[Mlir-commits] [mlir] [mlir][nvgpu] update commit group and wait async ops (PR #130482)

[Guray Ozen]

grypp wrote:

I think the way your PR implements it is also fine. But as we read in the previous discussion, folks chose the current approach. I don’t see much difference between the two. The current way is easier to follow in terms of dependencies, while your PR makes that a little harder.

I’m wondering—does the current approach block you from implementing an algorithm? If that’s the case, can you elaborate what is this case?

https://github.com/llvm/llvm-project/pull/130482