[Mlir-commits] [mlir] [mlir][gpu] reverse parallel loop to gpu dimension mapping order. (PR #79592)

[Jungwook Park]

jungpark-mlir wrote:

Switched to draft.
As pointed by reviewers, there are more factors to be considered and there seems little benefit to make this minor change.
A design for the loop analysis and transforms need to be considered first and potentially some hints can be propagated to the mapping. We can discuss further in the discourse when someone or myself get an idea for it.

https://github.com/llvm/llvm-project/pull/79592