←back to thread

ChibiHash: Small, Fast 64 bit hash function

(nrk.neocities.org)
104 points thunderbong | 1 comments | 18 Nov 24 00:23 UTC | HN request time: 0.213s | source
Show context
aappleby ◴[18 Nov 24 04:58 UTC] No.42169830[source]▶
SMHasher/Murmurhash author here - I don't see anything fundamentally wrong with this hash function, it uses basically the same operations as the Murmur family (and a lot of other hashes at this point).

The handling of the "tail" of the key (the last < 32 bytes) is slightly odd (the "if (l & 1) { mix 1 byte of key }" happens before 8-byte chunks and 2-byte chunks), but if it passes SMHasher it should be fine for general use.

replies(1): >>42171063 #
HexDecOctBin ◴[18 Nov 24 09:49 UTC] No.42171063[source]▶
> if it passes SMHasher it should be fine for general use.

Author of the post writes:

> I kept making changes until the tests passed.

If SMHasher has become a target, shouldn't it be considered a bad measure now? [1]

[1]: https://en.wikipedia.org/wiki/Goodhart%27s_law

replies(2): >>42171485 #>>42186477 #
1. aappleby ◴[19 Nov 24 18:12 UTC] No.42186477[source]▶
Hi, don't know if you'll see this but the tests in SMHasher are designed to break hash functions by generating low-entropy key sets that try to trigger weak bit-mixing paths in the hash function.

When a hash function has weak mixing they tend to fail catastrophically on one of those key sets, so you can think of SMHasher as more like "attempts to test the hash function to the point of failure" instead of a "target" as described in Goodhart's Law.