[llvm-branch-commits] [llvm] [AMDGPU] Model the cost of the expanded integer to/from floating point casts (PR #223572)

[Dmitry Sidorov via llvm-branch-commits]

https://github.com/MrSidims edited https://github.com/llvm/llvm-project/pull/223572