[llvm] [AMDGPU] In promote-alloca, if index is dynamic, sandwich load with bitcasts to reduce number of extractelements as they have large expansion in the backend. (PR #171253)

[Nicolai Hähnle via llvm-commits]

https://github.com/nhaehnle commented:

Agreed that the alignment test is required.

https://github.com/llvm/llvm-project/pull/171253