Messages in this thread | | | Date | Fri, 5 Oct 2018 20:40:03 +0200 | From | "Jason A. Donenfeld" <> | Subject | Re: [PATCH net-next v6 19/23] zinc: Curve25519 ARM implementation |
| |
Hey Dan,
On Fri, Oct 05, 2018 at 03:05:38PM -0000, D. J. Bernstein wrote: > Of course, there are other ARM microarchitectures, and there are many > cases where different microarchitectures prefer different optimizations. > The kernel already has boot-time benchmarks for different optimizations > for raid6, and should do the same for crypto code, so that implementors > can focus on each microarchitecture separately rather than living in the > barbaric world of having to choose which CPUs to favor.
I've been playing a bit with some code to do this sort of thing, choosing a set of implementations to enable or disable by trying all the combinations, and then calculating a quick median. I don't know if I'll submit that for the initial merge of this patchset -- and in fact all the current implementations I'm proposing are pretty much okay on microarchitectures -- but down the line this could be useful as a mechanism.
Jason [unhandled content-type:application/pgp-signature] | |