王一 WangYi

Thanks for your code. However, I prefer to trust memcpy guys as I know little about the hardware and I think they know better than me.

these results are a bit old @data-man . https://github.com/ktprime/emhash is a good place for hashmap benchmark. I stick on ska::bytell_hash_map for speed and space.

@human-0 have you tested this on wyhash?

well, it depends on you. just an additional benchmark that comes for free.

A simple modification is that to use (sum&32)+1 as length. Since each time the seed is different, it is not lockable, otherwise the hash function will fail later quality tests.

> Also, what if `len == 0` returns `-1`? it will be fine. sum will be ~0ull or 0xffffffffffffffff due to the propety of uint64_t

> This wouldn't change the issue that the serie of lengths produced by each hash would be different, > making a speed comparison between them more difficult. At least we...

It is just an extreme case test to see if these hashes works properly. if they do memory access, they will fail. xxh3 leaks its key if 0 byte string...

> * wyhash fails `BadSeeds` even with `WYHASH_CONDOM 2` ([smhasher output](https://gist.github.com/flyingmutant/7159d434f20067e548576501ef6e7262)) In fact I dont't buy the BadSeeds test as it "invades hasher's personal space". The BadSeeds test make use...

师兄好，我是2000级复旦生科的。wyhash是小弟开发的。 ^_^