[Mlir-commits] [mlir] [mlir][Vector] Infer mask and pass_thru types for maskedload/store (PR #131482)

[Andrzej Warzyński]

banach-space wrote:

> Do you have a strong preference here? I do have a strong preference here for having less redundent information in the IR here.

I do, but this shouldn’t be about personal preferences. To help unblock this, we agreed in today’s "Tensor Compiler" call that:

* This should be split into two PRs (verifier changes + syntax changes).
* @dcaballe, our masking expert, will review it to ensure broader input.
* If "mask type" is removed from the syntax, it should be done consistently across all Ops that take masks.

Regardless of personal preferences, **let’s maintain consistency**.

Thanks for your contributions to `Vector`! 🙏🏻

https://github.com/llvm/llvm-project/pull/131482