Elastic Hashing vs Linear SSE3 Probing

[ocodista]

Hey,

I came across this article ["Undergraduate upends a 40 year old data science conjecture"](https://www.quanta magazine.org/undergraduate-upends-a-40-year-old-data-science-conjecture-20250210/), which claims to break a 40-year conjecture around linear probing and introduces a new theoretical technique called elastic hashing.

I was wondering if the performance improvements could be applied to the current Swiss Table implementation.

While the Swiss Table uses the power of SSE3 to leverage parallelism, improving the standard linear probing, combining both strategies may result in a better performance, however, this is an assumption, I haven't run any benchmark yet.

Have you (contributors) came across this strategy/article before? Would love to hear some ideas.