[llvm] [AMDGPU] Add DAG mutation to improve scheduling before barriers (PR #142716)

[Carl Ritson via llvm-commits]

perlfu wrote:

> Can you add an explanation that articulates what is the sweet spot achieved by this mutation?

I have reworded the existing description to try and be more clear on the object of this mutation.



https://github.com/llvm/llvm-project/pull/142716