[llvm] [AMDGPU] Multi dword spilling for unaligned tuples (PR #183701)

[Christudasan Devadasan via llvm-commits]

================

----------------
cdevadas wrote:

These combinations arise as the EltSize is redefined each time to represent the actual spilled size. I think we should restructure the for loop and simplify these unique combinations.

https://github.com/llvm/llvm-project/pull/183701