[Mlir-commits] [mlir] [MLIR][XeGPU] Xegpu distribution patterns for load_nd, store_nd, and create_nd_tdesc. (PR #112945)

[Stella Laurenzo]

stellaraccident wrote:

Drive-by suggestion: The vector dialect is shared by many pieces, and it is generally easier on everyone to land such changes separately (if I scanned this right, the vector changes would even qualify for an NFC, which helps on review scope/latency even further). This doesn't just help review latency, but it also helps integrators because in case of issues, we tend to not expect that patches labeled for a specific target (XeGPU) can affect anything else.

https://github.com/llvm/llvm-project/pull/112945