Crystal: Digging into struct initialization performance

[NewsBot]

A new Crystal blog post/announcement has been posted!

I started toying with BLAKE3 and I quickly wanted to assess how much faster it was compared to SHA256 in Crystal, and see if it would bring some free improvement. Luckily there is a shard wrapping the official library that is highly optimized with custom assembly for different CPU features (SSE, AVX2, NEON …).

To my surprise, performance was pretty bad. This post is the written legacy of the quest to understand why. Be prepared to dig into technical details and how language design can impact performance.

Here’s a shard.yml if you want to try to run the benchmarks at home:

name: <span...

Get the full details here: Digging into struct initialization performance - The Crystal Programming Language