Feature gate: #![feature(stdarch_const_x86)]
This is a tracking issue for unstable const versions of x86(64) intrinsics in stdarch.
Public API
~50% of x86(64) intrinsics in stdarch are amenable to constification. Intel has some weird pecularities in floating point, especially in comparison and conversion, which will be nontrivial to constify. Most integer intrinsics have been constified.
Steps / History
Unresolved Questions
- There are a few very specialized float intrinsics (mostly in AVX512), do we want to constify them?