Mostly a general tracking issue so that I don't forget about this. The Mono amd64 intrinsics implementation should use some of the lines-of-code-reducing stuff introduced as part of the implementation of arm64 intrinsics. While SSE and the numerous revisions thereof aren't as uniform as NEON, there's still a good deal of cleanup that can be done. Less .text, more .rodata, fewer special cases, etc.