I have Wolf0's AES merged into cpuminer-opt and am running the first test.
The locally reported hashrate issue seems to persist but is constant. I'm not
sure why it's happening, it's only with hodl.
First impression is about a 10% increase over the non-aes version. If testing goes well
cpuminer-opt v3.1.10 will be released later today with support for hodl on CPUs with,
and without AES_NI, as well as all the other optimized algos supported by cpuminer-opt.
Sounds very good.
Thank you (and Wolf0 too!) for your work!