I’ve written my own TLS 1.3 implementation (for fun). I would like to keep this up to date when post quantum algorithms come around. I’m guessing a supported_groups extension will be added for one of the algorithms, maybe Kyber.
I understand how NTRU works but haven’t looked into Kyber or other solutions.
What might I benefit from being aware of? Have any proposals been made? Will hybrid implementations be considered? Is there a timeline for this?
For elliptic curves, Montgomery modular multiplication is a somewhat essential optimisation. What similar optimisations are needed when going from pedagogical to performant Kyber implementations?