Closed o0101 closed 1 year ago
I can confirm that it's working: https://paste.ee/p/euV9K
System:
Processor: AMD Ryzen 5 1400 Memory: 2133 DDR4 Cmake version: 3.25.1 GCC version: (Debian 12.2.0-14) 12.2.0 Host system: Debian testing amd64 64-bits
@rurban DONE! :)
No, you cannot just call it crypto without cryptoanalysis, and esp. not with 64bit. Submit it to a crypto competition first with the 128bit variant. Like this: https://csrc.nist.gov/projects/lightweight-cryptography Or rename it to discohash512.
Sure!
@rurban OK! DONE! :) thank you for review ❤️
I"m afraid, but discohash v2 has tons of bad seeds, I forgot to add. the 512 variant also. also the old version try ./testseeds.sh discohash
@rurban Thank you! :)
@rurban that should fix the bad seeds...mostly! :)
@rurban That looks good! :)
@rurban BTW - can you suggest a cheaper or better way to do the alignment of key? I'm not a C++ expert so I'm sort of flying by the seat of my pants on things like that :)
@rurban BTW - can you suggest a cheaper or better way to do the alignment of key? I'm not a C++ expert so I'm sort of flying by the seat of my pants on things like that :)
You did fine with alignas()
the latest is much faster and better. testing today
Had to provide a bad seeds checker.
@rurban thanks! I read somewhere that bad seeds in Disco were due to not the algorithm but data being incorrectly accessed across threads somehow. I still have a few minor updates to make, as well as a new hash to push. I'll make a new PR in a few days. I appreciate your work on this, thank you again! :)
No, threads certainly not. It's a bad mixer, mostly
OK well let's see how it works out in the next couple days! :) Thanks again, @rurban for your work on this -- I really appreciate it! Hope you're well! :)
Results for the new variant: