Sergey G. Brester

> the answer is that you can find two keys that always collide. With 100% probability. @lemire I think her question was not whether a collision is possible (it's pretty...

> Can i use different seed for every time reshuffling to do that? Yes, but ... is not really clear for which purposes you need it... > the way that...

I hope I've correct understood what you meant with "reshuffling"... Because in this sense you cannot really talk about "reshuffling" using any hash-function. **Hashes are not a bit-mixers. Hash-function acts...

> whether it is possible to use your source code, that is marked as Public Domain, under the BSD-2 license? Basically the work under public domain is compatible with any...

It does not compromise the hash quality. If the hash function viewed as pure hash (without cryptographic sense), the quality of the hash dependend on good distribution and cannot be...

Still again, why in your opinion one of the following values is better? ``` 88c4adec 54d201b9 54d201b9 54d201b9 385e023f 865a0e29 c8bd436f 0b9dada7 ``` If one don't forget, both are just...

This is possible from scratch for MM3: - you can build MM3-x86-128 and MM3-x64-128 targeting 32-bit and 64-bit systems (or even under certain circumstances even 16-bit systems); - MM3-x86-128 built...

> to have both 32bit and 64bit versions of my application having equal values? Or even MM3-x64-128 - if it is compile-able for you on x86 and you can live...

I guess it would then not as much deviate from SHA-256, according to performance and distribution quality (note I said SHA, not SHA3). Anyway the performance comparison of both looks...

> Would it make sense (and would it make any qualitative difference) to allow a 64-bit seed value instead? Shortly - no. For longer answer see https://github.com/aappleby/smhasher/issues/56#issuecomment-380782966.