rurban / smhasher

Hash function quality and speed tests
https://rurban.github.io/smhasher/
Other
1.85k stars 178 forks source link

wyhash fv4.1 #277

Closed tansy closed 1 year ago

tansy commented 1 year ago

V4.1 to make it more stream friendly

Change is cosmetic from a tester point of view. Just so you know.

rurban commented 1 year ago

But that's still the bad wyhash version, which is slower and as has more bad seeds.

wangyi-fudan commented 1 year ago

4.1 has bug due to make_secret function which came from someone's PR. 4.2 should work

tansy commented 1 year ago

4.2 should work

Didn't know it's already changed.

But that's still the bad wyhash version, which is slower and as has more bad seeds.

After taking a look I conclude the difference is in secret only. Is it that much of a difference?

wangyi-fudan commented 1 year ago

yes. V4.1 seeds are generated by a "simplified" make_secret function. This function is from someone's early PR, and I didn't notice till now that he removed an essential code line (to make sure secrets are prime numbers). Thus your PR awake me to fixed this bug :-)

rurban commented 1 year ago

I"m working on the 4.2 update instead