rust-cuckoofilter icon indicating copy to clipboard operation
rust-cuckoofilter copied to clipboard

Poor false positive rate for memory used

Open DanielHeath opened this issue 5 years ago • 7 comments

I'm using rust-cuckoofilter to check entries in the 'Have I Been Pwned' master list (551509767 SHA1 hashes).

I've been taking the first 64 bits of the SHA1 hash as the key.

According to this bloom filter calculator, a 1gb bloom filter should be able to give me a 0.1% false positive rate on the full dataset.

However, when I load this data into rust-cuckoofilter, I get >2% false positives with a 1gb cuckoo filter.

I'm happy to provide any assistance with diagnosing further.

DanielHeath avatar Mar 18 '19 09:03 DanielHeath