My first #blog post in 6 years! It comes with some updates, like Mastodon comments
"Maybe don't use Blake3 on Short Inputs"
@jszym We talked a bit in the HN comments, and I think the main takeaway from that was that most of what's being measured for short inputs here is Python function call overhead and e.g. the cost of constructing a 32-byte string. Python does fine for hashing long inputs, because it can spend the vast majority of its time in C/Rust code, but if you really care about performance for lots of short inputs then you probably shouldn't use Python.