An Optimal Bloom Filter Replacement Based on Matrix Solving — mentioned by Freaky

Wed, 30 May 2018 18:45:05 GMT

> Oh right, I was wrongly thinking you'd have to memcmp the whole size of the filter. Yeah, it's just k single-bit lookups - ideally you do something to get them into clusters, like dividing the database into sub-filters, so you're doing random lookups into, say, a 32KB chunk instead of a whole 2GB filter. > Did you look into Cuckoo filters as well? Cuckoo filters look like an interesting alternative and looking at them more closely is on the to-do. I don't think they'd have any significant space savings, though - they're similarly about 75% the size of the equivalent bloom filter. Maybe…

Paper and discussion history. A mention is not an endorsement.

An Optimal Bloom Filter Replacement Based on Matrix Solving · HackerPapers discussions

An Optimal Bloom Filter Replacement Based on Matrix Solving — mentioned by Freaky