[llvm] [VectorCombine][AMDGPU] Shrink demanded vector loads (PR #202501)

[via llvm-commits]

tianhbai wrote:

Confirmed. I have read the LLVM AI Tool Use Policy, LLVM Code-Review Policy and Practices, LLVM Developer Policy, and InstCombine Contributor Guide. AI tools were used while working on this PR, and that usage has been noted in the PR description.

https://github.com/llvm/llvm-project/pull/202501