[Cryptography] sha1sum speed

[Bill Cox]

On Sat, Apr 30, 2016 at 9:00 AM, Henry Baker <hbaker1 at pipeline.com> wrote:

> Questions:
>
> A.  Are there HW speedups today for crypto hash functions?
>

SHA256 has some Intel HW instructions.  However, from my latest benchmarks,
it is still a lot slower than BLAKE2b on CPUs with SSE or AVS enabled.


> B.  Are there side-channel issues that are slowing down these hash
> functions?
>

I don't think so.  IIRC, the BLAKE2b implementations are side-channel
resistant.


> C.  Are hash function speeds considered important these days?
>

They are to me!  People still use MD5 for the speed, when they should
upgrade to more secure hashes.  That's why I am particularly excited about
BLAKE2b (and BLAKE2's other flavors).

I'm working a bit with the HighwayHash authors, playing with what we can do
using AVX2's parallel multiplication and byte shuffling.  It's very fast.
However, the BLAKE2 functions are very well analyzed and are ready for
secure deployment now.  Any new function will take years to get there.
 b2sum runs faster than md5sum on all my machines.  The parallel versions
(BLAKE2bp and BLAKE2sp) are even faster.

Bill
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20160430/5bcd9305/attachment.html>