning SIMD in general, so wrote a small pattern scanner with it.
My super unscientific benchmark on a platform I barely know how to profile in had a ~481% performance increase on scanning 1 GB random data.
I don't usually release code I write (it's bad) but thought I'd try it for once.
https://codeberg.org/Sharkfie/Notepad/src/branch/main/neon.c
