[Mlir-commits] [mlir] [AMDGPU] fold `memref.subview` into `amdgpu.gather_to_lds` (PR #149851)

[Jakub Kuderski]

https://github.com/kuhar commented:

I have similar concerns to @qedawkins -- it seems weird for memref transforms to know about amdgpu ops. This entails a build-system dep (for bazel, cmake is not that strict) and seems like a future headache

https://github.com/llvm/llvm-project/pull/149851