[Mlir-commits] [mlir] [MLIR][XeGPU] Enhance XeGPU lane layout to support "wrap-around" distribution (PR #186958)

[Artem Kroviakov]

https://github.com/akroviakov commented:

The PR description says:

> replication of lane-level tensor tiles across all lanes when the tile size **matches** lane_data along a given dimension

But the code says that replication of some dim is the case when the distribution unit does not equally divide the source shape in that dim.

https://github.com/llvm/llvm-project/pull/186958