[Mlir-commits] [mlir] [mlir][vector] Add support for yielding loop bounds in `scf.for` distribution. (PR #163443)

[Artem Kroviakov]

akroviakov wrote:

Yes, the fix is fine. I was just confused by your reply

> but unfortunately that does not help us here.

to my comment

>>  I'd expect scalar usages inside warpOp to be "forwarded" to outer definitions beforehand

> I meant that scalar values coming from the outside may be referenced directly, not necessarily via the warpOp arguments. Meaning that some pass, similar to hoisting in terms of when it should be called, would clean up things beforehand


https://github.com/llvm/llvm-project/pull/163443