[cperciva]

I'd love some context (maybe from cperciva himself?) around the performance enhancement of integrating new Intel AESNI instructions.

I was using OpenSSL for that (which was using a software implementation). The code (you can see it in spiped) now detects the CPU feature and selects between AESNI or OpenSSL automatically. Given that the tarsnap server code was spending about 40% of its time running AES, it's a nontrivial CPU time saving.

I should probably have been clearer in my writeup though -- using AESNI was never a "once I roll this out everything will be good" fix. Rather, it was a case of "I have this well-tested code available which will help a bit while I finish testing the real fixes".

[gonzo]

One wonders why you aren't using a version of OpenSSL that has the AESNI bits already in it.

[cperciva]

I've learned that wondering about OpenSSL internals is detrimental to my sanity.