[llvm] [X86] Fold FMSUBADD into VFCMULC for fp16 complex conjugate multiply (PR #224533)

[via llvm-commits]

ganeshgit wrote:

> @ganeshgit Please can we clarify the purpose of the `optimization:AMDCPU` label? I understood it as an x86 optimization that will mainly benefit AMDs CPUs much more than Intel CPUs, but that's not the case here. Some of your tags suggest you're using it to track for downstream merges or something?
> 
> Ideally the label description needs to be filled in (but that's true for many labels....).

We are tracking the PRs which are of interest to AMD CPUs (includes middle end optimizations as well and not restrictive to codegen). In addition, we are tracking the X86 community contribution from AMD. Downstream merges are definitely not of interest/tracked using this label.

https://github.com/llvm/llvm-project/pull/224533