[llvm] [AMDGPU] Don't cluster lds instrs together (PR #180908)

[Matt Arsenault via llvm-commits]

arsenm wrote:

> > Was there a benefit on older targets?
> 
> I think since gfx6, lds access is not backed by cache: 

Then should you add an arch check for when this changed? 


https://github.com/llvm/llvm-project/pull/180908