[clang] [llvm] [CIR] X86 vector masked load builtins (PR #169464)

[Andy Kaylor via llvm-commits]

andykaylor wrote:

> I find #163802 modify int_masked_load Intrinsic to not use mask as args, which means I need to modify cir codes to emit call intrisic(not use mask as args), Is it right? Any suggestion!

@woruyu Definitely do not revert the LLVM intrinsic back to its previous state.

We should probably add masked load and masked store operations to CIR. The LLVM dialect already has these and they take alignment as an attribute. It's possible that the LLVM dialect lowering will need to be updated.

@bcardosolopes Has any work been done towards having masked loads and stores in CIR?

https://github.com/llvm/llvm-project/pull/169464