Sha2 became slower

[tailhook]

Hi,

I use sha256 in vagga a lot. And I just noticed that current implementation of sha256 is much slower (2x-3x) than old sha2 = 0.1.2.

I do not post benchmarks here because I can't find repository for old code. Do you have a link?

Is there any chance to fix it?

[newpavlov]

Thank you for this performance regression report!

Problem was in the fake-simd crate, because now it in the separate crate, operations on simd types weren't inlined by compiler. I've added to fake-simd inline annotations and it fixed this issue.

[tailhook]

Thank you for the quick response! I'm confirming that performance fixed.