ristretto/z: use hash/memhash for hashing

[rogpeppe]

This is cleaner, but unfortunately slower.

Reusing the Hash value in KeyToHash may help there,
but wouldn't get back to the original speed, so maybe
this isn't a reasonable change.

name             old time/op    new time/op    delta
MemHash-4          12.4ns ± 7%    43.1ns ± 1%  +248.30%  (p=0.008 n=5+5)
MemHashString-4    19.6ns ± 7%    70.9ns ±20%  +262.00%  (p=0.008 n=5+5)

name             old speed      new speed      delta
MemHash-4        5.17GB/s ± 7%  1.48GB/s ± 0%   -71.31%  (p=0.008 n=5+5)
MemHashString-4  6.29GB/s ± 6%  1.75GB/s ±18%   -72.12%  (p=0.008 n=5+5)

name             old alloc/op   new alloc/op   delta
MemHash-4           0.00B          0.00B           ~     (all equal)
MemHashString-4     0.00B          0.00B           ~     (all equal)

name             old allocs/op  new allocs/op  delta
MemHash-4            0.00           0.00           ~     (all equal)
MemHashString-4      0.00           0.00           ~     (all equal)

---

This change is 

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[manishrjain]

Thanks for the PR, @rogpeppe . Yeah, the speeds don't look encouraging, though I'm wondering why that's the case. My understanding was that Go 1.14 exposed the same functions.

[rogpeppe]

Yeah, me too. I did it in a rush over about 20 minutes before bed last night so didn't have time to investigate more. I'm interested why it's so much slower - I might take another look today.

[nes1983]

Just peeking at your code, I don't think you're using maphash the way it's meant to be used. You're supposed to use one maphash struct per goroutine, you're not supposed to create a fresh struct for each hash call. But that's what the code does. This leads to a ton of unneeded allocation. So I'm not surprised this winds up being slower than the original.

[github-actions]

This PR has been stale for 60 days and will be closed automatically in 7 days. Comment to keep it open.