[Mlir-commits] [mlir] [MLIR] [Vector] Linearization patterns for vector.load and vector.store (PR #145115)

[Diego Caballero]

dcaballe wrote:

Hey, could we add more context about what we are trying to do here, preferably to the PR description? Why only unit dims are supported? What is the long-term plan to support more generic cases? What are vector linearization responsibilities vs applying drop unit dims or other transformations? I think elaborating on this aspect of the design would be helpful to understand where this is going and set users' expectations. Thanks!

https://github.com/llvm/llvm-project/pull/145115