[noise] Specifying Blake2sp / Blake2bp

[Jason A. Donenfeld]

On Tue, May 2, 2017 at 1:50 PM, Jean-Philippe Aumasson
<jeanphilippe.aumasson at gmail.com> wrote:
> The parallel versions are mostly beneficial when hashing long (say, 1k+)
> messages and when faster hashing is noticeable. Not sure it's the case here.

Seems reasonable. So Trevor's conclusion holds.

(WireGuard ships a avx1-accelerated Blake2s compression function [1],
which I suppose is more than necessary for Noise then.)

Thanks for the info!

Jason

[1] https://git.zx2c4.com/WireGuard/tree/src/crypto/blake2s-avx-x86_64.S