[Mlir-commits] [mlir] [MLIR][XeGPU] Allow some nd ops to have argument shapes mismatch for … (PR #120566)

[Jianhui Li]

Jianhui-Li wrote:

Actually I find this code example. nvgpu ldmatrix also need to deal with the issue.  

https://github.com/llvm/llvm-project/blob/e5de2a2df4f983e1db83deaaa1d21457c922404f/mlir/test/Dialect/NVGPU/optimize-shared-memory.mlir#L31

  %mat = nvgpu.ldmatrix %shm[%fragRow, %fragCol] {numTiles = 4 : i32, transpose = false}
      : memref<128x32xf16, 3> -> vector<4x2xf16>
      

https://github.com/llvm/llvm-project/pull/120566