[pittma]

ymms were used here on purpose! With full-width registers, the IFMA insns have a deleterious effect on frequency, at least in the Icelake timeframe.

[anematode]

Ye, hence a separate version for CPUs which don't have that problem. Although, maintaining so many of these RSA kernels does seem like a pain. Didn't realize u wrote that code; super cool that it's used in practice!

[pittma]

I am not the original author—this is adapted from an implementation by Shay Gueron, the author of that paper I linked, but I do agree that it's cool!

[hnuser123456]

zen5 can run avx512 at near full boost clocks: https://chipsandcheese.com/p/zen-5s-avx-512-frequency-behavi...