[llvm] AMDGPU/GlobalISel: Fold (trunc (lshr x, 16)) to unmerge high half (PR #217633)

[Krzysztof Drewniak via llvm-commits]

https://github.com/krzysz00 approved this pull request.

Looks broadly fine, one question, but I'd give @arsenm a second look

https://github.com/llvm/llvm-project/pull/217633