[llvm] AMDGPU: Remove flat/global fmin/fmax intrinsics (PR #105642)

[Matt Arsenault via llvm-commits]

arsenm wrote:

> I would still like to see something like the functions I suggested in my previous comment. Something that will be equivalent with the intrinsic that is being eliminated. These functions would use the new atomicRMW method AND also set the attributes accordingly such that it matches the behavior of the intrinsic.

There won't be a new function, but @yxsamliu is working on source level controls for these fields for the atomic builtins 

https://github.com/llvm/llvm-project/pull/105642