Benchmarks for concurrent hash map implementations in Go

A Go developer just dropped a mega-benchmark pitting five ways to share a dictionary across many threads — and yes, the author wrote one of the contenders. Cue instant drama: one camp cheers the transparency, the other yells “fox guarding the henhouse,” while everyone argues over what actually matters. The standout vibe? Respect for the write-up’s trade-offs and clear tests, with top commenters applauding that it didn’t just crown a winner. Still, people went spicy. Some insist the standard library’s sync.Map is finally “grown-up” thanks to Go 1.24’s new design, others swear the sharded old favorite feels safer, and the lock‑free crowd flexes like they just found free speed. The charts show ops-per-second across different map sizes (from “fits in the tiny chip cache” to “spills into RAM”) and worker counts; one lock‑free library even had to tap out at big sizes, fueling the “benchmarks aren’t real life” chorus. Jokes flew: “GOMAXDRAMA,” “HashTrie sounds like a dinosaur,” and “zero allocations? My memory leak says hi.” A practical thread grumbled that most apps aren’t 99% reads, while a performance‑purist shot back that read-heavy paths are where speed wins. Through it all, the mood is: thank you for the data, now let’s argue about it.