Looks great! Wondering how it would compare with @cmuratori MeowHashhttps://github.com/cmuratori/meow_hash …
-
-
Replying to @aras_p @cmuratori
I would expect MeowHash to feature higher bandwidth on large inputs, as it leverages dedicated hardware capabilities. The difference is unlikely to be "large" though, since both algorithms basically saturate memory bandwidth.
1 reply 0 retweets 2 likes -
Replying to @Cyan4973
Some folks here recently found places where we do 128 bit file checksums using MD4 (why? I've no idea why) and were looking for an upgrade. I'm telling them to look at XXH3 now (other contenders are SpookyV2 & Meow)
2 replies 0 retweets 6 likes -
Fair warning: there's a new version of Meow hash coming soon with closer to true 128-bit hashing (there are some issues with AES hashing that make it not true 128-bit if you don't do it just so). There's also a new hash called Minipaw that will be included for simplicity.
2 replies 1 retweet 12 likes -
You might be interested to hear that Meow hash has been added to https://github.com/injinj/smhasher and that its maintainer has found a test it fails: https://news.ycombinator.com/item?id=19406374 …
1 reply 0 retweets 1 like -
We'd love to know if our new version passes that test - I assume I can just download the latest GitHub repository for SMHasher and it's in there?
1 reply 0 retweets 0 likes -
Should be, I guess (I'm not the maintainer, I just saw it in the HN comments.)
1 reply 0 retweets 1 like -
I can now confirm that Meow v0.5 indeed passes the upgraded smhasher, whereas v0.4 didn't!
2 replies 0 retweets 11 likes -
nice! push to github! :)
1 reply 0 retweets 2 likes -
I just cleaned up an RC this week and
@NoHatCoder needs some time to look it over and ensure I didn't break anything crypto-sensitive :) SMHasher is still not a very good test suite and, at least as far as I can tell, the only way to know if a hash is good is to manually inspect.1 reply 0 retweets 0 likes
Crypto ain't my area of expertise, but it seems to me that most functions of non-trivial complexity are just very hard to "randomly" find collisions for. You have to look at the algorithm and see how collisions are likely to occur, and SMHasher doesn't do that (obviously).
-
-
Replying to @cmuratori @aras_p and
So it ends up being more like a virus checker. It has a bunch of tests designed to find problems that _other_ hashes have had, but if it doesn't have a test designed to find problems _your_ hash has, well... you get the idea.
2 replies 0 retweets 2 likes -
Replying to @cmuratori @aras_p and
Anyway, long story short, I'll push to GitHub once Jacob's satisfied with it, and then we can start another round of testing.
0 replies 0 retweets 1 like
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.