[Mlir-commits] [mlir] [MLIR][GPU] Added FoldSelfCopy pattern to gpu memcpy (PR #221216)

[llvmlistbot at llvm.org]

Edward-EJ wrote:

> Actually, rethinking about it: "gpu.wait async" should be semantically equivalent...

As in replacing the memcpy with a "gpu.wait async" that has no async dependencies or following the replacement strategy in the PR as is?

https://github.com/llvm/llvm-project/pull/221216