[Mlir-commits] [mlir] [mlir][vector] Support n-D vectors in i8 to i4 trunci emulation (PR #94946)

[Lei Zhang]

https://github.com/antiagainst requested changes to this pull request.

Core arith/vector dialect ops are upper layer abstractions and meant to support both llvm and spirv lowering paths. So I think we should add impl for both to make sure we harden the design choices to be robust and have the generalicity. For patterns and passes it's similar, unless we have optionality to enable. I understand everbody have their own priorities and goals though. We can help to impl for this time; but need to block until that's done. Hopefully that's okay. :)

https://github.com/llvm/llvm-project/pull/94946