Adam Stylinski

> Apparently [you _can_ tell NEON](https://developer.arm.com/documentation/den0018/a/NEON-Instruction-Set-Architecture/Alignment) that you promise an access is aligned and it'll then optimize based on it and fault if you lied, but we don't appear to....

So, while it undoubtedly is probably a stronger hash, I do wonder if those AES rounds with AES-NI are comparably as fast as the SSE/AVX accelerated fletcher 4 checksums, which...

Isn't 512e basically 4k sectors with firmware absorbing the RMW write cycles (poorly)? I think zfs is doing the right thing here, you really want 512n drives if you have...

Arguably your point 3 would maybe be better addressed with compression on a dataset, letting zfs burn some CPU cycles decompressing the metadata rather than letting drive firmware grab the...

Static libraries are the stage just prior to LTO, I think. When you link the final library, that's when the IPOs and other clever things that happen that cause trouble.

@kloczek that's not exactly what's happening here. They are specifically disabling LTO for the sake of producing a portable binary package. What's there to disable it now is a fairly...

For architectures where unaligned access is not allowed you will see a SIGBUS abort of the process well before you see any data corruption. The decision as to whether or...

Ok, so the _ABI_ doesn't define behavior for what to do for unaligned access, but for most architectures I've dealt with the ISA defines some sort of behavior to handle...

Also the "arch" string being pulled out of OpenBSD's uname is likely why your configure script isn't using any of the altivec code, despite it clearly being a G4. Perhaps...

There are big endian aarch64 implementations floating around? I've yet to see that, hah.