It's an interesting use-case where the data is all known up-front. I'd be interested to see an incrementally-increasing benchmark. I'd imagine the cache-miss

It's an interesting use-case where the data is all known up-front. I'd be interested to see an incrementally-increasing benchmark. I'd imagine the cache-miss penalty from the HashSet is countered by the continuous re-sorting/copying/moving from the Array solution. But that's why we measure, right?